Cognitive behavioural therapy added to standard care for first‐episode and recent‐onset psychosis

Abstract

Background

Cognitive behavioural therapy (CBT) can be effective in the general population of people with schizophrenia. It is still unclear whether CBT can be effectively used in the population of people with a first‐episode or recent‐onset psychosis.

Objectives

To assess the effects of adding cognitive behavioural therapy to standard care for people with a first‐episode or recent‐onset psychosis.

Search methods

We conducted a systematic search on 6 March 2022 in the Cochrane Schizophrenia Group's Study‐Based Register of Trials, which is based on CENTRAL, MEDLINE, Embase, CINAHL, PsycINFO, PubMed, ClinicalTrials.gov, ISRCTN, and WHO ICTRP.

Selection criteria

We included randomised controlled trials (RCTs) comparing CBT added to standard care vs standard care in first‐episode or recent‐onset psychosis, in patients of any age.

Data collection and analysis

Two review authors (amongst SFM, CC, LK and IB) independently screened references for inclusion, extracted data from eligible studies and assessed the risk of bias using RoB2. Study authors were contacted for missing data and additional information. Our primary outcome was general mental state measured on a validated rating scale. Secondary outcomes included other specific measures of mental state, global state, relapse, admission to hospital, functioning, leaving the study early, cognition, quality of life, satisfaction with care, self‐injurious or aggressive behaviour, adverse events, and mortality.

Main results

We included 28 studies, of which 26 provided data on 2407 participants (average age 24 years). The mean sample size in the included studies was 92 participants (ranging from 19 to 444) and duration ranged between 26 and 52 weeks.

When looking at the results at combined time points (mainly up to one year after start of the intervention), CBT added to standard care was associated with a greater reduction in overall symptoms of schizophrenia (standardised mean difference (SMD) ‐0.27, 95% confidence interval (CI) ‐0.47 to ‐0.08, 20 RCTs, n = 1508, I² = 68%, substantial heterogeneity, low certainty of the evidence), and also with a greater reduction in positive (SMD ‐0.22, 95% CI ‐0.38 to ‐0.06, 22 RCTs, n = 1565, I² = 52%, moderate heterogeneity), negative (SMD ‐0.20, 95% CI ‐0.30 to ‐0.11, 22 RCTs, n = 1651, I² = 0%) and depressive symptoms (SMD ‐0.13, 95% CI ‐0.24 to ‐0.01, 18 RCTs, n = 1182, I² = 0%) than control. CBT added to standard care was also associated with a greater improvement in the global state (SMD ‐0.34, 95% CI ‐0.67 to ‐0.01, 4 RCTs, n = 329, I² = 47%, moderate heterogeneity) and in functioning (SMD ‐0.23, 95% CI ‐0.42 to ‐0.05, 18 RCTs, n = 1241, I² = 53%, moderate heterogeneity, moderate certainty of the evidence) than control.

We did not find a difference between CBT added to standard care and control in terms of number of participants with relapse (relative risk (RR) 0.82, 95% CI 0.57 to 1.18, 7 RCTs, n = 693, I² = 48%, low certainty of the evidence), leaving the study early for any reason (RR 0.87, 95% CI 0.72 to 1.05, 25 RCTs, n = 2242, I² = 12%, moderate certainty of the evidence), adverse events (RR 1.29, 95% CI 0.85 to 1.97, 1 RCT, n = 43, very low certainty of the evidence) and the other investigated outcomes.

Authors' conclusions

This review synthesised the latest evidence on CBT added to standard care for people with a first‐episode or recent‐onset psychosis.

The evidence identified by this review suggests that people with a first‐episode or recent‐onset psychosis may benefit from CBT additionally to standard care for multiple outcomes (overall, positive, negative and depressive symptoms of schizophrenia, global state and functioning).

Future studies should better define this population, for which often heterogeneous definitions are used.

Plain language summary

Cognitive behavioural therapy plus standard care for first‐episode and recent‐onset psychosis

Key messages

Cognitive behavioural therapy (CBT) is effective for reducing symptoms of schizophrenia and improving functioning in people at the beginning of the disorder.

Very little information was available on the potential adverse effects connected with this intervention.

Introduction

Schizophrenia is a severe mental disorder because of the heavy effect of the symptoms on the daily lives of those affected. People with this illness struggle to differentiate between their own thoughts, beliefs and ideas versus reality. For example, they may be hearing voices in their head, but it feels like someone is really talking to them. The treatment of the beginning phase of the illness is of critical importance, in order to prevent or reduce the risk of a chronic course. This phase is usually defined as 'first episode', and the term 'recent onset' is also used to describe the period of three to five years from the first episode.

A psychological intervention, cognitive behavioural therapy, can be effective for treating the symptoms in people with schizophrenia in general; it is still not clear if this intervention could be helpful also for people in the initial phases of the illness.

What did we want to find out?

We wanted to find out the effects of cognitive behavioural therapy added to standard care for people with a first episode or recent onset of schizophrenia.

What did we do?

We searched for studies that examined cognitive behavioural therapy given in addition to standard care (usually drugs) compared with standard care alone or other psychosocial interventions.

We compared and summarised the results of the studies and rated our confidence in the evidence. For example, some factors in the studies (such as too few participants or results not showing clear effects) were rated with a lower confidence in the evidence.

What did we find?

We found 28 studies, of which 26 had usable data on 2407 participants with recent‐onset or first‐episode schizophrenia. The studies had a duration of between 26 and 52 weeks for the intervention phase and, in 18 studies, participants were contacted again to collect further data after the end of the treatment. The studies were conducted in the UK, Europe, Australia, Canada, USA and China. They were mainly funded by public institutions, and a small number also by pharmaceutical companies.

When looking at the results about one year after the start of the intervention, we found that CBT added to standard care:

‐ is more effective than control conditions in reducing the overall symptoms of schizophrenia

‐ is more effective than control conditions in reducing positive symptoms of schizophrenia (e.g. hearing voices)

‐ is more effective than control conditions in reducing negative symptoms of schizophrenia (e.g. apathy, loss of interest and motivation, lack of concentration)

‐ is more effective than control conditions in reducing depressive symptoms of schizophrenia

‐ is more effective than control conditions in improving the global state (number of participants with symptoms improvement, number of participants with symptoms worsening, scores on scales that measure the general level of symptoms of the participants)

‐ is more effective than control conditions in improving functioning (e.g. participating in social life, taking part in everyday activities).

What are the limitations of the evidence?

The certainty of the evidence is between moderate and very low, meaning that we have moderate to very little confidence in the effect estimates. This is due to the fact that, for some of the studies, the quality ('risk of bias') was low, and that the results of the various studies sometimes slightly differed from each other. Moreover, for some outcomes, the results were imprecise, and a limited number of participants contributed to the data.

How up‐to‐date is this evidence?

The evidence is up‐to‐date until March 2022.

Summary of findings

Summary of findings 1. Summary of findings table ‐ Cognitive behavioural therapy plus standard care compared to control for first‐episode or recent‐onset psychosis.

Cognitive behavioural therapy plus standard care compared to control for first‐episode or recent‐onset psychosis
Patient or population: first‐episode or recent‐onset psychosis Setting: Intervention: Cognitive behavioural therapy plus standard care Comparison: control
Outcomes	Anticipated absolute effects* (95% CI)		Relative effect (95% CI)	№ of participants (studies)	Certainty of the evidence (GRADE)	Comments
	Risk with control	Risk with Cognitive behavioural therapy plus standard care
Mental state ‐ general: mean endpoint overall symptom scales (high = poor) (combined time points)	‐	SMD 0.27 SD lower (0.47 lower to 0.08 lower)	‐	1508 (20 RCTs)	⊕⊕⊝⊝ Lowa,b
Global state ‐ number of participants with relapse/exacerbations of psychosis (combined time points)	358 per 1000	293 per 1000 (204 to 422)	RR 0.82 (0.57 to 1.18)	693 (7 RCTs)	⊕⊕⊝⊝ Lowc
Leaving the study early ‐ for any reason ‐ overall acceptability (combined time points)	214 per 1000	186 per 1000 (154 to 225)	RR 0.87 (0.72 to 1.05)	2242 (25 RCTs)	⊕⊕⊕⊝ Moderated
Functioning ‐ mean endpoint functioning scale (high = good) (combined time points)	‐	SMD 0.23 SD lower (0.42 lower to 0.05 lower)	‐	1241 (18 RCTs)	⊕⊕⊕⊝ Moderatee
Adverse events ‐ participants with at least one adverse event ‐ < 12 months	591 per 1000	762 per 1000 (502 to 1000)	RR 1.29 (0.85 to 1.97)	43 (1 RCT)	⊕⊝⊝⊝ Very lowf
*The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). CI: confidence interval; RR: risk ratio; SMD: standardised mean difference
GRADE Working Group grades of evidence High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.
See interactive version of this table: https://gdt.gradepro.org/presentations/#/isof/isof_question_revman_web_439588391841774344.

^a Downgraded by one level for risk of bias: 5 out of 20 studies were at overall high risk of bias 
^b Downgraded by one level for serious inconsistency (I‐squared = 68%)
^c Downgraded by two levels for imprecision: inadequate information size (< 1000 participants for a dichotomous outcome) and CI includes both better outcome for CBT + TAU and no difference between CBT + TAU and control
^d Downgraded by one level for imprecision: CI includes both better outcome for CBT + TAU and no difference between CBT + TAU and control 
^e Downgraded by one level for risk of bias: 2 out of 18 studies were at overall high risk of bias
^f Downgraded by three levels for imprecision: inadequate information size (< 1000 participants for a dichotomous outcome, N = 43) and CI includes both better outcome for control and no difference between CBT + TAU and control

Background

Description of the condition

The estimate of the lifetime prevalence of psychotic illnesses is 7 per 1000 of the population; the incidence of the first episode of psychosis (FEP) is estimated at 34 new cases per 100,000 persons/year (Kirkbride 2012; Kirkbride 2017). Psychosis usually starts in late adolescence and early adulthood, with an average age of onset at the beginning of the second decade of life (Kirkbride 2017). Psychotic symptoms fall into two broad categories: a) positive symptoms, which are abnormal by their presence (such as hallucinations, delusions, and disordered thinking), and b) negative symptoms, which involve a loss or reduction of normal functions that should be normally present (social withdrawal, flat or blunted affect, poverty of speech, restricted range or intensity of emotional expressions). Psychotic illnesses include a range of different diagnoses, such as schizophrenia, schizoaffective disorder, and delusional disorder (WHO 2019). The prevalence of schizophrenia is similar in men and women, higher in developed countries compared with the least developed countries, and higher in migrants compared with native‐born individuals (Saha 2005).

There is no consensus on the definition of FEP (Breitborde 2009; Puntis 2020) and different criteria are used in clinical and research settings to define this term. These include duration criteria (e.g. less than three or five years since first onset of symptoms), contact with mental health services criteria (e.g. first contact with mental health services), or initiation of antipsychotic medication criteria (e.g. no more than six months of antipsychotic prescription (Puntis 2020)). In this review, we will focus on people with a first episode of psychosis as defined in the included studies. In order to capture this uncertainty, we will also accept studies including participants with a recent‐onset psychosis, defined with a time criterion of five years (see Types of participants).

A study of outcomes of FEP found that 58% of patients with FEP met criteria for remission (absent psychotic symptoms for at least six months) and 38% met criteria for recovery (both symptomatic and functional improvement ‐ including social, educational and occupational domains ‐ for at least two years) over mean follow‐up periods of 5.5 years and 7.2 years, respectively (Lally 2017). In a review of predictors of treatment resistance in FEP, the following were found to influence outcome: lower premorbid functioning; lower level of education; negative symptoms from first psychotic episode; comorbid substance use; younger age at onset; lack of early response; non‐adherence to treatment; and longer duration of untreated psychosis (Bozzatello 2019). Evidence regarding the longer‐term course of clinical symptomatology, social and occupational functioning following FEP remains lacking, with significant heterogeneity between studies limiting comparability. A longitudinal study of an FEP cohort (n = 326) in the UK identified four broad trajectories of psychotic symptoms over a ten‐year follow‐up period; the majority of patients were categorised as either remitting‐improving (58%) or persistent (30.6%), while smaller groups displayed late decline (5.6%), and late improvement (5.4%) (Morgan 2022).

The impact on the individual is usually significant, because the onset of the illness typically occurs at an age when the person is engaged in many important life tasks, such as completing their education, establishing oneself in the job market, and forming significant relationships. A psychotic illness has wide‐ranging implications for quality of life, functioning, social inclusion, education, and employment (Marwaha 2004; Meltzer 2002). For these reasons, the burden of schizophrenia and related psychotic illnesses is dramatic for the person, his or her relatives, and society. According to the 2019 Global Burden of Disease Report, schizophrenia ranks 20th in years lived with disability (YLD (GBD 2019)). With estimated total costs of more than EUR 29 billion per year, schizophrenia is also amongst the most expensive illnesses in the European Union (EU (Gustavsson 2010)). Annual societal costs per person with schizophrenia vary across countries, and have been estimated to be USD 5818 in Thailand, USD 30,140 in Germany, USD 33,851 in Japan, and USD 94,587 in Norway (Jin 2017). The main reasons for these costs are high hospitalisation rates, high unemployment, and loss of productivity; 80% to 90% of people with schizophrenia do not work (Evensen 2016; Marwaha 2004; Marwaha 2007). In addition, there are high rates of physical morbidity, and lower life expectancy (Hjorthøj 2017).

Improving the ability of people with schizophrenia to function in society and, therefore, being active in their educational, professional, and social life can help reduce the burden connected with first‐episode and recent‐onset psychosis. Furthermore, treatment of people with their first‐episode and recent‐onset psychosis is of fundamental importance, in order to prevent an exacerbation of the symptomatology (relapse), and the development of chronicity of the illness.

Description of the intervention

The mainstay of treatment for first‐episode psychosis is antipsychotic medication, even though about 26% of people in their first episode continue to experience psychotic symptoms (Ceraso 2020), and may experience unwanted and unpleasant side effects (Schneider‐Thoma 2019). Psychotherapy can be additional interventions, provided in conjunction with medication.

Cognitive behavioural therapy (CBT) is a type of psychotherapy, during which people are encouraged to establish links between their thoughts, feelings, or actions, and their current or past symptoms or functioning (or both). People are encouraged to re‐evaluate their perceptions, beliefs, and reasoning in relation to the target symptoms. Therapy promotes alternative ways of coping with the target symptom, with the aim of reducing stress and improving functioning (NICE 2014).

CBT for psychosis was developed with the primary aim of reducing symptom severity, and there is evidence that it may be effective in this regard for the general population of people with schizophrenia (Bighelli 2018; Jones 2018a; Jones 2018b). There is also evidence that CBT could be beneficial for improving other outcomes, such as insight, well‐being, and functioning in this population (Jones 2018a; Laws 2018).

The National Institute for Health and Care Excellence (NICE) guidelines recommend offering CBT in conjunction with antipsychotic medication to people with first‐episode psychosis for at least 16 sessions (NICE 2014). NICE guidelines for psychosis and schizophrenia in children and young people also recommend offering CBT in addition to pharmacotherapy (NICE 2013).

How the intervention might work

CBT promotes the creation of links between thoughts, feelings, actions, and symptoms, with the goal of changing the way in which the person interprets and evaluates their experiences and reflects on their meanings.

One goal of CBT for psychosis is to reduce the occurrence of symptoms. People with schizophrenia are taught to identify and monitor their thoughts and assumptions in specific situations, and to evaluate and correct these thoughts and assumptions by considering objective external evidence and actual circumstances. This encourages people to challenge their delusional beliefs and their perceptions of hallucinations (Hagen 2010).

Other goals of CBT include reducing the distress associated with symptoms and their interference with the person's well‐being and quality of life, for example, by improving coping skills (Hagen 2010).

Why it is important to do this review

A prior Cochrane Review of early interventions for psychosis found only two eligible randomised controlled trials on CBT (Marshall 2011). Recent Cochrane Reviews on cognitive behavioural therapy for people with schizophrenia identified nine studies on people with first‐episode schizophrenia, but did not analyse them separately; so currently, there is no reliable summary of the evidence for CBT in this important population (Jones 2018a; Jones 2018b).

Therefore, a high‐quality synthesis is needed and relevant for all stakeholders, including clinicians, consumers, and funders.

Objectives

To assess the effects of adding cognitive behavioural therapy (CBT) to standard care for participants with a first‐episode and recent‐onset schizophrenia.

Methods

Criteria for considering studies for this review

Types of studies

We considered all relevant randomised controlled trials (RCTs). We included RCTs that met our inclusion criteria and reported data that were usable for either the qualitative or quantitative synthesis. We considered both open‐label and blind‐outcome assessor studies. Given the nature of cognitive behavioural therapy (CBT), it may not have been possible to blind the participant to the treatment; on the contrary, the active engagement of the person in the therapy was required. As a result, we did not expect to find double‐blind studies; if we identified any, we included them. The researcher collecting outcome data could have been unaware of the allocation of the individual participant (blind outcome assessor). We examined the role of a blind outcome assessor in a Sensitivity analysis, excluding the studies that had not used blinding for this role.

We excluded quasi‐randomised studies, such as those that allocated participants to interventions on alternate days of the week. When people were given additional treatments, such as pharmacological treatment with antipsychotics, we only included data if the adjunct treatment was evenly distributed between groups, and it was only the cognitive behavioural therapy that was randomised.

Types of participants

The target of this review were people with a first‐episode or recent‐onset psychosis. We included studies with participants who exhibited symptoms that matched the criteria for an episode of a psychotic disorder (schizophrenia, however defined; or related disorders, including schizophreniform disorder, schizoaffective disorder, and delusional disorder, however diagnosed).

A first episode of psychosis is defined as the first manifestation of the disorder that meets the defining diagnostic symptoms and time criteria (APA 2013). Studies that explicitly recruited participants with a first episode were included. Studies that use terms such as "recent‐onset psychosis" were also included if it was defined as no longer than five years from first contact/treatment/diagnosis.

We also included studies that recruited participants from early intervention services, since these centres usually recruit participants within three to five years of the illness, so matching our inclusion criteria.

We included studies independently of the age of the participants (except those over the age of 65 years), in order to account for the variability in age of onset for psychosis. We excluded trials that reported outcomes from participants with a second or subsequent episode, deemed to be at risk of developing schizophrenia, trials that focused on participants with a late‐onset psychosis (e.g. first episode around the age of 65), and trials that focused on people older than 65 as part of their inclusion criteria.

We also excluded studies specifically focused on participants with a first episode of mania, psychotic depression, and psychosis related to substance abuse; studies in which participants had organic psychoses; or studies that recruited participants with prodromal symptoms, or at risk mental states.

If a study included participants with other diagnoses, we included the study only if participants with psychosis (as defined above) comprised at least 50% of the participants, or if we were able to extract the data on the eligible participants separately. If a study included participants with recent‐onset psychosis, without further defining this term, we included the study only if participants with a first episode of psychosis (as defined above) were at least 75% of the sample population.

Types of interventions

1. Cognitive behavioural therapy (CBT)

Cognitive behavioural therapy is used as a broad label that can include a variety of interventions, such as manualised CBT programmes, CBT integrated with other treatments (for example motivational interviewing), or CBT combined with third‐wave components (mindfulness, acceptance, and commitment therapy). Therefore, it is difficult to provide a single definition of CBT.

For the aims of this review, any CBT‐based interventions were eligible for inclusion. The role of different subtypes of CBT interventions was ascertained by analysing them separately, in a subgroup analysis (see Subgroup analysis and investigation of heterogeneity).

2. Control

We included studies that compared CBT to non‐CBT interventions:

• standard care, or treatment‐as‐usual (TAU), defined as the treatment that a person with a first episode of psychosis would normally receive had they not been involved in the trial. This may include antipsychotic medication;

• waiting list;

• inactive treatments (e.g. psychological placebo), defined as any inactive intervention intended to control for the time spent with the therapist (e.g. recreational activities) plus standard care; or

• other psychosocial interventions, such as family therapy or psychoeducation, plus standard care.

Types of outcome measures

We divided all outcomes into short‐term (up to six months since the onset of therapy), medium‐term (up to 12 months), and long‐term (over 12 months). This grouping reflects the recommendations of the National Institute for Health and Care Excellence (NICE), according to which, CBT should be delivered for at least 16 sessions and usually once a week (NICE 2014).

Data were presented both for the above‐mentioned time points separately and for the time points combined. When combining time points, we gave priority to medium‐term, then, if not available, short‐term, then long‐term. For outcomes such as relapse and adverse events, we gave priority to long‐term, then medium‐term, then short‐term. This priority was determined following the predefined time points for evaluating Risk of Bias 2 (see Assessment of risk of bias in included studies).

For outcomes, such as 'clinically important change', 'any change', and 'relapse', we used the definition used by each of the trials.

For valid scales, please see Data extraction and management.

Primary outcomes

1. Mental state

1.1 General

1.1.1 Average endpoint or change score on a general mental state scale (e.g. Positive and Negative Syndrome Scale (PANSS), Brief Psychiatric Rating Scale (BPRS))

Secondary outcomes

1. Mental state

1.1 General

1.1.1 Clinically important change in general mental state (study defined)

1.2 Specific

1.2.1 Clinically important change in positive symptoms (delusions, hallucinations, disordered thinking), as defined by individual studies

1.2.2 Average endpoint or change score on a published scale or subscale that addresses positive symptoms (e.g. PANSS positive, Scale for the Assessment of Positive Symptoms (SAPS))

1.2.3 Clinically important change in negative symptoms (avolition, poor self‐care, blunted affect), as defined by individual studies

1.2.4 Average endpoint or change score on a published scale or subscale that addresses negative symptoms (e.g. PANSS negative, Scale for the Assessment of Negative Symptoms (SANS))

1.2.5 Clinically important change in depressive symptoms, as defined by individual studies

1.2.6 Average endpoint or change score on a published scale or subscale that addresses depressive symptoms (e.g. Montgomery‐Asberg Depression Rating Scale (MADRS), Beck Depression Inventory (BDI))

2. Global state

2.1 Relapse. If more than one measure of relapse is reported, we will give priority to: 1. relapse defined with operationalised criteria (e.g. worsening on a symptom scale); 2. clinical judgement.

2.2 Recovery

2.3 Remission

2.4 Clinically important change in global state (e.g. global impression of much improved, or more than 50% improvement on a rating scale, such as the Clinical Global Impression scale (CGI))

2.5 Average endpoint or change score on a global state scale

3. Service use

3.1 Admission to hospital

3.2 Number of days in hospital

4. Leaving the study early

4.1 For any reason

4.2 Due to inefficacy

4.3 Due to adverse effect(s)

5. Functioning

5.1 Clinically important change in functioning: number of participants with a clinically important change in functioning, as defined in each study

5.2 Average endpoint or change score on a functioning scale. We will accept any published rating scales, such as the Global Assessment of Functioning (GAF), or the Psychosocial Performance Scale (PSP).

6. Cognitive functioning

6.1 Clinically important change in cognitive functioning

6.2 Average endpoint or change score on a cognitive functioning scale assessing multiple cognitive domains. We will accept any published cognitive functioning scale (e.g. the MATRICS^TM consensus cognitive battery, or the Neuropsychological Assessment Battery).

7. Quality of life

7.1 Clinically important change in quality of life

7.2 Average endpoint or change score on a quality of life scale. We will accept any published quality of life scale (e.g. Heinrich‐Carpenter quality of life scale).

8. Satisfaction with care

8.1 Recipient

8.1.1 Recipient satisfied with care

8.1.2 Average endpoint or change score on satisfaction with care scale. We will accept any published rating scale measuring satisfaction with care.

8.2 Carers

8.2.1 Carers satisfied with care

8.2.2 Average endpoint or change score on satisfaction with care scale

9. Behaviour

9.1 Occurrence of violent incidents (to self, others, or property)

9.2 Self‐injury

9.3 Suicide attempt

10. Adverse events

10.1 At least one adverse event

10.2 Specific adverse events. We will use the classification proposed by Linden and colleagues to record adverse events potentially connected with CBT (Linden 2014).

11. Mortality

11.1 Overall mortality

11.2 Mortality due to natural causes

11.3 Mortality due to suicide

Search methods for identification of studies

Electronic searches

The Cochrane Schizophrenia Information Specialist searched Cochrane Schizophrenia's register, using the following search strategy:

*Cognit* in Intervention Field of STUDY

In a study‐based register, searching for the major concept retrieved all the synonyms and relevant studies. This is because the studies have already been organised, based on their interventions, and linked to the relevant topics (Shokraneh 2017). This allowed rapid and accurate searches that reduced waste in the next steps of systematic reviewing (Shokraneh 2019).

Following standard Cochrane methodology, the information specialist compiled this register from systematic searches of major resources and their monthly updates (unless otherwise specified (Lefebvre 2020)).

• Cochrane Central Register of Controlled Trials (CENTRAL) in the Cochrane Library

• MEDLINE

• Embase

• Allied and Complementary Medicine (AMED)

• BIOSIS

• Cumulative Index to Nursing and Allied Health Literature (CINAHL)

• PsycINFO

• PubMed

• US National Institute of Health Ongoing Trials Register (ClinicalTrials.gov)

• World Health Organization International Clinical Trials Registry Platform (ICTRP; www.who.int/ictrp)

• ProQuest Dissertations and Theses A & I and its quarterly update

The register also included handsearches and conference proceedings (see the group's website). It did not place any limitations on language, date, document type or publication status.

Searching other resources

1. Reference searching

We inspected the references of all included studies for further relevant studies.

2. Personal contact

We contacted the first author of each included study for information regarding unpublished trials and additional data. We noted the outcome of this contact in the included studies or studies awaiting classification tables.

Data collection and analysis

Selection of studies

At least two review authors (IB, CC, or EA) have independently inspected citations from the searches and identified relevant abstracts. When disputes arose, we acquired the full report for more detailed scrutiny. At least two review authors (IB, CC, or EA) then obtained and independently inspected full reports of the abstracts or reports that met, or appeared to meet, the review criteria. When it was not possible to resolve disagreement by discussion, we discussed the issue with the senior author of the review team to resolve it. If, following discussion with the senior author, disagreement still existed, we attempted to contact the authors of the study concerned for clarification. We further documented all decisions.

Data extraction and management

1. Extraction

At least two review authors (amongst SFM, CC, LK or, IB) independently extracted data from all included studies. Data presented only in graphs and figures were also extracted with the same strategy. We discussed any disagreement. When it was not possible to resolve disagreements by discussion, we discussed the matter with the senior author. We documented all decisions. If necessary, we attempted to contact authors, and used an open‐ended request to obtain missing information, or clarification. IB and SL helped clarify issues regarding any remaining problems, and we documented these final decisions.

2. Management

2.1 Forms

We extracted data onto simple, pre‐designed forms, which we specifically designed for this review.

2.2 Scale‐derived data

We included continuous data from rating scales only if: (1) the psychometric properties of the measuring instrument have been described in a peer‐reviewed journal (Marshall 2000); (2) the measuring instrument has not been written or modified by one of the trialists for that particular trial; and (3) the instrument is a global assessment of an area of functioning and not a subscore, which is not, in itself, validated or shown to be reliable. However, we included subscores of scales if they were validated, or if they were predefined on a scale, such as the positive symptom, negative symptom, and general symptom scores of the PANSS (Kay 1986).

Ideally, the measuring instrument should be either: (1) a self‐report, or (2) completed by an independent rater or relative (not the therapist). However, we realised that this is often not clearly reported.

2.3 Endpoint versus change data

There are advantages of both endpoint and change data: change data can remove a component of between‐person variability from the analysis; however, calculation of change needs two assessments (baseline and endpoint), which can be difficult to obtain in unstable and difficult‐to‐measure conditions, such as schizophrenia. However, we decided to primarily use change data, and only use endpoint data if the former were not available. If necessary, we combined endpoint and change data in the analysis. This procedure was possible when using both mean differences (MDs), or standardised mean differences (SMDs (Higgins 2020)). Although theoretically, the combination of change and endpoint data when SMDs are used can be problematic, meta‐epidemiological research has shown that, on average, no major over‐ or underestimations can have been expected (da Costa 2013). We analysed endpoint and change data separately in a Sensitivity analysis for the primary outcome, mental state.

2.4 Skewed data

Continuous data on clinical and social outcomes are often not normally distributed. To avoid the pitfall of applying parametric tests to non‐parametric data, we applied the following checks to relevant continuous data before including them.

For endpoint data from studies including fewer than 200 participants, we calculated the observed mean minus the lowest possible value of the scale, and divided this by the standard deviation (Higgins 2020).

For example, in a scale that has a minimum possible score higher than 0 (such as the PANSS, which can have values from 30 to 210 (Kay 1986)), we subtracted the minimum score (in this case, 30) from the observed mean, and then divided by the standard deviation (SD). In a scale that has 0 as the minimum possible score, we divided the observed mean by the standard deviation.

For this calculation, we checked the original publication of the scales referenced in the studies, to determine if they can have a minimum possible score different from 0, and whether the adjustment described above is needed or not.

A ratio that is lower than one strongly suggests that the data are skewed. A ratio that is higher than one but less than two suggests that the data are skewed; if the ratio is larger than two, it is less likely that they are skewed (Altman 1996).

Where there was a suggestion of skewness (ratio < 2), we excluded the relevant studies in a sensitivity analysis, to check if they have an impact on the results (see Sensitivity analysis).

We reported skewed results in other data tables.

We entered all relevant data from studies with more than 200 participants in the analysis, regardless of the above rules, because skewed data pose less of a problem in large studies. We also entered all relevant change data, since it was difficult to tell whether data were skewed or not when continuous data were presented on a scale that includes a possibility of negative values (such as change data).

2.5 Common measurement

To facilitate comparison between trials, we converted variables that can be reported in different metrics, such as days in hospital (mean days per year, per week, or per month) to a common metric (e.g. mean days per month), where relevant.

2.6 Conversion from continuous to binary

Where possible, we tried to convert outcome measures to dichotomous data. We have done this by identifying cut‐off points on rating scales, and divided participants accordingly into 'clinically improved' or 'not clinically improved'. It is generally assumed that, if there is a 50% reduction in a scale‐derived score, such as the BPRS (Overall 1962), or the PANSS (Kay 1986), which corresponds to 'much improved' according to the clinical global impressions (CGI) of raters (Guy 1976), it could be considered a clinically significant response, especially for people who are acutely ill (Leucht 2005a; Leucht 2005b). Since our population, people with first‐episode psychosis, usually responded well to treatment, we planned to keep the cut‐off of 50% improvement to convert continuous to binary outcomes. If data based on these thresholds were not available, we would have used the primary cut‐off presented by the original authors, because the exact cut‐off was not as important in a meta‐analysis that uses risk ratios or odds ratios as effect sizes (Furukawa 2010). Since many studies provided data only for the 25% cut‐off, we took this one also from the other studies, in order to present consistent results.

2.7 Direction of graphs

When possible, we entered data so that the area to the left of the line of no effect indicates a favourable outcome for CBT. When keeping to this makes it impossible to avoid outcome titles with clumsy double‐negatives (e.g. not unimproved), we reported data for which the estimates left of the line indicate an unfavourable outcome, and noted this in the relevant graphs.

Assessment of risk of bias in included studies

Review authors SFM, CC, LM and IB worked independently to assess the risk of bias, using the RoB 2 tool and the criteria described in chapter 8 of the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2020; Sterne 2019).

This set of criteria was based on the judgement of the following domains:

1. bias arising from the randomisation process;

2. bias due to deviations from intended interventions;

3. bias due to missing outcome data;

4. bias in measurement of the outcome; and

5. bias in selection of the reported result.

For each domain, we rated the available signalling questions to reach a judgement (high, some concerns, low) following the algorithms implemented in the RoB 2 Excel tool (available on the riskofbiasinfo.org website).

RoB 2 generally allows one to address studies from two angles: a) the effect of assignment to the interventions at baseline, regardless of whether the interventions were received as intended, i.e. the intention‐to‐treat effect; and b) the adherence to the interventions, i.e. the per‐protocol effect, or completer analysis (chapter 8.2.2 in Higgins 2020). For the purpose of this review, we planned to assess the intention‐to‐treat effect, when possible.

We assessed the risk of bias for the following outcomes:

1. Mental state (primary outcome): average endpoint or change score on a general mental state scale; medium‐term

2. Global state: relapse; long‐term

3. Leaving the study early: leaving the study early for any reason; medium‐term

4. Functioning: average endpoint or change score on functioning scale; medium‐term

5. At least one adverse event, measured in the long term

For cluster‐randomised trials, we used the specific version of the RoB 2 tool, provided at sites.google.com/site/riskofbiastool/welcome/rob-2-0-tool.

Since we only used data from the first phase for cross‐over trials, we used the standard version of RoB 2.

If there was disagreement amongst the review authors, we made the final rating by consensus. When inadequate details of randomisation and other characteristics of trials were provided, we attempted to contact the authors of the studies to obtain further information.

We noted the risk of bias in both the text of the review, the relevant forest plots, and the summary of findings table(s).

Measures of treatment effect

1. Binary data

For binary outcomes, we calculated a standard estimation of the risk ratio (RR) and its 95% confidence interval (CI), as it has been shown that RR is more intuitive than odds ratios (Boissel 1999), and odds ratios tend to be interpreted as RR by clinicians (Deeks 2000). Although the number needed to treat for an additional beneficial outcome (NNTB) and the number needed to treat for an additional harmful outcome (NNTH), with their CIs, are intuitively attractive to clinicians, they are problematic to calculate and interpret in meta‐analyses (Hutton 2009). For binary data presented in the summary of findings table(s), we calculated the anticipated absolute effects for the intervention and control, where possible.

2. Continuous data

For continuous outcomes, we estimated the mean difference (MD) between groups, in particular, when natural units (such as days, kilograms, etc.) were used. We preferred not to calculate effect size measures (standardised mean difference (SMD)). However, if trials use outcome scales of considerable similarity (such as PANSS and BPRS), we presumed there was a small difference in measurement, and we calculated the SMD (Higgins 2020).

Unit of analysis issues

1. Cluster‐randomised trials

Studies increasingly use cluster‐randomisation (such as randomisation by clinician or practice), but analysis and pooling of clustered data pose problems. Authors often fail to account for intra‐class correlation in clustered studies, leading to a unit‐of‐analysis error, whereby P values are spuriously low, CIs unduly narrow, and statistical significance overestimated (Divine 1992). This causes type I errors (Bland 1997; Gulliford 1999).

Where clustering has been incorporated into the analysis of included studies, we presented these data as if they were from a non‐cluster‐randomised study, and adjusted for the clustering effect.

Where clustering was not accounted for in the included studies, we presented data in a table, with a (*) symbol to indicate the presence of a probable unit of analysis error. We contacted first authors of studies to obtain intra‐class correlation coefficients (ICC) for their clustered data, and adjusted for this by using accepted methods (Gulliford 1999).

We have sought statistical advice, and have been advised that the binary data from cluster trials presented in a report should be divided by a 'design effect'. This was calculated using the mean number of participants per cluster (m) and the ICC: thus, design effect = 1 + (m − 1) * ICC (Donner 2002). If the ICC was not reported, we assumed it to be 0.1 (Ukoumunne 1999).

If cluster‐randomised studies had been appropriately analysed, and intra‐class correlation coefficients and relevant data documented in the report had been taken into account, we used the generic inverse variance technique to synthesise the results with results of other studies.

2. Cross‐over trials

A major concern of cross‐over trials was the carry‐over effect. This occurs if an effect (e.g. pharmacological, physiological, or psychological) of the treatment in the first phase was carried over to the second phase. As a consequence, participants can differ significantly from their initial state at entry to the second phase, despite a wash‐out phase. For the same reason, cross‐over trials were not appropriate if the condition of interest is unstable (Elbourne 2002). As both carry‐over and unstable conditions were very likely in severe mental illness, we only used data from the first phase of cross‐over studies.

3. Studies with multiple treatment groups

When a study involved more than two treatment arms, we presented the additional treatment arms in the same comparisons, if relevant. If data were binary, we simply added and combined them in the two‐by‐two table. If data were continuous, we combined data following the formula in the Cochrane Handbook, which was implemented in the Review Manager Web (RevMan Web) calculator (Higgins 2020; Review Manager Web 2022). When additional treatment arms were not relevant, we did not use these data.

Dealing with missing data

1. Overall loss of credibility

Although, at some degree of loss of follow‐up, data lose credibility, we did not exclude studies based on this (Xia 2009).

However, if more than 50% of data were unaccounted for (lost to follow‐up), we excluded these studies in a Sensitivity analysis. If more than 50% of participants in one arm of a study were lost to follow‐up, but the total loss was less than 50%, we addressed this in the summary of findings table(s) by downgrading the certainty of the evidence (and not excluded the study in the sensitivity analysis). Finally, we also downgraded the certainty of the evidence in the summary of findings table(s), should the total loss be 25% to 50%.

2. Binary

We presented data using an intention‐to‐treat analysis (ITT). We undertook a Sensitivity analysis, excluding studies that analysed only the data from participants who remained in the study at the point of measurement.

3. Continuous

3.1 Standard deviations

If standard deviations (SDs) were not reported, we tried to obtain the missing values from the authors. If SDs were still not available, and there were missing measures of variance for continuous data, but an exact standard error (SE) and CI was available for group means, and either a P value or t value was available for differences in the mean, we calculated SDs according to the rules described in the Cochrane Handbook (Higgins 2020).

When only the SE was reported, SDs were calculated with the formula SD = SE * √(n). The Cochrane Handbook presents detailed formulae for estimating SDs from P, t, or F values, CIs, ranges, or other statistics (Higgins 2020). If these formulae did not apply, we calculated the SDs according to a validated imputation method, which was based on the SDs of the other included studies (Furukawa 2006). Although some of these imputation strategies could introduce error, the alternative would be to exclude a given study’s outcome and, thus, lose information. Nevertheless, we examined the validity of the imputations in a sensitivity analysis that excluded trials that imputed values.

3.2 Assumptions about participants who left the trials early or were lost to follow‐up

Various methods were available to account for participants who left the trials early, or were lost to follow‐up. Some trials just presented the results of participants who completed the study; others used the method of last‐observation‐carried‐forward (LOCF). More recently, methods, such as multiple imputation or mixed‐effects models for repeated measurements (MMRM), have become more of a standard. While the latter methods seemed to be more precise than LOCF, we decided that the core problems in randomised trials of schizophrenia were often the high percentage of participants who left the studies early, and the differences between groups in their reasons for doing so (Leon 2006). Therefore, we did not exclude studies based on the statistical approach used. However, by preference, we used the more sophisticated approaches; i.e. we used MMRM or multiple imputation before LOCF, and we only presented completer analyses if ITT data were not available. We also addressed this issue in the 'Missing outcome data' domain of RoB 2.

Assessment of heterogeneity

1. Clinical heterogeneity

Initially, we considered all the included studies, without seeing the comparison data, to judge clinical heterogeneity. We inspected all studies for participants or settings that were clearly different to the participants or settings in the other included studies, and discussed such differences or participant groups.

2. Methodological heterogeneity

Initially, we considered all the included studies, without seeing comparison data, to judge methodological heterogeneity. We inspected all studies for clearly outlying methods, which we had not predicted would arise, and discussed them.

3. Statistical heterogeneity

3.1 Visual inspection

We inspected graphs visually to investigate the possibility of statistical heterogeneity (e.g. observing the overlap of confidence intervals).

3.2 Using the I² statistic

We investigated heterogeneity between studies by considering the I² statistic alongside the Chi² P value. The I² statistic provides an estimate of the percentage of inconsistency thought to be due to chance (Higgins 2003). The importance of the observed value of I² depends on the magnitude and direction of effects, as well as the strength of evidence for heterogeneity (e.g. P value from Chi² test, or a confidence interval for I²). We interpreted the I² estimate as follows (chapter 10; Cochrane Handbook (Higgins 2020)):

1. 0% to 40%: might not be important;

2. 30% to 60%: may represent moderate heterogeneity;

3. 50% to 90%: may represent substantial heterogeneity;

4. 75% to 100%: may represent considerable heterogeneity.

When we found substantial levels of heterogeneity (> 50%) for the primary outcome, we explored reasons for heterogeneity (Subgroup analysis and investigation of heterogeneity).

Assessment of reporting biases

Reporting biases arise when the dissemination of research findings is influenced by the nature and direction of results (Egger 1997). These are described in Chapter 13 of the Cochrane Handbook (Higgins 2020).

1. Protocol versus full study

We tried to locate protocols of the included randomised trials. If the protocol was available, we compared outcomes in the protocol and in the published report. If the protocol was not available, we compared outcomes listed in the Methods section of the trial reported with the actual reported results.

2. Funnel plot

Funnel plots may be useful in investigating reporting biases, but are of limited power to detect small‐study effects. We did not use funnel plots for outcomes for which there were fewer than ten studies, or when all studies were of similar size. When funnel plots were possible, we sought statistical advice on their interpretation. We applied the Egger's test for funnel plot asymmetry to support the visual inspection of the forest plot with a statistical test, using R software (R).

Data synthesis

We undertook a meta‐analysis only if we judged that participants, interventions, comparisons, and outcomes were sufficiently similar to ensure an answer that was clinically meaningful.

We used a random‐effects model for analyses. The random‐effects method incorporated an assumption that the different studies were estimating different, yet related, intervention effects. This often seemed to be true to us, and the random‐effects model took into account differences between studies, even if there was no statistically significant heterogeneity. The disadvantage of this model was that it puts added weight onto small studies.

To check the effect of the decision to use the random‐effects model, we applied the fixed‐effect model in a sensitivity analysis of the primary outcome.

Subgroup analysis and investigation of heterogeneity

1. Subgroup analyses

When we found substantial levels of heterogeneity (> 50%) for the primary outcome, we explored reasons for heterogeneity by conducting the following subgroup analyses.

1.1 Different types of CBT

We analysed different kinds of CBT as different subgroups, such as manualised CBT programmes, CBT integrated with other treatments (for example, motivational interviewing), or combined with third wave components (mindfulness, acceptance, and commitment therapy).

1.2 Modality of treatment

We analysed CBT administered in individual or group sessions separately.

1.3 Method of delivery

We analysed CBT delivered face‐to‐face or with other methods of delivery (online, telephone) separately.

1.4 Control group

In order to investigate the role of the comparator, we analysed CBT compared with different control conditions separately, such as treatment‐as‐usual, waiting list, psychological placebo, or other psychological treatments.

1.5 Therapist experience

We planned to investigate the role of the experience of the therapist providing the CBT intervention, by separately analysing trials that used only expert therapists, and trials that also included therapists in training.

Sensitivity analysis

Where possible, we undertook sensitivity analyses for the primary outcome – mental state – to explore the influence of the following factors on effect size. If there were substantial differences in the direction or precision of effect estimates in any of the sensitivity analyses listed below, we discussed them in the discussion section.

1. Blinding of outcome assessor

We excluded trials that did not blind the outcome assessor.

2. Assumptions for missing data

We excluded studies that used completer analyses only.

3. Loss to follow‐up

We planned to exclude studies in which the overall loss of data was greater than 50%, but this was never the case.

4. Risk of bias

We analysed the effects of excluding trials that were at overall high risk of bias.

5. Imputed values

We excluded trials in which we used imputed values for ICC to calculate the design effect in cluster‐randomised trials, or in which we imputed SDs.

6. Fixed‐effect

We re‐synthesised data using a fixed‐effect model.

7. Separating endpoint and change data

We planned to analyse studies that provided data as endpoint scores and change scores separately.

8. Skewed data

We planned to exclude studies for which there was a suggestion of skewness (mean/SD ratio lower than 2). If the results were different from the main analysis (from superiority of the intervention to superiority of the control, or vice‐versa), we would have excluded these studies from the main analysis, and presented their data in other data tables.

9. Outliers

We performed an additional post hoc sensitivity analysis excluding an outlier study (Palma‐Sevillano 2019).

Summary of findings and assessment of the certainty of the evidence

We used the GRADE approach to interpret findings, and GRADEpro GDT software to export data from our RevMan Web file to create a summary of findings table (GRADEpro GDT; Higgins 2020; Review Manager Web 2022). A summary of findings table provided outcome‐specific information concerning the overall certainty of evidence from the included studies in the comparison, the magnitude of effect of the interventions examined, and the sum of available data on all outcomes we rated as important to healthcare and decision‐making. We used the overall RoB 2 judgements to feed into the GRADE assessment.

We created a summary of findings table for the comparison, cognitive behavioural therapy plus standard care versus control.

We included results for these outcomes in the summary of findings table.

1. Mental state: average endpoint or change score on a general mental state scale (medium‐term)

2. Global state: relapse (long‐term)

3. Leaving the study early: leaving the study early for any reason (medium‐term)

4. Functioning: average endpoint or change score on functioning scale (medium‐term)

5. At least one adverse event (long‐term)

If data were not available for these prespecified outcomes, but were available for ones that are similar, we presented the closest outcome to the prespecified one in the summary of findings table, but took this into account when grading the evidence.

Results

Description of studies

For details, see Characteristics of included studies and Characteristics of excluded studies.

Results of the search

Our search of the Cochrane Schizophrenia Group's Study‐Based Register of Trials identified 260 articles for full‐text screening (corresponding to 74 studies) (Figure 1).We included a total of 28 studies in the review and 26 studies in the quantitative synthesis.

1.

Included studies

We included 28 studies (2498 participants) in the review, of which 26 studies (2407 participants) contributed to the meta‐analyses and are described below.

1. Design and duration

All included studies were RCTs. Of the studies contributing data to the meta‐analyses, 18 studies had a duration of treatment of up to 26 weeks (Bucci 2018 (Actissist); Edwards 2006; Edwards 2011; Gleeson 2009 (EPISODE II); Haddock 1999; Hjorthoj 2013 (CapOpus); Jackson 2008 (ACE); Jackson 2009; Lecomte 2008; Lepage 2022; Liu 2019; Madigan 2013; Morrison 2020 (MAPS); Pos 2019; Lewis 2002 (SOCRATES); Puig‐Navarro 2020; Sönmez 2020; Uzenoff 2008); seven studies lasted between 27 and 52 weeks (Barrowclough 2014; GetUp; Gonzalez‐Ortega 2021; Morrison 2018 (COMPARE); Müller 2020; Palma‐Sevillano 2019; Supereden3); and no studies lasted more than 52 weeks. Eighteen studies provided data at a longer follow‐up after the end of treatment. For one study, the assessment took place at six months but the study duration remained unclear (Jolley 2003).

2. Participants

Of all the studies which contributed data to the meta‐analyses, four studies did not provide further information on the diagnostic criteria (did not refer to a diagnostic manual) (Bucci 2018 (Actissist); Lecomte 2008; Lepage 2022; Supereden3). In 12 studies, participants could be diagnosed according to the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition (DSM‐IV) criteria (Barrowclough 2014; Edwards 2006; Edwards 2011; Gleeson 2009 (EPISODE II); Haddock 1999; Liu 2019; Madigan 2013; Müller 2020; Palma‐Sevillano 2019; Lewis 2002 (SOCRATES); Sönmez 2020; Uzenoff 2008). In three studies, the researchers used Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition, Text Revision (DSM‐IV‐TR) diagnostic criteria (Gonzalez‐Ortega 2021; Jackson 2008 (ACE); Pos 2019). In one study, participants were diagnosed according to DSM‐V (Puig‐Navarro 2020). In six studies, the authors used the ICD‐10 (GetUp; Hjorthoj 2013 (CapOpus); Jackson 2009; Jolley 2003; Morrison 2018 (COMPARE); Morrison 2020 (MAPS)). The average age of participants in the studies contributing to meta‐analyses was about 24.39 years old.

3. Size

The mean number of participants in the included studies was 92. Furthermore, the smallest study included only 19 participants (Uzenoff 2008), whereas the largest study randomised 444 participants (GetUp).

4. Setting

In only one study (Haddock 1999), the participants were recruited in an inpatient setting; in eight studies (Bucci 2018 (Actissist); Edwards 2006; GetUp; Gleeson 2009 (EPISODE II); Lecomte 2008; Lepage 2022; Liu 2019; Puig‐Navarro 2020) in an outpatient setting; and in eleven studies in both in‐ and outpatient settings (Edwards 2011; Gonzalez‐Ortega 2021; Jackson 2008 (ACE); Jolley 2003; Madigan 2013; Morrison 2018 (COMPARE); Müller 2020; Palma‐Sevillano 2019; Pos 2019; Lewis 2002 (SOCRATES); Uzenoff 2008). For six studies (Barrowclough 2014; Hjorthoj 2013 (CapOpus); Jackson 2009; Morrison 2020 (MAPS); Sönmez 2020; Supereden3) this information was not available. Most of the studies were conducted in the UK (Barrowclough 2014; Bucci 2018 (Actissist); Haddock 1999; Jackson 2009; Jolley 2003; Morrison 2018 (COMPARE); Morrison 2020 (MAPS); Lewis 2002 (SOCRATES); Supereden3). Four studies were conducted in Australia (Edwards 2006; Edwards 2011; Gleeson 2009 (EPISODE II); Jackson 2008 (ACE)); two studies were conducted in Canada (Lecomte 2008; Lepage 2022); one in the USA (Uzenoff 2008); and one in China (Liu 2019). The remaining nine studies were all conducted in Europe (GetUp; Gonzalez‐Ortega 2021; Hjorthoj 2013 (CapOpus); Madigan 2013; Müller 2020; Palma‐Sevillano 2019; Pos 2019; Puig‐Navarro 2020; Sönmez 2020).

