Cognitive behavioural therapy without medication for schizophrenia

Abstract

Background

Cognitive behavioural therapy (CBT) can be effective in people with schizophrenia when provided in combination with antipsychotic medication. It remains unclear whether CBT could be safely and effectively offered in the absence of concomitant antipsychotic therapy.

Objectives

To investigate the effects of CBT for schizophrenia when administered without concomitant pharmacological treatment with antipsychotics.

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, and WHO ICTRP.

Selection criteria

We included randomised controlled trials (RCTs) in people with schizophrenia comparing CBT without antipsychotics to standard care, standard care without antipsychotics, or the combination of CBT and antipsychotics.

Data collection and analysis

Two review authors independently screened references for inclusion, extracted data from eligible studies, and assessed risk of bias using Cochrane's RoB 2 tool. We contacted study authors for missing data and additional information. Our primary outcome was general mental state measured with a validated rating scale. Key secondary outcomes were specific symptoms of schizophrenia, relapse, service use, number of participants leaving the study early, functioning, quality of life, and number of participants actually receiving antipsychotics during the trial. We also assessed behaviour, adverse effects, and mortality.

Main results

We included 4 studies providing data for 300 participants (average age 21.94 years). The mean sample size was 75 participants (range 61 to 90 participants). Study duration was between 26 and 39 weeks for the intervention period and 26 to 104 weeks for the follow‐up period. Three studies employed a blind rater, while one study was triple‐blind.

All analyses included data from a maximum of three studies. The certainty of the evidence was low or very low for all outcomes.

For the primary outcome overall symptoms of schizophrenia, results showed a difference favouring CBT without antipsychotics when compared to no specific treatment at long term (> 1 year mean difference measured with the Positive and Negative Syndrome Scale (PANSS MD) −14.77, 95% confidence interval (CI) −27.75 to −1.79, 1 RCT, n = 34). There was no difference between CBT without antipsychotics compared with antipsychotics (up to 12 months PANSS MD 3.38, 95% CI −2.38 to 9.14, 2 RCTs, n = 63) (very low‐certainty evidence) or compared with CBT in combination with antipsychotics (up to 12 months standardised mean difference (SMD) 0.30, 95% CI −0.06 to 0.65, 3 RCTs, n = 125).

Compared with no specific treatment, CBT without antipsychotics was associated with a reduction in overall symptoms (as described above) and negative symptoms (PANSS negative MD −4.06, 95% CI −7.50 to −0.62, 1 RCT, n = 34) at longer than 12 months. It was also associated with a lower duration of hospital stay (number of days in hospital MD −22.45, 95% CI −28.82 to −16.08, 1 RCT, n = 74) and better functioning (Personal and Social Performance Scale MD −12.42, 95% CI −22.75 to −2.09, 1 RCT, n = 40, low‐certainty evidence) at up to 12 months.

We did not find a difference between CBT and antipsychotics in any of the investigated outcomes, with the exception of adverse events measured with the Antipsychotic Non‐Neurological Side‐Effects Rating Scale (ANNSERS) at both 6 and 12 months (MD −4.94, 95% CI −8.60 to −1.28, 2 RCTs, n = 48; MD −6.96, 95% CI −11.55 to −2.37, 2 RCTs, n = 42).

CBT without antipsychotics was less effective than CBT combined with antipsychotics in reducing positive symptoms at up to 12 months (SMD 0.40, 95% CI 0.05 to 0.76, 3 RCTs, n = 126). CBT without antipsychotics was associated with a lower number of participants experiencing at least one adverse event in comparison with CBT combined with antipsychotics at up to 12 months (risk ratio 0.36, 95% CI 0.17 to 0.80, 1 RCT, n = 39, low‐certainty evidence).

Authors' conclusions

This review is the first attempt to systematically synthesise the evidence about CBT delivered without medication to people with schizophrenia. The limited number of studies and low to very low certainty of the evidence prevented any strong conclusions. An important limitation in the available studies was that participants in the CBT without medication group (about 35% on average) received antipsychotic treatment, highlighting the challenges of this approach. Further high‐quality RCTs are needed to provide additional data on the feasibility and efficacy of CBT without antipsychotics.

Plain language summary

Cognitive behavioural therapy without antipsychotics for people with schizophrenia

Key messages

• There is not enough information to allow any strong conclusions about cognitive behavioural therapy (CBT) without medication for people with schizophrenia.

• Additional studies are needed to look at the effectiveness and safety of CBT without medication for people with schizophrenia.

Introduction

Schizophrenia is a severe mental disorder. People with the illness struggle to differentiate between their own thoughts, beliefs, and ideas versus reality. For example, they may hear voices in their head, but it feels like someone is really talking to them. CBT is a psychological intervention that can be effective for treating the symptoms of schizophrenia when it is offered with antipsychotic drugs. However, it remains unclear if CBT is effective and safe when used without these drugs. This is important to find out because the use of antipsychotics is often associated with unwanted side effects.