5. Interventions

In all included studies, cognitive behavioural therapy plus standard care was compared with a control group.

Nevertheless, different types of CBT can be distinguished: four studies (Bucci 2018 (Actissist); GetUp; Palma‐Sevillano 2019; Supereden3) comprised an integrated form of CBT (CBT integrated in an app, CBT combined with family intervention and case management, CBT combined with psychoeducation and family intervention, CBT combined with social recovery therapy). Moreover, four studies (Barrowclough 2014; Edwards 2006; Hjorthoj 2013 (CapOpus); Madigan 2013) specifically focused on a special form of CBT and Motivational Interviewing (MI) for cannabis cessation. Additionally, three studies (Gleeson 2009 (EPISODE II); Jackson 2009; Uzenoff 2008) used special types of CBT such as Relapse Prevention Therapy (RPT), Cognitive Recovery Intervention (CRI) and Adherence‐Coping‐Education (ACE). The remaining studies provided a form of CBT that could be considered standard (Edwards 2011; Gonzalez‐Ortega 2021; Haddock 1999; Jackson 2008 (ACE); Jolley 2003; Lecomte 2008; Lepage 2022; Liu 2019; Morrison 2018 (COMPARE); Morrison 2020 (MAPS); Müller 2020; Pos 2019; Puig‐Navarro 2020; Santos 2008; Sönmez 2020).

Most of the studies (17: Barrowclough 2014; Edwards 2011; GetUp; Gleeson 2009 (EPISODE II); Gonzalez‐Ortega 2021; Hjorthoj 2013 (CapOpus); Jackson 2009; Jolley 2003; Liu 2019; Madigan 2013; Morrison 2018 (COMPARE); Morrison 2020 (MAPS); Müller 2020; Palma‐Sevillano 2019; Pos 2019; Sönmez 2020; Supereden3) compared CBT plus standard care with treatment as usual (TAU). Additionally, seven studies compared CBT plus standard care with active control (Bucci 2018 (Actissist); Edwards 2006; Haddock 1999; Jackson 2008 (ACE); Lepage 2022; Puig‐Navarro 2020; Uzenoff 2008). Two studies compared CBT plus standard care both with active control and TAU (Lecomte 2008; Lewis 2002 (SOCRATES)).

Except for one study (Bucci 2018 (Actissist)), all studies took place within a face‐to‐face setting. Moreover, three studies (Lecomte 2008; Lepage 2022; Madigan 2013) took place in a group setting whereas, in 20 studies, CBT was delivered individually. Three studies (Pos 2019; Puig‐Navarro 2020; Supereden3) had both group and individual elements.

These aspects of the intervention were investigated in the subgroup analyses (Analysis 1.82; Analysis 1.83; Analysis 1.84; Analysis 1.85).

1.82. Analysis.

Comparison 1: CBT plus standard care versus control, Outcome 82: Subgroup analysis ‐ Different types of CBT

1.83. Analysis.

Comparison 1: CBT plus standard care versus control, Outcome 83: Subgroup analysis ‐ Modality of treatment

1.84. Analysis.

Comparison 1: CBT plus standard care versus control, Outcome 84: Subgroup analysis ‐ Method of delivery

1.85. Analysis.

Comparison 1: CBT plus standard care versus control, Outcome 85: Subgroup analysis ‐ Control group

Standard care usually comprised antipsychotic medication, but details on dose and type of antipsychotic were rarely provided in the studies. The percentage of participants taking antipsychotic medication, where available, is reported in Included studies.

6. Outcomes

Different studies used various scales to measure the following outcomes. As study reporting has often been incomplete, authors were contacted for missing outcome data as well as clarifications via email. Some study authors provided additional information and data (see Notes in Characteristics of included studies).

6.1. Outcome scales

6.1.1 Mental state

Positive and Negative Symptom Scale (PANSS) (Kay 1986)

PANSS is a clinician‐rated scale containing 30 items, where each item is scored on a 7‐point Likert scale ranging from 1 'absent' to 7 'extreme'. PANSS originally consisted of three subscales assessing both positive, negative, and general psychopathological symptoms of schizophrenia. By summing the scores of all items, a total score can be calculated which represents a measure of overall symptoms of schizophrenia (ranging from 30 to 210, with a higher score corresponding to higher severity of symptoms). Fifteen studies used PANSS (Barrowclough 2014; Bucci 2018 (Actissist); GetUp; Gonzalez‐Ortega 2021; Hjorthoj 2013 (CapOpus); Liu 2019; Morrison 2018 (COMPARE); Morrison 2020 (MAPS); Müller 2020; Palma‐Sevillano 2019; Pos 2019; Lewis 2002 (SOCRATES); Supereden3; Sönmez 2020; Uzenoff 2008), but Sönmez 2020 only had data for PANSS positive and negative. There are also different versions of PANSS, but none of the included studies used them.

The Psychotic Symptom Rating Scales (PSYRATS) (Haddock 1999)

PSYRATS consists of 11 items evaluating the severity of psychotic symptoms over different dimensions. Only one study (Liu 2019) used PSYRATS for assessing psychotic symptoms.

Brief Psychiatric Rating Scale (BPRS) (Overall 1962)

BPRS is a scale that measures the severity of psychiatric symptoms, including psychotic symptoms. It is rated by the clinician, and the most frequently used version of the scale is composed of 18 items, including positive, negative, and affective symptoms. Each item is scored on a 7‐point Likert scale from 1 'not present' to 7 'extremely severe'. It is possible to calculate a total score by adding the scores of all items as a measure of overall symptoms of schizophrenia (ranging from 18 to 126, where higher scores mean more severe symptoms). Four studies used BPRS (Edwards 2006; Gleeson 2009 (EPISODE II); Haddock 1999; Lecomte 2008). Two studies assessed only BPRS positive (Edwards 2011; Jackson 2008 (ACE)).

Scale for the Assessment of Positive Symptoms (SAPS) (Andreasen 1984a)

Positive symptoms of schizophrenia can be assessed on the SAPS rating scale. The scale can be divided into the following four domains: hallucinations, delusions, bizarre behaviour, and positive formal thought disorder. All the items are rated from 0 (none) to 5 (severe). Two studies used SAPS (Lepage 2022; Madigan 2013).

Scale for the Assessment of Negative Symptoms (SANS) (Andreasen 1984b)

SANS is a clinician‐rated scale consisting of 25 items to assess negative symptoms of patients with schizophrenia. The scale comprises five domains (affective flattening or blunting, alogia, avolition/apathy, anhedonia/associality, attention), and each item is rated on a scale from 0 (none) to 5 (severe). Six studies used SANS (Edwards 2006; Edwards 2011; Gleeson 2009 (EPISODE II); Jackson 2008 (ACE); Lepage 2022; Supereden3).

Brief Negative Symptom Scale (BNSS) (Kirkpatrick 2011)

BNSS is composed of 13 items measuring the five domains of blunted affect, alogia, associality, anhedonia, and avolition. The items can be rated from 0 to 6, with higher scores meaning more severe symptoms. One study used BNSS (Pos 2019).

Beck Depression Inventory‐Short Form (BDI‐SF) (Beck 1972)

The BDI‐SF is a short form of the BDI, and comprises 13 items which are rated from 0 to 3, with higher scores corresponding to more severe depression. Two studies used BDI‐SF (Edwards 2006; Edwards 2011).

Beck Depression Inventory (BDI‐II) (Beck 1996a, Beck 1996b)

BDI‐II consists of 21 symptom‐attitude categories, that describe specific behavioural manifestations of depression. Each category consists of four to five self‐evaluative statements, to be evaluated with a score from 0 to 3, with higher scores indicating more severe depression. Two studies used BDI‐II (Sönmez 2020; Supereden3).

Hospital Anxiety and Depression Scale (HADS‐D) (Zigmond 1983)

HADS is a self‐assessment scale that measures anxiety and depression. The fourteen‐item HADS includes the Anxiety subscale (HADS‐A) and the Depression subscale (HADS‐D), with each one composed of seven items. Each item can be rated from 0 to 3. By adding the subscale scores, a total of 0 to 21 points can be obtained for each of the depression and anxiety subscales. It is stated that a total of 0 to 7 points for each subscale are in the normal range, 8 to 10 points suggest the presence of a mood disorder, and a score of 11 and above indicates possible mood disorder. Two studies used HADS‐D (Morrison 2018 (COMPARE); Morrison 2020 (MAPS)).

Calgary Depression Scale for Schizophrenia (CDSS) (Addington 1993)

CDSS is a clinician‐rated scale measuring depressive symptoms in people with schizophrenia. It includes 9 items (8 structured questions and one observation item) considering depressive symptoms that can be scored on a 4‐point Likert scale from 0 'absent' to 3 'severe'. By summing the score of all items, a total score can be computed which ranges from 0 to 27, with a higher score corresponding to a higher severity of symptoms. Nine studies used CDSS (Barrowclough 2014; Bucci 2018 (Actissist); Jackson 2009; Jolley 2003; Lepage 2022; Madigan 2013; Müller 2020; Sönmez 2020; Uzenoff 2008).

Montgomery–Åsberg Depression Rating Scale (MADRS) (Montgomery 1979)

MADRS consists of ten items measuring depressive symptoms such as reduced sleep, reduced appetite, apparent sadness, reported sadness and suicidal thoughts. The items are rated on a score from 0 to 6, where 6 represents the highest severity of symptoms. One study used MADRS (Gleeson 2009 (EPISODE II)).

Hamilton Depression Rating Scale (HDRS) (Hamilton 1960)

HDRS consists of 17 variables plus four additional variables. The variables/items can be rated either on a 3‐point rating scale (from 0 to 2) or on a 5‐point rating scale (from 0 to 4), with higher scores meaning more severe symptoms. Two studies used HDRS (GetUp; Gonzalez‐Ortega 2021).

Centre for Epidemiologic Studies Depression scale ‐ Revised (CESD‐R) (Eaton 2001)

CESD‐R can be used to screen for depression as well as depressive disorders. It was initially developed by Laurie Radloff in 1977 (Radloff 1977) and revised by Eaton in 2001.Only one study used CESD‐R (Jackson 2008 (ACE)).

6.1.2 Global state

Clinical Global Impression (CGI) (Guy 1976)

CGI scales are 7‐point clinician‐rated scales, composed of two scales measuring global severity of illness (CGI‐Severity, or CGI‐S) and global clinical improvement (CGI‐Improvement, or CGI‐I). A lower score means lower severity of illness, more improvement or less deterioration, respectively. A CGI‐I score of 1 'very much improved' or 2 'much improved' corresponds to a clinically important improvement. One study used CGI (Palma‐Sevillano 2019). Three studies used CGI‐S (Edwards 2011; Gonzalez‐Ortega 2021; Morrison 2018 (COMPARE)). Two studies used CGI‐I (Gonzalez‐Ortega 2021; Morrison 2018 (COMPARE)).

6.1.3 Functioning

Global Assessment of Functioning (GAF)(APA 1987, Lehman 1983)

GAF is a clinician‐rated scale measuring the impact of a patient's severity of illness on their daily life. It is brief and easily administered, using a numeric scale from 0 to 100, broken into 10 intervals, with a higher score indicating better functioning.Ten studies used GAF (Barrowclough 2014; Bucci 2018 (Actissist); GetUp; Gonzalez‐Ortega 2021; Hjorthoj 2013 (CapOpus); Madigan 2013; Müller 2020; Palma‐Sevillano 2019; Pos 2019; Sönmez 2020).

Social and Occupational Functioning Assessment Scale (SOFAS) (APA 1994, Goldman 1992)

SOFAS was developed to measure the level of social and occupational functioning, taking into account the impact on a person's general health. The scale, which is found in DSM‐IV, is especially used to determine 5th axis features and to measure functionality. It can be rated from 0 to 100 points, with higher scores indicating good functioning. Five studies used SOFAS (Edwards 2006; Edwards 2011; Gleeson 2009 (EPISODE II); Jackson 2008 (ACE); Lepage 2022).

Personal and Social Performance Scale (PSP) (Morosini 2000)

PSP is a clinician‐rated scale that measures personal and social functioning. It is rated on a single‐item, from 1 to 100 and divided into 10 equal intervals, with a higher score meaning better functioning. The total score is calculated considering four indicators rated on a 6‐point scale from 1 'absent' to 6 'very severe' difficulties: 1) socially useful activities, including work and study, 2) personal and social relationships, 3) self‐care, and 4) disturbing and aggressive behaviours.Three studies used PSP (Bucci 2018 (Actissist); Liu 2019; Morrison 2018 (COMPARE)).

Functioning Assessment Short test (FAST) (Rosa 2007)

The interviewer‐administered scale is composed of 24 items that assess impairment and disability in six areas of functioning: autonomy, occupational functioning, cognitive functioning, financial issues, interpersonal relationships, and leisure time. The items are rated on a 4‐point scale, with 0 meaning no difficulty and 3 meaning severe difficulty. One study used FAST (Gonzalez‐Ortega 2021).

First‐Episode Social Functioning Scale (FESFS) (Lecomte 2014)

FESFS assesses nine functional domains: 'Life skills', 'Interaction with people', 'Friends and activities', 'Intimacy', 'Family', 'Work relationships and social activities', 'Work skills', 'Interpersonal relationships and social activities at school', and 'Educational skills'. Each domain or area of functioning is rated by the participant on a scale from 1 to 4 (e.g. 'strongly agree' to 'strongly disagree' or 'strongly disagree' to 'always agree'), considering their perceived competencies and frequency of involvement. Higher scores indicate better functioning. One study used FESFS (Morrison 2020 (MAPS)).

6.1.4 Quality of life

World Health Organization Quality of Life abbreviated form (WHOQoL‐BREF) (O'Carroll 2000)

WHOQoL‐BREF is a self‐rated scale measuring quality of life. It is composed of 26 questions regarding satisfaction with health, psychological functioning, social relationships, and environmental opportunities within the last 2 weeks. The participant is asked to rate each question on a 5‐point Likert scale from 1 to 5. A total score can be calculated by adding the scores of all items, with higher scores indicating a better quality of life.Three studies used WHOQoL‐BREF (Gleeson 2009 (EPISODE II); Madigan 2013; Morrison 2018 (COMPARE)).

EuroQol‐5 Dimensions three‐level version (EQ‐5D‐3L) (The EuroQol Group 1990)

EQ‐5D‐3L is a self‐rated scale used to measure health‐related quality of life. It is composed of two parts: 1) a descriptive system with five questions about mobility, self‐care, usual activities, pain/discomfort, and anxiety/depression, each rated from 1 'no problem' to 3 'extreme problems'; and 2) a visual analogue scale rated from 0 'worst imaginable' to 100 'best imaginable' health.The scores of the first part can be converted to a single summary index of health‐related quality of life (EQ‐5D‐HRQoL) according to the preferences of the general population of a region/country, and can range from less than 0 (worse than death) to 1 (perfect health). Three studies used EQ‐5D‐3L (Bucci 2018 (Actissist); Morrison 2020 (MAPS); Pos 2019).

Heinrich‐Carpenter Quality of Life Scale (QLS) (Heinrichs 1984)

QLS is a clinician‐rated scale administered as a semi‐structured interview. It is composed of 21 items in four domains: 1) interpersonal relations, 2) instrumental role, 3) intrapsychic foundations, and 4) common objects and activities. Each item is rated regarding the last four weeks on a 7‐point Likert scale from 0 to 6, with a higher score indicating less impairment. A total score can be calculated by adding the scores of all items. Two studies used QLS (Edwards 2011; Uzenoff 2008).

Modular System for Quality of Life (MSQoL) (Pukrop 2003)

The self‐report scale consists of 46 items in seven domains, and a total score can also be calculated. The items are rated from 1 to 7, with higher scores indicating a higher quality of life. One study used MSQoL (Müller 2020).

Manchester Short Assessment of Quality of Life (MANSA)(Priebe 1999)

MANSA is a clinician‐rated scale representing a shortened and slightly adapted version of the Lancashire Quality of Life Profile (LQLP) consisting of 16 items. Four items can be answered with 'yes/no', and 12 items are rated on a scale ranging from 1 to 7, with higher scores meaning better quality of life. Only one study used MANSA (Hjorthoj 2013 (CapOpus)).

7. Funding sources

One study was sponsored by a private foundation (Sönmez 2020); 17 studies reported public funding; three studies were jointly funded by public institutions and pharmaceutical companies; and five studies did not provide clear information on funding.

Excluded studies

We excluded 30 studies based on full‐text assessment. The reasons for exclusion were:

• ineligible design: not randomised (Browning 2013; Edwards 2008; Jackson 2005; Lally 2019; Newton 2005);

• ineligible population: (Drury 1996; Fowler 2002; Fowler 2009 (Isrep); Gumley 2003; Haddock 2018; Kemp 2007; Lecomte 2020; Lysaker 2007; McGorry 1998; McGorry 2000; Morrison 2016; Morrison 2018 (FOCUS); Tarrier 2014; Van der Gaag 2011; Wölwer 2022);

• ineligible intervention: no CBT plus standard care (Amann 2005; Gaebel 2005; Gleeson 2012; Griffiths 2019; Kidd 2020; Morrison 2014 (ACTION); Ostergaard Christensen 2014; Vohs 2018; Weller 2018 (EBIRT));

• other reasons: cancelled (NCT00722163).

Ongoing studies

We identified eight ongoing studies that matched our inclusion criteria (Barrowclough 2001; Haddock 2000; ISRCTN60855021; ISRCTN65536352; NCT00161408; NCT03491852; NCT04180709; NCT04916626).

Studies awaiting classification

There are seven studies awaiting classification (Gonzalez‐Ortega 2022; Lewis 2000; NCT02653729; NCT04889911; Renton 2004; Suri 2001; Wood 2022).

Risk of bias in included studies

The risk of bias assessment for each of the predefined outcomes is located in the risk of bias section (see Characteristics of included studies), including all domain judgements and support for judgements, and at the side of the relevant forest plots. Detailed risk of bias assessments are available upon request.

In general, the assessment of risk of bias ranged from low to high across studies and outcomes.

For the primary outcome, the risk of bias across all studies was predominantly assessed as having some concerns. In most of the studies, the allocation of participants was described as randomised and details were given on how the randomisation sequence was generated. However, in one study, the randomisation process could be criticised (GetUp). Furthermore, baseline differences could only be detected in two studies (Lecomte 2008; Uzenoff 2008) indicating potential problems with the randomisation process.

In all studies examining the primary outcome, participants and personnel were aware of the intervention, but there were no apparent deviations because the trial context could be identified. In 14 out of 20 studies, an appropriate analysis was used. In six studies (Bucci 2018 (Actissist); Gonzalez‐Ortega 2021; Haddock 1999; Jackson 2008 (ACE); Jolley 2003; Palma‐Sevillano 2019), it was either unclear or no appropriate analysis was used.

Referring to missing data, there were only four studies (Jolley 2003; Lecomte 2008; Morrison 2020 (MAPS); Supereden3) for which data on the primary outcome were not available for more than 20% of the participants and which had high risk of bias in this domain.

Except for one study (GetUp), in all studies, the primary outcome was measured on a validated rating scale by blind outcome assessors.

In eleven out of 20 studies the primary outcome was mentioned in the prespecified plan of the study, whereas in nine studies (Barrowclough 2014; Edwards 2006; Gleeson 2009 (EPISODE II); Haddock 1999; Jackson 2008 (ACE); Jolley 2003; Müller 2020; Palma‐Sevillano 2019; Uzenoff 2008), the primary outcome was either not mentioned in a prespecified plan or there was no prespecified plan available.

For the outcome, relapse, risk of bias across the studies ranged from overall low (Lewis 2002 (SOCRATES); Liu 2019; Morrison 2018 (COMPARE)), having some concerns (Barrowclough 2014; Gleeson 2009 (EPISODE II); Haddock 1999) to high (Morrison 2020 (MAPS)).

For all studies except one (GetUp), the risk of bias for the outcome, leaving the study early for any reason, was judged as being overall low or having some concerns.

For the outcome, functioning, the risk of bias in all studies was judged as being either low overall or having some concerns overall, except for two studies (GetUp; Morrison 2020 (MAPS)) that had an overall high risk of bias.

Finally, for the outcome of adverse events, the only study (Morrison 2020 (MAPS)) providing data was judged as having some concerns.

Effects of interventions

See: Table 1

See Table 1 and forest plots for detailed results.

Comparison 1: CBT plus TAU versus TAU

Primary outcomes

1. Mental state

1.1. General

1.1.1 Average endpoint or change score on a general mental state scale

Twenty studies provided data on this outcome. CBT + TAU was associated with a greater reduction in overall symptoms of schizophrenia (SMD ‐0.27, 95% CI ‐0.47 to ‐0.08, 20 RCTs, n = 1508, I² = 68%, substantial heterogeneity, low certainty of the evidence) (Analysis 1.5). No differences emerged between separate time points (P = 0.96) (Analysis 1.6).

1.5. Analysis.

Comparison 1: CBT plus standard care versus control, Outcome 5: Mental state ‐ general: mean endpoint overall symptom scales (high = poor) (combined time points)

1.6. Analysis.

Comparison 1: CBT plus standard care versus control, Outcome 6: Mental state ‐ general: mean endpoint overall symptom scales (high = poor) (separated time points)

SENSITIVITY ANALYSES

1.1.2 Sensitivity analysis ‐ Blinding of outcome assessor

After excluding one trial (GetUp) that did not employ a blind outcome assessor, the results remained similar to the main analysis (SMD ‐0.28, 95% CI ‐0.50 to ‐0.06, 19 RCTs, n = 1229, I² = 69%, substantial heterogeneity) (Analysis 1.76).

1.76. Analysis.

Comparison 1: CBT plus standard care versus control, Outcome 76: Sensitivity analysis ‐ Blinding of outcome assessor

1.1.3 Sensitivity analysis ‐ Assumptions for missing data

After excluding five studies (Bucci 2018 (Actissist); Gonzalez‐Ortega 2021; Haddock 1999; Jackson 2008 (ACE); Jolley 2003) that used completer analyses only, results did not essentially change from the main analysis (SMD ‐0.29, 95% CI ‐0.52 to ‐0.05, 15 RCTs, n = 1215, I² = 73%, substantial heterogeneity) (Analysis 1.77).

1.77. Analysis.

Comparison 1: CBT plus standard care versus control, Outcome 77: Sensitivity analysis ‐ Assumptions for missing data

1.1.4 Sensitivity analysis ‐ Loss to follow‐up

None of the studies had more than a 50% dropout (loss to follow‐up) rate and, therefore, no sensitivity analysis was conducted.

1.1.5 Sensitivity analysis ‐ Risk of bias

After removing five studies with an overall high risk of bias (GetUp; Jolley 2003; Lecomte 2008; Morrison 2020 (MAPS); Supereden3), the results did not essentially change (SMD ‐0.29, 95% CI ‐0.56 to ‐0.01, 15 RCTs, n = 1006, I² = 76%, substantial heterogeneity) (Analysis 1.78).

1.78. Analysis.

Comparison 1: CBT plus standard care versus control, Outcome 78: Sensitivity analysis ‐ Risk of Bias

1.1.6 Sensitivity analysis ‐ Imputed values

After excluding two trials in which we used imputed values for ICC to calculate the design effect in cluster‐randomised trials, or in which we imputed SDs (GetUp; Jolley 2003), the results remained similar to the main analysis (SMD ‐0.28, 95% CI ‐0.50 to ‐0.05, 18 RCTs, n = 1214, I² = 71%, substantial heterogeneity) (Analysis 1.79).

1.79. Analysis.

Comparison 1: CBT plus standard care versus control, Outcome 79: Sensitivity analysis ‐ Imputed values

1.1.7 Sensitivity analysis ‐ Fixed‐effect

When re‐synthesising data using a fixed‐effect model, results remained similar to the main analysis (SMD ‐0.25, 95% CI ‐0.35 to ‐0.14, 20 RCTs, n = 1508, I² = 68%, substantial heterogeneity) (Analysis 1.80).

1.80. Analysis.

Comparison 1: CBT plus standard care versus control, Outcome 80: Sensitivity analysis ‐ Fixed‐effect

1.1.8 Sensitivity analysis ‐ Separating endpoint and change data

No study provided only change data; therefore, this analysis was not conducted.

1.1.9 Sensitivity analysis ‐ Skewed data

No evidence of skewness (mean/SD ratio lower than 2) was found for any of the studies, so this analysis was not conducted.

1.1.10 Sensitivity analysis without outliers

Given the substantial heterogeneity in the primary outcome (I² = 68%), we performed a post hoc additional sensitivity analysis excluding an outlier study (Palma‐Sevillano 2019). Results did not substantially change (SMD ‐0.19, 95% CI ‐0.30 to ‐0.09, 19 RCTs, n = 1446) and heterogeneity was reduced (I² = 0%) (Analysis 1.81).

1.81. Analysis.

Comparison 1: CBT plus standard care versus control, Outcome 81: Sensitivity analysis excluding outliers

SUBGROUP ANALYSES

1.1.11 Subgroup analysis ‐ Different types of CBT

Eleven studies investigated CBT (SMD ‐0.19, 95% CI ‐0.35 to ‐0.04, 11 RCTs, n = 775, I² = 9%), four studies investigated CBT integrated with other psychosocial interventions (SMD ‐0.87, 95% CI ‐1.73 to ‐0.01, 4 RCTs, n = 494, I² = 93%, considerable heterogeneity), three studies investigated CBT and motivational Interviewing for cannabis use cessation (SMD ‐0.04, 95% CI ‐0.38 to 0.30, 3 RCTs, n = 139, I² = 0%), and two studies used a special type of CBT (SMD ‐0.26, 95% CI ‐0.82 to 0.31, 2 RCTs, n = 100, I² = 32%) (Analysis 1.82). Both CBT and CBT integrated with other psychosocial interventions were associated with greater reduction in symptoms compared to control, while for the subgroups CBT and motivational interviewing for cannabis use cessation and special types of CBT, the confidence interval included the possibility of no difference. The test for subgroup differences did not show a difference between different types of CBT (P = 0.37).

1.1.12 Subgroup analysis ‐ Modality of treatment

17 studies investigated individual CBT (SMD ‐0.29, 95% CI ‐0.53 to ‐0.05, 17 RCTs, n = 1246, I² = 73%, substantial heterogeneity), whereas only three studies (Lecomte 2008; Pos 2019; Supereden3) investigated group CBT (SMD ‐0.18, 95% CI ‐0.42 to 0.07, 3 RCTs, n = 262, I² = 0%) (Analysis 1.83).

CBT offered in an individual setting was associated with a reduction in overall symptoms; however, for CBT applied in a group setting, the confidence interval includes the possibility of no difference. The test for subgroup differences did not show a difference between different types of CBT (P = 0.52).

1.1.13 Subgroup analysis ‐ Method of delivery

Except one study which used a smartphone app (Bucci 2018 (Actissist)) (SMD ‐0.60, 95% CI ‐1.39 to 0.20, 1 RCT, n = 30), all studies took place in a face‐to‐face setting (SMD ‐0.26, 95% CI ‐0.47 to ‐0.06, 19 RCTs, n = 1478, I² = 69%, substantial heterogeneity) (Analysis 1.84). The test for subgroup differences did not show a difference between the two methods of delivery (P = 0.43).

1.1.14 Subgroup analysis ‐ Control group

Fifteen studies compared CBT to TAU, and seven studies compared CBT to another psychological intervention (Analysis 1.85). CBT + TAU was still more effective in reducing symptoms of schizophrena when compared to TAU (SMD ‐0.34, 95% CI ‐0.59 to ‐0.09, 15 RCTs, n = 1211, I² = 73%, substantial heterogeneity), but not when compared to other psychological interventions (SMD ‐0.08, 95% CI ‐0.32 to 0.17, 7 RCTs, n = 297, I² = 6%). However, no differences were found between subgroups (test for subgroups differences P = 0.14) (Analysis 1.85).

1.1.15 Subgroup analysis ‐ Therapist experience

CBT in all studies that provided data on the therapist was conducted by expert therapists. No study included therapists in training to deliver the intervention.

Secondary outcomes

1. Mental state

1.1 General

1.1.1 Clinically important change in general mental state (study defined)

Four studies reported on this outcome. There was a trend in the direction of more participants reaching a clinically important change with CBT + TAU, but the confidence interval included the possibility of no difference (RR 1.50, 95% CI 0.95 to 2.39, 4 RCTs, n = 301, I² = 38%) (Analysis 1.7). No differences were found between separate time points (P = 0.72) (Analysis 1.8).

1.7. Analysis.

Comparison 1: CBT plus standard care versus control, Outcome 7: Mental state ‐ general: number of participants with clinically important change in general mental state (combined time points)

1.8. Analysis.

Comparison 1: CBT plus standard care versus control, Outcome 8: Mental state ‐ general: number of participants with clinically important change in general mental state (separated time points)

1.2 Specific

1.2.1 Clinically important change in positive symptoms (delusions, hallucinations, disordered thinking), as defined by individual studies

Based on two studies, there was a trend in the direction of more participants reaching a clinically important change in positive symptoms with CBT + TAU, but the confidence interval included the possibility of no difference (RR 1.38, 95% CI 0.98 to 1.93, 2 RCTs, n = 99, I² = 0%) (Analysis 1.9). Subgroup analysis did not reveal a difference between separate time points (P = 0.52) (Analysis 1.10).

1.9. Analysis.

Comparison 1: CBT plus standard care versus control, Outcome 9: Mental state ‐ specific: number of participants with clinically important change in positive symptoms (combined time points)

1.10. Analysis.

Comparison 1: CBT plus standard care versus control, Outcome 10: Mental state ‐ specific: number of participants with clinically important change in positive symptoms (separated time points)

1.2.2 Average endpoint or change scores on a published scale or subscale that addresses positive symptoms (e.g. PANSS positive, Scale for the Assessment of Positive Symptoms (SAPS))

Twenty‐two studies provided data on this outcome. Positive symptoms were reduced in the CBT + TAU group compared to control (SMD ‐0.22, 95% CI ‐0.38 to ‐0.06, 22 RCTs, n = 1565, I² = 52%, moderate heterogeneity) (Analysis 1.15). Analysis of separate time points revealed no differences (P = 0.64) (Analysis 1.16).

1.15. Analysis.

Comparison 1: CBT plus standard care versus control, Outcome 15: Mental state ‐ specific: mean endpoint positive symptoms scales (high = poor) (combined time points)

1.16. Analysis.

Comparison 1: CBT plus standard care versus control, Outcome 16: Mental state ‐ specific: mean endpoint positive symptoms scales (high = poor) (separated time points)

1.2.3 Average endpoint or change score on a published scale or subscale that addresses negative symptoms (e.g. PANSS negative, Scale for the Assessment of Negative Symptoms (SANS))

Based on twenty‐two studies, CBT + TAU was associated with greater reduction in negative symptoms than control (SMD ‐0.20, 95% CI ‐0.30 to ‐0.11, 22 RCTs, n = 1651, I² = 0%) (Analysis 1.22). No differences were found between separate time points (P = 0.22) (Analysis 1.23).

1.22. Analysis.

Comparison 1: CBT plus standard care versus control, Outcome 22: Mental state ‐ specific: mean endpoint negative symptoms scale (high = poor) (combined time points)

1.23. Analysis.

Comparison 1: CBT plus standard care versus control, Outcome 23: Mental state ‐ specific: mean endpoint negative symptoms scale (high = poor) (separated time points)

1.2.4 Average endpoint or change score on a published scale or subscale that addresses depressive symptoms (e.g. Montgomery‐Asberg Depression Rating Scale (MADRS), Beck Depression Inventory (BDI))

Eighteen studies provided data for this outcome. CBT + TAU was associated with greater reduction in depressive symptoms than control (SMD ‐0.13, 95% CI ‐0.24 to ‐0.01, 18 RCTs, n = 1182, I² = 0%) (Analysis 1.32). Testing for subgroup differences did not reveal a difference between time points (P = 0.16) (Analysis 1.33).

1.32. Analysis.

Comparison 1: CBT plus standard care versus control, Outcome 32: Mental state ‐ specific: mean endpoint depressive symptoms (high = poor) (combined time points)

1.33. Analysis.

Comparison 1: CBT plus standard care versus control, Outcome 33: Mental state ‐ specific: mean endpoint depressive symptoms (high = poor) (separated time points)

2. Global state

2.1 Relapse

Based on seven studies, there was no evidence of a difference in relapse rates between CBT + TAU and the control group (RR 0.82, 95% CI 0.57 to 1.18, 7 RCTs, n = 693, I² = 48%, low certainty of the evidence) (Analysis 1.34). No differences were found between separate time points in subgroup analysis (P = 0.47) (Analysis 1.35).

1.34. Analysis.

Comparison 1: CBT plus standard care versus control, Outcome 34: Global state ‐ number of participants with relapse/exacerbations of psychosis (combined time points)

1.35. Analysis.

Comparison 1: CBT plus standard care versus control, Outcome 35: Global state ‐ number of participants with relapse/exacerbations of psychosis (separated time points)

2.2 Remission

Two studies reported on remission, with no difference found between CBT + TAU and control (RR 1.16, 95% CI 0.70 to 1.91, 2 RCTs, n = 73, I² = 0%) (Analysis 1.36). No differences were found across separate time points (P = 0.74) (Analysis 1.37).

1.36. Analysis.

Comparison 1: CBT plus standard care versus control, Outcome 36: Global state ‐ number of participants with remission (combined time points)

1.37. Analysis.

Comparison 1: CBT plus standard care versus control, Outcome 37: Global state ‐ number of participants with remission (separated time points)

2.3 Clinically important change in global state (e.g. global impression of much improved, or more than 50% improvement on a rating scale, such as the Clinical Global Impression scale (CGI))

Based on one study, results showed no difference in this outcome between intervention and control groups at six months (RR 1.58, 95% CI 0.82 to 3.01, 1 RCT, n = 99) or twelve months (RR 1.53, 95% CI 0.83 to 2.83, 1 RCT, n = 99) (Analysis 1.38).

1.38. Analysis.

Comparison 1: CBT plus standard care versus control, Outcome 38: Global state ‐ number of participants with clinically important change in global state

2.4 Average endpoint or change score on a global state scale

Four studies reported on this outcome. CBT + TAU was associated with a greater reduction in global state scales than control (SMD ‐0.34, 95% CI ‐0.67 to ‐0.01, 4 RCTs, n = 329, I² = 47%, moderate heterogeneity) (Analysis 1.42). No differences were found for separate time points (P = 0.22) (Analysis 1.43).

1.42. Analysis.

Comparison 1: CBT plus standard care versus control, Outcome 42: Global state ‐ mean endpoint/change global state scales (high = poor) (combined time points)

1.43. Analysis.

Comparison 1: CBT plus standard care versus control, Outcome 43: Global state ‐ mean endpoint/change global state scales (high = poor) (separated time points)

3. Service Use

3.1 Admission to hospital

Based on ten studies, there was no difference in hospital admission rates between CBT + TAU and control (RR 1.00, 95% CI 0.80 to 1.25, 10 RCTs, n = 1121, I² = 0%) (Analysis 1.44). No differences were found for subgroup analysis of separate time points (P = 0.81) (Analysis 1.45).

1.44. Analysis.

Comparison 1: CBT plus standard care versus control, Outcome 44: Service use ‐ admission to hospital (combined time points)

1.45. Analysis.

Comparison 1: CBT plus standard care versus control, Outcome 45: Service use ‐ admission to hospital (separated time points)

3.2 Number of days in hospital

Based on seven studies, there was no evidence of a difference in number of days in hospital between groups (MD ‐8.18, 95% CI ‐24.74 to 8.38, 7 RCTs, n = 571, I² = 94%) (Analysis 1.46). No differences were found for separate time points (P = 0.32) (Analysis 1.47).

1.46. Analysis.

Comparison 1: CBT plus standard care versus control, Outcome 46: Service use ‐ number of days in hospital (combined time points)

1.47. Analysis.

Comparison 1: CBT plus standard care versus control, Outcome 47: Service use ‐ number of days in hospital (separated time points)

4. Leaving the study early

4.1 For any reason

Twenty‐five studies provided data on leaving the study early for any reason, with no difference between groups (RR 0.87, 95% CI 0.72 to 1.05, 25 RCTs, n = 2242, I² = 12%, moderate certainty of the evidence) (Analysis 1.48). No difference was found between separate time points (P = 0.87) (Analysis 1.49).

1.48. Analysis.

Comparison 1: CBT plus standard care versus control, Outcome 48: Leaving the study early ‐ for any reason ‐ overall acceptability (combined time points)

1.49. Analysis.

Comparison 1: CBT plus standard care versus control, Outcome 49: Leaving the study early ‐ for any reason ‐ overall acceptability (separated time points)

5. Functioning

Eighteen studies reported on functioning. CBT + TAU was associated with better functioning than control (SMD ‐0.23, 95% CI ‐0.42 to ‐0.05, 18 RCTs, n = 1241, I² = 53%, moderate heterogeneity, moderate certainty of the evidence) (Analysis 1.56). There was no evidence of a difference between the separate time points (test for subgroup differences: P = 0.69) (Analysis 1.57).

1.56. Analysis.

Comparison 1: CBT plus standard care versus control, Outcome 56: Functioning ‐ mean endpoint functioning scale (high = good) (combined time points)

1.57. Analysis.

Comparison 1: CBT plus standard care versus control, Outcome 57: Functioning ‐ mean endpoint functioning scale (high = good) (separated time points)

6. Quality of life

Nine studies reported on quality of life. There was no evidence of a difference between CBT + TAU and control (SMD ‐0.13, 95% CI ‐0.33 to 0.07, 9 RCTs, n = 386, I² = 0%) (Analysis 1.64).

1.64. Analysis.

Comparison 1: CBT plus standard care versus control, Outcome 64: Quality of life ‐ mean endpoint quality of life scale (high = good) (combined time points)

No differences were found between the separate time‐points (P = 0.82) (Analysis 1.65).

1.65. Analysis.

Comparison 1: CBT plus standard care versus control, Outcome 65: Quality of life ‐ mean endpoint quality of life scale (high = good) (separated time points)

7. Behaviour

7.1 Occurrence of violent incidents (to self, others, or property)

Two studies provided data on this outcome. There was no evidence of a difference between CBT + TAU and control at < 12 months (RR 0.96, 95% CI 0.15 to 6.20, 2 RCTs, n = 92, I² = 0%) (Analysis 1.66).

1.66. Analysis.

Comparison 1: CBT plus standard care versus control, Outcome 66: Behaviour ‐ occurence of violent incidents

7.2 Self‐injury

Two studies provided data on participants with self‐injury. There was no evidence of a difference between CBT + TAU and control at < 12 months (RR 1.89, 95% CI 0.95 to 3.80, 2 RCTs, n = 92, I² = 0%) (Analysis 1.67).

1.67. Analysis.

Comparison 1: CBT plus standard care versus control, Outcome 67: Behaviour ‐ participants with self‐injury

7.3 Suicide attempts

Two studies provided data on suicide attempts. There was no evidence of a difference in the number of events between CBT + TAU and control at < 12 months (RR 0.97, 95% CI 0.10 to 9.41, 2 RCTs, n = 68, I² = 5%). However, at > 1 year, one study provided data on this outcome (RR 0.31, 95% CI 0.01 to 6.94, 1 RCT, n = 25); the confidence interval includes the possibility of no difference between CBT + TAU and control. No difference was found between separate time points (test for subgroup differences P = 0.56) (Analysis 1.68).

1.68. Analysis.

Comparison 1: CBT plus standard care versus control, Outcome 68: Behaviour ‐ participants with suicide attempts

8. Adverse events

8.1 At least one adverse event

One study provided data on this outcome. There was no evidence of a difference in terms of participants with at least one adverse event between CBT + TAU and control (RR 1.29, 95% CI 0.85 to 1.97, 1 RCT, n = 43, very low certainty of the evidence) (Analysis 1.69).

1.69. Analysis.

Comparison 1: CBT plus standard care versus control, Outcome 69: Adverse events ‐ participants with at least one adverse event

8.2 Specific adverse events: Non‐compliance

Two studies provided data on this outcome. No differences were found at < 6 months (RR 1.06, 95% CI 0.47 to 2.37, 2 RCTs, n = 190, I² = 0%), at < 12 months (RR 0.33, 95% CI 0.04 to 3.07, 1 RCT, n = 80) and at more than one year (RR 0.33, 95% CI 0.04 to 3.07, 1 RCT, n = 80). No difference was found between separate time points (test for subgroup differences P = 0.44) (Analysis 1.70).

1.70. Analysis.

Comparison 1: CBT plus standard care versus control, Outcome 70: Adverse events ‐ specific: Non Compliance

8.3 Specific adverse events: Deterioration in symptoms

Four studies provided data on this outcome. There was no evidence of a difference at < 6 months (RR 1.08, 95% CI 0.17 to 7.03, 2 RCTs, n = 92, I² = 0%), at < 12 months (RR 0.58, 95% CI 0.12 to 2.79, 3 RCTs, n = 117, I² = 0%) and at more than one year (RR 0.53, 95% CI 0.27 to 1.02, 2 RCTs, n = 87). The test for subgroup difference did not reveal a difference between time points (P = 0.78) (Analysis 1.71).

1.71. Analysis.

Comparison 1: CBT plus standard care versus control, Outcome 71: Adverse events ‐ specific: Deterioration of symptoms

9. Mortality

9.1 Overall mortality

Four studies provided data on this outcome. There was no evidence of a difference between CBT + TAU and control (RR 1.22, 95% CI 0.41 to 3.66, 4 RCTs, n = 900, I² = 6%) (Analysis 1.72). Also, no differences were found between the separate time points (P = 0.79) (Analysis 1.73).

1.72. Analysis.

Comparison 1: CBT plus standard care versus control, Outcome 72: Mortality ‐ overall mortality (combined time points)

1.73. Analysis.

Comparison 1: CBT plus standard care versus control, Outcome 73: Mortality ‐ overall mortality (separated time points)

9.2 Mortality due to natural causes

Two studies provided data on this outcome. There was no evidence of a difference between CBT + TAU and control (RR 2.00, 95% CI 0.22 to 17.99, 2 RCTs, n = 635, I² = 30%) (Analysis 1.74). The test for subgroup differences did not show a difference between the time points (P = 0.09).

1.74. Analysis.

Comparison 1: CBT plus standard care versus control, Outcome 74: Mortality ‐ due to natural causes

9.3 Mortality due to suicide

Based on the data of two studies, results did not show a difference between CBT + TAU and control in terms of mortality due to suicide (RR 1.08, 95% CI 0.05 to 24.00, 2 RCTs, n = 388, I² = 50%, moderate heterogeneity) (Analysis 1.75). The test for subgroup differences did not show a difference between the time points (P = 0.16).

1.75. Analysis.

Comparison 1: CBT plus standard care versus control, Outcome 75: Mortality ‐ due to suicide

10. Outcomes without available data

The following outcomes were pre‐planned in the protocol; however, it was not possible to derive usable data from the included studies:

‐ Global state ‐ recovery

‐ Cognitive functioning

‐ Satisfaction with care

Discussion

Summary of main results

We identified 28 studies eligible for inclusion in the review, of which 26 (2407 participants) provided data for the meta‐analysis.

For our primary outcome, general mental state, our results show that CBT + TAU was associated with a reduction in overall symptoms of schizophrenia in comparison with control, with low certainty of the evidence (Analysis 1.5).

When excluding studies not using a blind outcome assessor (Analysis 1.76), studies that used completer analyses (Analysis 1.77), studies with high overall risk of bias (Analysis 1.78), studies with imputed values (Analysis 1.79), analysing the data with a fixed‐effect model (Analysis 1.80) and excluding one outlier study (Analysis 1.81), results did not substantially change from the main analysis.

When looking at different types of CBT, the confidence intervals excluded the possibility of no difference in comparison with TAU for all, except for 'CBT and motivational interviewing for cannabis use cessation' and 'special types of CBT'. However, it must be noted that only a few studies contributed to these categories (Analysis 1.82). Moreover, testing for subgroup differences did not reveal a difference within the investigated subgroups ‐ types of CBT (P = 0.37).

CBT offered in an individual setting was associated with a greater reduction in overall symptoms of schizophrenia than control, while for group CBT the confidence interval did not exclude the possibility of no difference to control. However, also here it must be noted that only three studies investigated group CBT (Analysis 1.83), and testing for subgroup differences did not reveal a difference between the two treatment modalities (P = 0.52).

We found that CBT was efficacious when delivered in a face‐to‐face setting, while when using a smartphone app, the confidence interval did not exclude the possibility of no difference, with only one study investigating this (Analysis 1.84). A test for subgroup difference did not show a difference between these two methods of delivery (P = 0.43).

CBT + TAU was associated with a greater symptom reduction when compared to TAU, but not when compared to other psychosocial interventions (Analysis 1.85). However, testing for subgroup differences did not reveal a difference due to the control group (P = 0.14).

CBT + TAU was also associated with a greater reduction in positive (Analysis 1.15), negative (Analysis 1.22) and depressive (Analysis 1.32) symptoms of schizophrenia, and a greater improvement in global state (Analysis 1.42) and in functioning (moderate certainty of the evidence) than control (Analysis 1.56).

We did not find a difference between CBT + TAU and control in terms of number of participants with relapse (low certainty of the evidence) (Analysis 1.34), leaving the study early for any reason (moderate certainty of the evidence) (Analysis 1.48), adverse events (very low certainty of the evidence) (Analysis 1.69) and the other investigated outcomes.

Overall completeness and applicability of evidence

In spite of the multiple systematic reviews and meta analyses investigating CBT for schizophrenia, we could not identify a recent systematic review and meta‐analysis on CBT specifically offered to the special population of participants with a recent‐onset or first‐episode schizophrenia.

The present review synthesises the available evidence on CBT + TAU for people in the early stage of the disorder, with a recent‐onset or first‐episode schizophrenia. The presented results are therefore applicable to people in the early stage of the disorder, meaning within three to five years from first treatment/first diagnosis/first contact with services, with no age limits.

Caution is advised in the interpretation of results due to the different definitions for this particular population being used in different studies, locations and countries. For example, Early Intervention Psychosis (EIP) services providing specialist assertive multidisciplinary input, including pharmacological and psychosocial interventions for first‐episode psychosis, were introduced in England in 1999 (Department of Health and Social Care 1999). In 2016, it was recommended that there be no upper age limit and that there be rapid assessment. In contrast, national early psychosis services in Australia are run through primary care youth mental health services, generally catering for a specific age range of 15 to 25 years. Generally, EIP services aim to treat the disorder early to reduce DUP (duration of untreated psychosis) and improve outcomes by providing a comprehensive range of services including CBT, family interventions and employment support. Similar services to England are provided in Wales, Scotland and Northern Ireland. This is particularly relevant, since most of the included studies were conducted in the UK (N = 9) (Included studies).

We did our best to operationalise the definition of the target population as described above and in Types of participants, and we contacted the study authors whenever clarification was needed.

The results are not applicable to the elderly population which was specifically excluded from this review (even if we found no study on recent‐onset psychosis focusing on this population).

Concerning the CBT treatment, it must be noted that, on average, the participants in the CBT + TAU arms of the included studies received about 15.6 sessions. This is in line with the recommendations of the National Institute for Health and Care Excellence (NICE), according to which CBT should be delivered for at least 16 sessions (NICE 2014).

It must be noted that very scarce information was available in the included studies about adverse events potentially connected with CBT, even if this is also an important outcome to be investigated for psychological interventions (Linden 2014).

Lastly, the last update of the search was conducted in March 2022. However, we are not aware of potentially eligible studies published after that date.

Quality of the evidence

Using the GRADE approach, we assessed the certainty of the evidence as ranging from very low for the outcome, adverse events, to moderate for the outcomes, leaving the study early and functioning (see Table 1).

We evaluated the certainty of the evidence as low for the outcome, mental state ‐ general: mean endpoint overall symptoms scale, as some of the studies were at overall high risk of bias and there was moderate heterogeneity.

We evaluated the certainty of the evidence as low for the outcome, global state ‐ relapse, because of imprecision; the confidence interval included both appreciable benefit with CBT + TAU and no difference between CBT + TAU and control, and the number of participants providing data for this outcome was low.

We evaluated the certainty of the evidence as moderate for the outcome, leaving the study early, due to imprecision; the confidence interval included both appreciable benefit for CBT + TAU and no difference between CBT + TAU and control.

We evaluated the certainty of the evidence as moderate for the outcome functioning, because some of the studies contributing data were at high overall risk of bias.

We evaluated the certainty of the evidence as very low for the outcome adverse events, because the confidence interval included both appreciable harm with CBT + TAU and no difference between CBT + TAU and control; moreover, the number of participants providing data for this outcome was extremely low.

We did not detect potential publication bias for any of the outcomes investigated with funnel plots. Visual inspection of the funnel plots suggested no asymmetry, and this was confirmed by Egger's test (mental state: P = 0.6320 (Figure 2), leaving the study early P = 0.4223 (Figure 3), functioning P = 0.8043 (Figure 4)).

2.

Egger's test for funnel plot asymmetry did not reveal asymmetry in the funnel plot (P = 0.6320).

3.

Egger's test for funnel plot asymmetry did not reveal asymmetry in the funnel plot (P = 0.4223).

4.

Egger's test for funnel plot asymmetry did not reveal asymmetry in the funnel plot (P = 0.8043).

Potential biases in the review process

We have documented and justified modifications to our published protocol in the Differences between protocol and review section.

The current review has some limitations.

Firstly, the definition of the population of participants with a first‐episode or recent‐onset psychosis is very heterogeneously described in the included studies. In recognition of this lack of agreement in the definitions used, there have been attempts to operationalise the terms, but these are still inconsistently used (Breitborde 2009). Different decisions could be taken when investigating this population; for example, concerning the number of years after the first manifestation of the illness, including or excluding second episodes, and so on. We reported our criteria and decisions transparently, so that our approach can be replicated.

Furthermore, it was not possible to conduct some of the pre‐planned sensitivity and subgroup analyses, because no studies presented the necessary characteristics (exclusion of studies with more than 50% dropout, exclusion of studies providing only change data, exclusion of studies with skewed data, subgroup analysis of different levels of therapists' experience), therefore, it is not possible to say anything on these potential effect moderators. On the other hand, we conducted a post hoc sensitivity analysis excluding a study that had strong outlier results in the direction of a benefit with CBT + TAU (Palma‐Sevillano 2019). Results remained consistent with the ones of the main analysis.

It must be noted, that the number of sessions provided when CBT was conducted in the different studies varied, and this might have contributed to some degree of heterogeneity.

Moreover, the reader should consider that some standard deviations of the continuous outcomes were imputed (and consequently excluded in the relevant sensitivity analysis). It should also be noted that, for binary outcomes, the denominator we used in the analyses was the number of randomised participants. This would imply that all missing participants were considered not to experience the outcomes.

Agreements and disagreements with other studies or reviews

This review is of special interest, as a timely and effective treatment of patients with their first‐episode or recent‐onset psychosis is of particular importance in order to prevent a chronification of the disease and furthermore reduce the enormous socioeconomic burden (Evensen 2016; Marwaha 2004).

A meta‐analysis published in 2010 investigating early intervention services, CBT and family interventions for people with early psychosis included only four studies on CBT (Bird 2010). Similar to our results, they found a higher reduction in positive and negative symptoms of schizophrenia with CBT, and no difference with control in terms of relapse or hospital admissions. In this review, early psychosis was defined as "a clinical diagnosis of psychosis within 5 years of the first psychotic episode or presentation to mental health services", so overlapping with the definition used in the present review.

In 2021, Frawley and colleagues published a meta‐analysis on psychosocial interventions for social and occupational functioning in early stage psychosis, but they included in this definition also the high‐risk stage (Frawley 2021). Six studies investigated CBT, of which five assessed participants in the high‐risk stage of the disease and only one assessed those in the first‐episode stage, which was the focus of the present review (Jackson 2008 (ACE)).

As a result, it can be said that no reliable summary of evidence on CBT for this important population has been identified to date.

Related to the general population of patients with schizophrenia, it has already been shown in previous meta‐analyses and network meta‐analyses that CBT + TAU can be an effective therapy in comparison to TAU for treating positive symptoms (Bighelli 2018), for reducing overall symptoms of schizophrenia (McGlanaghy 2021), for preventing relapses (Bighelli 2021) and for improving functioning (Bighelli 2023). A recent Cochrane Review comparing cognitive behavioural therapy to treatment‐as‐usual for people with schizophrenia only identified nine studies on people with a first episode, but did not analyse them separately (Jones 2018a; Jones 2018b).

A Cochrane Review is currently in development, investigating the role of CBT when provided without concomitant medication with antipsychotics (Bighelli 2022a). The included studies are mainly recruiting people in the early phase of the disorder, so that the results could be relevant also for the population addressed by the present review.

Authors' conclusions

Implications for practice.

The evidence identified by this review suggests that people with a first‐episode or recent‐onset psychosis may benefit from CBT additionally to standard care for multiple outcomes (overall, positive, negative and depressive symptoms of schizophrenia, global state and functioning).

CBT is already recommended by clinical guidelines for participants with schizophrenia, including the first‐episode and recent‐onset phase, in both the adult and young population (NICE 2013; NICE 2014). The present results confirm the benefits that can be achieved by the population of people with a first‐episode or recent‐onset psychosis with CBT. Policymakers may consider CBT for this population.

The present data could help patients and clinicians in a shared decision‐making process about the benefits for individuals of starting therapy with CBT. The potential adverse effects connected with this intervention, unfortunately not adequately covered in the studies found for this review, also need to be taken in consideration.

Implications for research.

At best, common terms and clinically based criteria should be developed and consistently applied in research on people in the early stage of the disorder.

In the meantime, given the heterogeneous descriptions used for this population, future studies should describe in detail the included sample in terms of first‐episode and recent‐onset psychosis, providing not only clear inclusion criteria, but also percentages of participants with a first episode. Furthermore, services for the early stages of psychosis, on which recruitment for clinical trials often relies, are organised differently in various countries. Therefore, a detailed description of the recruited sample is of crucial importance in the studies in order to allow meaningful comparisons.

Future studies should also report adverse events potentially connected to the treatment of CBT.

Finally, a network meta‐analysis could investigate the active comparisons of CBT with other psychological interventions used in this population, such as family interventions.

History

Protocol first published: Issue 5, 2022

Risk of bias

Risk of bias for analysis 1.5 Mental state ‐ general: mean endpoint overall symptom scales (high = poor) (combined time points).

Study

Bias

Randomisation process

Deviations from intended interventions

Missing outcome data

Measurement of the outcome

Selection of the reported results

Overall

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Barrowclough 2014

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Some concerns

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status, but this is not likely.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Some concerns

The outcome is not mentioned in the prespecified plan of this study.

Some concerns

Calculated based on RoB2 algorithm.

Bucci 2018 (Actissist)

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Some concerns

Participants and personnel were aware of the intervention, no apparent deviation because of the trial context. Not clear if an appropriate analysis was used, but probably this had no impact on the analysis results.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

Some concerns

Calculated based on RoB2 algorithm.

Edwards 2006

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Some concerns

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status, but this is not likely.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Some concerns

No prespecified plan available.

Some concerns

Calculated based on RoB2 algorithm.

GetUp

High risk of bias

Centres were randomised, but some centres were forced into the intervention arm. No baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Low risk of bias

Data on this outcome are given for all randomised participants.

Some concerns

Outcome measured with an appropriate method, no blind outcome assessors, the assessment could have been influenced, but this is unlikely.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

High risk of bias

Calculated based on RoB2 algorithm.

Gleeson 2009 (EPISODE II)

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Some concerns

The outcome is not mentioned in the prespecified plan of this study.

Some concerns

Calculated based on RoB2 algorithm.

Gonzalez‐Ortega 2021

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Some concerns

Participants and personnel were aware of the intervention, no apparent deviation because of the trial context. No appropriate analysis was used, but probably this had no impact on the analysis results.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

Some concerns

Calculated based on RoB2 algorithm.

Haddock 1999

Low risk of bias

Participants were randomised, there is no mention to how the random sequence was generated but allocation sequence was concealed. No information about baseline differences to suggest a problem with the randomisation process.

Some concerns

Participants and personnel were aware of the intervention, no apparent deviation because of the trial context. No appropriate analysis was used, but probably this had no impact on the analysis results.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Some concerns

No prespecified plan available.

Some concerns

Calculated based on RoB2 algorithm.

Hjorthoj 2013 (CapOpus)

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Some concerns

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status, but this is not likely.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

Some concerns

Calculated based on RoB2 algorithm.

Jackson 2008 (ACE)

Low risk of bias

Participants were randomised, there is no mention to how the random sequence was generated but allocation sequence was concealed. No information about baseline differences to suggest a problem with the randomisation process.

Some concerns

No information whether participants were aware of the intervention, personnel were aware of the intervention, no apparent deviations because of the trial context. No appropriate analysis was used, but probably this had no impact on the analysis results.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Some concerns

No prespecified plan available.

Some concerns

Calculated based on RoB2 algorithm.

Jolley 2003

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Some concerns

Participants and personnel were aware of the intervention, no apparent deviation because of the trial context. Not clear if an appropriate analysis was used, but probably this had no impact on the analysis results.

High risk of bias

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Some concerns

The outcome is not mentioned in the prespecified plan of this study.

High risk of bias

Calculated based on RoB2 algorithm.

Lecomte 2008

Some concerns

Participants were randomised, there is no mention to how the random sequence was generated and no information about allocation sequence concealment. No baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

High risk of bias

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

High risk of bias

Calculated based on RoB2 algorithm.

Lecomte 2008

Some concerns

Participants were randomised, there is no mention to how the random sequence was generated and no information about allocation sequence concealment. No baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

High risk of bias

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

High risk of bias

Calculated based on RoB2 algorithm.

Lewis 2002 (SOCRATES)

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Some concerns

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status, but this is not likely.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

Some concerns

Calculated based on RoB2 algorithm.

Lewis 2002 (SOCRATES)

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Some concerns

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status, but this is not likely.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

Some concerns

Calculated based on RoB2 algorithm.

Liu 2019

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

Low risk of bias

Calculated based on RoB2 algorithm.

Morrison 2018 (COMPARE)

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

Low risk of bias

Calculated based on RoB2 algorithm.

Morrison 2020 (MAPS)

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

High risk of bias

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

High risk of bias

Calculated based on RoB2 algorithm.

Müller 2020

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Some concerns

No prespecified plan available.

Some concerns

Calculated based on RoB2 algorithm.

Palma‐Sevillano 2019

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Some concerns

Participants and personnel were aware of the intervention, no apparent deviation because of the trial context. Not clear if an appropriate analysis was used, but probably this had no impact on the analysis results.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Some concerns

No prespecified plan available.

Some concerns

Calculated based on RoB2 algorithm.

Pos 2019

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Some concerns

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status, but this is not likely.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

Some concerns

Calculated based on RoB2 algorithm.

Supereden3

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

High risk of bias

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

High risk of bias

Calculated based on RoB2 algorithm.

Uzenoff 2008

Some concerns

Participants were randomised, there is no mention to how the random sequence was generated and no information about allocation sequence concealment. No baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Some concerns

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status, but this is not likely.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Some concerns

No prespecified plan available.

Some concerns

Calculated based on RoB2 algorithm.

Risk of bias for analysis 1.34 Global state ‐ number of participants with relapse/exacerbations of psychosis (combined time points).

Study

Bias

Randomisation process

Deviations from intended interventions

Missing outcome data

Measurement of the outcome

Selection of the reported results

Overall

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Barrowclough 2014

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured with an appropriate method by blind assessors.

Some concerns

The outcome is not mentioned in the prespecified plan of this study.

Some concerns

Calculated based on RoB2 algorithm.

Gleeson 2009 (EPISODE II)

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Some concerns

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status, but this is not likely.

Low risk of bias

Outcome measured with an appropriate method by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

Some concerns

Calculated based on RoB2 algorithm.

Haddock 1999

Low risk of bias

Participants were randomised, there is no mention to how the random sequence was generated but allocation sequence was concealed. No information about baseline differences to suggest a problem with the randomisation process.

Some concerns

Participants and personnel were aware of the intervention, no apparent deviation because of the trial context. No appropriate analysis was used, but probably this had no impact on the analysis results.

Low risk of bias

Data on this outcome are given for all randomised participants.

Some concerns

Outcome measured with an appropriate method, not clear whether the assessors were blind; the assessment could have been influenced but this is not likely.

Some concerns

No prespecified plan available.

Some concerns

Calculated based on RoB2 algorithm.

Lewis 2002 (SOCRATES)

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured with an appropriate method by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

Low risk of bias

Calculated based on RoB2 algorithm.

Lewis 2002 (SOCRATES)

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured with an appropriate method by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

Low risk of bias

Calculated based on RoB2 algorithm.

Liu 2019

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured with an appropriate method by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

Low risk of bias

Calculated based on RoB2 algorithm.

Morrison 2018 (COMPARE)

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured with an appropriate method by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

Low risk of bias

Calculated based on RoB2 algorithm.

Morrison 2020 (MAPS)

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

High risk of bias

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status.

Low risk of bias

Outcome measured with an appropriate method by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

High risk of bias

Calculated based on RoB2 algorithm.

Risk of bias for analysis 1.48 Leaving the study early ‐ for any reason ‐ overall acceptability (combined time points).

Study

Bias

Randomisation process

Deviations from intended interventions

Missing outcome data

Measurement of the outcome

Selection of the reported results

Overall

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Barrowclough 2014

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured with an appropriate method, not clear whether the assessors were blind, but the assessment could not have been influenced (hard outcome).

Low risk of bias

The outcome is not mentioned in the prespecified plan, but according to CONSORT is expected to be reported.

Low risk of bias

Calculated based on RoB2 algorithm.

Bucci 2018 (Actissist)

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Some concerns

Participants and personnel were aware of the intervention, no apparent deviation because of the trial context. Not clear if an appropriate analysis was used, but probably this had no impact on the analysis results.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured with an appropriate method by blind assessors.

Low risk of bias

The outcome is not mentioned in the prespecified plan, but according to CONSORT is expected to be reported.

Some concerns

Calculated based on RoB2 algorithm.

Edwards 2006

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured with an appropriate method, not clear whether the assessors were blind, but the assessment could not have been influenced (hard outcome).

Low risk of bias

The outcome is not mentioned in the prespecified plan, but according to CONSORT is expected to be reported.

Low risk of bias

Calculated based on RoB2 algorithm.

GetUp

High risk of bias

Centres were randomised, but some centres were forced into the intervention arm. No baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured with an appropriate method, no blind outcome assessors, but the assessment could not have been influenced (hard outcome).

Low risk of bias

The outcome is not mentioned in the prespecified plan, but according to CONSORT is expected to be reported.

High risk of bias

Calculated based on RoB2 algorithm.

Gleeson 2009 (EPISODE II)

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured with an appropriate method by blind assessors.

Low risk of bias

The outcome is not mentioned in the prespecified plan, but according to CONSORT is expected to be reported.

Low risk of bias

Calculated based on RoB2 algorithm.

Gonzalez‐Ortega 2021

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Some concerns

Participants and personnel were aware of the intervention, no apparent deviation because of the trial context. No appropriate analysis was used, but probably this had no impact on the analysis results.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured with an appropriate method by blind assessors.

Low risk of bias

The outcome is not mentioned in the prespecified plan, but according to CONSORT is expected to be reported.

Some concerns

Calculated based on RoB2 algorithm.

Haddock 1999

Low risk of bias

Participants were randomised, there is no mention to how the random sequence was generated but allocation sequence was concealed. No information about baseline differences to suggest a problem with the randomisation process.

Some concerns

Participants and personnel were aware of the intervention, no apparent deviation because of the trial context. No appropriate analysis was used, but probably this had no impact on the analysis results.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured with an appropriate method, not clear whether the assessors were blind, but the assessment could not have been influenced (hard outcome).

Low risk of bias

The outcome is not mentioned in the prespecified plan, but according to CONSORT is expected to be reported.

Some concerns

Calculated based on RoB2 algorithm.

Hjorthoj 2013 (CapOpus)

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured with an appropriate method by blind assessors.

Low risk of bias

The outcome is not mentioned in the prespecified plan, but according to CONSORT is expected to be reported.

Low risk of bias

Calculated based on RoB2 algorithm.

Jackson 2008 (ACE)

Low risk of bias

Participants were randomised, there is no mention to how the random sequence was generated but allocation sequence was concealed. No information about baseline differences to suggest a problem with the randomisation process.

Some concerns

No information whether participants were aware of the intervention, personnel were aware of the intervention, no apparent deviations because of the trial context. Not clear if an appropriate analysis was used, but probably this had no impact on the analysis results.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured with an appropriate method by blind assessors.

Low risk of bias

The outcome is not mentioned in the prespecified plan, but according to CONSORT is expected to be reported.

Some concerns

Calculated based on RoB2 algorithm.

Jackson 2009

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured with an appropriate method, not clear whether the assessors were blind, but the assessment could not have been influenced (hard outcome).

Low risk of bias

The outcome is not mentioned in the prespecified plan, but according to CONSORT is expected to be reported.

Low risk of bias

Calculated based on RoB2 algorithm.

Jolley 2003

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Some concerns

Participants and personnel were aware of the intervention, no apparent deviation because of the trial context. Not clear if an appropriate analysis was used, but probably this had no impact on the analysis results.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured with an appropriate method by blind assessors.

Low risk of bias

The outcome is not mentioned in the prespecified plan, but according to CONSORT is expected to be reported.

Some concerns

Calculated based on RoB2 algorithm.

Lecomte 2008

Some concerns

Participants were randomised, there is no mention to how the random sequence was generated and no information about allocation sequence concealment. No baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured with an appropriate method by blind assessors.

Low risk of bias

The outcome is not mentioned in the prespecified plan, but according to CONSORT is expected to be reported.

Some concerns

Calculated based on RoB2 algorithm.

Lecomte 2008

Some concerns

Participants were randomised, there is no mention to how the random sequence was generated and no information about allocation sequence concealment. No baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured with an appropriate method by blind assessors.

Low risk of bias

The outcome is not mentioned in the prespecified plan, but according to CONSORT is expected to be reported.

Some concerns

Calculated based on RoB2 algorithm.

Lepage 2022

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured with an appropriate method, not clear whether the assessors were blind, but the assessment could not have been influenced (hard outcome).

Low risk of bias

The outcome is not mentioned in the prespecified plan, but according to CONSORT is expected to be reported.

Low risk of bias

Calculated based on RoB2 algorithm.

Lewis 2002 (SOCRATES)

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured with an appropriate method by blind assessors.

Low risk of bias

The outcome is not mentioned in the prespecified plan, but according to CONSORT is expected to be reported.

Low risk of bias

Calculated based on RoB2 algorithm.

Lewis 2002 (SOCRATES)

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured with an appropriate method by blind assessors.

Low risk of bias

The outcome is not mentioned in the prespecified plan, but according to CONSORT is expected to be reported.

Low risk of bias

Calculated based on RoB2 algorithm.

Liu 2019

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured with an appropriate method by blind assessors.

Low risk of bias

The outcome is not mentioned in the prespecified plan, but according to CONSORT is expected to be reported.

Low risk of bias

Calculated based on RoB2 algorithm.

Madigan 2013

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured with an appropriate method by blind assessors.

Low risk of bias

The outcome is not mentioned in the prespecified plan, but according to CONSORT is expected to be reported.

Low risk of bias

Calculated based on RoB2 algorithm.

Morrison 2018 (COMPARE)

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured with an appropriate method by blind assessors.

Low risk of bias

The outcome is not mentioned in the prespecified plan, but according to CONSORT is expected to be reported.

Low risk of bias

Calculated based on RoB2 algorithm.

Morrison 2020 (MAPS)

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured with an appropriate method, not clear whether the assessors were blind, but the assessment could not have been influenced (hard outcome).

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

Low risk of bias

Calculated based on RoB2 algorithm.

Müller 2020

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured with an appropriate method by blind assessors.

Low risk of bias

The outcome is not mentioned in the prespecified plan, but according to CONSORT is expected to be reported.

Low risk of bias

Calculated based on RoB2 algorithm.

Palma‐Sevillano 2019

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Some concerns

Participants and personnel were aware of the intervention, no apparent deviation because of the trial context. Not clear if an appropriate analysis was used, but probably this had no impact on the analysis results.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured with an appropriate method by blind assessors.

Low risk of bias

The outcome is not mentioned in the prespecified plan, but according to CONSORT is expected to be reported.

Some concerns

Calculated based on RoB2 algorithm.

Pos 2019

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured with an appropriate method by blind assessors.

Low risk of bias

The outcome is not mentioned in the prespecified plan, but according to CONSORT is expected to be reported.

Low risk of bias

Calculated based on RoB2 algorithm.

Puig‐Navarro 2020

Some concerns

Participants were randomised, there is no mention to how the random sequence was generated and no information about allocation sequence concealment. No information about baseline differences to suggest a problem with the randomisation process.

Low risk of bias

No information whether participants were aware of the intervention, personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured with an appropriate method, not clear whether the assessors were blind, but the assessment could not have been influenced (hard outcome).

Low risk of bias

The outcome is not mentioned in the prespecified plan, but according to CONSORT is expected to be reported.

Some concerns

Calculated based on RoB2 algorithm.

Supereden3

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured with an appropriate method, not clear whether the assessors were blind, but the assessment could not have been influenced (hard outcome).

Low risk of bias

The outcome is not mentioned in the prespecified plan, but according to CONSORT is expected to be reported.

Low risk of bias

Calculated based on RoB2 algorithm.

Sönmez 2020

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Some concerns

Participants and personnel were aware of the intervention, no apparent deviation because of the trial context. Not clear if an appropriate analysis was used, but probably this had no impact on the analysis results.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured with an appropriate method by blind assessors.

Low risk of bias

The outcome is not mentioned in the prespecified plan, but according to CONSORT is expected to be reported.

Some concerns

Calculated based on RoB2 algorithm.

Uzenoff 2008

Some concerns

Participants were randomised, there is no mention to how the random sequence was generated and no information about allocation sequence concealment. No baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured with an appropriate method by blind assessors.

Low risk of bias

The outcome is not mentioned in the prespecified plan, but according to CONSORT is expected to be reported.

Some concerns

Calculated based on RoB2 algorithm.

Risk of bias for analysis 1.56 Functioning ‐ mean endpoint functioning scale (high = good) (combined time points).

Study

Bias

Randomisation process

Deviations from intended interventions

Missing outcome data

Measurement of the outcome

Selection of the reported results

Overall

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Barrowclough 2014

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Some concerns

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status, but this is not likely.

Low risk of bias

Outcome measured with an appropriate method by blind assessors.

Some concerns

The outcome is not mentioned in the prespecified plan of this study.

Some concerns

Calculated based on RoB2 algorithm.

Bucci 2018 (Actissist)

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Some concerns

Participants and personnel were aware of the intervention, no apparent deviation because of the trial context. Not clear if an appropriate analysis was used, but probably this had no impact on the analysis results.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

Some concerns

Calculated based on RoB2 algorithm.

Edwards 2006

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Some concerns

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status, but this is not likely.

Low risk of bias

Outcome measured with an appropriate method by blind assessors.

Some concerns

No prespecified plan available.

Some concerns

Calculated based on RoB2 algorithm.

Edwards 2011

Some concerns

Participants were randomised, there is no mention to how the random sequence was generated and no information about allocation sequence concealment. No baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured with an appropriate method by blind assessors.

Some concerns

No prespecified plan available.

Some concerns

Calculated based on RoB2 algorithm.

GetUp

High risk of bias

Centres were randomised, but some centres were forced into the intervention arm. No baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Low risk of bias

Data on this outcome are given for all randomised participants.

Some concerns

Outcome measured with an appropriate method, no blind outcome assessors, the assessment could have been influenced, but this is unlikely.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

High risk of bias

Calculated based on RoB2 algorithm.

Gleeson 2009 (EPISODE II)

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured with an appropriate method by blind assessors.

Some concerns

The outcome is not mentioned in the prespecified plan of this study.

Some concerns

Calculated based on RoB2 algorithm.

Gonzalez‐Ortega 2021

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Some concerns

Participants and personnel were aware of the intervention, no apparent deviation because of the trial context. No appropriate analysis was used, but probably this had no impact on the analysis results.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured with an appropriate method by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

Some concerns

Calculated based on RoB2 algorithm.

Hjorthoj 2013 (CapOpus)

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Some concerns

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status, but this is not likely.

Low risk of bias

Outcome measured with an appropriate method by blind assessors.

Some concerns

The outcome is not mentioned in the prespecified plan of this study.

Some concerns

Calculated based on RoB2 algorithm.

Jackson 2008 (ACE)

Low risk of bias

Participants were randomised, there is no mention to how the random sequence was generated but allocation sequence was concealed. No information about baseline differences to suggest a problem with the randomisation process.

Low risk of bias

No information whether participants were aware of the intervention, personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured with an appropriate method by blind assessors.

Some concerns

No prespecified plan available.

Some concerns

Calculated based on RoB2 algorithm.

Lepage 2022

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Some concerns

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status, but this is not likely.

Low risk of bias

Outcome measured with an appropriate method by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

Some concerns

Calculated based on RoB2 algorithm.

Liu 2019

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured with an appropriate method by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

Low risk of bias

Calculated based on RoB2 algorithm.

Madigan 2013

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Some concerns

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status, but this is not likely.

Low risk of bias

Outcome measured with an appropriate method by blind assessors.

Some concerns

No prespecified plan available.

Some concerns

Calculated based on RoB2 algorithm.

Morrison 2018 (COMPARE)

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured with an appropriate method by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

Low risk of bias

Calculated based on RoB2 algorithm.

Morrison 2020 (MAPS)

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

High risk of bias

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status.

Low risk of bias

Outcome measured with an appropriate method by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

High risk of bias

Calculated based on RoB2 algorithm.

Müller 2020

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured with an appropriate method by blind assessors.

Some concerns

No prespecified plan available.

Some concerns

Calculated based on RoB2 algorithm.

Palma‐Sevillano 2019

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Some concerns

Participants and personnel were aware of the intervention, no apparent deviation because of the trial context. Not clear if an appropriate analysis was used, but probably this had no impact on the analysis results.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured with an appropriate method by blind assessors.

Some concerns

No prespecified plan available.

Some concerns

Calculated based on RoB2 algorithm.

Pos 2019

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Some concerns

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status, but this is not likely.

Low risk of bias

Outcome measured with an appropriate method by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

Some concerns

Calculated based on RoB2 algorithm.

Sönmez 2020

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Some concerns

Participants and personnel were aware of the intervention, no apparent deviation because of the trial context. Not clear if an appropriate analysis was used, but probably this had no impact on the analysis results.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured with an appropriate method by blind assessors.

Some concerns

The outcome is not mentioned in the prespecified plan of this study.

Some concerns

Calculated based on RoB2 algorithm.

Risk of bias for analysis 1.69 Adverse events ‐ participants with at least one adverse event.

Study

Bias

Randomisation process

Deviations from intended interventions

Missing outcome data

Measurement of the outcome

Selection of the reported results

Overall

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Morrison 2020 (MAPS)

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Low risk of bias

Data on this outcome are given for all randomised participants.

Some concerns

Outcome measured with an appropriate method, not clear whether the assessors were blind, the assessment could have been influenced, but this is unlikely.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

Some concerns

Calculated based on RoB2 algorithm.

Risk of bias for analysis 1.76 Sensitivity analysis ‐ Blinding of outcome assessor.

Study

Bias

Randomisation process

Deviations from intended interventions

Missing outcome data

Measurement of the outcome

Selection of the reported results

Overall

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Barrowclough 2014

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Some concerns

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status, but this is not likely.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Some concerns

The outcome is not mentioned in the prespecified plan of this study.

Some concerns

Calculated based on RoB2 algorithm.

Bucci 2018 (Actissist)

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Some concerns

Participants and personnel were aware of the intervention, no apparent deviation because of the trial context. Not clear if an appropriate analysis was used, but probably this had no impact on the analysis results.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

Some concerns

Calculated based on RoB2 algorithm.

Edwards 2006

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Some concerns

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status, but this is not likely.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Some concerns

No prespecified plan available.

Some concerns

Calculated based on RoB2 algorithm.

Gleeson 2009 (EPISODE II)

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Some concerns

The outcome is not mentioned in the prespecified plan of this study.

Some concerns

Calculated based on RoB2 algorithm.

Gonzalez‐Ortega 2021

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Some concerns

Participants and personnel were aware of the intervention, no apparent deviation because of the trial context. No appropriate analysis was used, but probably this had no impact on the analysis results.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

Some concerns

Calculated based on RoB2 algorithm.

Haddock 1999

Low risk of bias

Participants were randomised, there is no mention to how the random sequence was generated but allocation sequence was concealed. No information about baseline differences to suggest a problem with the randomisation process.

Some concerns

Participants and personnel were aware of the intervention, no apparent deviation because of the trial context. No appropriate analysis was used, but probably this had no impact on the analysis results.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Some concerns

No prespecified plan available.

Some concerns

Calculated based on RoB2 algorithm.

Hjorthoj 2013 (CapOpus)

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Some concerns

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status, but this is not likely.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

Some concerns

Calculated based on RoB2 algorithm.

Jackson 2008 (ACE)

Low risk of bias

Participants were randomised, there is no mention to how the random sequence was generated but allocation sequence was concealed. No information about baseline differences to suggest a problem with the randomisation process.

Some concerns

No information whether participants were aware of the intervention, personnel were aware of the intervention, no apparent deviations because of the trial context. No appropriate analysis was used, but probably this had no impact on the analysis results.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Some concerns

No prespecified plan available.

Some concerns

Calculated based on RoB2 algorithm.

Jolley 2003

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Some concerns

Participants and personnel were aware of the intervention, no apparent deviation because of the trial context. Not clear if an appropriate analysis was used, but probably this had no impact on the analysis results.

High risk of bias

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Some concerns

The outcome is not mentioned in the prespecified plan of this study.

High risk of bias

Calculated based on RoB2 algorithm.

Lecomte 2008

Some concerns

Participants were randomised, there is no mention to how the random sequence was generated and no information about allocation sequence concealment. No baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

High risk of bias

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

High risk of bias

Calculated based on RoB2 algorithm.

Lecomte 2008

Some concerns

Participants were randomised, there is no mention to how the random sequence was generated and no information about allocation sequence concealment. No baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

High risk of bias

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

High risk of bias

Calculated based on RoB2 algorithm.

Lewis 2002 (SOCRATES)

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Some concerns

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status, but this is not likely.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

Some concerns

Calculated based on RoB2 algorithm.

Lewis 2002 (SOCRATES)

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Some concerns

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status, but this is not likely.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

Some concerns

Calculated based on RoB2 algorithm.

Liu 2019

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

Low risk of bias

Calculated based on RoB2 algorithm.

Morrison 2018 (COMPARE)

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

Low risk of bias

Calculated based on RoB2 algorithm.

Morrison 2020 (MAPS)

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

High risk of bias

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

High risk of bias

Calculated based on RoB2 algorithm.

Müller 2020

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Some concerns

No prespecified plan available.

Some concerns

Calculated based on RoB2 algorithm.

Palma‐Sevillano 2019

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Some concerns

Participants and personnel were aware of the intervention, no apparent deviation because of the trial context. Not clear if an appropriate analysis was used, but probably this had no impact on the analysis results.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Some concerns

No prespecified plan available.

Some concerns

Calculated based on RoB2 algorithm.

Pos 2019

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Some concerns

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status, but this is not likely.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

Some concerns

Calculated based on RoB2 algorithm.

Supereden3

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

High risk of bias

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

High risk of bias

Calculated based on RoB2 algorithm.

Uzenoff 2008

Some concerns

Participants were randomised, there is no mention to how the random sequence was generated and no information about allocation sequence concealment. No baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Some concerns

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status, but this is not likely.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Some concerns

No prespecified plan available.

Some concerns

Calculated based on RoB2 algorithm.

Risk of bias for analysis 1.77 Sensitivity analysis ‐ Assumptions for missing data.

Study

Bias

Randomisation process

Deviations from intended interventions

Missing outcome data

Measurement of the outcome

Selection of the reported results

Overall

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Barrowclough 2014

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Some concerns

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status, but this is not likely.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Some concerns

The outcome is not mentioned in the prespecified plan of this study.

Some concerns

Calculated based on RoB2 algorithm.

Edwards 2006

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Some concerns

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status, but this is not likely.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Some concerns

No prespecified plan available.

Some concerns

Calculated based on RoB2 algorithm.

GetUp

High risk of bias

Centres were randomised, but some centres were forced into the intervention arm. No baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Low risk of bias

Data on this outcome are given for all randomised participants.

Some concerns

Outcome measured with an appropriate method, no blind outcome assessors, the assessment could have been influenced, but this is unlikely.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

High risk of bias

Calculated based on RoB2 algorithm.

Gleeson 2009 (EPISODE II)

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Some concerns

The outcome is not mentioned in the prespecified plan of this study.

Some concerns

Calculated based on RoB2 algorithm.

Hjorthoj 2013 (CapOpus)

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Some concerns

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status, but this is not likely.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

Some concerns

Calculated based on RoB2 algorithm.

Lecomte 2008

Some concerns

Participants were randomised, there is no mention to how the random sequence was generated and no information about allocation sequence concealment. No baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

High risk of bias

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

High risk of bias

Calculated based on RoB2 algorithm.

Lecomte 2008

Some concerns

Participants were randomised, there is no mention to how the random sequence was generated and no information about allocation sequence concealment. No baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

High risk of bias

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

High risk of bias

Calculated based on RoB2 algorithm.

Lewis 2002 (SOCRATES)

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Some concerns

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status, but this is not likely.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

Some concerns

Calculated based on RoB2 algorithm.

Lewis 2002 (SOCRATES)

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Some concerns

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status, but this is not likely.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

Some concerns

Calculated based on RoB2 algorithm.

Liu 2019

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

Low risk of bias

Calculated based on RoB2 algorithm.

Morrison 2018 (COMPARE)

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

Low risk of bias

Calculated based on RoB2 algorithm.

Morrison 2020 (MAPS)

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

High risk of bias

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

High risk of bias

Calculated based on RoB2 algorithm.

Müller 2020

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Some concerns

No prespecified plan available.

Some concerns

Calculated based on RoB2 algorithm.

Palma‐Sevillano 2019

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Some concerns

Participants and personnel were aware of the intervention, no apparent deviation because of the trial context. Not clear if an appropriate analysis was used, but probably this had no impact on the analysis results.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Some concerns

No prespecified plan available.

Some concerns

Calculated based on RoB2 algorithm.

Pos 2019

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Some concerns

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status, but this is not likely.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

Some concerns

Calculated based on RoB2 algorithm.

Supereden3

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

High risk of bias

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

High risk of bias

Calculated based on RoB2 algorithm.

Uzenoff 2008

Some concerns

Participants were randomised, there is no mention to how the random sequence was generated and no information about allocation sequence concealment. No baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Some concerns

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status, but this is not likely.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Some concerns

No prespecified plan available.

Some concerns

Calculated based on RoB2 algorithm.

Risk of bias for analysis 1.78 Sensitivity analysis ‐ Risk of Bias.

Study

Bias

Randomisation process

Deviations from intended interventions

Missing outcome data

Measurement of the outcome

Selection of the reported results

Overall

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Barrowclough 2014

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Some concerns

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status, but this is not likely.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Some concerns

The outcome is not mentioned in the prespecified plan of this study.

Some concerns

Calculated based on RoB2 algorithm.

Bucci 2018 (Actissist)

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Some concerns

Participants and personnel were aware of the intervention, no apparent deviation because of the trial context. Not clear if an appropriate analysis was used, but probably this had no impact on the analysis results.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

Some concerns

Calculated based on RoB2 algorithm.

Edwards 2006

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Some concerns

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status, but this is not likely.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Some concerns

No prespecified plan available.

Some concerns

Calculated based on RoB2 algorithm.

Gleeson 2009 (EPISODE II)

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Some concerns

The outcome is not mentioned in the prespecified plan of this study.

Some concerns

Calculated based on RoB2 algorithm.

Gonzalez‐Ortega 2021

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Some concerns

Participants and personnel were aware of the intervention, no apparent deviation because of the trial context. No appropriate analysis was used, but probably this had no impact on the analysis results.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

Some concerns

Calculated based on RoB2 algorithm.

Haddock 1999

Low risk of bias

Participants were randomised, there is no mention to how the random sequence was generated but allocation sequence was concealed. No information about baseline differences to suggest a problem with the randomisation process.

Some concerns

Participants and personnel were aware of the intervention, no apparent deviation because of the trial context. No appropriate analysis was used, but probably this had no impact on the analysis results.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Some concerns

No prespecified plan available.

Some concerns

Calculated based on RoB2 algorithm.

Hjorthoj 2013 (CapOpus)

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Some concerns

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status, but this is not likely.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

Some concerns

Calculated based on RoB2 algorithm.

Jackson 2008 (ACE)

Low risk of bias

Participants were randomised, there is no mention to how the random sequence was generated but allocation sequence was concealed. No information about baseline differences to suggest a problem with the randomisation process.

Some concerns

No information whether participants were aware of the intervention, personnel were aware of the intervention, no apparent deviations because of the trial context. No appropriate analysis was used, but probably this had no impact on the analysis results.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Some concerns

No prespecified plan available.

Some concerns

Calculated based on RoB2 algorithm.

Lewis 2002 (SOCRATES)

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Some concerns

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status, but this is not likely.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

Some concerns

Calculated based on RoB2 algorithm.

Lewis 2002 (SOCRATES)

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Some concerns

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status, but this is not likely.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

Some concerns

Calculated based on RoB2 algorithm.

Liu 2019

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

Low risk of bias

Calculated based on RoB2 algorithm.

Morrison 2018 (COMPARE)

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

Low risk of bias

Calculated based on RoB2 algorithm.

Müller 2020

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Some concerns

No prespecified plan available.

Some concerns

Calculated based on RoB2 algorithm.

Palma‐Sevillano 2019

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Some concerns

Participants and personnel were aware of the intervention, no apparent deviation because of the trial context. Not clear if an appropriate analysis was used, but probably this had no impact on the analysis results.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Some concerns

No prespecified plan available.

Some concerns

Calculated based on RoB2 algorithm.

Pos 2019

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Some concerns

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status, but this is not likely.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

Some concerns

Calculated based on RoB2 algorithm.

Uzenoff 2008

Some concerns

Participants were randomised, there is no mention to how the random sequence was generated and no information about allocation sequence concealment. No baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Some concerns

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status, but this is not likely.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Some concerns

No prespecified plan available.

Some concerns

Calculated based on RoB2 algorithm.

Risk of bias for analysis 1.79 Sensitivity analysis ‐ Imputed values.

Study

Bias

Randomisation process

Deviations from intended interventions

Missing outcome data

Measurement of the outcome

Selection of the reported results

Overall

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Barrowclough 2014

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Some concerns

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status, but this is not likely.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Some concerns

The outcome is not mentioned in the prespecified plan of this study.

Some concerns

Calculated based on RoB2 algorithm.

Bucci 2018 (Actissist)

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Some concerns

Participants and personnel were aware of the intervention, no apparent deviation because of the trial context. Not clear if an appropriate analysis was used, but probably this had no impact on the analysis results.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

Some concerns

Calculated based on RoB2 algorithm.

Edwards 2006

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Some concerns

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status, but this is not likely.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Some concerns

No prespecified plan available.

Some concerns

Calculated based on RoB2 algorithm.

Gleeson 2009 (EPISODE II)

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Some concerns

The outcome is not mentioned in the prespecified plan of this study.

Some concerns

Calculated based on RoB2 algorithm.

Gonzalez‐Ortega 2021

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Some concerns

Participants and personnel were aware of the intervention, no apparent deviation because of the trial context. No appropriate analysis was used, but probably this had no impact on the analysis results.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

Some concerns

Calculated based on RoB2 algorithm.

Haddock 1999

Low risk of bias

Participants were randomised, there is no mention to how the random sequence was generated but allocation sequence was concealed. No information about baseline differences to suggest a problem with the randomisation process.

Some concerns

Participants and personnel were aware of the intervention, no apparent deviation because of the trial context. No appropriate analysis was used, but probably this had no impact on the analysis results.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Some concerns

No prespecified plan available.

Some concerns

Calculated based on RoB2 algorithm.

Hjorthoj 2013 (CapOpus)

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Some concerns

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status, but this is not likely.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

Some concerns

Calculated based on RoB2 algorithm.

Jackson 2008 (ACE)

Low risk of bias

Participants were randomised, there is no mention to how the random sequence was generated but allocation sequence was concealed. No information about baseline differences to suggest a problem with the randomisation process.

Some concerns

No information whether participants were aware of the intervention, personnel were aware of the intervention, no apparent deviations because of the trial context. No appropriate analysis was used, but probably this had no impact on the analysis results.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Some concerns

No prespecified plan available.

Some concerns

Calculated based on RoB2 algorithm.

Lecomte 2008

Some concerns

Participants were randomised, there is no mention to how the random sequence was generated and no information about allocation sequence concealment. No baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

High risk of bias

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

High risk of bias

Calculated based on RoB2 algorithm.

Lecomte 2008

Some concerns

Participants were randomised, there is no mention to how the random sequence was generated and no information about allocation sequence concealment. No baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

High risk of bias

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

High risk of bias

Calculated based on RoB2 algorithm.

Lewis 2002 (SOCRATES)

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Some concerns

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status, but this is not likely.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

Some concerns

Calculated based on RoB2 algorithm.

Lewis 2002 (SOCRATES)

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Some concerns

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status, but this is not likely.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

Some concerns

Calculated based on RoB2 algorithm.

Liu 2019

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

Low risk of bias

Calculated based on RoB2 algorithm.

Morrison 2018 (COMPARE)

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

Low risk of bias

Calculated based on RoB2 algorithm.

Morrison 2020 (MAPS)

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

High risk of bias

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

High risk of bias

Calculated based on RoB2 algorithm.

Müller 2020

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Some concerns

No prespecified plan available.

Some concerns

Calculated based on RoB2 algorithm.

Palma‐Sevillano 2019

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Some concerns

Participants and personnel were aware of the intervention, no apparent deviation because of the trial context. Not clear if an appropriate analysis was used, but probably this had no impact on the analysis results.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Some concerns

No prespecified plan available.

Some concerns

Calculated based on RoB2 algorithm.

Pos 2019

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Some concerns

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status, but this is not likely.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

Some concerns

Calculated based on RoB2 algorithm.

Supereden3

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

High risk of bias

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

High risk of bias

Calculated based on RoB2 algorithm.

Uzenoff 2008

Some concerns

Participants were randomised, there is no mention to how the random sequence was generated and no information about allocation sequence concealment. No baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Some concerns

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status, but this is not likely.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Some concerns

No prespecified plan available.

Some concerns

Calculated based on RoB2 algorithm.

Risk of bias for analysis 1.80 Sensitivity analysis ‐ Fixed‐effect.

Study

Bias

Randomisation process

Deviations from intended interventions

Missing outcome data

Measurement of the outcome

Selection of the reported results

Overall

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Barrowclough 2014

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Some concerns

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status, but this is not likely.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Some concerns

The outcome is not mentioned in the prespecified plan of this study.

Some concerns

Calculated based on RoB2 algorithm.

Bucci 2018 (Actissist)

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Some concerns

Participants and personnel were aware of the intervention, no apparent deviation because of the trial context. Not clear if an appropriate analysis was used, but probably this had no impact on the analysis results.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

Some concerns

Calculated based on RoB2 algorithm.

Edwards 2006

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Some concerns

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status, but this is not likely.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Some concerns

No prespecified plan available.

Some concerns

Calculated based on RoB2 algorithm.

GetUp

High risk of bias

Centres were randomised, but some centres were forced into the intervention arm. No baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Low risk of bias

Data on this outcome are given for all randomised participants.

Some concerns

Outcome measured with an appropriate method, no blind outcome assessors, the assessment could have been influenced, but this is unlikely.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

High risk of bias

Calculated based on RoB2 algorithm.

Gleeson 2009 (EPISODE II)

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Some concerns

The outcome is not mentioned in the prespecified plan of this study.

Some concerns

Calculated based on RoB2 algorithm.

Gonzalez‐Ortega 2021

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Some concerns

Participants and personnel were aware of the intervention, no apparent deviation because of the trial context. No appropriate analysis was used, but probably this had no impact on the analysis results.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

Some concerns

Calculated based on RoB2 algorithm.

Haddock 1999

Low risk of bias

Participants were randomised, there is no mention to how the random sequence was generated but allocation sequence was concealed. No information about baseline differences to suggest a problem with the randomisation process.

Some concerns

Participants and personnel were aware of the intervention, no apparent deviation because of the trial context. No appropriate analysis was used, but probably this had no impact on the analysis results.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Some concerns

No prespecified plan available.

Some concerns

Calculated based on RoB2 algorithm.

Hjorthoj 2013 (CapOpus)

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Some concerns

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status, but this is not likely.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

Some concerns

Calculated based on RoB2 algorithm.

Jackson 2008 (ACE)

Low risk of bias

Participants were randomised, there is no mention to how the random sequence was generated but allocation sequence was concealed. No information about baseline differences to suggest a problem with the randomisation process.

Some concerns

No information whether participants were aware of the intervention, personnel were aware of the intervention, no apparent deviations because of the trial context. No appropriate analysis was used, but probably this had no impact on the analysis results.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Some concerns

No prespecified plan available.

Some concerns

Calculated based on RoB2 algorithm.

Jolley 2003

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Some concerns

Participants and personnel were aware of the intervention, no apparent deviation because of the trial context. Not clear if an appropriate analysis was used, but probably this had no impact on the analysis results.

High risk of bias

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Some concerns

The outcome is not mentioned in the prespecified plan of this study.

High risk of bias

Calculated based on RoB2 algorithm.

Lecomte 2008

Some concerns

Participants were randomised, there is no mention to how the random sequence was generated and no information about allocation sequence concealment. No baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

High risk of bias

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

High risk of bias

Calculated based on RoB2 algorithm.

Lecomte 2008

Some concerns

Participants were randomised, there is no mention to how the random sequence was generated and no information about allocation sequence concealment. No baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

High risk of bias

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

High risk of bias

Calculated based on RoB2 algorithm.

Lewis 2002 (SOCRATES)

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Some concerns

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status, but this is not likely.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

Some concerns

Calculated based on RoB2 algorithm.

Lewis 2002 (SOCRATES)

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Some concerns

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status, but this is not likely.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

Some concerns

Calculated based on RoB2 algorithm.

Liu 2019

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

Low risk of bias

Calculated based on RoB2 algorithm.

Morrison 2018 (COMPARE)

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

Low risk of bias

Calculated based on RoB2 algorithm.

Morrison 2020 (MAPS)

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

High risk of bias

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

High risk of bias

Calculated based on RoB2 algorithm.

Müller 2020

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Some concerns

No prespecified plan available.

Some concerns

Calculated based on RoB2 algorithm.

Palma‐Sevillano 2019

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Some concerns

Participants and personnel were aware of the intervention, no apparent deviation because of the trial context. Not clear if an appropriate analysis was used, but probably this had no impact on the analysis results.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Some concerns

No prespecified plan available.

Some concerns

Calculated based on RoB2 algorithm.

Pos 2019

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Some concerns

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status, but this is not likely.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

Some concerns

Calculated based on RoB2 algorithm.

Supereden3

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

High risk of bias

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

High risk of bias

Calculated based on RoB2 algorithm.

Uzenoff 2008

Some concerns

Participants were randomised, there is no mention to how the random sequence was generated and no information about allocation sequence concealment. No baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Some concerns

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status, but this is not likely.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Some concerns

No prespecified plan available.

Some concerns

Calculated based on RoB2 algorithm.

Risk of bias for analysis 1.81 Sensitivity analysis excluding outliers.

Study

Bias

Randomisation process

Deviations from intended interventions

Missing outcome data

Measurement of the outcome

Selection of the reported results

Overall

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Barrowclough 2014

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Some concerns

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status, but this is not likely.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Some concerns

The outcome is not mentioned in the prespecified plan of this study.

Some concerns

Calculated based on RoB2 algorithm.

Bucci 2018 (Actissist)

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Some concerns

Participants and personnel were aware of the intervention, no apparent deviation because of the trial context. Not clear if an appropriate analysis was used, but probably this had no impact on the analysis results.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

Some concerns

Calculated based on RoB2 algorithm.

Edwards 2006

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Some concerns

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status, but this is not likely.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Some concerns

No prespecified plan available.

Some concerns

Calculated based on RoB2 algorithm.

GetUp

High risk of bias

Centres were randomised, but some centres were forced into the intervention arm. No baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Low risk of bias

Data on this outcome are given for all randomised participants.

Some concerns

Outcome measured with an appropriate method, no blind outcome assessors, the assessment could have been influenced, but this is unlikely.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

High risk of bias

Calculated based on RoB2 algorithm.

Gleeson 2009 (EPISODE II)

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Some concerns

The outcome is not mentioned in the prespecified plan of this study.

Some concerns

Calculated based on RoB2 algorithm.

Gonzalez‐Ortega 2021

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Some concerns

Participants and personnel were aware of the intervention, no apparent deviation because of the trial context. No appropriate analysis was used, but probably this had no impact on the analysis results.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

Some concerns

Calculated based on RoB2 algorithm.

Haddock 1999

Low risk of bias

Participants were randomised, there is no mention to how the random sequence was generated but allocation sequence was concealed. No information about baseline differences to suggest a problem with the randomisation process.

Some concerns

Participants and personnel were aware of the intervention, no apparent deviation because of the trial context. No appropriate analysis was used, but probably this had no impact on the analysis results.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Some concerns

No prespecified plan available.

Some concerns

Calculated based on RoB2 algorithm.

Hjorthoj 2013 (CapOpus)

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Some concerns

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status, but this is not likely.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

Some concerns

Calculated based on RoB2 algorithm.

Jackson 2008 (ACE)

Low risk of bias

Participants were randomised, there is no mention to how the random sequence was generated but allocation sequence was concealed. No information about baseline differences to suggest a problem with the randomisation process.

Some concerns

No information whether participants were aware of the intervention, personnel were aware of the intervention, no apparent deviations because of the trial context. No appropriate analysis was used, but probably this had no impact on the analysis results.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Some concerns

No prespecified plan available.

Some concerns

Calculated based on RoB2 algorithm.

Jolley 2003

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Some concerns

Participants and personnel were aware of the intervention, no apparent deviation because of the trial context. Not clear if an appropriate analysis was used, but probably this had no impact on the analysis results.

High risk of bias

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Some concerns

The outcome is not mentioned in the prespecified plan of this study.

High risk of bias

Calculated based on RoB2 algorithm.

Lecomte 2008

Some concerns

Participants were randomised, there is no mention to how the random sequence was generated and no information about allocation sequence concealment. No baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

High risk of bias

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

High risk of bias

Calculated based on RoB2 algorithm.

Lecomte 2008

Some concerns

Participants were randomised, there is no mention to how the random sequence was generated and no information about allocation sequence concealment. No baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

High risk of bias

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

High risk of bias

Calculated based on RoB2 algorithm.

Lewis 2002 (SOCRATES)

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Some concerns

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status, but this is not likely.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

Some concerns

Calculated based on RoB2 algorithm.

Lewis 2002 (SOCRATES)

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Some concerns

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status, but this is not likely.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

Some concerns

Calculated based on RoB2 algorithm.

Liu 2019

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

Low risk of bias

Calculated based on RoB2 algorithm.

Morrison 2018 (COMPARE)

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

Low risk of bias

Calculated based on RoB2 algorithm.

Morrison 2020 (MAPS)

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

High risk of bias

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

High risk of bias

Calculated based on RoB2 algorithm.

Müller 2020

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Some concerns

No prespecified plan available.

Some concerns

Calculated based on RoB2 algorithm.

Pos 2019

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Some concerns

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status, but this is not likely.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

Some concerns

Calculated based on RoB2 algorithm.

Supereden3

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

High risk of bias

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

High risk of bias

Calculated based on RoB2 algorithm.

Uzenoff 2008

Some concerns

Participants were randomised, there is no mention to how the random sequence was generated and no information about allocation sequence concealment. No baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Some concerns

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status, but this is not likely.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Some concerns

No prespecified plan available.

Some concerns

Calculated based on RoB2 algorithm.

Risk of bias for analysis 1.82 Subgroup analysis ‐ Different types of CBT.

Study

Bias

Randomisation process

Deviations from intended interventions

Missing outcome data

Measurement of the outcome

Selection of the reported results

Overall

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Subgroup 1.82.1 CBT

Gonzalez‐Ortega 2021

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Some concerns

Participants and personnel were aware of the intervention, no apparent deviation because of the trial context. No appropriate analysis was used, but probably this had no impact on the analysis results.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

Some concerns

Calculated based on RoB2 algorithm.

Haddock 1999

Low risk of bias

Participants were randomised, there is no mention to how the random sequence was generated but allocation sequence was concealed. No information about baseline differences to suggest a problem with the randomisation process.

Some concerns

Participants and personnel were aware of the intervention, no apparent deviation because of the trial context. No appropriate analysis was used, but probably this had no impact on the analysis results.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Some concerns

No prespecified plan available.

Some concerns

Calculated based on RoB2 algorithm.

Jackson 2008 (ACE)

Low risk of bias

Participants were randomised, there is no mention to how the random sequence was generated but allocation sequence was concealed. No information about baseline differences to suggest a problem with the randomisation process.

Some concerns

No information whether participants were aware of the intervention, personnel were aware of the intervention, no apparent deviations because of the trial context. No appropriate analysis was used, but probably this had no impact on the analysis results.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Some concerns

No prespecified plan available.

Some concerns

Calculated based on RoB2 algorithm.

Jolley 2003

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Some concerns

Participants and personnel were aware of the intervention, no apparent deviation because of the trial context. Not clear if an appropriate analysis was used, but probably this had no impact on the analysis results.

High risk of bias

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Some concerns

The outcome is not mentioned in the prespecified plan of this study.

High risk of bias

Calculated based on RoB2 algorithm.

Lecomte 2008

Some concerns

Participants were randomised, there is no mention to how the random sequence was generated and no information about allocation sequence concealment. No baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

High risk of bias

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

High risk of bias

Calculated based on RoB2 algorithm.

Lecomte 2008

Some concerns

Participants were randomised, there is no mention to how the random sequence was generated and no information about allocation sequence concealment. No baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

High risk of bias

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

High risk of bias

Calculated based on RoB2 algorithm.

Lewis 2002 (SOCRATES)

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Some concerns

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status, but this is not likely.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

Some concerns

Calculated based on RoB2 algorithm.

Lewis 2002 (SOCRATES)

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Some concerns

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status, but this is not likely.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

Some concerns

Calculated based on RoB2 algorithm.

Liu 2019

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

Low risk of bias

Calculated based on RoB2 algorithm.

Morrison 2018 (COMPARE)

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

Low risk of bias

Calculated based on RoB2 algorithm.

Morrison 2020 (MAPS)

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

High risk of bias

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

High risk of bias

Calculated based on RoB2 algorithm.

Müller 2020

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Some concerns

No prespecified plan available.

Some concerns

Calculated based on RoB2 algorithm.

Pos 2019

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Some concerns

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status, but this is not likely.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

Some concerns

Calculated based on RoB2 algorithm.

Subgroup 1.82.2 CBT integrated with other psychosocial interventions

Bucci 2018 (Actissist)

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Some concerns

Participants and personnel were aware of the intervention, no apparent deviation because of the trial context. Not clear if an appropriate analysis was used, but probably this had no impact on the analysis results.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

Some concerns

Calculated based on RoB2 algorithm.

GetUp

High risk of bias

Centres were randomised, but some centres were forced into the intervention arm. No baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Low risk of bias

Data on this outcome are given for all randomised participants.

Some concerns

Outcome measured with an appropriate method, no blind outcome assessors, the assessment could have been influenced, but this is unlikely.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

High risk of bias

Calculated based on RoB2 algorithm.

Palma‐Sevillano 2019

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Some concerns

Participants and personnel were aware of the intervention, no apparent deviation because of the trial context. Not clear if an appropriate analysis was used, but probably this had no impact on the analysis results.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Some concerns

No prespecified plan available.

Some concerns

Calculated based on RoB2 algorithm.

Supereden3

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

High risk of bias

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

High risk of bias

Calculated based on RoB2 algorithm.

Subgroup 1.82.3 CBT and Motivational Interviewing for cannabis use cessation

Barrowclough 2014

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Some concerns

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status, but this is not likely.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Some concerns

The outcome is not mentioned in the prespecified plan of this study.

Some concerns

Calculated based on RoB2 algorithm.

Edwards 2006

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Some concerns

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status, but this is not likely.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Some concerns

No prespecified plan available.

Some concerns

Calculated based on RoB2 algorithm.

Hjorthoj 2013 (CapOpus)

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Some concerns

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status, but this is not likely.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

Some concerns

Calculated based on RoB2 algorithm.

Subgroup 1.82.4 Special types of CBT

Gleeson 2009 (EPISODE II)

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Some concerns

The outcome is not mentioned in the prespecified plan of this study.

Some concerns

Calculated based on RoB2 algorithm.

Uzenoff 2008

Some concerns

Participants were randomised, there is no mention to how the random sequence was generated and no information about allocation sequence concealment. No baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Some concerns

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status, but this is not likely.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Some concerns

No prespecified plan available.

Some concerns

Calculated based on RoB2 algorithm.

Risk of bias for analysis 1.83 Subgroup analysis ‐ Modality of treatment.

Study

Bias

Randomisation process

Deviations from intended interventions

Missing outcome data

Measurement of the outcome

Selection of the reported results

Overall

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Subgroup 1.83.1 Individual CBT

Barrowclough 2014

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Some concerns

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status, but this is not likely.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Some concerns

The outcome is not mentioned in the prespecified plan of this study.

Some concerns

Calculated based on RoB2 algorithm.

Bucci 2018 (Actissist)

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Some concerns

Participants and personnel were aware of the intervention, no apparent deviation because of the trial context. Not clear if an appropriate analysis was used, but probably this had no impact on the analysis results.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

Some concerns

Calculated based on RoB2 algorithm.

Edwards 2006

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Some concerns

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status, but this is not likely.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Some concerns

No prespecified plan available.

Some concerns

Calculated based on RoB2 algorithm.

GetUp

High risk of bias

Centres were randomised, but some centres were forced into the intervention arm. No baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Low risk of bias

Data on this outcome are given for all randomised participants.

Some concerns

Outcome measured with an appropriate method, no blind outcome assessors, the assessment could have been influenced, but this is unlikely.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

High risk of bias

Calculated based on RoB2 algorithm.

Gleeson 2009 (EPISODE II)

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Some concerns

The outcome is not mentioned in the prespecified plan of this study.

Some concerns

Calculated based on RoB2 algorithm.

Gonzalez‐Ortega 2021

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Some concerns

Participants and personnel were aware of the intervention, no apparent deviation because of the trial context. No appropriate analysis was used, but probably this had no impact on the analysis results.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

Some concerns

Calculated based on RoB2 algorithm.

Haddock 1999

Low risk of bias

Participants were randomised, there is no mention to how the random sequence was generated but allocation sequence was concealed. No information about baseline differences to suggest a problem with the randomisation process.

Some concerns

Participants and personnel were aware of the intervention, no apparent deviation because of the trial context. No appropriate analysis was used, but probably this had no impact on the analysis results.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Some concerns

No prespecified plan available.

Some concerns

Calculated based on RoB2 algorithm.

Hjorthoj 2013 (CapOpus)

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Some concerns

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status, but this is not likely.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

Some concerns

Calculated based on RoB2 algorithm.

Jackson 2008 (ACE)

Low risk of bias

Participants were randomised, there is no mention to how the random sequence was generated but allocation sequence was concealed. No information about baseline differences to suggest a problem with the randomisation process.

Some concerns

No information whether participants were aware of the intervention, personnel were aware of the intervention, no apparent deviations because of the trial context. No appropriate analysis was used, but probably this had no impact on the analysis results.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Some concerns

No prespecified plan available.

Some concerns

Calculated based on RoB2 algorithm.

Jolley 2003

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Some concerns

Participants and personnel were aware of the intervention, no apparent deviation because of the trial context. Not clear if an appropriate analysis was used, but probably this had no impact on the analysis results.

High risk of bias

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Some concerns

No prespecified plan available.

High risk of bias

Calculated based on RoB2 algorithm.

Lewis 2002 (SOCRATES)

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Some concerns

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status, but this is not likely.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

Some concerns

Calculated based on RoB2 algorithm.

Lewis 2002 (SOCRATES)

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Some concerns

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status, but this is not likely.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

Some concerns

Calculated based on RoB2 algorithm.

Liu 2019

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

Low risk of bias

Calculated based on RoB2 algorithm.

Morrison 2018 (COMPARE)

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

Low risk of bias

Calculated based on RoB2 algorithm.

Morrison 2020 (MAPS)

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

High risk of bias

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

High risk of bias

Calculated based on RoB2 algorithm.

Müller 2020

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Some concerns

No prespecified plan available.

Some concerns

Calculated based on RoB2 algorithm.

Palma‐Sevillano 2019

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Some concerns

Participants and personnel were aware of the intervention, no apparent deviation because of the trial context. Not clear if an appropriate analysis was used, but probably this had no impact on the analysis results.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Some concerns

No prespecified plan available.

Some concerns

Calculated based on RoB2 algorithm.

Uzenoff 2008

Some concerns

Participants were randomised, there is no mention to how the random sequence was generated and no information about allocation sequence concealment. No baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Some concerns

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status, but this is not likely.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Some concerns

No prespecified plan available.

Some concerns

Calculated based on RoB2 algorithm.

Subgroup 1.83.2 Group CBT

Lecomte 2008

Some concerns

Participants were randomised, there is no mention to how the random sequence was generated and no information about allocation sequence concealment. No baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

High risk of bias

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

High risk of bias

Calculated based on RoB2 algorithm.

Lecomte 2008

Some concerns

Participants were randomised, there is no mention to how the random sequence was generated and no information about allocation sequence concealment. No baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

High risk of bias

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

High risk of bias

Calculated based on RoB2 algorithm.

Pos 2019

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Some concerns

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status, but this is not likely.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

Some concerns

Calculated based on RoB2 algorithm.

Supereden3

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

High risk of bias

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

High risk of bias

Calculated based on RoB2 algorithm.

Risk of bias for analysis 1.84 Subgroup analysis ‐ Method of delivery.

Study

Bias

Randomisation process

Deviations from intended interventions

Missing outcome data

Measurement of the outcome

Selection of the reported results

Overall

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Subgroup 1.84.1 Face to face

Barrowclough 2014

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Some concerns

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status, but this is not likely.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Some concerns

The outcome is not mentioned in the prespecified plan of this study.

Some concerns

Calculated based on RoB2 algorithm.

Edwards 2006

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Some concerns

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status, but this is not likely.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Some concerns

No prespecified plan available.

Some concerns

Calculated based on RoB2 algorithm.

GetUp

High risk of bias

Centres were randomised, but some centres were forced into the intervention arm. No baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Low risk of bias

Data on this outcome are given for all randomised participants.

Some concerns

Outcome measured with an appropriate method, no blind outcome assessors, the assessment could have been influenced, but this is unlikely.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

High risk of bias

Calculated based on RoB2 algorithm.

Gleeson 2009 (EPISODE II)

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Some concerns

The outcome is not mentioned in the prespecified plan of this study.

Some concerns

Calculated based on RoB2 algorithm.

Gonzalez‐Ortega 2021

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Some concerns

Participants and personnel were aware of the intervention, no apparent deviation because of the trial context. No appropriate analysis was used, but probably this had no impact on the analysis results.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

Some concerns

Calculated based on RoB2 algorithm.

Haddock 1999

Low risk of bias

Participants were randomised, there is no mention to how the random sequence was generated but allocation sequence was concealed. No information about baseline differences to suggest a problem with the randomisation process.

Some concerns

Participants and personnel were aware of the intervention, no apparent deviation because of the trial context. No appropriate analysis was used, but probably this had no impact on the analysis results.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Some concerns

No prespecified plan available.

Some concerns

Calculated based on RoB2 algorithm.

Hjorthoj 2013 (CapOpus)

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Some concerns

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status, but this is not likely.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

Some concerns

Calculated based on RoB2 algorithm.

Jackson 2008 (ACE)

Low risk of bias

Participants were randomised, there is no mention to how the random sequence was generated but allocation sequence was concealed. No information about baseline differences to suggest a problem with the randomisation process.

Some concerns

No information whether participants were aware of the intervention, personnel were aware of the intervention, no apparent deviations because of the trial context. No appropriate analysis was used, but probably this had no impact on the analysis results.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Some concerns

No prespecified plan available.

Some concerns

Calculated based on RoB2 algorithm.

Jolley 2003

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Some concerns

Participants and personnel were aware of the intervention, no apparent deviation because of the trial context. Not clear if an appropriate analysis was used, but probably this had no impact on the analysis results.

High risk of bias

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Some concerns

The outcome is not mentioned in the prespecified plan of this study.

High risk of bias

Calculated based on RoB2 algorithm.

Lecomte 2008

Some concerns

Participants were randomised, there is no mention to how the random sequence was generated and no information about allocation sequence concealment. No baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

High risk of bias

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

High risk of bias

Calculated based on RoB2 algorithm.

Lecomte 2008

Some concerns

Participants were randomised, there is no mention to how the random sequence was generated and no information about allocation sequence concealment. No baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

High risk of bias

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

High risk of bias

Calculated based on RoB2 algorithm.

Lewis 2002 (SOCRATES)

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Some concerns

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status, but this is not likely.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

Some concerns

Calculated based on RoB2 algorithm.

Lewis 2002 (SOCRATES)

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Some concerns

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status, but this is not likely.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

Some concerns

Calculated based on RoB2 algorithm.

Liu 2019

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

Low risk of bias

Calculated based on RoB2 algorithm.

Morrison 2018 (COMPARE)

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

Low risk of bias

Calculated based on RoB2 algorithm.

Morrison 2020 (MAPS)

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

High risk of bias

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

High risk of bias

Calculated based on RoB2 algorithm.

Müller 2020

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Some concerns

No prespecified plan available.

Some concerns

Calculated based on RoB2 algorithm.

Palma‐Sevillano 2019

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Some concerns

Participants and personnel were aware of the intervention, no apparent deviation because of the trial context. Not clear if an appropriate analysis was used, but probably this had no impact on the analysis results.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Some concerns

No prespecified plan available.

Some concerns

Calculated based on RoB2 algorithm.

Pos 2019

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Some concerns

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status, but this is not likely.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

Some concerns

Calculated based on RoB2 algorithm.

Supereden3

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

High risk of bias

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

High risk of bias

Calculated based on RoB2 algorithm.

Uzenoff 2008

Some concerns

Participants were randomised, there is no mention to how the random sequence was generated and no information about allocation sequence concealment. No baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Some concerns

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status, but this is not likely.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Some concerns

No prespecified plan available.

Some concerns

Calculated based on RoB2 algorithm.

Subgroup 1.84.2 Smartphone app

Bucci 2018 (Actissist)

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Some concerns

Participants and personnel were aware of the intervention, no apparent deviation because of the trial context. Not clear if an appropriate analysis was used, but probably this had no impact on the analysis results.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

Some concerns

Calculated based on RoB2 algorithm.

Risk of bias for analysis 1.85 Subgroup analysis ‐ Control group.

Study

Bias

Randomisation process

Deviations from intended interventions

Missing outcome data

Measurement of the outcome

Selection of the reported results

Overall

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Subgroup 1.85.1 CBT compared to TAU

Barrowclough 2014

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Some concerns

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status, but this is not likely.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Some concerns

The outcome is not mentioned in the prespecified plan of this study.

Some concerns

Calculated based on RoB2 algorithm.

GetUp

High risk of bias

Centres were randomised, but some centres were forced into the intervention arm. No baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Low risk of bias

Data on this outcome are given for all randomised participants.

Some concerns

Outcome measured with an appropriate method, no blind outcome assessors, the assessment could have been influenced, but this is unlikely.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

High risk of bias

Calculated based on RoB2 algorithm.

Gleeson 2009 (EPISODE II)

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Some concerns

The outcome is not mentioned in the prespecified plan of this study.

Some concerns

Calculated based on RoB2 algorithm.

Gonzalez‐Ortega 2021

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Some concerns

Participants and personnel were aware of the intervention, no apparent deviation because of the trial context. No appropriate analysis was used, but probably this had no impact on the analysis results.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

Some concerns

Calculated based on RoB2 algorithm.

Hjorthoj 2013 (CapOpus)

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Some concerns

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status, but this is not likely.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

Some concerns

Calculated based on RoB2 algorithm.

Jolley 2003

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Some concerns

Participants and personnel were aware of the intervention, no apparent deviation because of the trial context. Not clear if an appropriate analysis was used, but probably this had no impact on the analysis results.

High risk of bias

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Some concerns

The outcome is not mentioned in the prespecified plan of this study.

High risk of bias

Calculated based on RoB2 algorithm.

Lecomte 2008

Some concerns

Participants were randomised, there is no mention to how the random sequence was generated and no information about allocation sequence concealment. No baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

High risk of bias

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

High risk of bias

Calculated based on RoB2 algorithm.

Lewis 2002 (SOCRATES)

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Some concerns

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status, but this is not likely.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

Some concerns

Calculated based on RoB2 algorithm.

Liu 2019

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

Low risk of bias

Calculated based on RoB2 algorithm.

Morrison 2018 (COMPARE)

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

Low risk of bias

Calculated based on RoB2 algorithm.

Morrison 2020 (MAPS)

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

High risk of bias

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

High risk of bias

Calculated based on RoB2 algorithm.

Müller 2020

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Some concerns

No prespecified plan available.

Some concerns

Calculated based on RoB2 algorithm.

Palma‐Sevillano 2019

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Some concerns

Participants and personnel were aware of the intervention, no apparent deviation because of the trial context. Not clear if an appropriate analysis was used, but probably this had no impact on the analysis results.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Some concerns

No prespecified plan available.

Some concerns

Calculated based on RoB2 algorithm.

Pos 2019

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Some concerns

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status, but this is not likely.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

Some concerns

Calculated based on RoB2 algorithm.

Supereden3

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

High risk of bias

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

High risk of bias

Calculated based on RoB2 algorithm.

Subgroup 1.85.2 CBT compared to other psychosocial interventions

Bucci 2018 (Actissist)

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Some concerns

Participants and personnel were aware of the intervention, no apparent deviation because of the trial context. Not clear if an appropriate analysis was used, but probably this had no impact on the analysis results.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

Some concerns

Calculated based on RoB2 algorithm.

Edwards 2006

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Some concerns

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status, but this is not likely.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Some concerns

No prespecified plan available.

Some concerns

Calculated based on RoB2 algorithm.

Haddock 1999

Low risk of bias

Participants were randomised, there is no mention to how the random sequence was generated but allocation sequence was concealed. No information about baseline differences to suggest a problem with the randomisation process.

Some concerns

Participants and personnel were aware of the intervention, no apparent deviation because of the trial context. No appropriate analysis was used, but probably this had no impact on the analysis results.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Some concerns

No prespecified plan available.

Some concerns

Calculated based on RoB2 algorithm.

Jackson 2008 (ACE)

Low risk of bias

Participants were randomised, there is no mention to how the random sequence was generated but allocation sequence was concealed. No information about baseline differences to suggest a problem with the randomisation process.

Some concerns

No information whether participants were aware of the intervention, personnel were aware of the intervention, no apparent deviations because of the trial context. No appropriate analysis was used, but probably this had no impact on the analysis results.

Low risk of bias

Data on this outcome are given for all randomised participants.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Some concerns

No prespecified plan available.

Some concerns

Calculated based on RoB2 algorithm.

Lecomte 2008

Some concerns

Participants were randomised, there is no mention to how the random sequence was generated and no information about allocation sequence concealment. No baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

High risk of bias

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

High risk of bias

Calculated based on RoB2 algorithm.

Lewis 2002 (SOCRATES)

Low risk of bias

Participants were randomised, details are given about the generation of the randomisation sequence, no baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Some concerns

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status, but this is not likely.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Low risk of bias

The outcome is mentioned in the prespecified plan of this study.

Some concerns

Calculated based on RoB2 algorithm.

Uzenoff 2008

Some concerns

Participants were randomised, there is no mention to how the random sequence was generated and no information about allocation sequence concealment. No baseline differences to suggest a problem with the randomisation process.

Low risk of bias

Participants and personnel were aware of the intervention, no apparent deviations because of the trial context. Appropriate analysis was used.

Some concerns

Data on this outcome are not available for more than 20% of the participants; missingness of the outcome could be related to participants’ health status, but this is not likely.

Low risk of bias

Outcome measured on a validated rating scale by blind assessors.

Some concerns

No prespecified plan available.

Some concerns

Calculated based on RoB2 algorithm.

Data and analyses

Comparison 1. CBT plus standard care versus control.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1.1 Mental state ‐ general: mean endpoint PANSS total (high = poor)	15		Mean Difference (IV, Random, 95% CI)	Subtotals only
1.1.1 < 6 months	12	857	Mean Difference (IV, Random, 95% CI)	‐7.48 [‐12.79, ‐2.17]
1.1.2 < 12 months	10	808	Mean Difference (IV, Random, 95% CI)	‐6.66 [‐13.36, 0.05]
1.1.3 > 1 year	6	555	Mean Difference (IV, Random, 95% CI)	‐10.81 [‐21.23, ‐0.39]
1.2 Mental state ‐ general: mean change PANSS total (high = poor)	2		Mean Difference (IV, Random, 95% CI)	Subtotals only
1.2.1 < 6 months	1	15	Mean Difference (IV, Random, 95% CI)	‐1.40 [‐12.59, 9.79]
1.2.2 < 12 months	1	25	Mean Difference (IV, Random, 95% CI)	10.90 [‐0.28, 22.08]
1.2.3 > 1 year	1	22	Mean Difference (IV, Random, 95% CI)	8.00 [‐2.30, 18.30]
1.3 Mental state ‐ general: mean endpoint PSYRATS (high = poor)	1	240	Mean Difference (IV, Random, 95% CI)	‐7.44 [‐9.65, ‐5.22]
1.3.1 < 6 months	1	80	Mean Difference (IV, Random, 95% CI)	‐5.75 [‐9.66, ‐1.84]
1.3.2 < 12 months	1	80	Mean Difference (IV, Random, 95% CI)	‐8.00 [‐11.66, ‐4.34]
1.3.3 > 1 year	1	80	Mean Difference (IV, Random, 95% CI)	‐8.50 [‐12.45, ‐4.55]
1.4 Mental state ‐ general: mean endpoint BPRS total (high = poor)	5		Mean Difference (IV, Random, 95% CI)	Subtotals only
1.4.1 < 6 months	4	214	Mean Difference (IV, Random, 95% CI)	‐1.05 [‐4.39, 2.29]
1.4.2 < 12 months	3	179	Mean Difference (IV, Random, 95% CI)	‐1.99 [‐4.96, 0.97]
1.4.3 > 1 year	1	81	Mean Difference (IV, Random, 95% CI)	1.60 [0.95, 2.25]
1.5 Mental state ‐ general: mean endpoint overall symptom scales (high = poor) (combined time points)	20	1508	Std. Mean Difference (IV, Random, 95% CI)	‐0.27 [‐0.47, ‐0.08]
1.6 Mental state ‐ general: mean endpoint overall symptom scales (high = poor) (separated time points)	20		Std. Mean Difference (IV, Random, 95% CI)	Subtotals only
1.6.1 < 6 months	16	1071	Std. Mean Difference (IV, Random, 95% CI)	‐0.33 [‐0.57, ‐0.09]
1.6.2 < 12 months	13	987	Std. Mean Difference (IV, Random, 95% CI)	‐0.32 [‐0.60, ‐0.03]
1.6.3 > 1 year	7	636	Std. Mean Difference (IV, Random, 95% CI)	‐0.42 [‐1.03, 0.20]
1.7 Mental state ‐ general: number of participants with clinically important change in general mental state (combined time points)	4	301	Risk Ratio (IV, Random, 95% CI)	1.50 [0.95, 2.39]
1.8 Mental state ‐ general: number of participants with clinically important change in general mental state (separated time points)	4		Risk Ratio (IV, Random, 95% CI)	Subtotals only
1.8.1 < 6 months	3	221	Risk Ratio (IV, Random, 95% CI)	1.89 [1.26, 2.85]
1.8.2 < 12 months	3	221	Risk Ratio (IV, Random, 95% CI)	1.42 [0.78, 2.59]
1.8.3 > 1 year	1	80	Risk Ratio (IV, Random, 95% CI)	1.90 [1.01, 3.56]
1.9 Mental state ‐ specific: number of participants with clinically important change in positive symptoms (combined time points)	2	99	Risk Ratio (IV, Random, 95% CI)	1.38 [0.98, 1.93]
1.10 Mental state ‐ specific: number of participants with clinically important change in positive symptoms (separated time points)	2		Risk Ratio (IV, Random, 95% CI)	Subtotals only
1.10.1 > 6 months	2	99	Risk Ratio (IV, Random, 95% CI)	1.78 [1.11, 2.83]
1.10.2 > 12 months	1	80	Risk Ratio (IV, Random, 95% CI)	1.33 [0.93, 1.91]
1.10.3 > 1 year	1	80	Risk Ratio (IV, Random, 95% CI)	1.31 [1.01, 1.70]
1.11 Mental state ‐ specific: mean endpoint PANSS positive (high = poor)	15		Mean Difference (IV, Random, 95% CI)	Subtotals only
1.11.1 < 6 months	13	913	Mean Difference (IV, Random, 95% CI)	‐1.76 [‐3.09, ‐0.43]
1.11.2 < 12 months	9	684	Mean Difference (IV, Random, 95% CI)	‐2.03 [‐4.18, 0.11]
1.11.3 > 1 year	6	502	Mean Difference (IV, Random, 95% CI)	‐2.83 [‐5.44, ‐0.22]
1.12 Mental state ‐ specific: mean change PANSS positive (high = poor)	2		Mean Difference (IV, Random, 95% CI)	Subtotals only
1.12.1 < 6 months	1	15	Mean Difference (IV, Random, 95% CI)	0.80 [‐3.94, 5.54]
1.12.2 < 12 months	1	25	Mean Difference (IV, Random, 95% CI)	‐1.90 [‐4.39, 0.59]
1.12.3 > 1 year	1	22	Mean Difference (IV, Random, 95% CI)	2.60 [‐2.67, 7.87]
1.13 Mental state ‐ specific: mean endpoint BPRS positive (high = poor)	5		Mean Difference (IV, Random, 95% CI)	Subtotals only
1.13.1 < 6 months	4	253	Mean Difference (IV, Random, 95% CI)	‐0.85 [‐1.78, 0.08]
1.13.2 < 12 months	2	98	Mean Difference (IV, Random, 95% CI)	0.25 [‐1.62, 2.12]
1.13.3 > 1 year	2	143	Mean Difference (IV, Random, 95% CI)	‐0.20 [‐0.40, ‐0.01]
1.14 Mental state ‐ specific: mean endpoint SAPS (high = poor)	2		Mean Difference (IV, Random, 95% CI)	Subtotals only
1.14.1 < 6 months	2	105	Mean Difference (IV, Random, 95% CI)	‐0.13 [‐2.06, 1.81]
1.14.2 < 12 months	2	86	Mean Difference (IV, Random, 95% CI)	‐0.12 [‐2.40, 2.16]
1.15 Mental state ‐ specific: mean endpoint positive symptoms scales (high = poor) (combined time points)	22	1565	Std. Mean Difference (IV, Random, 95% CI)	‐0.22 [‐0.38, ‐0.06]
1.16 Mental state ‐ specific: mean endpoint positive symptoms scales (high = poor) (separated time points)	22		Std. Mean Difference (IV, Random, 95% CI)	Subtotals only
1.16.1 < 6 months	19	1271	Std. Mean Difference (IV, Random, 95% CI)	‐0.26 [‐0.43, ‐0.10]
1.16.2 < 12 months	13	868	Std. Mean Difference (IV, Random, 95% CI)	‐0.19 [‐0.44, 0.06]
1.16.3 > 1 year	8	645	Std. Mean Difference (IV, Random, 95% CI)	‐0.40 [‐0.76, ‐0.04]
1.17 Mental state ‐ specific: mean endpoint PANSS negative (high = poor)	15		Mean Difference (IV, Random, 95% CI)	Subtotals only
1.17.1 < 6 months	12	706	Mean Difference (IV, Random, 95% CI)	‐0.91 [‐1.70, ‐0.13]
1.17.2 < 12 months	9	682	Mean Difference (IV, Random, 95% CI)	‐0.94 [‐1.81, ‐0.08]
1.17.3 > 1 year	6	501	Mean Difference (IV, Random, 95% CI)	‐2.54 [‐4.04, ‐1.04]
1.18 Mental state ‐ specific: mean change PANSS negative (high = poor)	2		Mean Difference (IV, Random, 95% CI)	Subtotals only
1.18.1 < 6 months	1	15	Mean Difference (IV, Random, 95% CI)	‐1.30 [‐6.53, 3.93]
1.18.2 < 12 months	1	25	Mean Difference (IV, Random, 95% CI)	0.70 [‐2.72, 4.12]
1.18.3 > 1 year	1	22	Mean Difference (IV, Random, 95% CI)	‐0.50 [‐5.37, 4.37]
1.19 Mental state ‐ specific: mean endpoint BPRS negative (high = poor)	1		Mean Difference (IV, Random, 95% CI)	Subtotals only
1.19.1 < 6 months	1	98	Mean Difference (IV, Random, 95% CI)	0.50 [‐1.26, 2.26]
1.19.2 < 12 months	1	65	Mean Difference (IV, Random, 95% CI)	‐0.21 [‐1.80, 1.38]
1.20 Mental state ‐ specific: mean endpoint SANS (high = poor)	6		Mean Difference (IV, Random, 95% CI)	Subtotals only
1.20.1 < 6 months	4	196	Mean Difference (IV, Random, 95% CI)	‐3.28 [‐6.36, ‐0.20]
1.20.2 < 12 months	4	272	Mean Difference (IV, Random, 95% CI)	‐2.34 [‐6.18, 1.51]
1.20.3 > 1 year	2	166	Mean Difference (IV, Random, 95% CI)	‐2.60 [‐7.38, 2.18]
1.21 Mental state ‐ specific: mean endpoint BNSS (high = poor)	1		Mean Difference (IV, Random, 95% CI)	Subtotals only
1.21.1 < 6 months	1	79	Mean Difference (IV, Random, 95% CI)	‐2.60 [‐7.25, 2.05]
1.21.2 < 12 months	1	74	Mean Difference (IV, Random, 95% CI)	0.50 [‐4.32, 5.32]
1.22 Mental state ‐ specific: mean endpoint negative symptoms scale (high = poor) (combined time points)	22	1651	Std. Mean Difference (IV, Random, 95% CI)	‐0.20 [‐0.30, ‐0.11]
1.23 Mental state ‐ specific: mean endpoint negative symptoms scale (high = poor) (separated time points)	22		Std. Mean Difference (IV, Random, 95% CI)	Subtotals only
1.23.1 < 6 months	17	1000	Std. Mean Difference (IV, Random, 95% CI)	‐0.20 [‐0.32, ‐0.07]
1.23.2 < 12 months	14	1019	Std. Mean Difference (IV, Random, 95% CI)	‐0.19 [‐0.32, ‐0.06]
1.23.3 > 1 year	8	667	Std. Mean Difference (IV, Random, 95% CI)	‐0.42 [‐0.65, ‐0.18]
1.24 Mental state ‐ specific: mean endpoint BDI short form (high = poor)	2		Mean Difference (IV, Random, 95% CI)	Subtotals only
1.24.1 < 6 months	2	93	Mean Difference (IV, Random, 95% CI)	‐0.95 [‐3.58, 1.68]
1.24.2 < 12 months	1	33	Mean Difference (IV, Random, 95% CI)	1.20 [‐3.41, 5.81]
1.25 Mental state ‐ specific: mean endpoint BDI‐II (high = poor)	2		Mean Difference (IV, Random, 95% CI)	Subtotals only
1.25.1 < 6 months	1	56	Mean Difference (IV, Random, 95% CI)	‐1.70 [‐7.31, 3.91]
1.25.2 < 12 months	1	117	Mean Difference (IV, Random, 95% CI)	‐1.66 [‐6.77, 3.45]
1.25.3 > 1 year	2	146	Mean Difference (IV, Random, 95% CI)	‐0.69 [‐4.61, 3.22]
1.26 Mental state ‐ specific: mean endpoint HADS Depression (high = poor)	2		Mean Difference (IV, Random, 95% CI)	Subtotals only
1.26.1 < 6 months	2	59	Mean Difference (IV, Random, 95% CI)	‐2.95 [‐5.70, ‐0.20]
1.26.2 < 12 months	2	53	Mean Difference (IV, Random, 95% CI)	‐0.15 [‐2.94, 2.64]
1.27 Mental state ‐ specific: mean endpoint CDS (high = poor)	9		Mean Difference (IV, Random, 95% CI)	Subtotals only
1.27.1 < 6 months	8	337	Mean Difference (IV, Random, 95% CI)	0.28 [‐0.70, 1.26]
1.27.2 < 12 months	5	215	Mean Difference (IV, Random, 95% CI)	‐0.22 [‐1.37, 0.92]
1.27.3 > 1 year	3	134	Mean Difference (IV, Random, 95% CI)	0.71 [‐0.81, 2.24]
1.28 Mental state ‐ specific: mean change CDS (high = poor)	2		Mean Difference (IV, Random, 95% CI)	Subtotals only
1.28.1 > 6 months	1	15	Mean Difference (IV, Random, 95% CI)	0.30 [‐3.58, 4.18]
1.28.2 < 12 months	1	25	Mean Difference (IV, Random, 95% CI)	2.20 [‐1.37, 5.77]
1.28.3 > 1 year	1	22	Mean Difference (IV, Random, 95% CI)	2.10 [‐1.88, 6.08]
1.29 Mental state ‐ specific: mean endpoint MADRS (high = poor)	1	162	Mean Difference (IV, Random, 95% CI)	‐0.20 [‐2.07, 1.68]
1.29.1 < 12 months	1	81	Mean Difference (IV, Random, 95% CI)	‐2.50 [‐7.00, 2.00]
1.29.2 > 1 year	1	81	Mean Difference (IV, Random, 95% CI)	0.20 [‐0.54, 0.94]
1.30 Mental state ‐ specific: mean endpoint Hamilton Depression (high = poor)	2	467	Mean Difference (IV, Random, 95% CI)	‐1.55 [‐2.81, ‐0.30]
1.30.1 < 6 months	1	177	Mean Difference (IV, Random, 95% CI)	‐1.41 [‐2.97, 0.15]
1.30.2 < 12 months	1	290	Mean Difference (IV, Random, 95% CI)	‐1.81 [‐3.93, 0.31]
1.31 Mental state ‐ specific: mean endpoint CESD‐R (high = poor)	1	48	Mean Difference (IV, Random, 95% CI)	‐3.94 [‐13.24, 5.36]
1.31.1 < 6 months	1	48	Mean Difference (IV, Random, 95% CI)	‐3.94 [‐13.24, 5.36]
1.32 Mental state ‐ specific: mean endpoint depressive symptoms (high = poor) (combined time points)	18	1182	Std. Mean Difference (IV, Random, 95% CI)	‐0.13 [‐0.24, ‐0.01]
1.33 Mental state ‐ specific: mean endpoint depressive symptoms (high = poor) (separated time points)	18		Std. Mean Difference (IV, Random, 95% CI)	Subtotals only
1.33.1 < 6 months	14	714	Std. Mean Difference (IV, Random, 95% CI)	‐0.11 [‐0.27, 0.06]
1.33.2 < 12 months	11	789	Std. Mean Difference (IV, Random, 95% CI)	‐0.13 [‐0.27, 0.02]
1.33.3 > 1 year	5	313	Std. Mean Difference (IV, Random, 95% CI)	0.13 [‐0.10, 0.35]
1.34 Global state ‐ number of participants with relapse/exacerbations of psychosis (combined time points)	7	693	Risk Ratio (IV, Random, 95% CI)	0.82 [0.57, 1.18]
1.35 Global state ‐ number of participants with relapse/exacerbations of psychosis (separated time points)	7		Risk Ratio (IV, Random, 95% CI)	Subtotals only
1.35.1 < 6 months	3	172	Risk Ratio (IV, Random, 95% CI)	0.43 [0.09, 2.07]
1.35.2 < 12 months	5	363	Risk Ratio (IV, Random, 95% CI)	0.54 [0.22, 1.32]
1.35.3 > 1 year	5	601	Risk Ratio (IV, Random, 95% CI)	0.86 [0.62, 1.19]
1.36 Global state ‐ number of participants with remission (combined time points)	2	73	Risk Ratio (IV, Random, 95% CI)	1.16 [0.70, 1.91]
1.37 Global state ‐ number of participants with remission (separated time points)	2		Risk Ratio (IV, Random, 95% CI)	Subtotals only
1.37.1 < 6 months	1	48	Risk Ratio (IV, Random, 95% CI)	0.99 [0.52, 1.88]
1.37.2 < 12 months	1	25	Risk Ratio (IV, Random, 95% CI)	1.48 [0.67, 3.27]
1.37.3 > 1 year	1	25	Risk Ratio (IV, Random, 95% CI)	1.11 [0.45, 2.70]
1.38 Global state ‐ number of participants with clinically important change in global state	1		Risk Ratio (IV, Random, 95% CI)	Subtotals only
1.38.1 < 6 months	1	99	Risk Ratio (IV, Random, 95% CI)	1.58 [0.82, 3.01]
1.38.2 < 12 months	1	99	Risk Ratio (IV, Random, 95% CI)	1.53 [0.83, 2.83]
1.39 Global state ‐ mean endpoint CGI (high = poor) (separated time points)	1		Mean Difference (IV, Random, 95% CI)	Subtotals only
1.39.1 < 6 months	1	62	Mean Difference (IV, Random, 95% CI)	‐1.00 [‐1.37, ‐0.63]
1.39.2 < 12 months	1	62	Mean Difference (IV, Random, 95% CI)	‐0.80 [‐1.25, ‐0.35]
1.39.3 > 1 year	1	62	Mean Difference (IV, Random, 95% CI)	‐1.30 [‐1.85, ‐0.75]
1.40 Global state ‐ mean endpoint/change CGI‐S (high = poor) (separated time points)	3		Mean Difference (IV, Random, 95% CI)	Subtotals only
1.40.1 < 6 months	3	268	Mean Difference (IV, Random, 95% CI)	‐0.30 [‐0.65, 0.05]
1.40.2 < 12 months	1	42	Mean Difference (IV, Random, 95% CI)	‐0.23 [‐0.89, 0.43]
1.41 Global state ‐ mean endpoint CGI‐I (high = poor) (separated time points)	2		Mean Difference (IV, Random, 95% CI)	Subtotals only
1.41.1 < 6 months	2	220	Mean Difference (IV, Random, 95% CI)	‐0.29 [‐1.78, 1.20]
1.41.2 < 12 months	1	42	Mean Difference (IV, Random, 95% CI)	‐0.20 [‐0.86, 0.46]
1.42 Global state ‐ mean endpoint/change global state scales (high = poor) (combined time points)	4	329	Std. Mean Difference (IV, Random, 95% CI)	‐0.34 [‐0.67, ‐0.01]
1.43 Global state ‐ mean endpoint/change global state scales (high = poor) (separated time points)	4		Std. Mean Difference (IV, Random, 95% CI)	Subtotals only
1.43.1 < 6 months	4	330	Std. Mean Difference (IV, Random, 95% CI)	‐0.51 [‐1.02, 0.01]
1.43.2 < 12 months	2	104	Std. Mean Difference (IV, Random, 95% CI)	‐0.55 [‐1.19, 0.09]
1.43.3 > 1 year	1	62	Std. Mean Difference (IV, Random, 95% CI)	‐1.12 [‐1.66, ‐0.58]
1.44 Service use ‐ admission to hospital (combined time points)	10	1121	Risk Ratio (IV, Random, 95% CI)	1.00 [0.80, 1.25]
1.45 Service use ‐ admission to hospital (separated time points)	10		Risk Ratio (IV, Random, 95% CI)	Subtotals only
1.45.1 < 6 months	4	240	Risk Ratio (IV, Random, 95% CI)	1.36 [0.42, 4.41]
1.45.2 < 12 months	5	590	Risk Ratio (IV, Random, 95% CI)	0.90 [0.56, 1.45]
1.45.3 > 1 year	5	586	Risk Ratio (IV, Random, 95% CI)	0.92 [0.68, 1.23]
1.46 Service use ‐ number of days in hospital (combined time points)	7	571	Mean Difference (IV, Random, 95% CI)	‐8.18 [‐24.74, 8.38]
1.47 Service use ‐ number of days in hospital (separated time points)	7		Mean Difference (IV, Random, 95% CI)	Subtotals only
1.47.1 < 6 months	3	98	Mean Difference (IV, Random, 95% CI)	‐21.56 [‐60.13, 17.02]
1.47.2 < 12 months	4	488	Mean Difference (IV, Random, 95% CI)	2.95 [‐9.42, 15.33]
1.47.3 > 1 year	4	199	Mean Difference (IV, Random, 95% CI)	‐12.39 [‐36.62, 11.84]
1.48 Leaving the study early ‐ for any reason ‐ overall acceptability (combined time points)	25	2242	Risk Ratio (IV, Random, 95% CI)	0.87 [0.72, 1.05]
1.49 Leaving the study early ‐ for any reason ‐ overall acceptability (separated time points)	25		Risk Ratio (IV, Random, 95% CI)	Subtotals only
1.49.1 < 6 months	21	1655	Risk Ratio (IV, Random, 95% CI)	0.87 [0.70, 1.07]
1.49.2 < 12 months	15	1396	Risk Ratio (IV, Random, 95% CI)	0.86 [0.66, 1.13]
1.49.3 > 1 year	10	1076	Risk Ratio (IV, Random, 95% CI)	0.81 [0.69, 0.96]
1.50 Functioning ‐ mean endpoint GAF (high = good)	10		Mean Difference (IV, Random, 95% CI)	Subtotals only
1.50.1 < 6 months	8	570	Mean Difference (IV, Random, 95% CI)	‐4.62 [‐9.35, 0.10]
1.50.2 < 12 months	7	592	Mean Difference (IV, Random, 95% CI)	‐2.91 [‐8.15, 2.33]
1.50.3 > 1 year	4	196	Mean Difference (IV, Random, 95% CI)	‐9.63 [‐18.70, ‐0.56]
1.51 Functioning ‐ mean change GAF (high = good)	1		Mean Difference (IV, Random, 95% CI)	Subtotals only
1.51.1 < 12 months	1	25	Mean Difference (IV, Random, 95% CI)	‐0.30 [‐8.18, 7.58]
1.51.2 > 1 year	1	22	Mean Difference (IV, Random, 95% CI)	‐0.40 [‐11.39, 10.59]
1.52 Functioning ‐ mean endpoint SOFAS (high = good)	5		Mean Difference (IV, Random, 95% CI)	Subtotals only
1.52.1 < 6 months	4	196	Mean Difference (IV, Random, 95% CI)	‐2.78 [‐6.13, 0.57]
1.52.2 < 12 months	3	151	Mean Difference (IV, Random, 95% CI)	0.02 [‐4.82, 4.87]
1.52.3 > 1 year	2	143	Mean Difference (IV, Random, 95% CI)	4.51 [‐5.39, 14.41]
1.53 Functioning ‐ mean endpoint PSP (high = good)	3		Mean Difference (IV, Random, 95% CI)	Subtotals only
1.53.1 < 6 months	3	153	Mean Difference (IV, Random, 95% CI)	‐4.59 [‐9.73, 0.54]
1.53.2 < 12 months	2	122	Mean Difference (IV, Random, 95% CI)	‐4.80 [‐9.66, 0.07]
1.53.3 > 1 year	1	80	Mean Difference (IV, Random, 95% CI)	‐7.88 [‐13.85, ‐1.91]
1.54 Functioning ‐ mean endpoint FAST (high = poor)	1		Mean Difference (IV, Random, 95% CI)	Subtotals only
1.54.1 < 6 months	1	177	Mean Difference (IV, Random, 95% CI)	‐4.90 [‐9.57, ‐0.23]
1.55 Functioning ‐ mean endpoint FESFS (high = good)	1		Mean Difference (IV, Random, 95% CI)	Subtotals only
1.55.1 < 6 months	1	9	Mean Difference (IV, Random, 95% CI)	‐1.15 [‐4.91, 2.61]
1.55.2 < 12 months	1	3	Mean Difference (IV, Random, 95% CI)	0.27 [‐7.22, 7.76]
1.56 Functioning ‐ mean endpoint functioning scale (high = good) (combined time points)	18	1241	Std. Mean Difference (IV, Random, 95% CI)	‐0.23 [‐0.42, ‐0.05]
1.57 Functioning ‐ mean endpoint functioning scale (high = good) (separated time points)	18		Std. Mean Difference (IV, Random, 95% CI)	Subtotals only
1.57.1 < 6 months	15	898	Std. Mean Difference (IV, Random, 95% CI)	‐0.32 [‐0.54, ‐0.10]
1.57.2 < 12 months	13	868	Std. Mean Difference (IV, Random, 95% CI)	‐0.18 [‐0.43, 0.07]
1.57.3 > 1 year	7	419	Std. Mean Difference (IV, Random, 95% CI)	‐0.15 [‐1.16, 0.87]
1.58 Quality of life ‐ mean endpoint WHO QoL‐bref (high = good)	3		Mean Difference (IV, Random, 95% CI)	Subtotals only
1.58.1 < 6 months	2	87	Mean Difference (IV, Random, 95% CI)	‐3.15 [‐10.79, 4.49]
1.58.2 < 12 months	3	166	Mean Difference (IV, Random, 95% CI)	‐0.24 [‐1.06, 0.57]
1.59 Quality of life ‐ mean endpoint EQ5d (WHO) (high = good)	3	209	Std. Mean Difference (IV, Random, 95% CI)	‐0.38 [‐0.66, ‐0.10]
1.59.1 < 6 months	3	125	Std. Mean Difference (IV, Random, 95% CI)	‐0.50 [‐0.86, ‐0.13]
1.59.2 < 12 months	2	84	Std. Mean Difference (IV, Random, 95% CI)	‐0.21 [‐0.64, 0.22]
1.60 Quality of life ‐ mean endpoint QLS (high = good)	2		Mean Difference (IV, Random, 95% CI)	Subtotals only
1.60.1 < 6 months	2	67	Mean Difference (IV, Random, 95% CI)	1.63 [‐6.44, 9.69]
1.61 Quality of life ‐ mean endpoint MSQoL (high = good)	1		Mean Difference (IV, Random, 95% CI)	Subtotals only
1.61.1 < 12 months	1	22	Mean Difference (IV, Random, 95% CI)	‐5.80 [‐18.30, 6.70]
1.61.2 > 1 year	1	16	Mean Difference (IV, Random, 95% CI)	‐7.50 [‐24.63, 9.63]
1.62 Quality of life ‐ mean change MSQoL (high = good)	1		Mean Difference (IV, Random, 95% CI)	Subtotals only
1.62.1 < 12 months	1	25	Mean Difference (IV, Random, 95% CI)	‐3.70 [‐15.01, 7.61]
1.62.2 > 1 year	1	22	Mean Difference (IV, Random, 95% CI)	‐4.30 [‐14.09, 5.49]
1.63 Quality of life ‐ MANSA (high = good)	1		Mean Difference (IV, Random, 95% CI)	Subtotals only
1.63.1 < 6 months	1	36	Mean Difference (IV, Random, 95% CI)	2.90 [‐2.95, 8.75]
1.63.2 < 12 months	1	36	Mean Difference (IV, Random, 95% CI)	‐0.20 [‐7.72, 7.32]
1.64 Quality of life ‐ mean endpoint quality of life scale (high = good) (combined time points)	9	386	Std. Mean Difference (IV, Random, 95% CI)	‐0.13 [‐0.33, 0.07]
1.65 Quality of life ‐ mean endpoint quality of life scale (high = good) (separated time points)	9		Std. Mean Difference (IV, Random, 95% CI)	Subtotals only
1.65.1 < 6 months	7	296	Std. Mean Difference (IV, Random, 95% CI)	‐0.24 [‐0.51, 0.02]
1.65.2 < 12 months	7	308	Std. Mean Difference (IV, Random, 95% CI)	‐0.16 [‐0.39, 0.07]
1.65.3 > 1 year	1	16	Std. Mean Difference (IV, Random, 95% CI)	‐0.40 [‐1.40, 0.60]
1.66 Behaviour ‐ occurence of violent incidents	2		Risk Ratio (IV, Random, 95% CI)	Subtotals only
1.66.1 < 12 months	2	92	Risk Ratio (IV, Random, 95% CI)	0.96 [0.15, 6.20]
1.67 Behaviour ‐ participants with self‐injury	2		Risk Ratio (IV, Random, 95% CI)	Subtotals only
1.67.1 < 12 months	2	92	Risk Ratio (IV, Random, 95% CI)	1.89 [0.95, 3.80]
1.68 Behaviour ‐ participants with suicide attempts	2		Risk Ratio (IV, Random, 95% CI)	Subtotals only
1.68.1 < 12 months	2	68	Risk Ratio (IV, Random, 95% CI)	0.97 [0.10, 9.41]
1.68.2 > 1 year	1	25	Risk Ratio (IV, Random, 95% CI)	0.31 [0.01, 6.94]
1.69 Adverse events ‐ participants with at least one adverse event	1		Risk Ratio (IV, Random, 95% CI)	Subtotals only
1.70 Adverse events ‐ specific: Non Compliance	2		Risk Ratio (IV, Random, 95% CI)	Subtotals only
1.70.1 < 6 months	2	190	Risk Ratio (IV, Random, 95% CI)	1.06 [0.47, 2.37]
1.70.2 < 12 months	1	80	Risk Ratio (IV, Random, 95% CI)	0.33 [0.04, 3.07]
1.70.3 > 1 year	1	80	Risk Ratio (IV, Random, 95% CI)	0.33 [0.04, 3.07]
1.71 Adverse events ‐ specific: Deterioration of symptoms	4		Risk Ratio (IV, Random, 95% CI)	Subtotals only
1.71.1 < 6 months	2	92	Risk Ratio (IV, Random, 95% CI)	1.08 [0.17, 7.03]
1.71.2 < 12 months	3	117	Risk Ratio (IV, Random, 95% CI)	0.58 [0.12, 2.79]
1.71.3 > 1 year	2	87	Risk Ratio (IV, Random, 95% CI)	0.53 [0.27, 1.02]
1.72 Mortality ‐ overall mortality (combined time points)	4	900	Risk Ratio (IV, Random, 95% CI)	1.22 [0.41, 3.66]
1.73 Mortality ‐ overall mortality (separated time points)	4		Risk Ratio (IV, Random, 95% CI)	Subtotals only
1.73.1 < 12 months	2	481	Risk Ratio (IV, Random, 95% CI)	0.97 [0.18, 5.13]
1.73.2 > 1 year	2	419	Risk Ratio (IV, Random, 95% CI)	1.38 [0.20, 9.60]
1.74 Mortality ‐ due to natural causes	2	635	Risk Ratio (IV, Random, 95% CI)	2.00 [0.22, 17.99]
1.74.1 < 12 months	1	326	Risk Ratio (IV, Random, 95% CI)	0.21 [0.01, 5.13]
1.74.2 > 1 year	1	309	Risk Ratio (IV, Random, 95% CI)	6.12 [0.64, 58.07]
1.75 Mortality ‐ due to suicide	2	388	Risk Ratio (IV, Random, 95% CI)	1.08 [0.05, 24.00]
1.75.1 < 12 months	1	326	Risk Ratio (IV, Random, 95% CI)	0.21 [0.01, 5.13]
1.75.2 > 1 year	1	62	Risk Ratio (IV, Random, 95% CI)	5.00 [0.25, 100.08]
1.76 Sensitivity analysis ‐ Blinding of outcome assessor	19	1229	Std. Mean Difference (IV, Random, 95% CI)	‐0.28 [‐0.50, ‐0.06]
1.77 Sensitivity analysis ‐ Assumptions for missing data	15	1215	Std. Mean Difference (IV, Random, 95% CI)	‐0.29 [‐0.52, ‐0.05]
1.78 Sensitivity analysis ‐ Risk of Bias	15	1006	Std. Mean Difference (IV, Random, 95% CI)	‐0.29 [‐0.56, ‐0.01]
1.79 Sensitivity analysis ‐ Imputed values	18	1214	Std. Mean Difference (IV, Random, 95% CI)	‐0.28 [‐0.50, ‐0.05]
1.80 Sensitivity analysis ‐ Fixed‐effect	20	1508	Std. Mean Difference (IV, Fixed, 95% CI)	‐0.25 [‐0.35, ‐0.14]
1.81 Sensitivity analysis excluding outliers	19	1446	Std. Mean Difference (IV, Random, 95% CI)	‐0.19 [‐0.30, ‐0.09]
1.82 Subgroup analysis ‐ Different types of CBT	20	1508	Std. Mean Difference (IV, Random, 95% CI)	‐0.27 [‐0.47, ‐0.08]
1.82.1 CBT	11	775	Std. Mean Difference (IV, Random, 95% CI)	‐0.19 [‐0.35, ‐0.04]
1.82.2 CBT integrated with other psychosocial interventions	4	494	Std. Mean Difference (IV, Random, 95% CI)	‐0.87 [‐1.73, ‐0.01]
1.82.3 CBT and Motivational Interviewing for cannabis use cessation	3	139	Std. Mean Difference (IV, Random, 95% CI)	‐0.04 [‐0.38, 0.30]
1.82.4 Special types of CBT	2	100	Std. Mean Difference (IV, Random, 95% CI)	‐0.26 [‐0.82, 0.31]
1.83 Subgroup analysis ‐ Modality of treatment	20	1508	Std. Mean Difference (IV, Random, 95% CI)	‐0.27 [‐0.47, ‐0.08]
1.83.1 Individual CBT	17	1246	Std. Mean Difference (IV, Random, 95% CI)	‐0.29 [‐0.53, ‐0.05]
1.83.2 Group CBT	3	262	Std. Mean Difference (IV, Random, 95% CI)	‐0.18 [‐0.42, 0.07]
1.84 Subgroup analysis ‐ Method of delivery	20	1508	Std. Mean Difference (IV, Random, 95% CI)	‐0.27 [‐0.47, ‐0.08]
1.84.1 Face to face	19	1478	Std. Mean Difference (IV, Random, 95% CI)	‐0.26 [‐0.47, ‐0.06]
1.84.2 Smartphone app	1	30	Std. Mean Difference (IV, Random, 95% CI)	‐0.60 [‐1.39, 0.20]
1.85 Subgroup analysis ‐ Control group	20	1508	Std. Mean Difference (IV, Random, 95% CI)	‐0.27 [‐0.47, ‐0.08]
1.85.1 CBT compared to TAU	15	1211	Std. Mean Difference (IV, Random, 95% CI)	‐0.34 [‐0.59, ‐0.09]
1.85.2 CBT compared to other psychosocial interventions	7	297	Std. Mean Difference (IV, Random, 95% CI)	‐0.08 [‐0.32, 0.17]

1.1. Analysis.

Comparison 1: CBT plus standard care versus control, Outcome 1: Mental state ‐ general: mean endpoint PANSS total (high = poor)

1.2. Analysis.

Comparison 1: CBT plus standard care versus control, Outcome 2: Mental state ‐ general: mean change PANSS total (high = poor)

1.3. Analysis.

Comparison 1: CBT plus standard care versus control, Outcome 3: Mental state ‐ general: mean endpoint PSYRATS (high = poor)

1.4. Analysis.

Comparison 1: CBT plus standard care versus control, Outcome 4: Mental state ‐ general: mean endpoint BPRS total (high = poor)

1.11. Analysis.

Comparison 1: CBT plus standard care versus control, Outcome 11: Mental state ‐ specific: mean endpoint PANSS positive (high = poor)

1.12. Analysis.

Comparison 1: CBT plus standard care versus control, Outcome 12: Mental state ‐ specific: mean change PANSS positive (high = poor)

1.13. Analysis.

Comparison 1: CBT plus standard care versus control, Outcome 13: Mental state ‐ specific: mean endpoint BPRS positive (high = poor)

1.14. Analysis.

Comparison 1: CBT plus standard care versus control, Outcome 14: Mental state ‐ specific: mean endpoint SAPS (high = poor)

1.17. Analysis.

Comparison 1: CBT plus standard care versus control, Outcome 17: Mental state ‐ specific: mean endpoint PANSS negative (high = poor)

1.18. Analysis.

Comparison 1: CBT plus standard care versus control, Outcome 18: Mental state ‐ specific: mean change PANSS negative (high = poor)

1.19. Analysis.

Comparison 1: CBT plus standard care versus control, Outcome 19: Mental state ‐ specific: mean endpoint BPRS negative (high = poor)

1.20. Analysis.

Comparison 1: CBT plus standard care versus control, Outcome 20: Mental state ‐ specific: mean endpoint SANS (high = poor)

1.21. Analysis.

Comparison 1: CBT plus standard care versus control, Outcome 21: Mental state ‐ specific: mean endpoint BNSS (high = poor)

1.24. Analysis.

Comparison 1: CBT plus standard care versus control, Outcome 24: Mental state ‐ specific: mean endpoint BDI short form (high = poor)

1.25. Analysis.

Comparison 1: CBT plus standard care versus control, Outcome 25: Mental state ‐ specific: mean endpoint BDI‐II (high = poor)

1.26. Analysis.

Comparison 1: CBT plus standard care versus control, Outcome 26: Mental state ‐ specific: mean endpoint HADS Depression (high = poor)

1.27. Analysis.

Comparison 1: CBT plus standard care versus control, Outcome 27: Mental state ‐ specific: mean endpoint CDS (high = poor)

1.28. Analysis.

Comparison 1: CBT plus standard care versus control, Outcome 28: Mental state ‐ specific: mean change CDS (high = poor)

1.29. Analysis.

Comparison 1: CBT plus standard care versus control, Outcome 29: Mental state ‐ specific: mean endpoint MADRS (high = poor)

1.30. Analysis.

Comparison 1: CBT plus standard care versus control, Outcome 30: Mental state ‐ specific: mean endpoint Hamilton Depression (high = poor)

1.31. Analysis.

Comparison 1: CBT plus standard care versus control, Outcome 31: Mental state ‐ specific: mean endpoint CESD‐R (high = poor)

1.39. Analysis.

Comparison 1: CBT plus standard care versus control, Outcome 39: Global state ‐ mean endpoint CGI (high = poor) (separated time points)

1.40. Analysis.

Comparison 1: CBT plus standard care versus control, Outcome 40: Global state ‐ mean endpoint/change CGI‐S (high = poor) (separated time points)

1.41. Analysis.

Comparison 1: CBT plus standard care versus control, Outcome 41: Global state ‐ mean endpoint CGI‐I (high = poor) (separated time points)

1.50. Analysis.

Comparison 1: CBT plus standard care versus control, Outcome 50: Functioning ‐ mean endpoint GAF (high = good)

1.51. Analysis.

Comparison 1: CBT plus standard care versus control, Outcome 51: Functioning ‐ mean change GAF (high = good)

1.52. Analysis.

Comparison 1: CBT plus standard care versus control, Outcome 52: Functioning ‐ mean endpoint SOFAS (high = good)

1.53. Analysis.

Comparison 1: CBT plus standard care versus control, Outcome 53: Functioning ‐ mean endpoint PSP (high = good)

1.54. Analysis.

Comparison 1: CBT plus standard care versus control, Outcome 54: Functioning ‐ mean endpoint FAST (high = poor)

1.55. Analysis.

Comparison 1: CBT plus standard care versus control, Outcome 55: Functioning ‐ mean endpoint FESFS (high = good)

1.58. Analysis.

Comparison 1: CBT plus standard care versus control, Outcome 58: Quality of life ‐ mean endpoint WHO QoL‐bref (high = good)

1.59. Analysis.

Comparison 1: CBT plus standard care versus control, Outcome 59: Quality of life ‐ mean endpoint EQ5d (WHO) (high = good)

1.60. Analysis.

Comparison 1: CBT plus standard care versus control, Outcome 60: Quality of life ‐ mean endpoint QLS (high = good)

1.61. Analysis.

Comparison 1: CBT plus standard care versus control, Outcome 61: Quality of life ‐ mean endpoint MSQoL (high = good)

1.62. Analysis.

Comparison 1: CBT plus standard care versus control, Outcome 62: Quality of life ‐ mean change MSQoL (high = good)

1.63. Analysis.

Comparison 1: CBT plus standard care versus control, Outcome 63: Quality of life ‐ MANSA (high = good)

Characteristics of studies

Characteristics of included studies [ordered by study ID]

Barrowclough 2014.

Study characteristics
Methods	Study design: randomised controlled trial Study grouping: parallel group Antipsychotic medication allowed: yes Duration in weeks: 39 weeks maximum treatment, 78 weeks maximum follow‐up Number of study arms: 3 Type of blinding: rater blind Type of data analysis for overall efficacy: ITT Number of sites: 5 Number of countries: 1 Country: England Sponsorship source: public Trial registration ID: ISRCTN88275061 Publication year: 2014 Researcher's allegiance: The authors developed the manual for the intervention group.
Participants	Diagnosis: non‐affective psychotic disorder, cannabis dependence or abuse; diagnostic criteria: DSM‐IV Current clinical state: NA Definition of stability: NA Inclusion criteria: "Participants were recruited from Early Intervention Services in five mental health trusts in the North West of England. Inclusion criteria were: aged 16–35 years; meeting Diagnostic and Statistical Manual of Mental Disorders, fourth edition (DSM‐IV) criteria for non‐affective psychotic disorder; DSM‐IV diagnosis of cannabis dependence or abuse; cannabis use of at least 1 day per week in at least half the weeks in the 3 months prior to assessment; having stable accommodation; sufficient English to complete the assessments; no significant history of organic factors implicated in the aetiology of psychotic symptoms; and able to give informed consent." Exclusion criteria: NA Setting: NA Special subgroup: drug abuse N: 110 Gender: 98 men, 12 women Age: mean 24.2 years CBT plus standard care arm (brief): participants total: 38, participants male: 34, participants female: 4, age: mean 24.9 years (SD = 5.6), age range NA years, duration of illness: median 13.4 years, % taking antipsychotics: NA, % diagnosis of schizophrenia: 52.6, % diagnosis of schizoaffective disorder: 13.2, % sub‐induced disorder: 5.3, baseline PANSS total M = 66.1, SD = 9.2, N = 38 CBT plus standard care arm (long): participants total: 37, participants male: 34, participants female: 3, age: mean 24.1 years (SD = 5.4), age range NA years, duration of illness: median 17.5 years, % taking antipsychotics: NA, % diagnosis of schizophrenia: 43.2, % diagnosis of schizoaffective disorder: 13.5, % sub‐induced disorder: 2.7, baseline PANSS total M = 61.6, SD = 9.9, N = 37 Control arm: participants total: 35, participants male: 30, participants female: 5, age: mean 23.4 years (SD = 3.8), age range NA years, duration of illness: median 17.2 years, % taking antipsychotics: NA, % diagnosis of schizophrenia: 51.4, % diagnosis of schizoaffective disorder: 8.6, % sub‐induced disorder: 8.6, baseline PANSS total M = 61.7, SD = 6.8, N = 35
Interventions	1.a Integrated MI‐CBT (brief) plus standard care. N = 38 Description of treatment: "Phase 1 of the intervention – ‘motivation building’ – concerned engaging the patient, eliciting and understanding their perspective in relation to life goals, and exploring and resolving ambivalence so as to facilitate motivation for change. Information and feedback from assessments were incorporated into this process, utilising a motivational style, to support the formulation of a shared understanding in relation to each person’s concerns, their cannabis use and mental health difficulties. Adaptations from the 2010 intervention included youth‐friendly and cannabis‐focused information materials in the form of purpose‐made DVDs and a cannabis information booklet produced by lifeline.org.uk (http://www.exchangesupplies.org/shopdisp_A37.php). In phase 2 of the intervention, a plan for change was developed. Where the person was open to change in cannabis use, CBT techniques from both the psychosis and substance use evidence base were used to help the patient implement and maintain changes. For those not identifying substances as problematic, the intervention was sufficiently flexible to allow therapists to work with other patient‐led problems. In such cases, the therapist would continue to assist the patient to link substance use to their concerns using MI techniques. Both interventions attempted to progress through both phases. However, the long intervention was designed to allow more time in phase 2 to develop the change plan and particularly the use of CBT within the plan. Progress was communicated to the participants’ care co‐ordinator at two liaison meetings attended by both the therapist and the participant. Standard care from the Early Intervention Services involved in the study is compliant with the Mental Health Policy Implementation Guide (Department of Health 2001) and included intensive case management and crisis response." Number of sessions: 12 planned, 10.25 delivered (median) Sessions frequency and duration: 4 per month Therapist: NA Therapist experience: experts Supervision: yes Modality: individual Delivery: face‐to‐face Treatment manualised: yes ‐ manual by Barrowclough 1.b Integrated MI‐CBT (long) plus standard care. N = 37 Description of treatment: "Phase 1 of the intervention – ‘motivation building’ – concerned engaging the patient, eliciting and understanding their perspective in relation to life goals, and exploring and resolving ambivalence so as to facilitate motivation for change. Information and feedback from assessments were incorporated into this process, utilising a motivational style, to support the formulation of a shared understanding in relation to each person’s concerns, their cannabis use and mental health difficulties. Adaptations from the 2010 intervention included youth‐friendly and cannabis‐focused information materials in the form of purpose‐made DVDs and a cannabis information booklet produced by lifeline.org.uk (http://www.exchangesupplies.org/shopdisp_A37.php). In phase 2 of the intervention, a plan for change was developed. Where the person was open to change in cannabis use, CBT techniques from both the psychosis and substance‐use evidence base were used to help the patient implement and maintain changes. For those not identifying substances as problematic, the intervention was sufficiently flexible to allow therapists to work with other patient‐led problems. In such cases, the therapist would continue to assist the patient to link substance use to their concerns using MI techniques. Both interventions attempted to progress through both phases. However, the long intervention was designed to allow more time in phase 2 to develop the change plan and particularly the use of CBT within the plan. Progress was communicated to the participants’ care co‐ordinator at two liaison meetings attended by both the therapist and the participant. Standard care from the Early Intervention Services involved in the study is compliant with the Mental Health Policy Implementation Guide (Department of Health 2001) and included intensive case management and crisis response." Number of sessions: 24 planned, 11 delivered (median) Sessions frequency and duration: 3 per month Therapist: NA Therapist experience: experts Supervision: yes Modality: individual Delivery: face‐to‐face Treatment manualised: yes ‐ manual by Barrowclough 2. Control (Treatment as usual). N = 35 Description of treatment: intensive case management and crisis response Number of sessions: NA Sessions frequency and duration: NA Therapist: NA Therapist experience: NA Supervision: NA Modality: NA Delivery: NA Treatment manualised: NA
Outcomes	Mental state ‐ general ‐ PANNS total endpoint (< 6 months, < 12 months, > 1 year) Mental state ‐ specific ‐ PANNS positive endpoint (< 6 months, < 12 months, > 1 year) Mental state ‐ specific ‐ PANNS negative endpoint (< 6 months, < 12 months, > 1 year) Mental state ‐ specific ‐ Calgary Depression Scale endpoint (< 6 months, < 12 months, > 1 year) Global state ‐ Number of participants that relapsed (< 12 months, > 1 year) Service use ‐ Admission to hospital (< 12 months, > 1 year) Leaving the study early ‐ for any reason (dropout) (< 6 months, < 12 months, > 1 year) Functioning ‐ GAF endpoint (< 6 months, < 12 months, > 1 year) Adverse events ‐ Non compliance (< 6 months) Mortality ‐ Overall mortality (> 1 year)
Notes	We contacted the authors but received no reply.

Bucci 2018 (Actissist).

Study characteristics
Methods	Study design: randomised controlled trial Study grouping: parallel group Antipsychotic medication allowed: yes Duration in weeks: 12 weeks treatment phase, 22 weeks follow‐up Number of study arms: 2 Type of blinding: rater blind Type of data analysis for overall efficacy: probably completers Number of sites: NA Number of countries: 1 Country: UK Sponsorship source: public Trial registration ID: ISRCTN34966555 Publication year: 2018 Researcher's allegiance: The authors developed the app for both intervention and control.
Participants	Diagnosis: initial episode of psychosis; diagnostic criteria: NA Current clinical state: NA Definition of stability: NA Inclusion criteria: "Eligibility criteria were: (1) in current contact with an EIS in the North West of England; (2) capacity to provide informed consent; and (3) English language proficient. EIS are multidisciplinary community mental health services that provide psychosocial and pharmacological treatment and support to people in their first 3 years of their initial episode of psychosis." Exclusion criteria: "Exclusion criteria were: (1) aged less than 16 years at point of recruitment; (2) not capable of giving informed consent; (3) non‐English proficient; and (4) inpatient at point of recruitment." Setting: outpatient Special subgroup: first‐episode patients N: 36 Gender: 18 men, 18 women Age: NA CBT plus standard care arm: 24 participants, 15 male (16.25%), age: mean NA years, age range > 16 years, duration of illness: NA, duration of untreated psychosis: NA, % taking antipsychotics: NA, % diagnosis schizophrenia: NA, baseline PANSS total M = 65.9, SD = 12.9, N = 24 Control arm: 12 participants, 3 male (25%), age: mean NA years, age range > 16 years, duration of illness: NA, duration of untreated psychosis: NA, % taking antipsychotics: NA, % diagnosis schizophrenia: NA, baseline PANSS total M = 64.6, SD = 11.1, N = 12
Interventions	1. CBT plus standard care. N = 24 Description of treatment: "Actissist is a DHI that the user can engage with spontaneously or in response to being prompted. It then collects responses from the user and wirelessly uploads user responses to a server. Actissist is divided in 2 parts, although presented as a single app. Firstly, at 3 pseudo‐randomised time points per day, 6 days a week between 10.00 and 22.00, an auditory alert followed by a visual prompt is emitted from the app inviting participants to access the app. The notifications persist on the handset (i.e. no time out) until such point as they are accepted, dismissed, “snoozed” (up to 15 min), or another notification is received. The notifications serve merely as a reminder; the app also allows self‐initiated use at any point. If a user accepts a notification or initiates use, they are invited to select an intervention domain(s) and then complete a series of self‐assessment questions structured as question‐answer exchanges that focus on cognitive appraisals, belief conviction, emotions and associated behaviours. Depending on the appraisal selected, the exchange is followed by normalising messages and cognitive or behavioural strategies aimed at suggesting ways of coping with distressing experiences. Multiple messages and images associated with each exchange minimise boredom and repetition within the app. Alternatively, participants can report that they have had “no problems like this” since their last notification or (self‐initiated) interaction. Part 2 includes a menu of multimedia options that act in a stand‐alone fashion designed to complement and support the feedback from the intervention domains. This supplementary content contains information and activities including relaxation and mindfulness exercises, recovery stories (videos), a range of fact sheets (e.g. low mood, anxiety, self‐esteem), external links to web‐related content (e.g. TED talks), daily diary, and emergency contact resources. Furthermore, a graphical summary of data points entered over the previous 7 days allows users to track distressing experiences to support active self‐management of symptoms and shared decision‐making about treatment with clinicians. Users can customise the aesthetics of the Actissist interface; for example, personally meaningful images from the smartphone’ s local storage can be set as wallpaper to facilitate positive mood induction." "TAU involved regular clinician meetings, medication, risk monitoring, and psychosocial interventions. Actissist and ClinTouch are standalone apps that do not link with services." Number of sessions: 3 prompts per day, 6 days per week, for 12 weeks Sessions frequency and duration: NA Therapist: NA Therapist experience: NA Supervision: no Modality: individual Delivery: smartphone app Treatment manualised: yes 2. Control (Clintouch). N = 12 Description of treatment: "The ClinTouch app is a symptom‐monitoring app that triggers, collects, and wirelessly uploads symptom data to a server. As in the treatment condition, the app emits an alarm prompting participants to access the app at 3 pseudo‐randomised time points per day, 6 days a week between 10.00 and 22.00 for 12 weeks alongside usual treatment. The ClinTouch protocol is outlined in detail in Palmier‐Claus 2012; although, the number of prompts was altered for parity with Actissist alerts, such that participants submit one‐and‐a‐half data points daily with 10 branching items covering positive psychotic symptoms, anxiety, and mood. As each full data point was collected over 2 separate alerts, this equates to having received 3 alerts every day. The alert invites participants to use a touchscreen slider to rate the severity of 12 individual symptoms on a 1–7 scale. This takes an average of 70 s and the data are wirelessly uploaded to a secure server. Symptom items have been validated against corresponding items on the PANSS (Kay 1987). The aesthetics and interface mirror the Actissist interface. However, unlike the Actissist app, ClinTouch does not facilitate self‐initiated access; data entries must be in response to a notification. The notifications time out 30 mins after receipt, whereby the notification is no longer visible and the ClinTouch items are no longer accessible." "TAU involved regular clinician meetings, medication, risk monitoring, and psychosocial interventions. Actissist and ClinTouch are standalone apps that do not link with services." Number of sessions: 3 prompts per day, 6 days per week, for 12 weeks Sessions frequency and duration: NA Therapist: NA Therapist experience: NA Supervision: no Modality: individual Delivery: smartphone app Treatment manualised: yes
Outcomes	Mental state ‐ general ‐ PANSS total endpoint (< 6 months) Mental state ‐ specific ‐ PANSS positive endpoint (< 6 months) Mental state ‐ specific ‐ PANSS negative endpoint (< 6 months) Mental state ‐ specific ‐ Calgary Depression Scale endpoint (< 6 months) Leaving the study early ‐ for any reason (dropout) (< 6 months) Functioning ‐ GAF endpoint (< 6 months) Functioning ‐ PSP endpoint (< 6 months) Quality of Life ‐ EQ‐5D (WHO) endpoint (< 6 months)
Notes	We contacted the authors and obtained additional information.

Edwards 2006.

Study characteristics
Methods	Study design: randomised controlled trial Study grouping: parallel group Antipsychotic medication allowed: yes Duration in weeks: 39 (13 intervention + 26 follow‐up) Number of study arms: 2 Type of blinding: rater blind Type of data analysis for overall efficacy: ITT Number of sites: 1 (no information on number of academic sites) Number of countries: 1 Country: Australia Sponsorship source: NA Trial registration ID: NA Publication year: 2006 Researcher's allegiance: The authors developed the manual for the intervention group.
Participants	Diagnosis: psychotic disorder; diagnostic criteria: DSM‐IV Current clinical state: first episode, clinically stabilised Definition of stability: NA Inclusion criteria: "Study inclusion criteria were a DSM‐IV (APA 1994) diagnosis of a psychotic disorder (i.e. schizophrenia, schizophreniform, schizoaffective, delusional disorder, bipolar disorder, major depressive disorder with psychotic features, psychosis not otherwise stated, and brief reactive psychosis), informed consent for research participation and adequate English language comprehension. As approximately 50% of young people using cannabis at admission to early intervention services such as EPPIC can be expected to voluntarily cease use within the first 6–10 weeks (Edwards 2001), only patients who were continuing to use cannabis at 10 weeks post‐initial clinical stabilization were eligible for study inclusion." Exclusion criteria: "No additional exclusion criteria were applied." Setting: NA Special subgroup: drug abuse (cannabis) N: 47 Gender: 34 men, 13 women Age: mean 21.1 years CBT plus standard care arm: participants total: 23, participants male: NA, participants female: NA, age: mean 20.9 years (SD = 3.5), age range: 15‐29 years, duration of illness: NA, duration of untreated psychosis: NA, % taking antipsychotics: NA, % diagnosis schizophrenia: NA, baseline BPRS total M = 49.9, SD = 16.3, N = 23 Control arm: participants total: 24, age: mean 21.3 years (SD = 3.4), age range: 15‐29 years, duration of illness: NA, duration of untreated psychosis: NA, % taking antipsychotics: NA, % diagnosis schizophrenia: NA, baseline BPRS total M = 48.8, SD = 17, N = 24
Interventions	1. CBT plus standard care. N = 23 Description of treatment: "Cannabis and psychosis therapy is an individually delivered cognitive‐behavioral harm minimization approach to cannabis use in individuals with FEP. Young people are often traumatized by psychosis and its implications, and the desire to avoid further episodes can encourage reconsideration of cannabis use, providing a window of opportunity to influence behavior change (). The intervention is delivered over 3 months, ideally involving 10 weekly sessions (20–60 min in duration). A booster telephone call 3 months after end of treatment is designed as a generalized relapse prevention, and serves to emphasize gains made in treatment and reinforce strategies to manage potential relapse. CAP commences with a detailed assessment and attention to engagement, followed by education about cannabis and psychosis and building motivation to change (Miller 2002). The focus of therapy is then determined by phase of commitment to change (Prochaska 1992) and may include further education about cannabis and psychosis, motivational interviewing, goal setting, goal achievement strategies, and discussion about relapse prevention. Feedback was sought from five international experts on an early CAP outline and their comments were incorporated within the final treatment manual (Hinton 2002)." In addition to the randomized condition, each participant received standard EPPIC care which includes case management, regular psychiatric review and medication; access to mobile assessment and treatment, family work, group programs, and a prolonged recovery clinic (McGorry 1996)." Number of sessions: 10 planned, 7.6 received Sessions frequency and duration: 4, 20‐60 minutes each Therapist: four clinical psychologists trained in cognitive‐behaviour therapy and experienced in the treatment of FEP Therapist experience: experts Supervision: yes Modality: individual Delivery: face‐to‐face Treatment manualised: yes 2. Control (Psychoeducation); N = 24 Description of treatment: "The 10 individual PE sessions were guided by standardized powerpoint slides (in electronic or printed form) which covered the nature of psychosis, medication and other treatments, and relapse prevention and stigma, but avoided explicit discussion of cannabis." "In addition to the randomized condition, each participant received standard EPPIC care which includes case management, regular psychiatric review and medication; access to mobile assessment and treatment, family work, group programs, and a prolonged recovery clinic (McGorry 1996)." Number of sessions: 10 planned, 8.4 received Sessions frequency and duration: NA Therapist: four clinical psychologists trained in cognitive‐behaviour therapy and experienced in the treatment of FEP Therapist experience: experts Supervision: yes Modality: individual Delivery: face‐to‐face Treatment manualised: NA
Outcomes	Mental state ‐ general ‐ BPRS‐E total endpoint (< 6 months, < 12 months) Mental state ‐ specific ‐ BPRS‐PS endpoint (< 6 months, < 12 months) Mental state ‐ specific ‐ SANS endpoint (< 6 months, < 12 months) Mental state ‐ specific ‐ BDI short form endpoint (< 6 months, < 12 months) Leaving the study early ‐ for any reason (dropout) (< 6 months, < 12 months) Functioning ‐ SOFAS endpoint (< 6 months, < 12 months)
Notes	We contacted the authors but received no reply.

Edwards 2011.

Study characteristics
Methods	Study design: randomised controlled trial Study grouping: parallel group Antipsychotic medication allowed: yes Duration in weeks: 12 treatment phase, 12‐week follow‐up phase Number of study arms: 4 Type of blinding: rater blind Type of data analysis for overall efficacy: ITT Number of sites: 1 Number of countries: 1 Country: Australia Sponsorship source: Victorian Government's Health Promotion Foundation and NOVARTIS Trial registration ID: NA Publication year: 2011 Researcher's allegiance: The authors developed the manual for the intervention group.
Participants	Diagnosis: schizophrenia, schizophreniform disorder, delusional disorder, psychotic disorder not otherwise specified; diagnostic criteria: DSM‐IV Current clinical state: acute/experiencing positive symptoms Definition of stability: NA Inclusion criteria: "experiencing a first treated episode of a psychotic disorder that fulfilled the DSM‐IV criteria for a diagnosis of schizophrenia, schizophreniform disorder, delusional disorder, or psychotic disorder not otherwise specified; being registered with EPPIC for 12 to 26 weeks; and continuing to experience moderate to severe positive symptoms, defined as a score ≥ 4 on at least one of the hallucinations, unusual thought content, and conceptual disorganisation items of the expanded version of the brief psychiatric rating scale (BPRS) (Ventura 1993), with a score of not less than 3 on these items for a period of 14 consecutive days or more during the preceding 12 weeks. All participants had been treated with at least one atypical antipsychotic (usually risperidone, olanzapine or quetiapine) at doses up to 500 mg chlorpromazine equivalence (if tolerated), with demonstrated medication compliance for at least the past 4 weeks." Exclusion criteria: "an organic mental disorder, pregnancy or lactation, requiring antidepressant medication, a mood stabiliser or ECT, and a history of drug‐induced granulocytopenia." Setting: both inpatient and outpatient Special subgroup: treatment‐resistant patients N: 48 Gender: 36 men, 14 women Age: mean 21.4 years CBT plus standard care arm 1 (CBT + CLZ): participants total: 11, participants male: 10, participants female: 1, age: mean 20.8 years (SD = 2.8), age range: 15‐29 years, duration of untreated psychosis: 13.75 months (SD = 29.2), % taking antipsychotics: 100, % diagnosis schizophrenia: 81.8, % diagnosis schizophreniform disorder: 18.2 CBT plus standard care arm 2 (CBT + TDZ): participants total: 12, participants male: 7, participants female: 5, age: mean 22 years (SD = 4.1), age range: 15‐29 years, duration of untreated psychosis: 15.55 months (SD = 15), % taking antipsychotics: 100, % diagnosis schizophrenia: 75, % diagnosis schizophreniform disorder: 16.7, % delusional disorder: 8.3 Control arm 1 (CLZ): participants total: 14, participants male: 9, participants female: 5, age: mean 20.5 years (SD = 3.5), age range: 15‐29 years, duration of untreated psychosis: 14.65 months (SD = 14.2), % taking antipsychotics: 100, % diagnosis schizophrenia: 78.6, % diagnosis schizophreniform disorder: 14.3 Control arm 2 (TDZ): participants total: 11, participants male: 8, participants female: 3, age: mean 22.5 years (SD = 3.4), age range: 15‐29 years, duration of untreated psychosis: 14.95 months (SD = 15.9), % taking antipsychotics: 100, % diagnosis schizophrenia: 90.9, % diagnosis schizophreniform disorder: 9.1
Interventions	1. CBT plus standard care. N = 23 Description of treatment: "Participants commenced treatment at a dose of 12.5 mg/day which was titrated upwards in 25 mg/day increments up to a maximum dose of 300 mg/day, depending on clinical response. During Week 4 of the study, patients who still showed an inadequate response had their medication increased to a maximum of 400 mg/day, which could be further increased to a maximum of 600 mg/daily if necessary during Week 5, after which all patients were maintained on their current medication dose." "A manualized CBT program, the systematic treatment of persistent psychosis (STOPP, Hermann‐Doig 2003) was devised to target enduring positive symptoms and related patient needs." "In addition, all participants received routine clinical care, which included access to a 24‐hour mobile assessment and treatment team, inpatient service, case management, and psychiatric review. Patients were seen weekly by a psychiatrist/psychiatry registrar for the duration of the trial, (...)". Number of sessions: 24 planned (CBT + CLZ: 15.25 done; CBT + TDZ: 13.45 done) Sessions frequency and duration: 8 Therapist: psychiatrist, psychiatry registrar (TAU) Therapist experience: NA Supervision: NA Modality: individual Delivery: face‐to‐face Treatment manualised: yes 2. Control (Treatment as usual). N = 25 Description of treatment: "Participants commenced treatment at a dose of 12.5 mg/day which was titrated upwards in 25 mg/day increments up to a maximum dose of 300 mg/day, depending on clinical response. During Week 4 of the study, patients who still showed an inadequate response had their medication increased to a maximum of 400 mg/day, which could be further increased to a maximum of 600 mg/daily if necessary during Week 5, after which all patients were maintained on their current medication dose." "In addition, all participants received routine clinical care, which included access to a 24‐hour mobile assessment and treatment team, inpatient service, case management, and psychiatric review. Patients were seen weekly by a psychiatrist/psychiatry registrar for the duration of the trial, and all participants not receiving CBT attended weekly case management sessions." Number of sessions: NA Sessions frequency and duration: NA Therapist: psychiatrist, psychiatry registrar (TAU) Therapist experience: NA Supervision: NA Modality: NA Delivery: NA Treatment manualised: yes
Outcomes	Mental state ‐ general ‐ BPRS‐P endpoint (< 6 months) Mental state ‐ specific ‐ SANS endpoint (< 6 months) Mental state ‐ specific ‐ BDI endpoint (< 6 months) Global state ‐ Remission (< 6 months) Global state ‐ CGI‐S endpoint (< 6 months) Functioning ‐ SOFAS endpoint (< 6 months) Quality of Life ‐ QLS endpoint (< 6 months)
Notes	We contacted the authors but received no reply.

GetUp.

Study characteristics
Methods	Study design: cluster‐randomised controlled trial Study grouping: parallel Antipsychotic medication allowed: yes Duration in weeks: 9‐month follow‐up (39.1 weeks) Number of study arms: 2 Type of blinding: open‐label Type of data analysis for overall efficacy: ITT Number of sites: 117 Number of countries: 1 Country: Italy Sponsorship source: public Trial registration ID: NCT01436331 Publication year: 2015 Researcher's allegiance: NA
Participants	Diagnosis: specific ICD‐10 codes for psychosis (F1x.4; F1x.5; F1x.7; F20–29; F30.2, F31.2, F31.5, F31.6, F32.3, F33.3); diagnostic criteria: ICD‐10 Current clinical state: clinically stable Definition of stability: Patients capable of collaborating in at least a brief clinical examination Inclusion criteria: "age 18–54 years, residence in catchment areas of CMHCs, presence of at least one of the following: hallucinations, delusions, qualitative speech disorder, qualitative psychomotor disorder, bizarre, or grossly inappropriate behavior, or two of the following: loss of interest, initiative, and drive; social withdrawal; episodic severe excitement; purposeless destructiveness; overwhelming fear; or marked self‐neglect, per the WHO Screening Schedule for Psychosis (Jablensky 1992), first lifetime contact with CMHCs, prompted by these symptoms." Exclusion criteria: "(1) antipsychotic medication (> 3 months) prescribed for an identical or similar mental disorder; (2) mental disorders due to general medical condition; (3) moderate‐severe mental retardation per a clinical functional assessment; and (4) psychiatric diagnosis other than International Classification of Diseases (ICD)‐10 for psychosis." Setting: outpatient Special subgroup: first‐episode patients N: 444 Gender: 260 men, 184 women Age: mean 30.2 years CBT plus standard care arm: 272 participants, 166 male, 106 female, mean age 29.3 years (SD = 9.8), age range 18‐54 years, duration of illness: NA, duration of untreated psychosis: NA, % taking antipsychotics: NA, % diagnosis nonaffective psychosis: 78.7, % diagnosis affective psychosis: 21.3, baseline PANSS total M = 2.37, SD = 0.67, N = 271 Control arm: 172 participants, 94 male, 78 female, mean age 31.5 years (SD = 9.2), age range 18‐54 years, duration of illness: NA, duration of untreated psychosis: NA, % taking antipsychotics: NA, % diagnosis nonaffective psychosis: 76.7, % diagnosis affective psychosis: 23.3, baseline PANSS total M = 2.32, SD = 0.68, N = 172
Interventions	1. CBT plus standard care. N = 272 Description of treatment: "The experimental treatment package was provided by routine public Community Mental Health Centers (CMHCs) which operate within the Italian National Health Service and consisted of standard care (treatment as usual, TAU) plus evidence‐based additional treatment. Specifically, the multi‐element psychosocial intervention, adjunctive to TAU, comprised: (i) Cognitive Behavioral Treatment for psychosis (CBTp) to patients; (ii) psychosis‐focused Family Intervention (FIp) to individual families; and (iii) Case Management (CM) to both parties. FIp was based on the model proposed by Leff 1985 and further developed by Kuipers 2002. It included an optimal number of 10–15 sessions over 9 months, with each individual family: 6 sessions in the first 3 months, and at least 1 session/month in the 6 months afterwards. Every patient/family had a dedicated CM, who coordinated all planned interventions (Burns 2002)" Number of sessions: 10‐15 sessions over 9 months; Number of CBT sessions: mean (SD): 17,8 (10,3); number of FI sessions: mean (SD): 9,3 (7,0); number of CM contacts: mean (SD): 21,7 (24,4) Sessions frequency and duration: 1‐2 sessions per month; "weekly sessions in the first 3 months, and at least 1 session/month in the 6 months afterwards" Therapist: psychiatrist/ psychologist, nurse/educator Therapist experience: experts (received CBTp and Fip training) Supervision: yes ("The intervention was provided by CMHC staff, trained in the previous 6 months and supervised by experts.") Modality: individual Delivery: face‐to‐face Treatment manualised: yes 2. Control (Treatment as usual). N = 172 Description of treatment: "Treatment as usual (TAU) was also provided by routine public Community Mental Health Centers (CMHCs) involved in the Trial. In Italy standard care for FEP patients typically consisted of personalized outpatient psychopharmacological treatment, combined with non‐specific supportive clinical management and non‐specific informal support/educational sessions for families." Number of sessions: NA Sessions frequency and duration: NA Therapist: NA Therapist experience: NA Supervision: NA Modality: individual Delivery: face‐to‐face Treatment manualised: NA
Outcomes	Mental state ‐ general ‐ PANSS total endpoint (< 12 months) Mental state ‐ specific ‐ PANSS positive endpoint (< 12 months) Mental state ‐ specific ‐ PANSS negative endpoint (< 12 months) Mental state ‐ specific ‐ Hamilton Depression endpoint (< 12 months) Service use ‐ admission to hospital (< 12 months) Service use ‐ number of days in hospital (< 12 months) Leaving the study early ‐ for any reason (dropout) (< 12 months) Functioning ‐ GAF endpoint (< 12 months) Mortality ‐ overall mortality (< 12 months) Mortality ‐ mortality due to natural causes (< 12 months) Mortality ‐ mortality due to suicide (< 12 months)
Notes	We contacted the authors and obtained additional data.

Gleeson 2009 (EPISODE II).

Study characteristics
Methods	Study design: randomised controlled trial Study grouping: parallel group Antipsychotic medication allowed: yes Duration in weeks: relapse prevention intervention (of approximately 6‐month duration according to trial registration) ‐ "The study included a total of 6 assessment time points spanning a 2.5‐year follow‐up period." Number of study arms: 2 Type of blinding: rater blind Type of data analysis for overall efficacy: ITT Number of sites: 2 Number of countries: 1 Country: Australia Sponsorship source: public and pharmaceutical Trial registration ID: ACTRN12605000514606 Publication year: 2009 Researcher's allegiance: The authors developed the manual for the intervention group.
Participants	Diagnosis: diagnosis of a first episode of a DSM‐IV psychotic disorder, less than 6 months of prior treatment with antipsychotic medications; diagnostic criteria: DSM‐IV Current clinical state: remission Definition of stability: "Remission was defined as 4 weeks or more of scores of 3 (mild) or below on the subscale items hallucinations, unusual thought disorder, conceptual disorganization, and suspiciousness on the expanded version of the Brief Psychiatric Rating Scale (BPRS)". (Lukoff 1986, Overall 1962) Inclusion criteria: "The study inclusion criteria were a diagnosis of a first episode of a DSM‐IV (APA 1994) psychotic disorder, less than 6 months of prior treatment with antipsychotic medications, age 15 to 25 years inclusive, and remission on positive symptoms of psychosis. Remission was defined as 4 weeks or more of scores of 3 (mild) or below on the subscale items hallucinations, unusual thought disorder, conceptual disorganization, and suspiciousness on the expanded version of the Brief Psychiatric Rating Scale (BPRS)." Exclusion criteria: "Exclusion criteria were ongoing active positive symptoms of psychosis, severe intellectual disability, inability to converse in or read English, and participation in previous CBT trials." Setting: outpatient ("outpatient case manager") Special subgroup: first episode, stable N: 81 Gender: 51 men, 30 women Age: mean 20.1 years (SD = 3.1) CBT plus standard care arm: participants total: 41, participants male: 27, participants female: 14, age: mean 20.1 years (SD = 2.9), age range: 15‐25 years, duration of untreated psychosis: 401.1 days (SD = 529.1), % taking antipsychotics: 94.6, % diagnosis schizophrenia: 34.1, % diagnosis schizophreniform disorder: 7.3, % diagnosis schizoaffective disorder: 7.3, % diagnosis bipolar disorder: 7.3, % MDE with psychotic features: 2.4, % psychotic disorder NOS: 34.1, baseline BPRS total M = 35.4, SD = 6.9, N = 41 Control arm: participants total: 40, participants male: 24, participants female: 16, age: mean 20.1 years (SD = 3.2), age range: 15‐29 years, duration of untreated psychosis: 368.6 days (SD = 611.9), % taking antipsychotics: 100, % diagnosis schizophrenia: 32.5, % diagnosis schizophreniform disorder: 15, % diagnosis schizoaffective disorder: 2.5, % diagnosis bipolar disorder: 2.5, % MDE with psychotic features: 10, % psychotic disorder NOS: 25, baseline BPRS total M = 34.3, SD = 8, N = 40
Interventions	1. CBT plus standard care (Relapse prevention therapy (RPT)). N = 41 Description of treatment: "Patients randomly assigned to RPT were introduced to their individual research therapist, who additionally adopted the role of outpatient case manager for the duration of their treatment at EPPIC. All patients randomly assigned to RPT continued their follow‐up treatment with their outpatient psychiatrist and had access to home‐based treatment and group interventions as indicated. The research therapists functioned as fully integrated members of the EPPIC treatment team and as visiting therapists to Barwon Health, which allowed the effectiveness of RPT to be evaluated within existing “real‐world” clinical roles. Key differences between TAU and RPT included (1) the shared, written individualized formulation regarding relapse risk; (2) the systematic and phased approach to relapse prevention via a range of cognitive behavioral interventions; (3) the parallel individual and family sessions focused on relapse prevention; and (4) supervision specifically focused on relapse prevention. The manualized individual therapy intervention comprises 5 phases of therapy underpinned by a CBT framework and informed by previous psychotherapy trials conducted at EPPIC (Edwards 2003, Jackson 2001) and by the collaborative therapy framework developed at the Mental Health Research Institute, Melbourne, Australia (Gilbert 2003a). The aim of the first phase of the individual therapy was to engage the patient and assess their extent of recovery and individual risk for relapse (e.g. substance use, medication noncompliance, stressful life events, and comorbid anxiety and depression). In the second phase, the formulation and agenda for therapy were agreed on with the patient and summarized in a letter, which was read out to the patient – a technique informed by cognitive analytic therapy (Ryle 2002). The therapeutic agenda was intended to address those risk factors that were identified in the initial assessment. The third phase focused on increasing awareness for the risk of setbacks and how to minimize them, and in the fourth phase, the potential early warning signs of relapse were identified and a relapse plan formulated (Birchwood 1989). The fifth phase included optional modules that addressed issues of nonadherence to treatment, substance abuse, coping with stress, and comorbid anxiety and depression. Selection of the intervention modules to be undertaken during this therapy phase was based on the collaborative formulation and therapy agenda accomplished after initial assessment. The final phase included a review and termination, and a booster session. The individual therapy phases were provided within a 7‐month therapy window, approximately fortnightly in order to match the recommended frequency of sessions in TAU (EPPIC 2001). The family intervention, also manualized and provided by a trained family therapist, was informed by cognitive behavioral family therapy for schizophrenia (Falloon 1988, Mueser 1999) and family interventions for FEP (Linszen 1996). The phases of family therapy were assessment and engagement, assessment of family communication, burden and coping, psychoeducation regarding relapse risk, and a review of early warning signs and documentation of a relapse prevention plan. Intensive communication skills training and problem solving were undertaken when indicated. Treatment manuals are available on request." Number of sessions: Participants randomly assigned to RPT completed a mean of 8.51 sessions; therapy completers (25 out of 41: 61%): mean of 11.84 sessions; family intervention: mean of 10,2 sessions; family completers (15 out of 24 completed: 62%): mean of 13.1 sessions Sessions frequency and duration: 2 per month Therapist: research therapist Therapist experience: experts Supervision: yes Modality: individual and family based Delivery: face‐to‐face Treatment manualised: yes 2. Control (Treatment as usual). N = 40 Description of treatment: "Patients randomly assigned to TAU continued with their routine treatment, via the EPPIC program or Barwon Health, which was coordinated via an outpatient case manager and outpatient consultant psychiatrist, with access to home‐based treatment and a range of psychosocial interventions (EPPIC 2001). During the course of the trial, the caseload of a full‐time case manager was approximately 35. A case management manual provided detailed guidelines for TAU (EPPIC 2001). All case managers were orientated to early psychosis treatment guidelines and were provided with a range of additional therapy manuals and standardized psychoeducation materials. Fidelity was managed via approximately fortnightly one‐to‐one supervision for all case managers with a senior clinician and via weekly multi‐disciplinary case review meetings, which all outpatient staff were required to attend. At entry into the service, all families were routinely offered access to a brief family psychoeducation group, and EPPIC families had access to a family peer support service." Number of sessions: unknown Sessions frequency and duration: 2 per month Therapist: NA Therapist experience: NA Supervision: yes Modality: individual and family psychoeducation group Delivery: face‐to‐face Treatment manualised: yes
Outcomes	Mental state ‐ general ‐ BPRS total endpoint (< 12 months, > 1 year) Mental state ‐ specific ‐ BPRS positive endpoint (> 1 year) Mental state ‐ specific ‐ SANS endpoint (< 12 months) Mental state ‐ specific ‐ MADRS endpoint (< 12 months, > 1year) Global state ‐ Number of participants that relapsed (< 12 months, > 1 year) Service use ‐ Number of days in hospital (< 12 months, > 1 year) Leaving the study early ‐ for any reason (dropout) (< 12 months, > 1 year) Functioning ‐ SOFAS endpoint (< 12 months, > 1 year) Quality of life ‐ WHOQoL‐BREF endpoint (< 12 months)
Notes	We contacted the authors but received no reply.

Gonzalez‐Ortega 2021.

Study characteristics
Methods	Study design: randomised controlled trial Study grouping: parallel group Antipsychotic medication allowed: yes Duration in weeks: 28 planned, 28 to 32 weeks actually Number of study arms: 2 Type of blinding: rater blind Type of data analysis for overall efficacy: completers Number of sites: 5 Number of countries: 1 Country: Spain Sponsorship source: public Trial registration ID: NCT01783457 Publication year: 2021 Researcher's allegiance: NA
Participants	Diagnosis: first episode psychosis; diagnostic criteria: DSM‐IV‐TR Current clinical state: first episode ‐ psychopathologically stable Definition of stability: NA Inclusion criteria: "The patients needed to be between 18–45 years old and to have received a diagnosis of FEP, according to the 4th edition text‐revised of the Diagnostic and Statistical Manual of Mental Disorders (DSM‐IV‐TR) (APA 2000), within the previous five years, which could include either schizophreniform disorder, schizoaffective disorder, brief psychotic disorder, delusional disorder, non‐specified psychotic disorder, bipolar disorder with psychotic symptoms, or major depressive disorder with psychotic symptoms." Exclusion criteria: "The exclusion criteria were intellectual disability, organic brain disorders or the presence of comorbidities that could hinder communication, and substance use disorder as the primary diagnosis." Setting: both inpatient and outpatient Special subgroup: first episode, stable N: 184 randomised, final sample: 177 Gender: 105 men, 72 women Age: mean 28.5 years (SD = 8.97) CBT plus standard care arm: participants total: 92, participants male: 47, participants female: 39, age: mean 27.74 years (SD = 7.66), age range: 18‐45 years, duration of untreated psychosis: NA, % taking antipsychotics: 98.8, % diagnosis nonspecified psychotic disorder: 67.4, % diagnosis schizophreniform disorder: 7, % diagnosis bipolar disorder: 15.1, % diagnosis affective psychosis: 1.2, baseline PANSS total M = 58.29, SD = 14.29, N = 86 Control arm: participants total: 92, participants male: 58, participants female: 33, age: mean 29.27 years (SD = 10.03), age range: 18‐45 years, duration of untreated psychosis: NA, % taking antipsychotics: 95.6, % diagnosis nonspecified psychotic disorder: 65.9, % diagnosis schizophreniform disorder: 12.1, % diagnosis bipolar disorder: 11, % diagnosis affective psychosis: 2.2, baseline PANSS total M = 57.61, SD = 12.3, N = 91
Interventions	1. CBT plus standard care. N = 92 Description of treatment: "The combined clinical treatment involved TAU plus an adjuvant individual CBT intervention. The intervention programme was implemented in all participating centres, addressing the same content and following the same structure. It was composed of 14 one‐hour sessions fortnightly (±3 days) for a period of 6.5–7.5 months. The first part of the program (sessions 1–9) was composed of psychoeducational sessions aimed at improving patients’ insight into their illness, treatment adherence, prodromal identification, early intervention to prevent relapses, and a healthy lifestyle. The second part of the intervention (sessions 10–14) included cognitive behavioural techniques for symptom and thought management (anxiety management techniques and social and problem‐solving skills). The patients were offered the possibility of using a telephone helpline between sessions if they had any questions related to the content of the sessions or their status. The program included the following sessions: 1. What is a first episode of psychosis? 2. Challenge and importance of insight into vulnerability. 3. Symptom recognition. 4. Prevention of relapses: protective and risk factors. 5. Detection of prodromes. 6. What can I do if I perceive that the symptoms are emerging again? 7. Treatment adherence. 8. Healthy lifestyles: sleep and sexuality. 9. Healthy lifestyles: substance use. 10. Anxiety management techniques (I). 11. Anxiety management techniques (II). 12. Social skills: assertiveness techniques. 13. Problem‐solving techniques. 14. Final doubts and farewell." Number of sessions: 14 Sessions frequency and duration: 2 per month, 60 minutes Therapist: NA Therapist experience: experts Supervision: NA (only during the training) Modality: individual Delivery: face‐to‐face Treatment manualised: NA 2. Control (Treatment as usual). N = 92 Description of treatment: "TAU refers to the standard treatment provided to FEP patients. This includes pharmacological treatment and regular sessions with the assigned psychiatrist and a multi‐disciplinary team in each centre. TAU includes physical care, career counselling, and the provision of unstructured information to families about disease symptoms, treatment, and prognosis." Number of sessions: NA Sessions frequency and duration: NA Therapist: NA Therapist experience: NA Supervision: NA Modality: NA Delivery: NA Treatment manualised: NA
Outcomes	Mental state ‐ general ‐ PANSS total endpoint (< 6 months) Mental state ‐ specific ‐ PANSS positive endpoint (< 6 months) Mental state ‐ specific ‐ PANSS negative endpoint (< 6 months) Mental state ‐ specific ‐ Hamilton Depression endpoint (< 6 months) Global state ‐ CGI‐S endpoint (< 6 months) Global state ‐ CGI‐I endpoint (< 6 months) Leaving the study early ‐ for any reason (dropout) (< 6 months) Funcitoning ‐ GAF endpoint (< 6 months) Functioning ‐ FAST endpoint (< 6 months)
Notes	We contacted the authors but received no reply.

Haddock 1999.

Study characteristics
Methods	Study design: randomised controlled trial Study grouping: parallel group Antipsychotic medication allowed: yes Duration in weeks: 5 weeks + booster sessions at 1, 2, 3 and 4 months following discharge Number of study arms: 2 Type of blinding: rater blind Type of data analysis for overall efficacy: completers Number of sites: NA Number of countries: 1 Country: UK Sponsorship source: NA Trial registration ID: NA Publication year: 1999 Researcher's allegiance: The authors developed the manual for both intervention and control.
Participants	Diagnosis: schizophrenia or schizoaffective disorder; diagnostic criteria: DSM‐IV Current clinical state: currently admitted to an acute general psychiatric ward for onset or relapse of psychotic symptoms Definition of stability: NA Inclusion criteria: "The study inclusion criteria were: 1. DSM‐IV (APA 1994) diagnosis of schizophrenia or schizo‐affective disorder confirmed by an independent psychiatrist attached to the project 2. First treatment for psychosis less than 5 years ago 3. Currently admitted to an acute general psychiatric ward for onset or relapse of psychotic symptoms" Exclusion criteria: NA Setting: inpatient Special subgroup: acute (currently admitted to an acute general psychiatric ward for onset or relapse of psychotic symptoms) N: 21 Gender: 19 men, 2 women Age: mean 29.1 years CBT plus standard care arm: participants total: 10, participants male: 10, participants female: 0, age: mean 28.1 years (SD = 7.24), age range: NA, duration of untreated psychosis: NA, % taking antipsychotics: 100, baseline BPRS total M = 53, SD = 7, N = 8 Control arm: participants total: 11, participants male: 9, participants female: 2, age: mean 30 years (SD = 7.9), age range: NA, duration of untreated psychosis: NA, % taking antipsychotics: 100, baseline BPRS total M = 53.2, SD = 8.2, N = 10
Interventions	1. CBT plus standard care. N = 10 Description of treatment: "The cognitive‐behavioural treatment followed a detailed treatment manual designed by the authors and used in earlier CBT for psychosis research studies. It was primarily focused on the treatment of auditory hallucinations and delusions, associated symptoms and problems (for example, anxiety, depression, self‐esteem) and relapse prevention. For some patients, psychotic symptoms rapidly remitted during the first few days of their admission. In these cases the cognitive‐behavioural intervention was focused on relapse prevention, treatment of associated problems and enhancing medication adherence. Treatment was designed to follow a cognitive‐behavioural approach shown to be effective with chronic psychotic patients, and involved the following elements: 1. Assessment and engagement 2. Formulation of key problems 3. Intervention directed at reducing the severity or occurrence of key problems 4. Relapse prevention/keeping well The following specific CBT strategies were used: formulation, guided discovery, symptom monitoring, exposure/focusing strategies for managing voices, hypothesis/reality testing, re‐framing attributions, rational responding, coping strategy enhancement, distraction techniques, anxiety management, depression/self‐esteem work, medication compliance/motivational interviewing, schema work, relapse prevention and keeping well strategies. The booster sessions involved review and consolidation of strategies used during inpatient sessions, with particular emphasis on keeping well strategies." Number of sessions: 10.2 + 1.76 booster sessions Sessions frequency and duration: twice monthly during inpatient stay, then monthly Therapist: clinical psychologists Therapist experience: experts Supervision: yes Modality: individual Delivery: face‐to‐face Treatment manualised: yes 2. Control (supportive counseling/psychoeducation treatment). N = 11 Description of treatment: "This treatment was also carried out according to a treatment manual designed by the authors and was matched in terms of therapy time and therapist to the CBT treatment. It aimed to control for non‐specific aspects of therapy by delivering basic assessment, psychoeducation and counselling in a supportive, warm, genuine, empathic and unstructured style. Specific techniques used involved: reflection, paraphrasing, summarising, feedback, avoidance of criticism, information about schizophrenia, medication and side effects." Number of sessions: 9.1 + 0.91 booster sessions Sessions frequency and duration: twice monthly during inpatient stay, then monthly Therapist: clinical psychologists Therapist experience: NA Supervision: yes Modality: individual Delivery: face‐to‐face Treatment manualised: yes
Outcomes	Mental state ‐ general ‐ BPRS total endpoint (< 6 months) Global state ‐ number of participants that relapsed (> 1 year ) Service use ‐ number of days in hospital (< 6 months, > 1 year) Leaving the study early ‐ for any reason (dropout) (< 6 months)
Notes	We contacted the authors but received no reply.

Hjorthoj 2013 (CapOpus).

Study characteristics
Methods	Study design: randomised controlled trial Study grouping: parallel Antipsychotic medication allowed: yes Duration in weeks: 26 weeks Number of study arms: 2 Type of blinding: rater blind Type of data analysis for overall efficacy: ITT Number of sites: NA Number of countries: 1 Country: Denmark Sponsorship source: "supported by the Lundbeck Foundation, the Municipality of Copenhagen, the Egmont Foundation, the Health Insurance Foundation, the Ministry of Social Welfare, Aase and Ejnar Danielsen’s Foundation and the Wørzner Foundation" Trial registration ID: NCT 00484302 Publication year: 2013 Researcher's allegiance: no
Participants	Diagnosis: psychosis and cannabis use disorder; diagnostic criteria: ICD‐10 Current clinical state: psychosis and cannabis use disorder Definition of stability: psychosis and cannabis use disorder Inclusion criteria: "ICD‐10 schizophrenia spectrum psychosis (F2) ; cannabis use disorder (F12); age 18–35 years (chosen so as to yield more homogeneous groups for the planned group sessions; a combination of slow influx of patients and interested participants just outside the inclusion age range led us to modify this after 5 months of inclusion and include patients aged 17–42 years); residence in the Copenhagen area; not requiring an interpreter; and the ability and willingness to give informed consent." We received data about participants with first episode from the authors. Exclusion criteria: NA Setting: both inpatient and outpatient Special subgroup: cannabis use disorder N: 36
Interventions	1. CBT plus standard care. With first episode: N = 18 Description of treatment: "CapOpus was an add‐on intervention to TAU, which all patients received. Full details are given in the design paper (Hjorthøj 2008) and full protocol (www.capopus.dk). The intervention was based on the EPPIC manual, and was similar to the Midas trial (Hinton 2002; Barrowclough 2010)." Number of sessions: Twenty‐four sessions were planned and, on average, 16 were achieved Sessions frequency and duration: 2 in first month, then 1 per month for 5 months Therapist: "Addiction consultants were trained and experienced in motivational interviewing and CBT. One consultant was a psychologist (handling 71% of participants), one a master student of psychology, and one an occupational therapist." Therapist experience: experts Supervision: yes Modality: individual Delivery: face‐to‐face Treatment manualised: yes 2. Control (Treatment as usual). With first episode: N = 18 Description of treatment: "TAU consisted of the treatment available to patients had they not participated in the trial, provided by staff not employed by CapOpus. TAU was carried out in Opus, CMHCs or ACT teams (Nordentoft 1996; Vendsborg 1999; Petersen 2005). No explicit manual exists regarding co‐occurring cannabis use disorder in TAU. Instead, these facilities primarily target the psychotic disorder using both antipsychotic medication and methods such as CBT (but generally not targeted at substance use)." Number of sessions: NA Sessions frequency and duration: NA Therapist: NA Therapist experience: NA Supervision: NA Modality: individual Delivery: face‐to‐face Treatment manualised: no
Outcomes	Mental state ‐ general ‐ PANSS total endpoint (< 6 months, < 12 months) Mental state ‐ specific ‐ PANSS positive endpoint (< 6 months, < 12 months) Mental state ‐ specific ‐ PANSS negative endpoint (< 6 months, < 12 months) Quality of fife ‐ MANSA (< 6 months, < 12 months) Functioning ‐ GAF endpoint (< 6 months, < 12 months)
Notes	We contacted the authors and obtained additional data.

Jackson 2008 (ACE).

Study characteristics
Methods	Study design: randomised controlled trial Study grouping: parallel group Antipsychotic medication allowed: yes Duration in weeks: 14 weeks; assessment period: 15 months Number of study arms: 2 Type of blinding: rater blind Type of data analysis for overall efficacy: ITT Number of sites: 1 Number of countries: 1 Country: Australia Sponsorship source: public Trial registration ID: NA Publication year: 2008 Researcher's allegiance: The authors developed the manual for both intervention and control.
Participants	Diagnosis: primary psychotic diagnoses or first‐episode psychosis; diagnostic criteria: Structured Clinical Interview for DSM‐IV‐TR Axis 1 Disorders‐Patient Edition (SCID) Current clinical state: first episode of psychosis Definition of stability: consecutive patients admitted to OYH within 4 weeks of their registration with the service Inclusion criteria: "EPPIC is a comprehensive treatment service for 15–25‐year‐old people experiencing a first episode of psychosis. It includes a 16‐bed in‐patient unit, an out‐patient case management system, family work, accommodation, prolonged recovery programmes and tailored group programmes (Edwards 2002)." Exclusion criteria: "Patients were excluded (before randomisation) if any of the following criteria were met: inability to speak English; intellectual disability (IQ < 70); psychosis due to a medical condition; change to a non‐psychotic diagnosis ; left the EPPIC catchment area; treatment from a private psychiatrist/psychologist ; participating in a first‐episode mania trial ; exhibiting violent behaviour; or being incarcerated (n = 111)." Setting: both inpatient and outpatient (16‐bed in‐patient unit, an outpatient case management system) Special subgroup: first episode N: 62 Gender: 45 men, 17 women Age: mean 22.3 years CBT plus standard care arm: participants total: 31, participants male: 19, participants female: 12, age at start of treatment: mean 22.13 years (SD = 3.30), age at onset of psychosis: mean 21.58 years (SD = 3.49), age range: 15‐25 years, duration of illness: 0.55 years, duration of untreated psychosis: 83 days, % taking antipsychotics: NA, % diagnosis schizophrenia: 12.90, % diagnosis schizophreniform disorder: 41.94, % diagnosis schizoaffective disorder: 12.90, % diagnosis affective disorder: 19.35, % delusional/psychotic (NOS): 12.90, baseline BPRS total M = 47.74, SD = 10.84, N = 31 Control arm: participants total: 31, participants male: 26, participants female: 5, age at start of treatment: mean 22.45 years (SD = 3.82), age at onset of psychosis: mean 21.67 years (SD = 4.20), age range: 15‐25 years, duration of illness: 0.78 years, duration of untreated psychosis: 107 days, % taking antipsychotics: NA, % diagnosis schizophrenia: 12.90, % diagnosis schizophreniform disorder: 38.71, % diagnosis schizoaffective disorder: 9.68, % diagnosis affective disorder: 22.58, % delusional/psychotic (NOS): 16.13, baseline BPRS total M = 50.19, SD = 10.82, N = 31
Interventions	1. CBT plus standard care. N = 31 Description of treatment: "ACE therapy involved an assessment of the presenting psychotic and non‐psychotic complaints followed by a formulation of the relationship between these complaints and the participant’s life history. Problems were prioritized according to a flowchart that directed the ACE therapy. Any issues of risk were considered a priority. Positive psychotic symptoms, if they were present and distressing, were the next priority, followed by co‐morbidities, negative symptoms, issues of identity and relapse prevention. Each area of difficulty was treated from a broadly cognitive behavioural perspective. Ongoing engagement was essential throughout the therapy process. This involved a flexible approach to the timing, location and content of therapy. For example, therapy might have included going for a bike ride, or being conducted at the participant’s home. The therapy was adapted from the work of Kingdon 1994, Chadwick 1996 and Fowler 1997. It also drew on other cognitive work conducted at EPPIC that aimed to assist the adaptation of the individual to psychosis (e.g. COPE; Henry 2002). A comprehensive description of ACE can be found in the ACE manual (Bendall 2005)." Number of sessions: 9 (SD = 4.93) Sessions frequency and duration: up to 6 sessions (up to 20 sessions over 14 weeks) 45 minutes each session (median time in therapy: 354 min) Therapist: clinical psychologist Therapist experience: experts (received 3 months of training in the treatments) Supervision: yes Modality: individual Delivery: face‐to‐face Treatment manualised: yes (ACE manual, Bendall 2005) 2. Control (Befriending). N = 31 Description of treatment: "Befriending aimed to control for time in therapy, participant expectations and positive experiences of therapy. Based on the Befriending therapy used by Sensky 2000, Befriending consisted of talking about neutral topics that interested the participant, such as music, sport, books, cooking and pets. If the participant found verbal interaction difficult, the participant and therapist engaged in activities such as board games, walking, or playing sport, with a view to using the activity as a tool to engage the participant in further neutral conversation during and after the activity. The therapist’s primary goals were to keep the participant engaged for the full duration of therapy and to keep the conversation or activity as close to a neutral ‘pleasant chat’ as possible. When emotionally loaded topics arose, such as symptoms or interpersonal problems, the therapist gently redirected the participant to more neutral topics. Befriending was used in this trial because it was directive; thus, therapists could control the conversation and redirect acutely psychotic participants from unstructured discussion of their psychotic symptoms, which may have been detrimental to them." Number of sessions: 7.21 (SD = 5.17) up to 6 sessions (up to 20 sessions over 14 weeks) Sessions frequency and duration: up to 6 sessions per month (up to 20 sessions over 14 weeks) 45 minutes each session (median time in therapy: 174 min) Therapist: clinical psychologist Therapist experience: experts (received 3 months of training in the treatments) Supervision: yes Modality: individual Delivery: face‐to‐face Treatment manualised: yes (Befriending manual, Bendall 2003)
Outcomes	Mental state ‐ general ‐ BPRS total endpoint (< 6 months) Mental state ‐ specific ‐ BPRS positive endpoint (< 6 months, > 1 year) Mental state ‐ SANS endpoint (< 6 months, > 1 year) Mental state ‐ CESD‐R endpoint (< 6 months) Service use ‐ admission to hospital (> 1 year) Service use ‐ number of days in hospital (> 1 year) Leaving the study early ‐ for any reason (dropout) (< 6 months, > 1 year) Functioning ‐ SOFAS endpoint (< 6 months, > 1 year) Mortality ‐ Mortality due to suicide (> 1 year)
Notes	We contacted the authors and obtained additional data.

Jackson 2009.

Study characteristics
Methods	Study design: randomised controlled trial Study grouping: parallel group Antipsychotic medication allowed: yes Duration in weeks: 26 treatment phase, 52 w follow‐up Number of study arms: 2 Type of blinding: rater blind Type of data analysis for overall efficacy: ITT Number of sites: 4 Number of countries: 1 Country: UK Sponsorship source: public Trial registration ID: ISRCTN43585723 Publication year: 2009 Researcher's allegiance: The authors developed the manual for the intervention group.
Participants	Diagnosis: first episode of non‐affective psychosis; diagnostic criteria: ICD‐10 Current clinical state: NA Definition of stability: NA Inclusion criteria: "Patients with a first episode of non‐affective psychosis conforming to ICD‐10 criteria (F20, F22, F23, F25) as verified by experienced Consultant Psychiatrists were recruited from four Mental Health Services throughout the West Midlands in the UK. The four sites served a total catchment area of approximately 1.5 million. All candidates for inclusion to the study were required to have experienced a first episode of psychosis within the previous 6–18 months and were aged between 16 and 35 years old." Exclusion criteria: "Patients were not admitted to the study if they could not speak English or were unable to give informed consent." Setting: NA Special subgroup: NA N: 66 Gender: 49 men, 17 women Age: mean 23.3 years (SD = 4.6) CBT plus standard care arm: participants total: 36, participants male: 31, participants female: 5, age: mean 24.1 years (SD = 4.7), age range: 16‐38 years, duration of untreated psychosis: 17.4 weeks (SD = 25.9), % taking antipsychotics: 86, baseline PANSS total M = 56.6, SD = 9.07, N = 36 Control arm: participants total: 30, participants male: 18, participants female: 12, age: mean 22.3 years (SD = 4.4), age range: 16‐31 years, duration of untreated psychosis: 23.7 weeks (SD = 58.4), % taking antipsychotics: 96.6, baseline PANSS total M = 55.6, SD = 10.18, N = 30
Interventions	1. CBT (cognitive therapy based recovery intervention (CRI)) plus standard care. N = 36 Description of treatment: "The cognitive therapy based recovery intervention (CRI) was designed to be delivered on a weekly basis over a 6 month period (i.e. it was limited to a maximum of 26 sessions) and followed a protocol based modular approach. In essence, the intervention arose from an individual formulation, which was translated, into an ‘‘individual recovery plan’’. This approach has been described previously in more detail in Jackson 2000. There were three key components: (a) engagement and formulation; (b) trauma processing; and (c) appraisals of psychotic illness (shame, loss and entrapment). The intervention, therefore, is not just designed for those who could be described as ‘traumatised’ by their experiences of psychosis. It is intended to be helpful for all first episode patients adjusting to and recovering from a first episode of psychosis. All participants completed the core component (engagement and formulation) and those pertinent to their problem list and goals (i.e. trauma processing and/or addressing appraisals of shame, loss and/or entrapment). Trauma processing evolved from the exploration of the primary appraisals of the first episode of psychosis including their symptoms (voices, paranoia), their management (hospital admission) and the social context in which they occurred (interpersonal reactions of others). A relapse prevention framework; ‘back in the saddle’ (Plaistow 1996) was used to aid this process. Appraisals of psychotic illness and its consequences were explored within the context of social rank theory as giving rise to shame (emotional reaction to perceived reduction of status or social rank; Gilbert 2003b), loss (of a valued role or goal) and/or entrapment (‘blocked escape’ or an inability to reaffirm an identity or sense of belonging; Rooke 1998)." "Standard cognitive therapy techniques (Socratic questioning, guided discovery, identifying and targeting beliefs and behaviours, developing alternative beliefs and reinforcing through behavioural change etc) were utilised in order to affect change in these appraisals." Number of sessions: 26 planned, 13 actually done (median) Sessions frequency and duration: 4 per month Therapist: four clinical psychologists and a cognitive behavioural psychotherapist Therapist experience: experts Supervision: yes Modality: individual Delivery: face‐to‐face Treatment manualised: yes 2. Control (Treatment as usual). N = 30 Description of treatment: NA Number of sessions: NA Sessions frequency and duration: NA Therapist: NA Therapist experience: NA Supervision: NA Modality: NA Delivery: NA Treatment manualised: NA
Outcomes	Mental state ‐ specific ‐ CDS endpoint (< 6 months, < 12 months) Leaving the study early ‐ for any reason (dropout) (< 6 months, < 12 months)
Notes	We attempted to contact the authors but it was not possible to retrieve a valid contact.

Jolley 2003.

Study characteristics
Methods	Study design: randomised controlled trial Study grouping: parallel group Antipsychotic medication allowed: yes Duration in weeks: unclear duration of treatment, assessment at 6 months Number of study arms: 2 Type of blinding: rater blind Type of data analysis for overall efficacy: completers Number of sites: NA Number of countries: 1 Country: UK Sponsorship source: other Trial registration ID: NA Publication year: 2003 Researcher's allegiance: NA
Participants	Diagnosis: schizophrenia spectrum disorder; diagnostic criteria: ICD‐10 Current clinical state: NA Definition of stability: NA Inclusion criteria: "an ICD‐10 schizophrenia spectrum disorder diagnosed < 5 years ago; 1st or 2nd episode (neither episode longer than 9 months duration); symptoms present within the previous 3 months, but not necessarily at entry." Exclusion criteria: "known organic aetiology; and alcohol or substance misuse as a primary diagnosis." Setting: both inpatient and outpatient Special subgroup: NA N: 21 Gender: 13 men, 8 women Age: mean 28.1 years (SD = 7.38 years) CBT plus standard care arm: participants total: 12, participants male: 8, participants female: 4, age: mean 25.1 years (SD = 6.2), age range: 16‐65 years, % diagnosis schizophrenia: 33.3, % diagnosis schizoaffective disorder: 8.3, % diagnosis unspecified non‐organic: 25.0, % diagnosis non‐organic: 8.3, % diagnosis persistent delusional disorder: 25.0, baseline PANSS total M = NA, SD = NA, N = NA Control arm: participants total: 9, participants male: 5, participants female: 4, age: mean 32.0 years (SD = 7.23), age range: 16‐65 years, % diagnosis schizophrenia: 22.2, % diagnosis schizoaffective disorder: 0, % diagnosis unspecified non‐organic: 44.4, % diagnosis non‐organic: 33.3, % diagnosis persistent delusional disorder: 0, baseline PANSS total M = NA, SD = NA, N = NA
Interventions	1. CBT plus standard care. N = 12 Description of treatment: "The CT comprised 18 sessions over 6 months and was adapted to place more emphasis on areas identified in pre‐pilot clinical work as important in early psychosis. These were i) processing the experiences of psychosis, and often of admission to hospital or other contact with services; Jackson 1996 work identifies a high prevalence of post‐traumatic stress symptomatology following the initial psychotic episode; ii) making sense of these experiences in the context of a personal formulation, with emphasis on developing as non‐stigmatizing as possible a view of the person’s difficulties; iii) coming to terms with the impact of psychosis on the person’s life – the losses and changes and the impact on the sense of self, the future, and relationships. Generating optimistic but realistic plans for the future, and discussing the possible course of the illness, and what this may mean in terms of future contact with services and care of self, but with an emphasis on recovery; iv) emphasis on the preservation, to the greatest extent possible, of social, occupational and educational links or rapid re‐engagement in community activities." Number of sessions: 18 sessions over 6 months planned; mean number of sessions at 6 months: 11.0 (SD = 4.7, range: 5‐17; median: 11) Sessions frequency and duration: NA Therapist: clinical psychologist Therapist experience: experts Supervision: yes Modality: individual Delivery: face‐to‐face Treatment manualised: yes (details in treatment manual by Fowler 1995) 2. Control (Treatment as usual). N = 9 Description of treatment: "Treatment as usual (TAU) was predominantly medical and social and varied from reasonably intensive community follow‐up from a team, including home visits, family contact and individual meetings to outpatient follow‐up only." Number of sessions: NA Sessions frequency and duration: NA Therapist: NA Therapist experience: NA Supervision: NA Modality: individual Delivery: face‐to‐face Treatment manualised: NA
Outcomes	Mental state ‐ general ‐ PANSS total endpoint (> 6 months) Mental state ‐ general ‐ PANSS total change (< 6 months) Mental state ‐ specific ‐ PANSS positive endpoint (< 6 months) Mental state ‐ specific ‐ PANSS positive change (< 6 months) Mental state ‐ specific ‐ PANSS negative endpoint (< 6 months) Mental state ‐ specific ‐ PANSS negative change (< 6 months) Mental state ‐ specific ‐ CDS endpoint (< 6 months) Mental state ‐ specific ‐ CDS change (< 6 months) Service use ‐ admission to hospital (< 6 months) Service use ‐ number of days in hospital (< 6 months) Leaving the study early ‐ for any reason (dropout) (< 6 months)
Notes	We contacted the authors and obtained additional data.

Lecomte 2008.

Study characteristics
Methods	Study design: randomised controlled trial Study grouping: parallel group Antipsychotic medication allowed: yes Duration in weeks: 13 w treatment, 39 w and 65 w follow‐up Number of study arms: 3 Type of blinding: rater blind Type of data analysis for overall efficacy: ITT Number of sites: 6 Number of countries: 1 Country: Canada Sponsorship source: public Trial registration ID: NCT00358709 Publication year: 2008 Researcher's allegiance: The authors developed the manual for both intervention and control.
Participants	Diagnosis: persistent or fluctuating psychotic symptoms; diagnostic criteria: NA Current clinical state: stable Definition of stability: considered “stabilised” by their psychiatrist Inclusion criteria: "Potential participants were recruited from Early Psychosis Intervention programs and community mental health clinics in Quebec and British Columbia, Canada (...). Individuals were eligible if aged between 18 and 35, fluent (verbally as well as reading and writing skills) in one of the official languages (English and French), currently presenting with persistent or fluctuating psychotic symptoms (defined as delusions or hallucinations appearing occasionally, such as in periods of stress), having consulted for the first time a mental health professional for psychotic symptoms in the past 2 years, and being followed by a psychiatrist (and therefore receiving antipsychotic medication). Individuals were only recruited once they had been discharged from the hospital and considered “stabilized” by their psychiatrist. Nonaffective psychosis was preferred but individuals with unclear diagnoses at the time of the referral were also accepted." Exclusion criteria: "Exclusion criteria included suffering from an organic disorder, having already received one of the interventions, and not being able to give informed consent." Setting: outpatient Special subgroup: first episode, stable, positive symptoms N: 129 Gender: 93 men, 36 women Age: mean 24.76 years CBT plus standard care arm: participants total: 48, participants male: 31, participants female: 17, age: mean 24.92 years, age range: 18‐35 years, duration of illness: 3.22 years, % taking antipsychotics: 100, % diagnosis schizophrenia: 53.6, % mood disorder with psychotic features: 21.4, % psychosis NOS: 25, baseline BPRS total M = 42.7, SD = 12.9, N = 48 Control arm (group social skills training): participants total: 54, participants male: 40, participants female: 14, age: mean 25.44 years, age range: 18‐35 years, duration of illness: 3.44 years, % taking antipsychotics: 100, % diagnosis schizophrenia: 58.7, % mood disorder with psychotic features: 20.8, % psychosis NOS: 20.8, baseline BPRS total M = 41, SD = 14.65, N = 54 Waiting list: participants total: 27, participants male: 22, participants female: 5, age: mean 23.1 years, age range: 18‐35 years, duration of illness: 1.4 years, % taking antipsychotics: 100, % diagnosis schizophrenia: 54.4, % mood disorder with psychotic features: 17.1, % psychosis NOS: 28.5, baseline BPRS total M = 41.3, SD = 10.11, N = 27
Interventions	1. CBT plus standard care (group CBT). N = 48 Description of treatment: "The CBT manual was developed by 3 of the authors (...) and integrates the principles and philosophy of individual CBT for psychosis, but adapted to a group format and tailored for first episodes. A detailed description of the intervention is described elsewhere (Lecomte 2003). The manual is built in 4 parts: (1) stress: how it affects me; (2) testing hypotheses and looking for alternatives, (3) drugs, alcohol, and how I feel; and (4) coping and competence. The manual follows a positive approach (rather than problem based) with specific emphasis on reaching personal goals, decreasing distress, and finding what works best for them; with the clients’ being the experts of their experience and the therapists acting as guides. The participants also learn to understand and use CBT techniques such as normalization, the ABC’s of CBT (linking events to thoughts and emotional and behavioral consequences), Socratic questioning, checking the facts, finding alternatives, trying new coping strategies, and modifying attributions." Number of sessions: 24 Sessions frequency and duration: 8 per month Therapist: occupational therapists, nurse, psychologist, social worker Therapist experience: average of six years of experience working with individuals with psychosis, though none had previous training in CBT for psychosis. The therapists received two days of intensive training. Supervision: yes Modality: group Delivery: face‐to‐face Treatment manualised: yes 2. Control (Group social skills training). N = 54 Description of treatment: "Social skills training is recognized as an evidence‐based intervention for acquiring skills necessary to live in the community (Kopelowicz 2006) and includes teaching skills such as symptom management and relapse prevention. The manual used in this study was the Symptom Management (SM) module developed by UCLA Psychiatric Rehabilitation Consultants (Liberman 1988). The treatment aims at building 4 skill areas: (1) identifying warning signs of relapse, (2) managing warning signs, (3) coping with persistent symptoms, and (4) avoiding alcohol and street drugs. Each section follows the exact same format: introduction to skill area, videotape questions and answers, roleplays, resource management, outcome problems (problem solving), in vivo exercises, and homework assignments. The intervention varies from the CBT intervention in many ways, namely by its prescriptive approach to encouraging the appropriate behaviors to diminish or cope with symptoms. The therapists are instructed to model appropriate interaction styles and behaviors, and to teach clients how to effectively use the skills by using repetition and encouragements. Though the manual was developed for more chronic clients, with one of the authors (...), it was revised to eliminate any references to chronicity or diagnoses." Number of sessions: 24 Sessions frequency and duration: 8 per month and NA duration Therapist:occupational therapists, nurse, psychologist, social worker Therapist experience: NA Supervision: yes Modality: group Delivery: face‐to‐face Treatment manualised: yes 3. Control (waiting list). N = 27 Description of the treatment: "Clients in the control group could receive 1 of the 2 treatments, should they still wish to, after being in the study for a minimum of 9 months."
Outcomes	Mental state ‐ general ‐ BPRS endpoint (< 6 months, < 12 months) Mental state ‐ specific ‐ number of participants with clinically important change in mental state (< 6 months, < 12 months) Mental state ‐ BPRS positive endpoint (< 6 months, < 12 months) Mental state ‐ BPRS negative endpoint (< 6 months, < 12 months) Leaving the study early ‐ for any reason (dropout) (< 6 months, < 12 months, > 1 year)
Notes	We contacted the authors and obtained additional information.

Lepage 2022.

Study characteristics
Methods	Study design: randomised controlled trial Study grouping: parallel group Antipsychotic medication allowed: yes Duration in weeks: 13 weeks intervention phase, 26 weeks follow‐up, 39 weeks follow‐up Number of study arms: 2 Type of blinding: rater blind Type of data analysis for overall efficacy: ITT Number of sites: 5 Number of countries: 1 Country: Canada Sponsorship source: public Trial registration ID: NCT02294409 Publication year: 2022 Researcher's allegiance: The authors developed the manual for both intervention and control.
Participants	Diagnosis: Psychotic disorder; diagnostic criteria: no information Current clinical state: stable Definition of stability: Exclusion criteria comprised: currently clinically unstable, defined as the presence of positive symptoms that are moderate‐to‐severe on the SAPS rating scale Inclusion criteria: "Participants within the first 2 years of treatment following the onset of a psychotic disorder were recruited from these programs. Inclusion criteria were: a diagnosis of a psychotic disorder; aged 18–35; ability to read and write English or French at an inter‐ mediate level (Education > 8 years); scores above predetermined cut‐offs on the three measures of SA [34 for the Social Interaction Anxiety Scale (Mattick 1998), 20 for the Social Phobia Inventory (Connor 2000) and 21 for the Brief Social Phobia Scale (Davidson 1991)]; and meet criteria for the diagnosis of SA disorder as determined with the SCID social phobia module." Exclusion criteria: "Exclusion criteria comprised: currently clinically unstable, defined as the presence of positive symptoms that are moderate to severe on the SAPS rating scale; IQ < 70; hospitalized at the time of recruitment; current diagnosis of substance dependence; lifetime history of a neurological condition." Setting: outpatient Special subgroup: first episode, stable, social anxiety N: 96 Gender: 63 men, 33 women Age: mean 24.62 years CBT plus standard care arm: participants total: 51, participants male: 31, participants female: 20, age: mean 25.33 years (SD = 4.39), age range: 18‐35 years, duration of illness: 1.56 years (SD = 1.46), % taking antipsychotics: NA, % diagnosis schizophrenia: 19.6, % diagnosis schizophreniform disorder: 13.7, % diagnosis schizoaffective disorder: 5.9, % diagnosis delusional disorder: 2, % diagnosis bipolar disorder: 13.7, % diagnosis affective disorder: 7.8 Control arm: participants total: 45, participants male: 32, participants female: 13, age: mean 23.82 years (SD = 4.36), age range: 18‐35 years, duration of illness: 1.44 years (SD = 1.5), % taking antipsychotics: NA, % diagnosis schizophrenia: 24.4, % diagnosis schizophreniform disorder: 15.5, % diagnosis schizoaffective disorder: 8.9, % diagnosis delusional disorder: 0, % diagnosis bipolar disorder: 11.1, % diagnosis affective disorder: 6.7
Interventions	1. CBT (group CBT for SA) plus standard care. N = 51 Description of treatment: "All CBT‐SA sessions were conducted in either English or French by trained clinicians with the aid of a group CBT‐SA manual that is described in Montreuil 2016. This intervention was delivered by a therapist (doctoral‐level psychologist) and a co‐therapist under the supervision of an experienced CBT therapist (...). This intervention included the five following modules: (i) Psychoeducation on SA disorder, stress, psychosis and self‐ stigma; (ii) Cognitive restructuring: identifying negative thoughts that occur before, during, or after anxiety‐provoking situations; (iii) Social Skills training, which provided an opportunity to practice interpersonal skills; (iv) Exposure component, which focused on the collection of information that would allow patients to revise their judgments about the degree of risk to which they are exposed in feared situations, and would challenge their dysfunctional beliefs; and (v) Relapse prevention and maintenance. Each of the 13 group sessions consisted of 4–8 participants and lasted for 1.5 h each. One session per week was provided for the 13‐week period." Number of sessions: 13 Sessions frequency and duration: 4 per month and 90 minutes Therapist: doctoral level psychologist and co‐therapist Therapist experience: experts Supervision: yes Modality: group Delivery: face‐to‐face Treatment manualised: no information 2. Control (Computer‐Assisted Cognitive Remediation Group Therapy (CR)). N = 45 Description of treatment: "This group intervention developed by Bowie 2012 devotes approximately 60% of the session to cognitive training activities on a computer, 20% to documenting and attempting new strategies for solving problems (‘strategic monitoring’), and 20% to doing ‘transfer activities’, which consist of simulated real‐world activities and discussion of how cognitive strategies might be applied in everyday life and potential compensatory strategies for overcoming cognitive challenges. This approach specifically targets processing speed, attention, memory, and executive function because they are domains that are commonly impaired in psychosis." Number of sessions: 13 Sessions frequency and duration: 4 per month and 90 minutes Therapist: therapist and co‐therapist Therapist experience: NA Supervision: NA Modality: group Delivery: face‐to‐face Treatment manualised: yes
Outcomes	Mental state ‐ specific ‐ SAPS endpoint (< 6 months, < 12 months) Mental state ‐ specific ‐ SANS endpoint (< 6 months, < 12 months) Mental state ‐ specific ‐ CDS (< 6 months, < 12 months) Service use ‐ admission to hospital (< 6 months) Leaving the study early ‐ for any reason (< 6 months) Functioning ‐ SOFAS endpoint (< 6 months, < 12 months)
Notes	We contacted the authors but received no reply.

Lewis 2002 (SOCRATES).

Study characteristics
Methods	Study design: randomised controlled trial Study grouping: parallel group Antipsychotic medication allowed: yes Duration in weeks: 5‐week CBT programme plus community‐based booster sessions (with an 18‐month follow‐up) Number of study arms: 3 Type of blinding: rater blind Type of data analysis for overall efficacy: ITT Number of sites: 11 Number of countries: 1 Country: England Sponsorship source: public Trial registration ID: ISRCTN65818527 Publication year: 2022 Researcher's allegiance: NA
Participants	Diagnosis: schizophrenia, schizophreniform disorder, schizoaffective disorder or delusional disorder; diagnostic criteria: DSM‐IV Current clinical state: NA Definition of stability: NA Inclusion criteria: "Inclusion criteria for subjects to enter the trial were: (a) either first or second admission (within 2 years of a first admission) to in‐patient or day patient unit for treatment of psychosis; (b) DSM–IV criteria for schizophrenia, schizophreniform disorder, schizoaffective disorder or delusional disorder (APA 1994); (c) positive psychotic symptoms for 4 weeks or more; (d) score of 4 or more (moderate or severe) on the PANSS (Kay 1989) target item either for delusions (P1) or hallucinations (P3)" Exclusion criteria: "neither substance misuse nor organic disorder judged to be the major cause of psychotic symptoms." Setting: both inpatient and outpatient Special subgroup: NA N: 309 Gender: 216 men, 93 women Age: median 24.4 years CBT plus standard care arm: participants total: 101, participants male: 72, participants female: 29, age: median 29.1 years, age range: 18‐65 years, % taking antipsychotics: NA, % diagnosis schizophrenia: 32.6, % diagnosis schizoaffective disorder: 16.8, % diagnosis schizophreniform: 37.6, % diagnosis delusional disorder: 10.9, % psychosis NOS: 2, baseline PANSS total M = 87.47, SD = 17.64, N = 101 Control (supportive counselling + treatment‐as‐usual): participants total: 106, participants male: 75, participants female: 31, age: median 27.2 years, age range: 18‐65 years, % taking antipsychotics: NA, % diagnosis schizophrenia: 45, % diagnosis schizoaffective disorder: 7.5, % diagnosis schizophreniform: 35.8, % diagnosis delusional disorder: 10.4, % psychosis NOS: 2, baseline PANSS total M = 89.22, SD = 17.53, N = 106 Control (treatment‐as‐usual): participants total: 102, participants male: 69, participants female: 33, age: median 27 years, age range: 18‐65 years, % taking antipsychotics: NA, % diagnosis schizophrenia: 41.2, % diagnosis schizoaffective disorder: 13.7, % diagnosis schizophreniform: 32.4, % diagnosis delusional disorder: 6.9, % psychosis NOS: 5.9, baseline PANSS total M = 87.01, SD = 16.81, N = 102
Interventions	1. CBT plus standard care. N = 101 Description of treatment: "In summary, treatment was conducted in four stages. The first stage was engagement and a detailed assessment of mental state and symptom dimensions (psychotic and nonpsychotic) to allow a cognitive–behavioural analysis of how symptoms might relate to cognitions, behaviour and coping strategies. Education about the nature and treatment of psychosis, using a stress vulnerability model to link biological and psychological mechanisms, was used to help engagement. Second, a problem list was generated collaboratively with the patient. This was then prioritised according to the degree of distress attached, feasibility and, where relevant, clinical risk involved. Prioritised problems were assessed in detail and a formulation was agreed which included such issues as trigger situations and cognitions. Third and fourth stages were intervention and monitoring. Interventions particularly addressed positive psychotic symptoms of delusions and hallucinations, generating alternative hypotheses for abnormal beliefs and hallucinations, identifying precipitating and alleviating factors and reducing associated distress." Number of sessions: 16.1 Sessions frequency and duration: 5‐week treatment envelope (3‐5 times per week up to 5 weeks), total therapy time: mean 8.6 hours; planned: 180‐240 min (15‐20 hours within a 5‐week treatment envelope, plus "booster" sessions at a further 2 weeks and 1, 2 and 3 months) Therapist: 5 therapists (3 clinical psychologists and 2 nurse therapists) Therapist experience: experts Supervision: yes Modality: individual Delivery: face‐to‐face Treatment manualised: yes 2. Control (Supportive counselling + treatment as usual). N = 106 Description of treatment: "Supportive counselling was used as a comparison intervention to control for non‐specific elements of therapist exposure. It was delivered in the same 5‐week format with three boosters, with the aim of matching the duration of total therapist contact time to that in the CBT arm. The supportive counselling was also manual‐based and supervised by an experienced counsellor. The same five research therapists administered both CBT and supportive counselling interventions, according to randomisation. Interventions were started within 3 days of randomisation. Patients were seen in hospital settings, family practitioner surgeries and in their own homes for treatment sessions. All treatment sessions, both for CBT and supportive counselling, were audiotaped where consent was given. After the acute phase of the study was completed, a random selection of 50 tapes were rated blindly by two independent raters, who were asked to classify them as CBT or supportive counselling sessions and to rate the quality of therapy on the Cognitive Therapy Scale for Psychosis (CTS–Psy; Haddock 2001)." Number of sessions: mean 15.7 Sessions frequency and duration: 5‐week with three boosters (3‐5 times per week up to 5 weeks), total therapy time: mean 7.1 hours; "It was delivered in the same 5‐week format with three boosters, with the aim of matching the duration of total therapist contact time to that in the CBT arm." Therapist: 5 therapists (3 clinical psychologists and 2 nurse therapists) Therapist experience: experts Supervision: yes Modality: individual Delivery: face‐to‐face Treatment manualised: yes 3. Control (treatment‐as‐usual). N = 102 Description of treatment: "Procedures to standardise routine clinical care, including drug treatment, were not used". "This means that the content of "routine care" is not specifiable, except that it always included day or in‐patient treatment and included antipsychotic drugs." Number of sessions: NA Sessions frequency and duration: NA Therapist: NA Therapist experience: NA Supervision: NA Modality: NA Delivery: NA Treatment manualised: NA
Outcomes	Mental state ‐ general ‐ PANSS total endpoint (< 6 months, > 12 months) Mental state ‐ specific ‐ PANSS positive endpoint (< 6 months, > 12 months) Mental state ‐ specific ‐ PANSS negative endpoint (< 6 months, > 12 months) Global state ‐ number of participants that relapsed (> 1 year) Service use ‐ admission to hospital (> 1 year) Leaving the study early ‐ for any reason (dropout) (< 6 months, > 1 year) Mortality ‐ overall mortality (> 1 year) Mortality ‐ mortality due to natural causes (> 1 year)
Notes

Liu 2019.

Study characteristics
Methods	Study design: randomised controlled trial Study grouping: parallel group Antipsychotic medication allowed: yes Duration in weeks: 10 weeks treatment phase, then data collection also at 36 m and 62 m follow‐up Number of study arms: 2 Type of blinding: rater blind Type of data analysis for overall efficacy: ITT Number of sites: 1 Number of countries: 1 Country: China Sponsorship source: public Trial registration ID: ChiCTR‐TRC‐13003929 Publication year: 2019 Researcher's allegiance: no
Participants	Diagnosis: schizophrenia; diagnostic criteria: DSM‐IV Current clinical state: first episode, with positive symptoms Definition of stability: NA Inclusion criteria: "(1) A diagnosis of schizophrenia as assessed by experienced research psychiatrists using the Structured Clinical Interview for DSM‐IV Axis I Disorders‐Clinician Version; (2) Age between 16–45 years; (3) The current illness episode was their first episode of schizophrenia, and any continuous treatment had been for less than 1 month or was inadequate treatment of less than 3 months verified through clinical interview and medical chart review; (4) Use of a single atypical antipsychotic drug for at least two weeks; (5) A score of 4 or more on the delusion (P1) or hallucination (P3) symptom scales of the Positive and Negative Syndrome Scale (PANSS); (6) Their first psychotic symptoms had occurred less than 3 years ago." Exclusion criteria: "(1) Co‐morbid diagnosis of mental disability or primary substance dependence; (2) Inability to communicate, or lack of spontaneity and flow of conversation (5 or above on PANSS in conceptual disorganization); (3) Electroconvulsive therapy within the 1 month prior to entry into the study; (4) Serious or unstable physical health condition; (5) Currently receiving any other form of systematic psychotherapy." Setting: outpatient Special subgroup: NA N: 80 Gender: 46 men, 34 women Age: mean 27.48 years CBT plus standard care arm: participants total: 40, participants male: 26, participants female: 14, age: mean 26.35 years, age range: 16‐45 years, % taking antipsychotics: 100, % diagnosis schizophrenia: 100, baseline PANSS total M = 68.48, SD = 15.67, N = 40 Control arm: participants total: 40, participants male: 20, participants female: 20, age: mean 28.6 years, age range: 16‐45 years, % taking antipsychotics: 100, % diagnosis schizophrenia: 100, baseline PANSS total M = 69.07, SD = 16.33, N = 40
Interventions	1. CBT plus standard care. N = 40 Description of treatment: "The main goals of this intervention were to apply the general principles of CBT, such as case formulation, goal setting, homework and the cognitive process and behavior experiment, with a focus on treatment of relapse prevention and primary positive symptoms. The intervention can be divided into three stages. In the first stage, therapists built a good rapport with patients, assessed patients’ life experiences, and developed patients’ problem list. In the intermediate stage, patients primarily learnt how to manage positive symptoms (such as delusions, hallucinations, thought disorder) via behavioural experiments, graded exposure, reattribution of symptoms and coping strategies. Emotional problems were also discussed. In the last stage, therapists focused on stress management, problem solving and medication adherence, as well as relapse prevention. Patients in stress management sessions discussed predisposing factors, stressors (e.g. sleep deprivation, trauma and posttraumatic stress symptoms), protective factors (e.g. social skills, social support and help seeking) and risk factors (e.g. social isolation, substance abuse). In relapse prevention sessions, patients learned: 1) 6 stages of relapse (quiet period, stress period, hopeless period, loss of control period, mental confusion and psychotic symptom occurrence); 2) the early signs of relapse; 3) how to make their own list of signs of relapse with unique features; 4) how to monitor their symptoms in daily life; 5) how to make emergency plans for early warning signs including grading signs, prevention strategies and plans for seeking help. At the beginning of each session, therapists reviewed patients’ homework to ensure patients had mastered the skills covered in previous sessions. In addition, all participants received treatment as usual; patients were seen monthly by a psychiatrist across the duration of the trial." Number of sessions: 10 planned, 7.4 (mean) actually done Sessions frequency and duration: 4 sessions per month; 45 minute duration Therapist: psychologist (delivered the CBT), psychiatrist (once a month) Therapist experience: experts Supervision: yes Modality: individual Delivery: face‐to‐face Treatment manualised: yes 2. Control (Treatment as usual). N = 40 Description of treatment: "TAU included medication and case management (including some psychological health education and social support). Psychiatrists evaluated all patients and formulated appropriate drug treatment, and saw patients and their family members in the outpatient unit each month, in order to supervise and encourage medication adherence and provide guidance to family members on managing adverse drug reactions. When patients presented with impulsiveness or serious adverse drug reactions, psychiatrists initiated an emergency response mechanism." "In addition, all participants received treatment as usual; patients were seen monthly by a psychiatrist across the duration of the trial." Number of sessions: NA Sessions frequency and duration: 1 per month Therapist: psychiatrist Therapist experience: NA Supervision: no Modality: individual and family Delivery: face‐to‐face Treatment manualised: no
Outcomes	Mental state ‐ general ‐ PANSS total endpoint (< 6 months, < 12 months, > 1 year) Mental state ‐ general ‐ PSYRATS endpoint (< 6 months, < 12 months, > 1 year) Mental state ‐ specific ‐ number of participants with clinically important change in mental state (> 1 year) Mental state ‐ specific ‐ participants with change in positive symptoms (< 6 months, < 12 months, > 1 year) Mental state ‐ specific ‐ PANSS positive endpoint (< 6 months, < 12 months, > 1 year) Mental state ‐ specific ‐ PANSS negative endpoint (< 6 months, < 12 months, > 1 year) Global state ‐ number of participants that relapsed (< 6 months, < 12 months, > 1 year) Service use ‐ admission to hospital (< 6 months, < 12 months, > 1 year) Leaving the study early ‐ for any reason (dropout) (< 6 months, < 12 months, > 1 year) Functioning ‐ PSP endpoint (< 6 months, < 12 months, > 1 year) Adverse events ‐ non‐compliance (< 6 months, < 12 months, > 1 year)
Notes	We contacted the authors but received no reply.

Madigan 2013.

Study characteristics
Methods	Study design: randomised controlled trial Study grouping: parallel group Antipsychotic medication allowed: yes Duration in weeks: 18 weeks (12 weeks plus a "booster" session 6 weeks later); reevaluation at 3 months and 1 year Number of study arms: 2 Type of blinding: rater blind Type of data analysis for overall efficacy: ITT Number of sites: 3 Number of countries: 1 Country: Ireland Sponsorship source: public Trial registration ID: NA Publication year: 2013 Researcher's allegiance: no
Participants	Diagnosis: diagnosis of psychosis with comorbid substance dependence; diagnostic criteria: DSM‐IV Current clinical state: patients either experiencing their first psychotic episode or within 3 years following the onset of non‐affective or affective psychosis Definition of stability: NA Inclusion criteria: "The subjects were 88 patients who were either experiencing their first psychotic episode or within 3 years following the onset of non‐affective or affective psychosis and who had presented to (i) the DETECT first‐episode psychosis service in South County Dublin/North Wicklow (population 375,000), (ii) the National Drug Treatment Centre Board in Dublin city center, or (iii) Cavan–Monaghan Mental Health Service (population 109,000); the study was approved by the Provincial Ethics Committee of the St. John of God Hospitaller Services, the Research Ethics Committee for the National Drug Treatment Centre Board and the Research Ethics Committee of the Health Service Executive Dublin North East Area, respectively. The recruitment procedure involved a poster and media campaign, including appearance on a prime‐time current affairs television program, letters of information to referred patients prior to consent, and flexible appointments for assessment at a time and venue of the patients choosing. Each patient's participation in assessment was acknowledged with the allocation of a token voucher to cover any costs of attendance. Patients were also offered the opportunity to meet with a service user who had previously participated in the psychological intervention, to clarify any outstanding issues and help alleviate any anxieties regarding involvement in the study. Once patients had given written, informed consent, patients who met inclusion criteria for the RCT were interviewed using the Structured Clinical Interview (SCID‐1) for DSM‐1V (First 1995) to confirm a diagnosis of psychosis with comorbid substance dependence. Both outpatients and inpatients aged between 16 and 65 years and without learning disability or organic brain damage were included." Exclusion criteria: "patients with learning disability or organic brain damage" Setting: both inpatient and outpatient Special subgroup: first‐episode patients, drug abuse N: 88 Gender: 69 men, 19 women Age: mean 27.8 years CBT plus standard care arm: participants total: 59, participants male: 46, participants female: 13, age: mean 27.6 years (SD = 8.4), age range: 16‐65 years, duration of untreated psychosis: 14.4 months (SD = 28.1), % taking antipsychotics: NA, % diagnosis schizophrenia: 42.37, % diagnosis schizophreniform disorder: 10.17, % diagnosis schizoaffective disorder: 1.69, % diagnosis bipolar disorder: 16.95, % diagnosis major depressive disorder: 8.47, % diagnosis NOS: 6.78 Control arm: participants total: 29, participants male: 23, participants female: 6, age: mean 28.2 years (SD = 7.7), age range: 16‐65 years, duration of untreated psychosis: 12.2 months (SD = 19.9), % taking antipsychotics: NA, % diagnosis schizophrenia: 44.83, % diagnosis schizophreniform: 10.34, % diagnosis schizoaffective disorder: 0, % diagnosis bipolar disorder: 13.79, % diagnosis major depressive disorder: 3.44, % diagnosis NOS: 0
Interventions	1. CBT plus standard care. N = 59 Description of treatment: "This arm also continued to receive care from a multidisciplinary team in their local service, similar to the TAU group. Ten patients in the GPI group had also availed of addiction counseling, primarily for opiate use, in excess of one year prior to consenting to the present trial. Group psychological intervention was provided by an experienced clinical psychologist; this specialist in cognitive therapy is a ratified member of the British Association for Behavioral and Cognitive Psychotherapies. Patients attended group sessions once per week for 12 weeks and were then invited back 6 weeks later for a ‘booster’ session. The interventions were held in community mental health centers and involved weekly sessions of anxiety management, motivational interviewing and CBT techniques (Edwards 2006); they consisted of six phases based upon motivational interviewing principles (Miller 1991; Edwards 2006), as per the Australian CAP trial, together with cognitive behavioral therapy. 1. Entry: engagement, gaining commitment to treatment and raising the issue of problem substance abuse (psychoeducation). 2. Commitment: building commitment to a goal of non‐problematic substance misuse (motivational interviewing; harm reduction). 3. Goal Setting: reinforcing commitment to change and development of goal achievement strategies. 4. Challenges: withdrawal counseling and psychoeducation. 5. Relapse prevention and lifestyle: motivational interviewing and psychoeducation. 6. Maintenance: coping skills and time management. 7. Exit: termination; provision of resources." Number of sessions: 13 (12 sessions + 1 booster) Sessions frequency and duration: 4 per month; duration NA Therapist: psychologist Therapist experience: experts Supervision: NA Modality: group Delivery: face‐to‐face Treatment manualised: no 2. Control (Treatment as usual). N = 29 Description of treatment: "This arm continued to receive care from a multidisciplinary team in their local service; this included antipsychotic treatment and regular review." Number of sessions: NA Sessions frequency and duration: NA Therapist: NA Therapist experience: NA Supervision: no Modality: individual Delivery: face‐to‐face Treatment manualised: no
Outcomes	Mental state ‐ specific ‐ SAPS endpoint (< 6 months, < 12 months) Mental state ‐ specific ‐ SANS endpoint (< 6 months, < 12 months) Mental state ‐ specific ‐ CDS endpoint (< 6 months, < 12 months) Leaving the study early ‐ for any reason (dropout) (< 6 months, < 12 months) Functioning ‐ GAF endpoint (< 6 months, < 12 months) Quality of life ‐ WHOQoL‐BREF (< 6 months, < 12 months)
Notes	We contacted the authors but received no reply.

Morrison 2018 (COMPARE).

Study characteristics
Methods	Three‐arm study (1. CBT + standard care, 2. standard care, 3. CBT alone); only arm 1 and 2 met the inclusion criteria and were used for this systematic review Study design: randomised controlled trial Study grouping: parallel group Antipsychotic medication allowed: yes Duration in weeks: 26 treatment phase + 36 weeks follow‐up Number of study arms: 3 Type of blinding: rater blind Type of data analysis for overall efficacy: ITT Number of sites: 4 Number of countries: 1 Country: UK Sponsorship source: public Trial registration ID: ISRCTN06022197 Publication year: 2018 Researcher's allegiance: The authors developed the manual for the intervention group.
Participants	Diagnosis: schizophrenia, schizoaffective disorder, or delusional disorder, or met the entry criteria for an early intervention for psychosis service; diagnostic criteria: ICD‐10 Current clinical state: NA Definition of stability: NA Inclusion criteria: "Eligible participants were aged 16 years or older; met ICD­10 criteria for schizophrenia, schizoaffective disorder, or delusional disorder, or met the entry criteria for an early intervention for psychosis service (operationally defined with the PANSS), because most individuals with first­ episode psychosis will receive care from specialist teams, as recommended by National Institute for Health and Care Excellence guidelines; were in contact with mental health services, under the care of a consultant psychiatrist; scored at least 4 on the PANSS delusions or hallucinations items, or at least 5 on suspiciousness, persecution, or grandiosity items; and had to have the capacity to consent and also had to be help ­seeking." Exclusion criteria: "Exclusion criteria were receipt of antipsychotic medication or structured CBT with a qualified therapist within the past 3 months, moderate‐­to‐­severe learning disabilities, organic impairment, a score of 5 or more on the PANSS conceptual disorganisation item, and a primary diagnosis of alcohol or substance dependence; patients who were an immediate risk to themselves or others, and those who did not speak English were also excluded." Setting: both inpatient and outpatient Special subgroup: first‐episode patients, positive symptoms N: 75 Gender: 43 men, 32 women Age: mean 23.61 years CBT plus standard care arm: participants total: 25, participants male: 14, participants female: 11, age: mean 24.44 years, age range > 16 years, duration of untreated psychosis: 39.43 weeks, % taking antipsychotics during the study: 72, baseline PANSS total M = 70.76, SD = 8.46, N = 25 Standard care arm: participants total: 24, participants male: 13, participants female: 11, age: mean 23.21 years, age range > 16 years, duration of untreated psychosis: 37.33 weeks, % taking antipsychotics during the study: 66.66, baseline PANSS total M = 70.13, SD = 10.11, N = 24 CBT only arm: participants total: 26, participants male: 16, participants female: 10, age: mean 23.19 years, age range > 16 years, duration of untreated psychosis: 44.48 weeks, % taking antipsychotics during the study: 30.7, baseline PANSS total M = 70.35, SD = 8.03, N = 26
Interventions	1. CBT plus standard care. N = 25 Description of treatment: "26 sessions of therapy based on a specific cognitive model (Morrison 2001) during a 6‐­month treatment window. Up to four optional booster sessions were available during the subsequent 6 months. Therapy was individualised and problem focused. Permissible interventions were described in the manualised treatment protocol (Morrison 2017). Therapy sessions were usually offered weekly and delivered by appropriately qualified psychological therapists. Fidelity to protocol was ensured by weekly supervision and regular rating of recorded sessions with the Cognitive Therapy Scale–Revised (Blackburn 2001)." Number of sessions: 26 (14.39 received mean) Sessions frequency and duration: 4 sessions per month with 60 min per session Therapist: appropriately qualified psychological therapist and psychiatrist Therapist experience: experts Supervision: weekly Modality: individual Delivery: face‐to‐face Treatment manualised: yes 2. Standard care. N = 24 Description of treatment: "Participants allocated to antipsychotics were prescribed medication by their responsible psychiatrist. Treatment was begun as soon as possible after randomisation. Prescribing mirrored standard clinical practice, and thus there were no restrictions on the antipsychotics that could be selected or their doses. Clinicians could switch antipsychotics and adjust doses as clinically indicated, but were encouraged to continue antipsychotic treatment for a minimum of 12 weeks, and preferably for at least 26 weeks." Number of sessions: NA Sessions frequency and duration: NA Therapist: psychiatrist Therapist experience: NA Supervision: NA Modality: NA Delivery: NA Treatment manualised: NA 3. CBT only. N = 26 Description of treatment: "26 sessions of therapy based on a specific cognitive model (Morrison 2001) during a 6‐­month treatment window. Up to four optional booster sessions were available during the subsequent 6 months. Therapy was individualised and problem focused. Permissible interventions were described in the manualised treatment protocol (Morrison 2017). Therapy sessions were usually offered weekly and delivered by appropriately qualified psychological therapists. Fidelity to protocol was ensured by weekly supervision and regular rating of recorded sessions with the Cognitive Therapy Scale–Revised (Blackburn 2001)." Number of sessions: 26 (14.39 received mean) Sessions frequency and duration: 4 sessions per month with 60 min per session Therapist: appropriately qualified psychological therapist and psychiatrist Therapist experience: experts Supervision: yes Modality: individual Delivery: face‐to‐face Treatment manualised: yes
Outcomes	Mental state ‐ general ‐ PANSS total endpoint (< 6 months, < 12 months) Mental state ‐ general ‐ number of participants with clinically important change in mental state (< 6 months, < 12 months) Mental state ‐ specific ‐ PANSS positive endpoint (< 6 months, < 12 months) Mental state ‐ specific ‐ PANSS negative endpoint (< 6 months, < 12 months) Mental state ‐ specific ‐ HADS Depression (< 6 months, < 12 months) Global state ‐ number of participants that relapsed (< 6 months, < 12 months) Global state ‐ CGI‐S endpoint (< 6 months, < 12 months) Global state ‐ CGI‐I change (< 6 months, < 12 months) Service use ‐ admission to hospital (< 12 months) Service use ‐ number of days in hospital (< 12 months) Leaving the study early ‐ for any reason (< 6 months, < 12 months) Functioning ‐ PSP endpoint (< 6 months, < 12 months) Quality of life ‐ WHOQoL‐BREF endpoint (< 6 months, < 12 months) Behaviour ‐ occurrence of violent incidents (< 12 months) Behaviour ‐ participants with self‐injury (< 12 months) Adverse events ‐ deterioration of symptoms (< 6 months, < 12 months)
Notes	We contacted the authors and obtained additional information.

Morrison 2020 (MAPS).

Study characteristics
Methods	Three‐arm study (1. CBT + standard care, 2. standard care, 3. CBT alone); only arm 1 and 2 met the inclusion criteria and were used for this systematic review Study design: randomised controlled trial Study grouping: parallel group Antipsychotic medication allowed: yes Duration in weeks: 26 weeks treatment phase + 26 weeks follow‐up Number of study arms: 3 Type of blinding: rater blind Type of data analysis for overall efficacy: ITT Number of sites: 7 Number of countries: 1 Country: UK Sponsorship source: public Trial registration ID: ISRCTN80567433 Publication year: 2020 Researcher's allegiance: The authors developed the manual for the intervention group.
Participants	Diagnosis: schizophrenia; diagnostic criteria: ICD‐10 Current clinical state: first‐episode Definition of stability: NA Inclusion criteria: "Eligible participants were aged 14–18 years; help‐seeking; presented with first‐episode psychosis (defined as being within 1 year of presentation to mental health services with psychosis symptoms); under the care of a psychiatrist within an Early Intervention in Psychosis (EIP) service or Child and Adolescent Mental Health Service (CAMHS); symptomatic at the time of randomisation (baseline), defined by a score of 4 or higher on the PANSS delusions or hallucinations subscales for at least 7 consecutive days; and met either the ICD‐10 criteria for schizophrenia, schizoaffective disorder, or delusional disorder, or the entry criteria for an EIP service for first‐episode psychosis at baseline. All participants had to have the capacity to provide informed, written consent to enter the trial. Participants aged 14–15 years also needed to have a parent or guardian willing to provide initial written consent for the research team to contact their child." Exclusion criteria: "Individuals who met any of the following criteria were excluded: receipt of antipsychotics or structured psychological intervention within the past 3 months; non‐English speaking; scored 5 or higher on the PANSS conceptual disorganisation item (to maximise the likelihood that those allocated to talking therapies would be able to engage in conversation with the therapist); were deemed an immediate risk to themselves or others by their psychiatrist or care coordinator; diagnoses of moderate‐to‐severe learning disabilities, ICD‐10 organic psychosis, or primary alcohol or substance dependence." Setting: NA Special subgroup: children or adolescents aged 14‐18 years with an early onset psychosis, positive symptoms N: 61 Gender: 30 male, 29 female, 2 non‐binary Age: mean 16.3 years CBT plus standard care arm: participants total: 21, participants male: 10, participants female: 10, age: mean 16.2 years (SD = 1.3), age range 14‐18 years, duration of untreated psychosis: 8 months, % taking antipsychotics: 57.1, baseline PANSS total M = 75.9, SD = 14.8, N = 21 Standard care arm: participants total: 22, participants male: 12, participants female: 9, age: mean 16.4 years (SD = 1.3), age range 14‐18 years, duration of untreated psychosis: 9 months, % taking antipsychotics: 72.7, baseline PANSS total M = 74.8, SD = 12.2, N = 22 CBT only arm: participants total: 18, participants male: 8, participants female: 10, age: mean 16.3 years (SD = 1.4), age range 14‐18 years, duration of untreated psychosis: 12 months, % taking antipsychotics: 33, baseline PANSS total M = 72.9, SD = 9.7, N = 18
Interventions	1. CBT plusstandard care. N = 21 Description of treatment: "Participants allocated to receive antipsychotics plus psychological intervention were offered all treatments as described for the monotherapy groups. All participants in the trial were able to receive any other concomitant therapies throughout the trial including mental health medications (this could include antipsychotics in the psychological intervention group), and psychological therapies (this could include CBT or family intervention in the antipsychotics group)." Number of sessions: 26 CBT + 6 family (optional) + 4 CBT (booster); 15 median actually received Sessions frequency and duration: 4 individual sessions and 1 family session per month Therapist: appropriately trained therapist Therapist experience: experts Supervision: yes Modality: individual Delivery: face‐to‐face Treatment manualised: yes 2. Standard care. N = 22 Description of treatment: "For participants allocated to receive drug intervention, antipsychotics were prescribed by the treating psychiatrist in their care team. Psychiatrists were asked to prescribe in line with NICE guideline CG155. They were encouraged to commence treatment as soon as possible following randomisation and to maintain treatment for at least 3 months, but preferably for 6 months or longer. Psychiatrists made decisions about the type and dose of antipsychotic consistent with their usual practice, and could change antipsychotic and dose as clinically required in response to monitoring of efficacy and adverse effects. The psychiatrists within the MAPS team (...) were available by phone or email to discuss antipsychotic prescribing with the participant’s psychiatrist." Number of sessions: NA Sessions frequency and duration: NA Therapist: psychiatrist Therapist experience: NA Supervision: NA Modality: individual Delivery: face‐to‐face Treatment manualised: NA 3. CBT only. N = 18 Description of treatment: "Participants allocated to receive psychological intervention were offered up to 26 h of individual CBT and up to six optional sessions of family intervention (plus regular communication with family members following CBT sessions for individuals who consented to the sharing of information) by appropriately trained therapists over a 6‐month treatment period, and up to 4 booster sessions of CBT following the treatment period. CBT sessions were typically once a week and family intervention once a month, and were generally delivered by the same therapist. Both interventions were informed by an integrative cognitive model. In the initial phase of CBT, patients and therapists collaboratively identified problems and agreed on goals to work on in CBT and an individualised maintenance formulation was developed. Subsequent phases focused on change strategies with interventions described in a published manual (Morrison 2017), historical formulations (ie, factors leading to the development of first‐episode psychosis), and a final consolidation phase focusing on relapse prevention. Family intervention was based on the behavioural family therapy approach (Fadden 2004). After an initial session involving assessment, formulation sharing, and agreeing goals and problems to be worked on, family intervention involved aspects such as psychoeducational work, provision of normalising information and recovery‐oriented information, problem solving, and relapse prevention planning." Number of sessions: 26 CBT + 6 family (optional) + 4 CBT (booster); 14 median actually received Sessions frequency and duration: 4 individual sessions and 1 family session per month Therapist: appropriately trained therapist Therapist experience: experts Supervision: yes Modality: individual Delivery: face‐to‐face Treatment manualised: yes
Outcomes	Mental state ‐ general ‐ PANSS total endpoint (< 6 months, < 12 months) Mental state ‐ specific ‐ number of participants with clinically important change in mental state (< 6 months, < 12 months) Mental state ‐ specific ‐ PANSS positive endpoint (< 6 months, < 12 months) Mental state ‐ specific ‐ PANSS negative endpoint (< 6 months, < 12 months) Mental state ‐ specific ‐ HADS Depression endpoint (< 6 months, < 12 months) Global state ‐ number of participants that relapsed (< 6 months, < 12 months) Service use ‐ admission to hospital (< 6 months) Leaving the study early ‐ for any reason (dropout) (< 6 months, < 12 months) Functioning ‐ FESFS endpoint (< 6 months, < 12 months) Quality of life ‐ EQ‐5D‐5L endpoint (< 6 months, < 12 months) Behaviour ‐ occurrence of violent incidents (< 12 months) Behaviour ‐ self‐injury (< 12 months) Behaviour ‐ suicide attempts (< 12 months) Adverse events ‐ at least one adverse event (< 12 months) Adverse events ‐ deterioration of symptoms (< 12 months)
Notes	We contacted the authors and obtained additional information.

Müller 2020.

Study characteristics
Methods	Study design: randomised controlled trial Study grouping: parallel group Antipsychotic medication allowed: yes Duration in weeks: 39 Number of study arms: 2 Type of blinding: rater blind Type of data analysis for overall efficacy: ITT Number of sites: 3 Number of countries: 1 Country: Sponsorship source: public Trial registration ID: NCT00465920 Publication year: 2020 Researcher's allegiance: The authors developed the manual for the intervention group.
Participants	Diagnosis: current diagnosis of schizophrenia, schizophreniform, schizoaffective (DSM IV 295.1‐0.9) or delusional disorder (DSM IV 297.1) according to the SCID‐P Diagnostic criteria: DSM‐IV Current clinical state: moderate‐to‐severe psychotic symptoms Definition of stability: moderate‐to‐severe psychotic symptoms Inclusion criteria: "Eligibility criteria for participants were: (1) current diagnosis of schizophrenia, schizophreniform, schizoaffective (DSM IV 295.1‐0.9) or delusional disorder (DSM IV 297.1) according to the SCID‐P; (2) moderate to severe positive symptoms as evidenced by a score of 4 or higher in at least one of the three items P1 (delusions), P3 (hallucinatory behavior), and G9 (unusual thought content) of the Positive and Negative Syndrome Scale (PANSS) (Kay 1987); (3) persistence of positive symptoms for at least three months; (4) age 14–20 years (it is a feature of the German health care system that young people can be treated in child and youth psychiatry up to the age of 20 years. Therefore, we have set the upper bound of our inclusion criteria to 20 years. Patients 19 years and older only are only included if they are still being treated in a youth setting); (5) sufficient fluency of the German language; and (6) ability to give informed consent." Exclusion criteria: "Exclusion criteria were defined as (1) organic disorder, (2) primary substance use disorder in the past 3 months, (3) verbal IQ below 70, (4) travel time to the study center of more than 1 hour, and (5) specific psychotherapy other than the study intervention during the study period." Setting: both inpatient and outpatient Special subgroup: children and adolescents with an early onset psychosis (EOP) N: 25 Gender: 14 male, 11 female Age: 17.28 years CBT plus standard care arm: participants total: 13, participants male: 7, participants female: 6, age: mean 17.46 years (SD = 1.51), age range 14‐20 years, duration of illness: 6.38 years, % taking antipsychotics: 100, baseline PANSS total M = 59.5, SD = 12.69, N = 13 Control arm: participants total: 12, participants male: 7, participants female: 5, age: mean 17.08 years (SD = 1.38), age range 14‐20 years, duration of illness: 3.66 years, % taking antipsychotics: 100, baseline PANSS total M = 68.1, SD = 13.48, N = 12
Interventions	1. CBT plus standard care. N = 13 Description of treatment: "CBT consisted of 20 individual sessions over a period of 9 months. The first four sessions were provided every week, followed by 16 sessions on a fortnightly basis. However, the timing could be adapted flexibly to specific needs of clients. After the initial assessment and engagement phase, the manual provided modules addressing a range of symptom areas (delusions, hallucinations, negative symptoms, and/or comorbid symptoms). Towards the end of treatment, strategies for stabilization and relapse prevention were offered. If parents or caregivers were available, five sessions were provided on an optional basis, with the last session addressing both parents/caregivers and the index patient. While the first three of these sessions focused on psychoeducation and were meant to take place in the early stages of the individual therapy, the last two of these sessions focused on relapse prevention and were to take place near the end of the individual therapy." Number of sessions: 20 Sessions frequency and duration: first four sessions were provided every week; the following 16 sessions on a fortnightly basis; "Participants of the study group attended a mean of 16.54 sessions (SD = 3.71, range 9–20), and 76.9% (n = 10) of CBT participants were considered to be completers (participated in a minimum of 15 sessions)." Therapist: four clinical psychologists Therapist experience: experts Supervision: yes Modality: individual Delivery: face‐to‐face Treatment manualised: yes 2. Control (Treatment as usual). N = 12 Description of treatment: "Patients assigned to the control group received standard outpatient care within the German health care system. TAU included antipsychotic medication as well as supportive counseling, i.e., regular appointments with the psychiatrist in charge. Patients may have received additional interventions such as occupational therapy or art therapy but specific psychotherapy sessions were not allowed." Number of sessions: NA Sessions frequency and duration: NA Therapist: NA Therapist experience: NA Supervision: NA Modality: individual Delivery: face‐to‐face Treatment manualised: no
Outcomes	Mental state ‐ general ‐ PANSS total endpoint (< 12 months, > 1 year) Mental state ‐ specific ‐ PANSS positive endpoint (< 12 months, > 1 year) Mental state ‐ specific ‐ PANSS positive change (< 12 months, > 1 year) Mental state ‐ specific ‐ PANSS negative endpoint (< 12 months, > 1 year) Mental state ‐ specific ‐ PANSS negative change (< 12 months, > 1 year) Mental state ‐ specific ‐ CDS endpoint (< 12 months, > 1 year) Mental state ‐ specific ‐ CDS change (< 12 months, > 1 year) Global state ‐number of participants with remission (< 12 months, > 1 year) Service use ‐ admission to hospital (< 12 months, > 1 year) Leaving the study early ‐ for any reason (dropout) (< 12 months, > 1 year) Functioning ‐ GAF endpoint (< 12 months, > 1 year) Functioning ‐ GAF change (< 12 months, > 1 year) Quality of Life ‐ MSQoL endpoint (< 12 months, > 1 year) Quality of Life ‐ MSQoL change (< 12 months, > 1 year) Behaviour ‐ suicide attempts (< 12 months, > 1 year) Adverse events ‐ deterioration of symptoms (< 12 months, > 1 year)
Notes	We contacted the authors and obtained additional information.

Palma‐Sevillano 2019.

Study characteristics
Methods	Study design: randomised controlled trial Study grouping: parallel group Antipsychotic medication allowed: yes Duration in weeks: 12 months intervention, 18 months and 5 years follow‐up Number of study arms: 2 Type of blinding: rater blind Type of data analysis for overall efficacy: probably ITT Number of sites: 1 Number of countries: 1 Country: Spain Sponsorship source: public Trial registration ID: NA Publication year: 2019 Researcher's allegiance: The authors developed the manual for the intervention group.
Participants	Diagnosis: initial phase of schizophrenia; diagnostic criteria: DSM‐IV Current clinical state: initial phase of schizophrenia Definition of stability: NA Inclusion criteria: "(a) patients had to be in the initial phase of schizophrenia (<3 years from the first episode) and must not have been diagnosed with schizoaffective disorder, intellectual disabilities or language difficulties; (b) they had to be users of the Mataró Mental Health Service (Barcelona). Diagnoses were established by two experienced professionals based on the Structured Clinical Interview for DSM‐IV Axis I Disorders (SCID) (APA 2000) and a review of the clinical history." Exclusion criteria: "the diagnosis of schizoaffective disorder, severe mental retardation and language difficulties were not included in the trial". Setting: both inpatient and outpatient Special subgroup: NA N: 62 Gender: 46 male, 16 female Age: mean 25.5 years CBT plus standard care arm: participants total: 35, participants male: 24, participants female: 11, age: mean 25.5 years (SD = 4.8), age range NA years, duration of untreated psychosis: 39.7 years (SD = 53.9), % taking antipsychotics: NA, % diagnosis schizophrenia: 100, baseline PANSS total M = 105.2, SD = 35.63, N = 35 Control arm: participants total: 27, participants male: 22, participants female: 5, age: mean 25.6 years (SD = 4.9), age range NA years, duration of untreated psychosis: 64.7 years (SD = 86.9), % taking antipsychotics: NA, % diagnosis schizophrenia: 100, baseline PANSS total M = 102.8, SD = 30.43, N = 27
Interventions	1. CBT plus standard care. N = 35 Description of treatment: "The intervention program lasted for 12 months, with a total of 34 (weekly) sessions (of 45 min each), carried out at the Mental Health Centre. The PIPE early intervention program consists of three different therapeutic components: psychoeducation and cognitive‐motivational therapy conducted with both the individual and the family. The PIPE program has been described elsewhere (Palma 2015). Motivational techniques were used to increase patients’ motivation to change, help them overcome their ambivalence about treatment, promote their adherence to both pharmacological and psychological treatment, change substance use habits (in the case of patients who used drugs), increase daily life activities, maintain hygiene habits (in the case of patients with difficulties in this area), treatment compliance, and increase their desire to start to engage in occupational activities. These techniques were combined with cognitive therapy (e.g., cognitive restructuration, exposure, coping strategies for delusions and/or hallucinations). The techniques of the IM promote change processes to generate an intrinsic motivation in its different phases (pre‐contemplative, contemplative, active phase, maintenance and relapse) (Palma 2005). Individual CBT was used to treat delusional ideas and beliefs, hallucinations, neurotic cognitive biases, and obsessive and co‐morbid symptoms (depression and anxiety) (Birchwood 1995). Family intervention was aimed at fostering attitudes conducive to communication (reduce expressed emotion) and at promoting improved adaptation to the illness within the family. As in individual therapy, the goal of the work with the family was to promote healthy and functional changes, in accordance with MI." Number of sessions: 34 Sessions frequency and duration: 4 sessions per month, 45 minutes each Therapist: psychologist Therapist experience: experts Supervision: yes Modality: individual and family Delivery: face‐to‐face Treatment manualised: yes 2. Control (Treatment as usual). N = 27 Description of treatment: "The patients received the usual treatment for the initial phase of schizophrenia offered by the Spanish Mental Health Care system, namely pharmacological treatment prescribed by a regular psychiatrist. Nursing care and monthly interventions by social workers were also included. The intervention consisted of quarterly psychiatric visits for the purposes of pharmacological follow‐up. The nursing visits were carried out monthly to improve patients’ health, diet and treatment compliance. Issues related to the increase of occupational activity were treated by social workers on a quarterly basis. Psychologists participated only where specific assessments or interventions were necessary." Number of sessions: NA Sessions frequency and duration: NA Therapist: psychiatrist, nurse, social workers Therapist experience: NA Supervision: NA Modality: individual Delivery: face‐to‐face Treatment manualised: no
Outcomes	Mental state ‐ general ‐ PANSS total endpoint (< 6 months, < 12 months, > 1 year) Mental state ‐ specific ‐ PANSS positive endpoint (< 6 months, < 12 months, > 1 year) Mental state ‐ specific ‐ PANSS negative endpoint (< 6 months, < 12 months, > 1 year) Global state ‐ CGI endpoint (< 6 months, < 12 months, > 1 year) Service use ‐ number of days in hospital (< 6 months, < 12 months, > 1 year) Leaving the study early ‐ for any reason (dropout) (< 6 months, < 12 months, > 1 year) Functioning ‐ GAF endpoint (< 6 months, < 12 months, > 1 year) Adverse events ‐ deterioration of symptoms (> 1 year)
Notes	We contacted the authors but received no reply.

Pos 2019.

Study characteristics
Methods	Study design: randomised controlled trial Study grouping: parallel group Antipsychotic medication allowed: yes Duration in weeks: maximum of 3 months and 6 months post‐treatment follow‐up Number of study arms: 2 Type of blinding: rater blind Type of data analysis for overall efficacy: ITT Number of sites: 5 Number of countries: 1 Country: Netherlands Sponsorship source: public Trial registration ID: NCT03217955 Publication year: 2019 Researcher's allegiance: The authors developed the manual for the intervention group.
Participants	Diagnosis: schizophrenia or a related disorder with onset of their first psychotic episode < 4 years prior to inclusion; diagnostic criteria: DSM‐IV‐TR Current clinical state: participants with a recent‐onset psychotic disorder Definition of stability: NA Inclusion criteria: "To be eligible to participate in the study, participants had to be between 18 and 36 years old and diagnosed with DSM–IV–TR (APA 2000) schizophrenia or a related disorder with onset of their first psychotic episode < 4 years prior to inclusion. Diagnosis was confirmed with the Structured Clinical Interview for DSM Disorders (First 2002). Only participants with at least a mild level of social withdrawal behavior, defined as a score of 3 on apathy/social withdrawal as measured with the negative scale of the Positive and Negative Syndrome Scale (PANSS; Kay 1988), or 2 on the social isolation items of the Brief Negative Symptom Scale (BNSS; Kirkpatrick 2011) were included." Exclusion criteria: "Patients with a comorbid diagnosis of bipolar disorder or autism spectrum disorder were not included in this study. The experience of positive symptoms was not an exclusion criterion. However, individuals were not eligible when negative symptoms were primarily the result of hallucinations or delusions (e.g., withdrawal due to paranoid delusions)." Setting: both inpatient and outpatient Special subgroup: primary negative symptoms, first‐episode patients N: 99 Gender: 80 men, 19 women Age: mean 25.43 years CBT plus standard care arm: participants total: 49, participants male: 37, participants female: 12, age: mean 25.14 years (SD = 4.47), age range 18‐36 years, % taking antipsychotics: 85.4, % diagnosis schizophrenia: 59.2, % diagnosis schizoaffective disorder: 14.3, % diagnosis psychotic disorder NOS: 20.4, % diagnosis other psychotic diagnosis: 8.2, baseline PANSS total M = 57.95, SD = 9.08, N = 49 Control arm: participants total: 50, participants male: 43, participants female: 7, age: mean 25.72 years (SD = 4.44), age range 18‐36 years, % taking antipsychotics: 98.0, % diagnosis schizophrenia: 68.0, % diagnosis schizoaffective disorder: 6.0, % diagnosis psychotic disorder NOS: 18.0, % diagnosis other psychotic diagnosis: 8.0, baseline PANSS total M = 55.85, SD = 8.79, N = 50
Interventions	1. CBT plus standard care. N = 49 Description of treatment: "The overall aim of the treatment was to modify dysfunctional beliefs and to increase engagement in constructive social activity in individuals with prominent negative symptoms." "During the first session of the group therapy, psychoeducation with a focus on negative symptoms (e.g., anhedonia, lack of energy, cognitive deficits, problem in social interaction and social withdrawal) was provided. Patients were asked to share their experiences and to define their individual social goals (e.g., [re] connect with friends, [re] engage in team sport) aimed to increase social interaction. To promote peer support as one of the major components of the group‐based intervention part, buddy couples were formed. These couples worked together during sessions and were encouraged to support each other between sessions in completing their homework (e.g., by sending reminders). At the start of the second session, goals with the highest personal value were selected and divided into smaller steps necessary to reach these goals. Participants were asked to rate which obstacles in reaching their goals they anticipated. Emphasis was placed on dysfunctional beliefs regarding one’s own cognitive functioning, skills, or the expected (lack of) pleasure or social devaluation (e.g., “I cannot enjoy things anymore,” “I am unable to concentrate and memorize anything and will not be able to have an interesting conversation”). Psychoeducation during this session was specifically focused on the cognitive model of negative symptoms, emphasizing the role of dysfunctional beliefs, avoidance behavior, and demoralization. Subsequent sessions focused on challenging anticipated and experienced obstacles using standardized material such as cognitive restructuring worksheets, role‐plays, and on‐site behavioral experiments. Specific actions to reach one’s personal goal were planned in detail during the sessions, and the participant was asked to carry out that step before the next planned session as a homework assignment. At the start of each following session, homework assignments were discussed. To maximize a sense of personal efficacy, each failure of the patient to execute a new behavior was interpreted as the therapist’s fault (e.g., due to insufficient preparation or setting goals too high). The underlying idea is that small successes are prerequisite to the experience of hope and personal efficacy. To facilitate engagement and generalization of learned skills to daily functioning, participants were encouraged to actively participate in discussions, role‐plays and behavior experiments." " Individual therapy sessions were aimed at the continuation of social activation and achievement of personal goals through a personalized focus on the person’s main dysfunctional beliefs and associated behaviors, and countering obstacles to these goals. Rather than providing a session‐by‐session protocol, the treatment manual provided material for psychoeducation and interventions that could be used if needed. Intervention tools were based on different elements of demoralization (e.g., cognitive techniques used to investigate dysfunctional beliefs and behavior, such as Socratic dialogue, behavioral and cognitive experiments; dimensional evaluation of negative or stereotype self‐image etc.; cognitive imagery techniques (imagining the steps needed to achieve new goals); and behavioral techniques (activity scheduling, exposure to new situations that trigger anxiety etc.). The choice for a particular combination of the above‐mentioned techniques was based on the individual needs and goals." Number of sessions: 22 Sessions frequency and duration: 8 group sessions of 1 hour/twice a week followed by six weekly individual sessions of 45 min Therapist: MSc level psychologists provided the group and individual CBT sessions. Therapist experience: experts Supervision: yes Modality: individual and group Delivery: face‐to‐face Treatment manualised: yes 2. Control (Treatment as usual). N = 50 Description of treatment: "Every participant in the TAU condition received either inpatient treatment, day‐treatment or outpatient treatment at one of the collaborating (local community) mental health centers. Across sites, TAU consisted of early intervention programs where patient’s symptoms, functioning and medication use are monitored for 3 years. At minimum, treatment as usual consisted of antipsychotic medication and supportive therapy. In addition, TAU could involve psycho‐education, family support, physical health care, psychomotor therapy and vocational therapy. The latter includes individual placement and support (IPS), intended to support patients in their efforts to achieve employment and reintegration." Number of sessions: NA Sessions frequency and duration: NA Therapist: psychiatrists, psychologists, psychiatric nurses and social workers Therapist experience: NA Supervision: no Modality: individual and group Delivery: face‐to‐face Treatment manualised: no
Outcomes	Mental state ‐ general ‐ PANSS total endpoint (< 6 months, < 12 months) Mental state ‐ specific ‐ PANSS positive endpoint (< 6 months, < 12 months) Mental state ‐ specific ‐ PANSS negative endpoint (< 6 months, < 12 months) Mental state ‐ specific ‐ BNSS endpoint (< 6 months, < 12 months) Global state ‐ number of participants with clinically important change in global state (< 6 months, < 12 months) Leaving the study early ‐ for any reason (dropout) (< 6 months, < 12 months) Functioning ‐ GAF endpoint (< 6 months, < 12 months) Quality of life ‐ EQ‐5D (WHO) (< 6 months, < 12 months)
Notes	We contacted the authors and obtained additional data.

Puig‐Navarro 2020.

Study characteristics
Methods	Study design: randomised controlled trial Study grouping: parallel group Antipsychotic medication allowed: yes Duration in weeks: 15 weeks Number of study arms: 2 Type of blinding: NA Type of data analysis for overall efficacy: ITT Number of sites: NA Number of countries: 1 Country: Spain Sponsorship source: Hospital Clinic of Barcelona Trial registration ID: NCT03261557 Publication year: 2020 Researcher's allegiance: NA
Participants	Diagnosis: schizophrenia spectrum disorder (schizophrenia, schizoaffective disorder, psychotic disorder not otherwise specified); diagnostic criteria: DSM‐V Current clinical state: clinically stabilised Definition of stability: "in non‐acute treatment" Inclusion criteria: "schizophrenia spectrum disorder (schizophrenia, schizoaffective disorder, psychotic disorder not otherwise specified), diagnosed before 18; being clinically stabilized (outpatients)" Exclusion criteria: "IQ < 70; having a substance dependence disorder; having a neurological disorder" Setting: outpatient Special subgroup: onset of a psychotic disorder before age 18 (EOP) N: 30 Gender: NA Age: NA CBT plus standard care arm: participants total: 16, age range 12‐22 years Control arm: participants total: 14, age range 12‐22 years
Interventions	1. CBT plus standard care (CBSST‐Adol). N = 16 Description of treatment: "The Cognitive‐Behvioral Social Skills Training (CBSST) is an evidence‐based psychological intervention for patients with schizophrenia." "The CBSST‐Adol version included 3 original modules (cognitive abilities, social skills, problem‐solving skills). It will consist of 1 individual sessions and 15 weekly periodicity group sessions distributed in 3 modules: cognitive skills module, which includes the basic principles of CBT for psychosis; social skills module, which includes the usual strategies in HHSS training; and problem solving module." Number of sessions: 16 (1 individual session + 15 weekly group sessions) Sessions frequency and duration: weekly group sessions Therapist: NA Therapist experience: NA Supervision: NA Modality: individual and group sessions Delivery: face‐to‐face Treatment manualised: NA 2. Control (PSYCHOED‐Adol). N = 14 Description of treatment: "The active control intervention, PSYCHOED‐Adol, included 3 modules (psychoeducation, relapse prevention, healthy life style)." Number of sessions: 16 (1 individual session + 15 weekly group sessions) Sessions frequency and duration: weekly group sessions Therapist: NA Therapist experience: NA Supervision: NA Modality: individual and group sessions Delivery: face‐to‐face Treatment manualised: NA
Outcomes	Leaving the study early ‐ for any reason (dropout) (< 6 months)
Notes	We contacted the authors and obtained additional information.

Santos 2008.

Study characteristics
Methods	Study design: randomised controlled trial Study grouping: parallel group Antipsychotic medication allowed: yes Duration in weeks: The intervention was for a 12‐month period. Assessments on the primary outcome measures were undertaken at baseline, 6‐month, 12 months and 18‐month follow‐up. Number of study arms: 2 Type of blinding: NA Type of data analysis for overall efficacy: NA Number of sites: 9 Number of countries: NA Country: Portugal Publication year: 2008 Researcher's allegiance: NA
Participants	Current clinical state: first‐episode schizophrenia N: 57 Gender: 39 men, 18 women Age: mean 28 years
Interventions	1. CBT plus standard care. Description of treatment: "Manualised cognitive‐behavioural treatment delivered within a therapeutic case management framework (CBCM) for young people with first‐episode schizophrenia and related psychotic disorders." Treatment manualised: yes 2. Control (Treatment as usual) Description of treatment: NA
Outcomes	No outcome data available for this study
Notes	We contacted the authors but received no reply.

Supereden3.

Study characteristics
Methods	Study design: randomised controlled trial Study grouping: parallel group Antipsychotic medication allowed: yes Duration in weeks: 39 intervention phase and 65, 104 follow‐up Number of study arms: 2 Type of blinding: rater blind Type of data analysis for overall efficacy: ITT Number of sites: 4 Number of countries: 1 Country: UK Sponsorship source: public Trial registration ID: ISRCTN61621571 Publication year: 2018 Researcher's allegiance: The authors developed the manual for the intervention group.
Participants	Diagnosis: non‐affective psychosis; diagnostic criteria: no information Current clinical state: "early non‐affective psychosis" Definition of stability: NA Inclusion criteria: "Eligible patients were aged 16–35 years; had non‐affective psychosis; were clients of early intervention services in Birmingham, Lancashire, Norfolk, and Sussex (for entry into early intervention services in the UK, individuals are required to have psychotic symptoms with a Positive and Negative Syndrome Scale [PANSS] score of ≥ 4); had been clients of early intervention services for 12–30 months; and had low levels of structured activity after at least 1 year of treatment in early intervention services (defined as ≤ 30 h/week on the Time Use Survey (Hodgekins 2015))." Exclusion criteria: "We excluded patients if they were part of the original National EDEN cohort (Birchwood 2014), did not speak adequate English to engage in the intervention, and were deemed too unwell to engage with the intervention." Setting: NA Special subgroup: low levels of structured activity N: 155 Gender: 116 men, 38 women Age: mean 24.78 years CBT plus standard care arm: participants total: 76, participants male: 56, participants female: 19, age: mean 24.87 years (SD = 6.28), age range 16‐35 years, duration of illness: 2.04 years, duration of untreated psychosis: 19.14 weeks, % taking antipsychotics: NA, baseline PANSS total M = 58.33, SD = 24.94, N = 75 Control arm: participants total: 79, participants male: 60, participants female: 19, age: mean 24.7 years (SD = 4.24), age range 16‐35 years, duration of illness: 2.33 years, duration of untreated psychosis: 19.14 weeks, % taking antipsychotics: NA, baseline PANSS total M = 63.3, SD = 20.38, N = 79
Interventions	1. CBT plus standard care. N = 76 Description of treatment: "In the intervention group, early intervention services were augmented by social recovery therapy delivered by a therapist who was trained and supervised by the trial team." "In brief, social recovery therapy is delivered in three stages. Stage one involves engagement and development of a formulation, which comprises establishment of a working therapeutic relationship to facilitate engagement and identify a problem list. Alongside this approach is a detailed assessment of personal motivation and pre‐morbid hopes, expectations, and goals, which might have changed with respect to the effect of illness. Specific behavioural assessment is done in vivo to assess how symptoms affect activity. Links are identified between personally meaningful values and goals and achievable day‐to‐day activity targets. Stage two involves preparing for new activities, whereby the client and therapist work together to identify pathways to meaningful new activities. This strategy includes referral to relevant vocational agencies, education providers, and community providers of social or sports activities. Cognitive work at this stage involves promoting a sense of agency and addressing hopelessness, feelings of stigma, and negative beliefs about self and others. Behavioural experiments start focusing on managing symptoms while engaging in activity. Stage three involves engagement in new activities, which involves the active promotion of social activity using behavioural experiments, and fostering feelings of mastery and agency. The behavioural experiments are progressively shaped to address specific problems presented by individuals. Therapists adopt an assertive outreach style of contact, most frequently visiting people at home or in community settings. Therapists are also encouraged to work systematically with family members, employers, and educational providers to discuss and overcome potential problems that could impede social recovery." Number of sessions: 16.49 mean Sessions frequency and duration: NA Therapist: therapists were either qualified clinical psychologists or cognitive behavioural therapists. Therapist experience: experts Supervision: yes Modality: individual and group Delivery: face‐to‐face Treatment manualised: yes (Fowler 2013) 2. Control (Treatment as usual). N = 79 Description of treatment: "These interventions include intensive and assertive recovery‐oriented case management, supported employment, peer support, group interventions, family work and CBT for psychosis, and psychiatric medications and medical and psychiatric monitoring. All participants had an early intervention services case manager who provided oversight of their care and remained in contact with the participant throughout the trial." Number of sessions: NA Sessions frequency and duration: NA Therapist: NA Therapist experience: experts Supervision: NA Modality: individual and group Delivery: face‐to‐face Treatment manualised: NA
Outcomes	Mental state ‐ general ‐ PANSS total endpoint (< 12 months, > 1 year) Mental state ‐ specific ‐ SANS endpoint (< 12 months, > 1 year) Mental state ‐ specific ‐ BDI endpoint (< 12 months, > 1 year) Leaving the study early ‐ for any reason (dropout) (< 12 months, > 1 year) Mortality ‐ overall mortality (< 12 months)
Notes	We contacted the authors and they provided information.

Sönmez 2020.

Study characteristics
Methods	Study design: randomised controlled trial Study grouping: parallel group Antipsychotic medication allowed: yes Duration in weeks: 65 weeks: 6 months (end of therapy) + 15 months (follow‐up) Number of study arms: 2 Type of blinding: rater blind Type of data analysis for overall efficacy: NA Number of sites: 4 Number of countries: 1 Country: Norway Sponsorship source: ExtraStiftelsen Trial registration ID: NCT01511406 Publication year: 2020 Researcher's allegiance: no
Participants	Diagnosis: primary diagnosis of psychosis spectrum disorder; diagnostic criteria: DSM‐IV Current clinical state: NA Definition of stability: NA Inclusion criteria: "All participants had a primary diagnosis of psychosis spectrum disorder according to the Diagnostic and Structural Manual of Mental Disorders, fourth edition (DSM‐IV), were aged 18–65, and had a maximum of two illness episodes or two years of adequate treatment for psychosis (not five years as registered in ClinicalTrials.gov). Furthermore, the patients must have had a diagnosable depressive episode within the past year or have a score of five or higher on the Calgary Depression Scale for Schizophrenia (CDSS)." Exclusion criteria: "Exclusion criteria were a history of severe head injury or of neurological or developmental disorders." Setting: NA Special subgroup: first episode, diagnosable depressive episode within the past year or have a score of five or higher on the Calgary Depression Scale for Schizophrenia (CDSS) N: 63 Gender: 37 men, 26 women Age: mean 27.9 years CBT plus standard care arm: participants total: 32, participants male: 17, participants female: 15, age: mean 28.6 years, age range 19‐51 years, duration of untreated psychosis: median 17 weeks, % taking antipsychotics: 84.4, % diagnosis schizophrenia: 44.0, % diagnosis schizoaffective disorder: 21.9, % diagnosis delusional disorder: 6.3, % diagnosis other psychosis: 28.1 Control arm: participants total: 31, participants male: 20, participants female: 11, age: mean 27.1 years, age range 18‐43 years, duration of untreated psychosis: median 20 weeks, % taking antipsychotics: 90.3, % diagnosis schizophrenia: 52.0, % diagnosis schizoaffective disorder: 9.7, % diagnosis delusional disorder: 9.7, % diagnosis other psychosis: 29.0
Interventions	1. CBT plus standard care. N = 32 Description of treatment: "The present study applied a CBT manual based on Kingdon and Turkington's CBT for psychosis (Kingdon 2005), and Fennel's CBT approach to overcoming low self‐esteem (Fennell 2007). The CBT intervention was designed with weekly individual sessions of 45–60 min, delivered over a six‐month period (maximum 26 sessions). The CBT treatment protocol was divided into three treatment stages. Stage 1 (sessions 1–5) focused on engagement and aimed to prepare the patient for CBT. During this stage the therapist informed the patient about the basic principles of CBT and explained how psychotic symptoms often develop based on the stress‐vulnerability model (Zubin 1977). Early sessions emphasized the importance of active participation and collaboration and the significance of homework and developing a list of problems and goals for the forthcoming sessions. A particular effort was made to clarify patients' expectations regarding therapy. Stage 2 (sessions 6–20) targeted depressive symptoms and low self‐esteem, which were the study's main focus. However, a problem list was prepared for each patient to address the patient's everyday psychological challenges and tailor therapy to patients' individual needs. The case formulation developed by Morrison was applied for each patient at this stage, developed in collaboration with the patients. The case formulation was updated and modified throughout the therapy period. Stage 3 (sessions 20–26) was dedicated to the termination of the therapy and relapse prevention. Specific attention was given to summarize the therapy and ensure that the patient had become “their own cognitive therapist.”." Number of sessions: maximum 26 sessions; mean number of 19.5 CBT sessions (median: 22 sessions) Sessions frequency and duration: 4 per month, duration 45‐60 min per session Therapist: clinical psychologists, psychiatrists, occupational therapist Therapist experience: trained Supervision: yes Modality: individual Delivery: face‐to‐face Treatment manualised: yes 2. Control (Treatment as usual). N = 31 Description of treatment: "All patients continued to receive their ongoing usual treatment from their therapists/case managers in various psychiatric units in Oslo, Norway. The core components of TAU entailed ongoing medication, regular psychiatric review, and regular follow‐ups by their case managers. Some of these patients had access to wider multidisciplinary community mental health services. The TAU patients most often received regular psychotherapy pertaining to different treatment methods, including some options of cognitive therapy. Treatment as usual specifications in terms of specific therapeutic interventions or number of sessions were not available to the research team." Number of sessions: NA Sessions frequency and duration: NA Therapist: therapists/case managers Therapist experience: NA Supervision: no Modality: individual Delivery: face‐to‐face Treatment manualised: no information
Outcomes	Mental state ‐ specific ‐ PANSS positive endpoint (< 6 months, > 1 year) Mental state ‐ specific ‐ PANSS negative endpoint (<6 months, > 1 year) Mental state ‐ specific ‐ CDS endpoint (< 6 months, > 1 year) Mental state ‐ specific ‐ BDI II endpoint (< 6 months, > 1 year) Leaving the study early ‐ for any reason (dropout) (< 6 months, > 1 year) Functioning ‐ GAF endpoint (< 6 months, > 1 year)
Notes	We contacted the authors and obtained additional information.

Uzenoff 2008.

Study characteristics
Methods	Study design: randomised controlled trial Study grouping: parallel group Antipsychotic medication allowed: yes Duration in weeks: 26 weeks Number of study arms: 2 Type of blinding: rater blind Type of data analysis for overall efficacy: ITT Number of sites: NA Number of countries: 1 Country: USA Sponsorship source: other Trial registration ID: NA Publication year: 2008 Researcher's allegiance: The authors developed the manual for the intervention group.
Participants	Diagnosis: schizophrenia, schizoaffective disorder, schizophreniform disorder; diagnostic criteria: DSM‐IV Current clinical state: clinical stable Definition of stability: after hospital discharge and clinical stabilisation Inclusion criteria: "Participants needed to be 16 years of age or older, meet DSM‐IV (APA 2000) diagnostic criteria for schizophrenia, schizoaffective disorder, or schizophreniform disorder, and have been in treatment for a first episode of psychosis for less than 12 months." Exclusion criteria: no information Setting: in and outpatient Special subgroup: no information N: 19 Gender: 12 men, 7 women Age: mean 22.69 years CBT plus standard care arm: participants total: 10, participants male: 6, participants female: 4, age: mean 25.30 years (SD = 5.95), age range 16 years or older, % taking antipsychotics: NA, baseline PANSS total M = 67.3, SD = 12.66, N = 10 Control arm: participants total: 11, participants male: 6, participants female: 3, age: mean 19.78 years (SD = 3.53), age range 16 years or older, % taking antipsychotics: NA, baseline PANSS total M = 54.44, SD = 11.43, N = 9
Interventions	1. CBT plus standard care. N = 10 Description of treatment: "ACE is a manual‐based psychotherapy consisting of 4 phases: (1) establishing therapeutic alliance; (2) promoting treatment adherence; (3) developing a plan for maintenance treatment; and (4) rehabilitation. In ACE Therapy, the therapist assumes an active role consistent with traditional cognitive‐behavioral therapies, and the therapeutic stance emphasizes empathy for the patient’s point of view. ACE Therapy draws upon the Health Belief Model (Becker 1975), a theoretical framework developed in preventative healthcare. This model posits that an individual’s decision to adhere to treatment is a dynamic process wherein patients weigh advantages and disadvantages of treatment, and that beliefs about advantages and disadvantages change over time (i.e., in response to education, or as symptom severity or treatment side effects change). Key therapeutic techniques used include motivational interviewing (e.g., linking taking medications or not to the client’s ability to reach important goals), inductive questioning, reframing, using normalizing rationales, selective validation of the patient’s beliefs and attitudes, and graded analysis of evidence to address beliefs about illness and treatment." Number of sessions: 14 therapy sessions planned; 13.8 sessions received Sessions frequency and duration: 6 weekly sessions followed by 8 biweekly sessions; 30‐45 minutes Therapist: clinicians with at least a Master's degree and working experience with individuals with first‐episode psychosis Therapist experience: trained Supervision: yes Modality: individual Delivery: face‐to‐face Treatment manualised: yes 2. Control (Supportive therapy plus treatment as usual). N = 11 Description of treatment: "ST was intended to control for the nonspecific aspects of therapy, such as therapeutic relationship and regular contact with a therapist. ST has 2 phases, which are: (1) establishing a therapeutic alliance, and (2) providing emotional support and discussing non‐illness issues or topics. ST therapists were instructed not to bring up or discuss issues related to medication or medication adherence. If patients brought up these issues in ST, the therapist was to listen with an empathic therapeutic stance, and advise the patient to discuss these issues with his or her treating physician." Number of sessions: 14 therapy sessions planned; 13.44 sessions received Sessions frequency and duration: 6 weekly sessions followed by 8 biweekly sessions Therapist: clinicians with at least a Master's degree and working experience with individuals with first‐episode psychosis Therapist experience: trained Supervision: yes Modality: individual Delivery: face‐to‐face Treatment manualised: NA
Outcomes	Mental state ‐ general ‐ PANSS total endpoint (< 6 months) Mental state ‐ specific ‐ participants with change in positive symptoms (< 6 months) Mental state ‐ specific ‐ PANSS positive endpoint (< 6 months) Mental state ‐ specific ‐ PANSS negative endpoint (< 6 months) Mental state ‐ specific ‐ CDS (< 6 months) Leaving the study early ‐ for any reason (dropout) (< 6 months) Quality of life ‐ QLS endpoint (< 6 months)
Notes	We contacted the authors but received no reply.

Weiden 2019.

Study characteristics
Methods	Study design: randomised controlled trial Study grouping: parallel group Antipsychotic medication allowed: yes Duration in weeks: NA Number of study arms: 2 Type of blinding: NA Type of data analysis for overall efficacy: NA Number of sites: NA Number of countries: NA Country: NA Sponsorship source: NA Trial registration ID: NA Publication year: 2019 Researcher's allegiance: NA
Participants	Diagnosis: schizophrenia; diagnostic criteria: Current clinical state: first psychotic symptoms <= 5 years Definition of stability: NA Inclusion criteria: "Subjects meeting diagnostic criteria of schizophrenia with 1st psychotic symptom ≤ 5 years with interest in receiving individual psychotherapy were enrolled from the UIC Psychosis Program." Exclusion criteria: NA Setting: NA Special subgroup: NA N: 34 Gender: 23 men, 11 women Age: mean 24 years CBT plus standard care arm: NA Control arm: NA
Interventions	1. CBT plus standard care (Health Dialogue Intervention). N = 16 Description of treatment: "The CBT Intervention was called Health Dialogue Intervention (HDI) and was based on the CBT Insight Model with an added section on medication adherence. " Number of sessions: 17.6 sessions Sessions frequency and duration: up to 5 sessions over 6 months to a maximum of 22 over a year Therapist: yes (not further described) Therapist experience: trained Supervision: yes Modality: individual Delivery: face‐to‐face Treatment manualised: yes 2. Control (Psychoeducation). N = 18 Description of treatment: Number of sessions: 9.2 sessions Sessions frequency and duration: up to 5 sessions over 6 months to a maximum of 22 over a year Therapist: yes (not further described) Therapist experience: trained Supervision: yes Modality: individual Delivery: face‐to‐face Treatment manualised: yes
Outcomes	No outcome data available for this study.
Notes	We contacted the authors but received no reply.

ACE = Adherence‐coping‐education

ACT = Acceptance and Committment Therapy

BDI = Beck depression inventory

BNSS = Brief Negative Symptom Scale

BPRS = Brief Psychiatric Rating Scale

BPRS‐E = Brief Psychiatric Rating Scale‐Expanded

BPRS‐P = BPRS total psychotic subscore

BPRS‐PS = BPRS positive symptom subscale

CAMHS = Child and Adolescent Mental Health Services

CAP = community‐acquired pneumonia

CBCM = Cognitive Behavioural Case Management

CBSST = Cognitive‐Behvioral Social Skills Training

CBSST‐Adol = Adolescent version of CBSST

CBT = Cognitive behavioural therapy

CDS = Calgary Depression Scale

CDSS = Calgary Depression Scale for Schizophrenia

CESD‐R = Centre for Epidemiologic Studies Depression scale ‐ Revised

CGI = Clinical Global Impression

CGI‐I = Clinical Gloabal Impression of Improvement

CGI‐S = Clinical Global Impression of Severity

CLZ = Clozapine

CM = case management

CMHC = community mental health center

CMHS = community mental health service

CR = cognitive remediation

CRI = Cognitive therapy based recovery intervention

CT = Cognitive therapy

CTS‐Psy = Cognitive Therapy Scale for Psychosis

DHI = Digital health intervention

DSM = Diagnostic and Statistical Manual of Mental Disorders

DVD = digital video disc

ECT = electro‐convulsive therapy

EIP = Early Intervention in Psychosis

EIS = Early intervention services

EOP = early onset psychosis

EPPIC = Early Psychosis Prevention and Intervention Centre

EQ‐5D = EuroQol‐5D

EQ‐5D‐5L = EuroQol‐5D‐5L

FAST = Functioning Assessment Short Test

FEP = First‐episode psychosis

FESFS = First Episode Social Functioning Scale

FIp = psychosis‐focused family intervention

GAF = Global Assessment of Functioning

GPI = group‐based psychological in‐
tervention

HADS = Hospital Anxiety and Depression Scale

HDI = Health Dialogue Intervention

HHSS = "Habilidades
sociales" (social skills)

ICD‐10 = International Classification of Diseases, Tenth Revision

IM = motivational interviews

IPS = individual placement and support

IQ = intelligence quotient

ITT = Intention‐to‐treat analysis

MADRS = Montgomery‐Åsberg Depression Rating Scale

MANSA = Manchester Short Assessment of Quality of Life

MAPS = Managing Adolescent first episode Psychosis

MDE = major depressive episode

MI = Motivational interviews

MI‐CBT = Motivational interviewing and cognitive behavioural therapy

MSQoL = Modular System for Quality of Life

NA = not available

NOS = not otherwise specified

OYH = Orygen Youth Health

PANSS = Positive and Negative Syndrome Scale

PE = Psychoeducation

PIPE = cognitive‐motivational therapy program

PSP = Personal and Social Performance Scale

PSYCHOED‐Adol = Psychoeducation for Adolescents

PSYRATS = The Psychotic Symptoms Rating Scales

QLS = Quality of life scale

RCT = Randomised controlled study

RPT = Relapse prevention therapy

SA = social anxiety

SANS = Scale for the assessment of negative symptoms

SAPS = Scale for the assessment of positive Symptoms

SCID‐P = Structured Clinical Interview for DSM disorders ‐ Patients

SD = standard deviation

SM = Symptom Management

SOFAS =Social and Occupational Functioning Assessment Scale

ST = supportive therapy

STOPP = systematic treatment of persistent psychosis

TAU = Treatment‐as‐usual

TDZ = Thioridazine

WHOQoL‐BREF = World Health Organization Quality of Life abbreviated form

WoL‐bref = World Health Organization Quality of Life abbreviated form

Characteristics of excluded studies [ordered by study ID]

Study	Reason for exclusion
Amann 2005	Design: randomised Participants: first episode Intervention: no CBT
Browning 2013	Design: not randomised
Drury 1996	Design: randomised Participants: no first episode
Edwards 2008	Design: not randomised
Fowler 2002	Design: randomised Participants: no first episode
Fowler 2009 (Isrep)	Design: randomised Participants: no first episode
Gaebel 2005	Design: randomised Participants: first episode Intervention: no CBT
Gleeson 2012	Design: randomised Participants: first episode Intervention: no CBT
Griffiths 2019	Design: randomised Participants: first episode Intervention: no CBT
Gumley 2003	Design: randomised Participants: no first episode
Haddock 2018	Design: randomised Participants: no first episode
Jackson 2005	Design: not randomised
Kemp 2007	Design: randomised Population: substance‐induced psychosis (communication with the author)
Kidd 2020	Design: randomised Participants: first episode Intervention: no CBT
Lally 2019	Design: not randomised
Lecomte 2020	Design: randomised Participants: no first episode
Lysaker 2007	Design: randomised Participants: no first episode
McGorry 1998	Design: randomised Participants: no first episode
McGorry 2000	Design: randomised Participants: no first episode
Morrison 2014 (ACTION)	Design: randomised Participants: unclear Intervention: no CBT plus standard care
Morrison 2016	Design: randomised Participants: no first episode
Morrison 2018 (FOCUS)	Design: randomised Participants: no first episode
NCT00722163	Study has been cancelled.
Newton 2005	Design: not randomised
Ostergaard Christensen 2014	Design: randomised Participants: first episode Intervention: no CBT
Tarrier 2014	Design: randomised Participants: no first episode
Van der Gaag 2011	Design: randomised Participants: no first episode
Vohs 2018	Design: randomised Participants: first episode Intervention: no CBT
Weller 2018 (EBIRT)	Design: randomised Participants: first episode Intervention: no CBT
Wölwer 2022	Design: randomised Participants: no first episode

CBT: cognitive behaviour therapy

Characteristics of studies awaiting classification [ordered by study ID]

Gonzalez‐Ortega 2022.

Methods	Study design: randomised controlled trial Trial registration ID: NCT02319746
Participants	Inclusion criteria: ‐ Diagnosed as FEP according to the revised fourth edition of the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition, Text Revision (DSM‐ IV‐TR) (APA 2000) (i.e. schizophreniform disorder, schizoaffective disorder, delusional disorder, bipolar disorder with psychotic symptoms, atypical psychosis, brief psychotic disorder, non‐specified psychotic disorder, or major depressive disorder with psychotic symptoms) ‐ Meeting dependence or abuse of cannabis criteria according to the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition, Text Revision (DSM‐IV‐TR) (APA 2000) and the scores of the European Addiction (Europ‐ASI) (Kokkevi 1995) (scores of 4 to 7: abuse; scores of 8 to 9: dependence) ‐ Aged between 15 and 40 years. In the case of minors (under 18 years of age), written informed consent was requested from their parents or guardians Exclusion criteria: ‐ Organic brain pathology and/or mental disability according to DSM‐IV‐TR (APA 2000) criteria
Interventions	1. CBT (Specific cognitive behavioural therapy (CBT) for cannabis cessation (CBT‐CC)) plus standard care. N = 34 2. Control (treatment‐as‐usual). N = 31
Outcomes	‐ Mental state ‐ Positive and Negative Syndrome Scale (PANSS positive, negative, total) ‐ Depression ‐ Hamilton Depression Rating Scale (HDRS) ‐ Anxiety ‐ Hamilton Anxiety Scale (HAM‐A) ‐ Manic symptoms ‐ Young Mania Rating Scale (YMRS) ‐ Illness awareness ‐ Scale to assess Unawareness in Mental Disorders (SUMD) ‐ Functioning ‐ Functioning Assessment Short Test (FAST)
Notes	Asked the authors for some more information about the diagnoses of the included participants We contacted the authors but received no reply.

Lewis 2000.

Methods	Study design: randomised controlled trial
Participants	Inclusion criteria: "recent onset schizophrenia sufferers with substance misuse problems" Exclusion criteria: NA
Interventions	1. CBT plus standard care 2. Control
Outcomes	‐ Relapse rate
Notes	Asked the authors to provide some more information about the population included in the study, especially referring to first‐episode and recent‐onset patients We contacted the authors but received no reply.

NCT02653729.

Methods	Study design: randomised controlled trial
Participants	Inclusion criteria: ‐ Patients diagnosed with psychosis by psychiatrist of the recruiting unit ‐ Patients with duration of illness till 3 years ‐ Patients with the minimum of 5 years of education ‐ Patients within the age range of 18 to 35 years ‐ Competent and willing to give informed consent ‐ Patients living in the study catchment area. Exclusion criteria: ‐ Patients with drug‐induced psychosis ‐ Patients with severe psychopathology, unable to give informed consent ‐ Patients suffering from organic or neurological disorder ‐ Patients suffering from chronic physical condition Age: 18 to 35 years
Interventions	1. CBT plus standard care (Espidone, Olepra, Donu) 2. Control (Espidone, Olepra, Donu) N = 50
Outcomes	‐ Mental state ‐ Positive and Negative Symptom Scale (PANSS) ‐ Mental state ‐ The PsychoSocial Remission Scale (PSRS) ‐ Insight ‐ Schedule for Assessment of Insight (SAI)
Notes	We attempted to contact the authors, but it was not possible to retrieve a valid contact.

NCT04889911.

Methods	Study design: randomised controlled trial
Participants	Inclusion criteria: "Inclusion criteria for the RCT phase will be assessed by an initial screening and the baseline interview. We will use relatively broad inclusion criteria to maximise the generalisability of the findings for the population of persons with FEP. Inclusion criteria will be: 1) meets criteria for FEP youth based on the following definition: age 15‐24, onset of psychotic symptoms within the last 5 years, and an absence of a primary substance use or mood disorder that could be causing the psychotic symptoms (confirmed by programme eligibility); 2) meets criteria for moderate (defined as a mean score of 1‐1.5 on the 0‐3 scale of the Internalized Stigma of Mental lllness Scale [ISMI]) or elevated (defined as a mean score of 1.5‐3 on the 0‐3 scale of the ISMI) self‐stigma; 3) speaks English well enough to complete assessments and participate in groups; 4) is able to provide informed consent to participate." Exclusion criteria: "Does not meet any of the above inclusion criteria."
Interventions	1. Narrative Enhancement and Cognitive Therapy‐ Young Adult, combined with Co‐ordinated Specialty Care 2. Control (Co‐ordinated Specialty Care). N = 40
Outcomes	Primary outcome measures: ‐ Service use ‐ treatment engagement ‐ Therapeutic alliance ‐ Working Alliance Inventory Short Form‐Client Version (WAI‐SF) ‐ Coping strategies ‐ Coping with Symptoms Checklist Secondary outcomes measures: ‐ Self‐esteem ‐ Rosenberg Self Esteem Scale ‐ Hopelessness ‐ Beck Hopelessness Scale ‐ Social functioning ‐ Quality of Life Scale
Notes	It was not clear whether the intervention met our inclusion criteria. We attempted to contact the authors, but it was not possible to retrieve a valid contact.

Renton 2004.

Methods	Unclear
Participants	Inclusion criteria: NA Exclusion criteria: NA
Interventions	1. CBT plus standard care (cognitive therapy) 2. Control (waiting‐list/treatment‐as‐usual)
Outcomes	‐ Mental state ‐ Positive and Negative Syndrome Scale (PANSS)
Notes	Asked the authors to provide some more information about the population included in the study, especially referring to first‐episode and recent‐onset patients. Furthermore, we asked whether the intervention also had a behavioural component. We contacted the authors but received no reply.

Suri 2001.

Methods	Study design: randomised controlled trial
Participants	Inclusion criteria: "patients with early onset or acute schizophrenia" Exclusion criteria: NA
Interventions	1. CBT plus standard care 2. Control
Outcomes	NA
Notes	We attempted to contact the authors, but it was not possible to retrieve a valid contact (no email address available, we contacted the hospital where the authors were affiliated, but they do not work there any more).

Wood 2022.

Methods	Study design: randomised controlled trial
Participants	Inclusion criteria: ‐ aged 18 and above ‐ meet criteria for schizophrenia‐spectrum diagnoses (schizophrenia, schizophreniform disorder, schizoaffective disorder, delusional disorder or psychotic disorder not otherwise specified; ICD‐10), or meet criteria for an early intervention service (EIS) for treatment of psychosis to allow for diagnostic uncertainty ‐ able to give informed consent and have the capacity to consent to participation in the study ‐ currently receiving care from an acute psychiatric inpatient team ‐ able to complete the research in English Exclusion criteria: ‐ non‐English speakers (due to translation costs and difficulty of producing valid translations of the research instruments and intervention) ‐ an acquired brain injury or substance misuse judged to be the acute cause of the psychotic experiences ‐ already undertaking a structured psychological intervention delivered by a psychologist or trained therapist at the time of the study
Interventions	1. CBT (crisis‐focused Cognitive Behavioural Therapy for psychosis) plus standard care 2. Control (treatment‐as‐usual) N = 60 (target number)
Outcomes	‐ Mental state ‐ Positive and Negative Syndrome Scale (PANSS) ‐ Depression ‐ Beck Depression Inventory (BDI‐7) ‐ Hopelessness ‐ Beck Hopelessness Scale (BHS‐9) ‐ Anxiety ‐ Generalised Anxiety Disorder Measure (GAD‐7) ‐ Personal recovery ‐ Process of Recovery Questionnaire (QPR) ‐ Subjective experiences of crisis ‐ Crisis Scale in Psychosis (CRISP) ‐ Quality of life ‐ Recovering Quality of Life (REQOL‐10) ‐ Service use ‐ Client Service Receipt Inventory (CSRI) ‐ Functioning ‐ Global Assessment of Functioning (GAF) ‐ Mental state ‐ Threshold Assessment Grid (TAG) ‐ Relapse ‐ Hospitalisation ‐ Adverse events
Notes	Asked the authors to provide some more information about the population included in the study, especially referring to first‐episode and recent‐onset patients We contacted the authors but received no reply.

BDI‐7 = Beck Depression Inventory

BHS‐9 = Beck Hopelessness Scale

BPRS = Brief Psychiatric Rating Scale

CBT(‐CC) = Specific cognitive behavioural therapy for cannabis cessation

CRISP = Crisis Scale in Psychosis

CSRI = Client Service Receipt Inventory

DSM‐IV‐TR = Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition, Text Revision

EIS = Early intervention service

Europ‐AST = European Addiction

FAST = Functioning Assessment Short Test

GAD‐7 = Generalised Anxiety Disorder Measure

GAF = Global Assessment of Functioning

HAM‐A = Hamilton Anxiety Scale

HDRS = Hamilton Depression Rating Scale

ICD‐10 = International Classification of Diseases,Tenth Revision

ISMI = Internalized Stigma of Mental Illness Scale

FEP = First‐episode psychosis

NA = not available

PANSS = Positive and Negative Syndrome Scale

PSRS = PsychoSocial Remission Scale

QPR = Process of Recovery Questionnaire

RCT = Randomised controlled trial

REQOL‐10 = Recovering Quality of Life Questionnaire

SAI = Schedule for Assessment of Insight

SUMD = Scale to assess Unawareness in Mental Disorders

TAG = Threshold Assessment Grid

WAI‐SF = Working Alliance Inventory Short Form‐Client Version

YMRS = Young Mania Rating Scale

Characteristics of ongoing studies [ordered by study ID]

Barrowclough 2001.

Study name	Barrowclough 2001
Methods	NA
Participants	Inclusion criteria: NA Exclusion criteria: NA
Interventions	1. CBT plus standard care 2. Control
Outcomes	NA
Starting date
Contact information
Notes

Haddock 2000.

Study name	Haddock 2000
Methods	Study design: randomised controlled trial
Participants	Inclusion criteria: "an ICD10 diagnosis of schizophrenia, schizoaffective disorder or delusional disorder, have less than five years since onset and have associated alcohol or drug abuse." Exclusion criteria: NA
Interventions	1. CBT plus standard care Description of the treatment: "The individual cognitive‐behaviour therapy will take place over 18 weekly sessions, followed by 6 fortnightly sessions. The individual intervention will include standardised assessment followed by techniques designed to reduce the severity and distress of persistent symptoms, such as distraction and coping skills, self‐monitoring, reality testing, monitoring of prodromal signs, techniques to enhance self‐esteem and interventions to improve knowledge regarding illness and medication." 2. Control N = 70
Outcomes	‐ Patient and carer functioning ‐ Number of days of substance use ‐ Satisfaction ‐ Severity of substance use ‐ Use of services ‐ Engagement with services
Starting date
Contact information
Notes

ISRCTN60855021.

Study name	ISRCTN60855021
Methods	Study design: randomised controlled trial
Participants	Inclusion criteria: ‐ Score above 30 on the Social Interaction Anxiety Scale (SIAS) ‐ Early Intervention Service user ‐ Aged 14 years and above, either sex Exclusion criteria: ‐ Current acute psychotic episode
Interventions	1. CBT plus standard care Description: "Cognitive behavioural intervention provided by graduate mental health workers. Follow‐up length: 6 months" 2. Control N = 48 (planned)
Outcomes	Primary outcome measure: ‐ Anxiety ‐ Social Anxiety Interaction Scale Secondary outcome measures: ‐ Paranoia ‐ Paranoia Scale ‐ Schemata ‐ Brief Core Schema Scale
Starting date
Contact information
Notes

ISRCTN65536352.

Study name	ISRCTN65536352
Methods	Study design: randomised controlled trial
Participants	Inclusion criteria: "12‐15 patients with first‐episode" Exclusion criteria: "Presence of organic mental disorder, mental retardation, or inadequate command of language."
Interventions	1. CBT (Cognitive Behavioural Case Management (CBCM)) plus standard care 2. Control (treatment‐as‐usual)
Outcomes	Primary outcome measures: ‐ Quality of Life Survey Secondary outcomes measures: ‐ Brief Psychiatric rating scale (expanded version) ‐ Montgomery‐Asberg Depression Rating Scale ‐ World Health Organization Quality of Life‐BREF (WHOQOL‐BREF) ‐ Social and Occupational Functioning Assessment Scale ‐ Addiction Severity Scale ‐ Illness Duration Interview ‐ Pathways to care item ‐ Premorbid Adjustment Scale ‐ Verona Service Satisfaction Scale ‐ An engagement measure ‐ Sociodemographic & Service Receipt Inventory ‐ Structured Clinical Interview for Diagnostic and Statistical Manual (DSM)‐III‐R (APA 1987) Patients (SCID‐P) ‐ Operational Criteria Checklist for Psychotic Disorders (OPCRIT)
Starting date
Contact information
Notes

NCT00161408.

Study name	NCT00161408
Methods	Study design: randomised controlled trial
Participants	Inclusion criteria: ‐ diagnosis of schizophrenic disorder according to ICD10: F20 ‐ first episode ‐ age between 18 and 55 ‐ willingness to give informed consent also to a double‐blind pharmacological treatment study Exclusion criteria: ‐ residence outside the catchment area ‐ insufficient knowledge of the German language ‐ substance abuse or addiction as primary clinical problem ‐ serious physical illness ‐ organic brain disease ‐ pregnancy ‐ contraindications to neuroleptic treatment Age: 18 to 55 years
Interventions	1. CBT plus standard care Description of treatment: "CBT includes 8 sessions of psychoeducation, 8 sessions of computer‐based cognitive training, 8 sessions with relatives and 20 sessions of focusing on stress management, relapse prevention and coping with persistent symptoms." 2. Control (psychoeducation) Description of treatment: "a short "information‐centred psychoeducation" (ICP) of 8 sessions" N = 106 (enrolment)
Outcomes	Primary outcome measure: ‐ Relapse Secondary outcome measures: ‐ Quantitative measures of symptoms ‐ Social functioning ‐ Cognitive functioning ‐ Quality of life
Starting date
Contact information
Notes

NCT03491852.

Study name	NCT03491852
Methods	Study design: randomised controlled trial
Participants	Inclusion criteria: "Involvement in an early psychosis clinic for a period of less than 3 years, due to the focus of this study being on the treatment of early psychosis. Participants must be between the ages of 16 and 65 years. Participants 16 years of age who are incapable to consent without a parent or guardian will be excluded. Participants must be fluent in English, as determined by referring clinicians or researchers (in the case of advertisement‐referred participants) in order to meaningfully participate in the BOOST intervention and complete the assessment tools." Exclusion criteria: "Potential participants who are unable to provide informed consent, as determined by the treatment team, will not be able to meaningfully participate in the BOOST intervention and will, therefore, be unable to participate in the research study. Individuals with a presence of intellectual disability or history of traumatic brain injury will also be excluded." Age: 16 to 65 years
Interventions	1. CBT (Be Outspoken and Overcome Stigmatizing Thoughts (BOOST)) plus standard care Description of treatment: "The BOOST intervention consists of 8 group‐based, weekly one‐hour sessions. While every BOOST session is different, in general they will focus on helping participants fight back against stigmatizing thoughts and develop a sense of self‐worth and empowerment. BOOST sessions are group‐based and facilitated by trained clinicians, with the aid of a peer support worker to provide unique insights on living with and overcoming self‐stigma. Content of sessions involve group discussions, exercises conducted in session, and between‐session "missions" (i.e. home practice activities)." 2. Control (waiting‐list control) Description of treatment: "Participants on the waitlist will still receive treatment as usual, which includes medical, psychosocial, and occupational interventions to help maximize patients' integration within the community and support recovery from a first episode of psychosis. Frequency of contact largely depends on the individual needs of patients. Waitlist controls will be offered the BOOST intervention 3 months post‐enrollment." N = 40 (estimated enrolment)
Outcomes	Primary outcome measure: ‐ Change in self‐stigma ‐ Internalized Stigma of Mental Illness Scale (ISMI) Secondary outcome measures: ‐ Change in self‐Esteem ‐ Rosenberg Self‐Esteem Scale (RSES) ‐ Change in quality of Life ‐ Satisfaction with Life Safe (SWLS) ‐ Change in stigma stress ‐ Cognitive Appraisal of Stigma Stress (CogApp) ‐ Change in depression ‐ Beck Depression Inventory‐II (BDI) ‐ Change in social anxiety ‐ Social Interaction Anxiety Scale (SIAS) ‐ Change in personal recovery ‐ Questionnaire about the Process of Recovery (QPR) ‐ Change in functioning ‐ Sheehan Disability Scale (SDS)
Starting date
Contact information
Notes	We contacted the authors and obtained additional information.

NCT04180709.

Study name	NCT04180709
Methods	Study design: randomised controlled trial
Participants	Inclusion criteria: ‐ Experiencing or having experienced a First Episode of Psychosis (FEP) within the past 5 years ‐ Currently participating in CAMEO Early Intervention services (north or south) and intending to continue TAU with CAMEO for at least the next 4 months ‐ Mental capacity for consent ‐ Currently experiencing threshold level (score ≤ 16 on SCI‐8) of disrupted sleep ‐ Adults 18 years or older ‐ Ability to understand and follow therapeutic instructions necessary for experiment and respond to online questionnaires ‐ Illness duration less than or equal to 5 years. Onset of illness was defined as first contact with psychiatric services for psychotic symptoms. ‐ Access to the internet Exclusion criteria: ‐ Too unwell to viably participate in study ‐ A diagnosis of drug‐induced psychosis or bipolar disorder ‐ Psychotic disorder due to a medical or physical disease (i.e. considered to have an organic basis) ‐ Current drug or alcohol dependency ‐ Currently taking prescribed sleep medication or intending to do so during study ‐ Currently doing shift work ‐ Travel over 2 time zones during or within two weeks prior to assessment period Age: 18 years and older
Interventions	1. CBT (web‐based CBT Intervention (Sleepio)) plus standard care 2. Control (treatment‐as‐usual) N = 44 (enrolment)
Outcomes	Primary outcome measure: ‐ Change from baseline Work and Social Adjustment Scale (WSAS) score at week 9 Secondary outcome measures: ‐ Social recovery ‐ Time Use Survey ‐ Structured Hours (TUS‐SH) ‐ Depression ‐ Patient Health Questionnaire (PHQ‐9) ‐ Sustained attention ‐ Rapid Visual Information Processing (RVP) ‐ Visual episodic memory ‐ Paired Associates Learning (PAL) ‐ Working memory and strategy ‐ Spatial Working Memory (SWM) ‐ Emotional recognition ‐ Emotion Recognition Task (ERT) ‐ Functioning ‐ Global Assessment of Functioning (GAF) ‐ Change from baseline Work and Social Adjustment Scale (WSAS) score at week 5, 13 and 17
Starting date
Contact information
Notes

NCT04916626.

Study name	NCT04916626
Methods	Study design: randomised controlled trial
Participants	Inclusion criteria: "1. Between 12 and 17 years of age (both inclusive) at trial inclusion. 2. First‐episode psychosis within F2 spectrum (F20 schizophrenia, F21 Schizotypal disorder, F22 delusional disorder, F23 acute and transient psychotic disorders, F25 schizoaffective disorders, F28/29 other or un‐specified non‐organic psychosis) or depression with psychotic symptoms (F32.3, F33.3) or substance‐induced psychosis (F1X.5) according to the International Statistical Classification of Diseases and Related Health Problems (ICD‐10). 3. Maximum 12 months since first prescription of antipsychotic treatment on the indication psychosis. 4. Speak and understand Danish. 5. Written informed consent from parents or legal caretakers. Participants who reach age 18 years during the trial will be asked to give personal written informed consent to continue their study participation." Exclusion criteria: "1. A diagnosis of mental retardation of at least moderate severity defined as an intelligence quotient (IQ) of 49 or below (F71, F72, F73 according to the International Statistical Classification of Diseases and Related Health Problems (ICD‐10). 2. Currently compulsory admission and/or treatment according to Danish legislation" Age: 12 to 17 years
Interventions	1. CBT (OPUS YOUNG) plus standard care Description of treatment: "OPUS YOUNG is a two‐year out‐patient specialized early intervention service for children and adolescents with a first episode psychosis. The OPUS YOUNG treatment consists of the following elements: modified assertive community treatment, low patients to case manager ratio, cognitive‐behavioural case management (CBCM), psychoeducational family treatment including multiple family groups (MFG) and psychoeducational siblings groups, Social Cognition and Interaction Training (SCIT), possible individual Cognitive Behavioural Therapy (CBT) in addition to CBCM, and manual based psychopharmacologic treatment. Additional, special transition support, individual school/employment support, and prevention and treatment of substance abuse." 2. Control (treatment‐as‐usual) Description of treatment: "The patients allocated to TAU will be offered non‐manualized treatment following national Danish guidelines and local guidelines. Treatment is provided by a multidisciplinary team and consists of case‐management (no defined upper case load), family support and psychoeducation, in addition to psychopharmacological treatment. In some cases, social skills training and CBT may be offered. In general, office visits take place in outpatient clinics." N = 284 (estimated enrolment)
Outcomes	Primary outcome measure: ‐ Functioning ‐ Personal and Social Performance Scale (PSP) Secondary outcome measures: ‐ Mental state ‐ Scale for Assessing Psychotic Symptoms in Schizophrenia (SAPS) ‐ Mental state ‐ Scale for Assessing Negative Symptoms in Schizophrenia (SANS) ‐ Satisfaction ‐ Client Satisfaction Questionnaire (CSQ) ‐ Quality of life ‐ Health Related Quality of Life Questionnaire for Children and Young People and their Parents (KIDSCREEN‐10)
Starting date
Contact information
Notes	We contacted the authors and obtained additional information.

BDI = Beck depression inventory

BOOST = Be Outspoken and Overcome Stigmatizing Thoughts

CAMEO = Cambridgeshire and Peterborough Assessing, Managing and Enhancing Outcomes

CBCM = Cognitive Behavioural Case Management

CBT = Cognitive behavioural therapy

CogApp = Cognitive Appraisal of Stigma Stress

CSQ = Client Satisfaction Questionnaire

DSM(‐III‐R) = Diagnostic and Statistical Manual of Mental Disorders, Third Edition, Revised

ERT = Emotion Recognition Task

FEP = First‐episode psychosis

GAF = Global Assessment of Functioning

ICD‐10 = International Classification of Diseases, Tenth Revision

ICP = information‐centred psychoeducation

ISMI = Internalized Stigma of Mental Illness Scale

IQ = intelligence quotient

KIDSCREEN‐10 = Health Related Quality of Life Questionnaire for Children and Young People and their Parents

MFG = multiple family groups

NA = not available

OPCRIT = Operational Criteria Checklist for Psychotic Disorders

PAL = Paired Associates Learning

PHQ‐9 = Patient Health Questionnaire

PSP = Personal and Social Performance Scale

QPR = Questionnaire about the Process of Recovery

RSES = Rosenberg Self‐Esteem Scale

RVP = Rapid Visual Information Processing

SANS = Scale for the assessment of negative symptoms

SAPS = Scale for the assessment of positive Symptoms

SCI(‐8) = Sleep Condition Indicator

SCID(‐P) = Structured Clinical Interview for Diagnostic and Statistical Manual (DSM)‐III‐R Patients

SCIT = Social Cognition and Interaction Training

SDS = Sheehan Disability Scale

SIAS = Social Interaction Anxiety Scale

Sleepio = fully automated digital sleep improvement program based on Cognitive Behavioral Therapy (CBT)

SWLS = Satisfaction with Life Safe

SWM = Spatial Working Memory

TAU = treatment‐as‐usual

TUS‐SH = Time Use Survey ‐ Structured Hours

WSAS = Work and Social Adjustment Scale

WHOQoL‐BREF = World Health Organization Quality of Life abbreviated form

Differences between protocol and review

The title of the review was originally "Cognitive behavioural therapy plus standard care for first episode psychosis" and has been changed to “Cognitive behavioural therapy added to standard care for first‐episode and recent‐onset psychosis“.

In the protocol, we planned to include participants with a first episode of psychosis, as defined in the single studies. During the screening of the studies, we realised that different and contradicting definitions were often used and, that strictly following the 'first episode' criterion would have resulted in the exclusion of relevant studies, where the population also included patients in their first episode, but that just used other definitions to describe this population. Therefore, we decided to also accept studies that included participants with a recent onset of the condition, but we limited this to a period of five years, with the idea of capturing the first episodes. In order to be consistent, we also decided to accept studies in which participants were recruited from centres focusing on early intervention for psychosis, because these services also recruit patients within their first three to five years of the illness. We added a threshold for inclusion of studies including participants with recent‐onset psychosis with no further definition of this term; where this occurred, we accepted the study if it was clear that at least 75% of the participants had a first‐episode or recent‐onset psychosis, as defined above.

We also accepted studies that included participants independently of their age (except for those aged more than 65 years), instead of a minimum of 15 years as previously planned. This decision was made to take into account the variability in the age of onset of schizophrenia.

Moreover, in the protocol, we planned that, if we found a study including participants with other diagnoses, we would have included the study if participants with a first episode of psychosis comprised at least 80% of the total. We altered this threshold to 50% in terms of diagnosis, mainly to have some more flexibility with studies that reported participants with a diagnosis of psychosis NOS.

In the protocol, we planned to create GRADE tables for cognitive behavioural therapy plus standard care versus different control conditions separately. However, a majority of the studies compared cognitive behavioural therapy plus standard care to standard care, so separate summary of findings tables would split the information and be less informative. Instead, the role of different comparators is investigated in a subgroup analysis.

The section on skewed data was adapted from the new template protocol of the Schizophrenia group.

We performed an additional post hoc sensitivity analysis excluding one strong outlier (Palma‐Sevillano 2019).

Contributions of authors

• Susanna Franziska Mayer: search and selection of studies for inclusion in the review; collection of data for the review; assessment of the risk of bias in the included studies; analysis of data; GRADE assessment; interpretation of data; writing of the review

• Ciaran Corcoran: search and selection of studies for inclusion in the review; collection of data for the review; assessment of the risk of bias in the included studies; interpretation of data; writing of the review

• Liam Kennedy: collection of data for the review; assessment of the risk of bias in the included studies; interpretation of data; writing of the review

• Stefan Leucht: conception of the review; design of the review; interpretation of data; writing of the review

• Irene Bighelli: conception of the review; design of the review; search and selection of studies for inclusion in the review; collection of data for the review; assessment of the risk of bias in the included studies; analysis of data; GRADE assessment; interpretation of data; writing of the review

Sources of support

Internal sources

• Freistaat Bayern, Germany

  Irene Bighelli and Stefan Leucht are employed by Freistaat Bayern.

• National Institute for Health and Care Research (NIHR), UK

  Provided funding for Cochrane Schizophrenia Group

External sources

• Evidence Synthesis Ireland (ESI) Fellowship, Ireland

  Ciaran Corcoran and Liam Kennedy were part supported by the Health Research Board (Ireland) and the HSC Public Health Agency (Grant number CBES‐2018‐001) through Evidence Synthesis Ireland and Cochrane Ireland.

Declarations of interest

Susanna Mayer: none

Ciaran Corcoran: none

Liam Kennedy: none

Stefan Leucht: In the last three years SL has received honoraria for advising/consulting and/or for lectures and/or for educational material from Angelini, Boehringer Ingelheim, Eisai, Ekademia, GedeonRichter, Janssen, Karuna, Kynexis, Lundbeck, Medichem, Medscape, Mitsubishi, Otsuka, NovoNordisk, Recordati, Rovi, Teva. SL is an editor of Cochrane Schizophrenia. He was not involved in the editorial process of the present review (see Acknowledgements for details about editorial process).

Irene Bighelli: IB is the Deputy Co‐ordinating editor of Cochrane Schizophrenia. She was not involved in the editorial process of the present review (see Acknowledgements for details about editorial process).

New