What did we want to find out?

We wanted to find out if CBT is effective and safe when used without medication for people with schizophrenia.

What did we do?

We searched for studies that examined CBT given without antipsychotics compared with no specific treatment, antipsychotics, or CBT plus antipsychotics.

We compared and summarised the results of the studies and rated our confidence in the evidence, based on factors such as study methods and sizes.

What did we find?

We found 4 studies that involved 300 participants with schizophrenia. Study duration was between 26 and 39 weeks for the intervention period, and participants were contacted again to collect further data between 26 and 104 weeks. The included studies were conducted in the UK and Australia, and were all sponsored by public institutions.

We found that compared to no specific treatment, CBT without antipsychotics may result in a reduction in overall symptoms (at least in the long term) and negative symptoms (e.g. apathy, loss of interest and motivation, lack of concentration). It may also result in better functioning and lower duration of hospital stay in comparison with no specific treatment.

CBT without antipsychotics may not differ from CBT plus antipsychotics in modifying overall symptoms of schizophrenia. CBT without antipsychotics may be less effective than CBT plus antipsychotics in reducing positive symptoms specifically (e.g. hearing voices), and may result in fewer adverse effects compared to CBT plus antipsychotics.

What are the limitations of the evidence?

The certainty of the evidence was low to very low, meaning that we have limited to very little confidence in the results. This is because study participants and therapists were aware of the treatment being received, which could have influenced the results. In addition, it was often the case that during the study participants received treatment with antipsychotics when this was not planned. Furthermore, not all of the studies provided data about everything that we were interested in, and the numbers of included studies and study participants were too small to be certain about the results.

How up‐to‐date is this evidence?

The evidence is current to March 2022.

Summary of findings

Summary of findings 1. Summary of findings table ‐ CBT without antipsychotics compared to no specific treatment for schizophrenia.

CBT without antipsychotics compared to no specific treatment for schizophrenia
Patient or population: schizophrenia Setting: Intervention: CBT without antipsychotics Comparison: no specific treatment
Outcomes	Anticipated absolute effects* (95% CI)		Relative effect (95% CI)	№ of participants (studies)	Certainty of the evidence (GRADE)	Comments
	Risk with no specific treatment	Risk with CBT without antipsychotics
Mental state ‐ general: mean endpoint PANSS total (high = poor) ‐ < 12 months		MD 9.77 lower (20.59 lower to 1.05 higher)	‐	39 (1 RCT)	⊕⊕⊝⊝ Lowa
Global state ‐ number of participants with relapse/exacerbations of psychosis (separated time points) ‐ > 1 year	54 per 1000	27 per 1000 (3 to 285)	RR 0.50 (0.05 to 5.28)	74 (1 RCT)	⊕⊕⊝⊝ Lowb
Leaving the study early ‐ for any reason ‐ overall acceptability (separated time points) ‐ < 12 months	135 per 1000	108 per 1000 (31 to 372)	RR 0.80 (0.23 to 2.75)	74 (1 RCT)	⊕⊕⊝⊝ Lowb
Functioning ‐ mean endpoint PSP (high = good) ‐ < 12 months		MD 12.42 lower (22.75 lower to 2.09 lower)	‐	40 (1 RCT)	⊕⊕⊝⊝ Lowa
Adverse events ‐ Participants with at least one AE ‐ > 1 year	0 per 1000	0 per 1000 (0 to 0)	Not estimable	(0 studies)	‐
*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; MD: mean difference; RR: risk ratio
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_439003076253823530.

^a Downgraded by two levels for imprecision: inadequate information size (< 800 participants for a continuous outcome) and CI includes both better outcome for CBT without antipsychotics and almost no difference
^b Downgraded by two levels for imprecision: inadequate information size (< 1000 participants for a dichotomous outcome) and CI includes both better outcome for CBT without antipsychotics and better outcome for standard care without antipsychotics

Summary of findings 2. Summary of findings table ‐ CBT without antipsychotics compared to antipsychotics for schizophrenia.

CBT without antipsychotics compared to antipsychotics for schizophrenia
Patient or population: schizophrenia Setting: Intervention: CBT without antipsychotics Comparison: antipsychotics
Outcomes	Anticipated absolute effects* (95% CI)		Relative effect (95% CI)	№ of participants (studies)	Certainty of the evidence (GRADE)	Comments
	Risk with antipsychotics	Risk with CBT without antipsychotics
Mental state ‐ general: mean endpoint PANSS total (high = poor) ‐ < 12 months		MD 3.38 higher (2.38 lower to 9.14 higher)	‐	63 (2 RCTs)	⊕⊝⊝⊝ Very lowa,b
Global state ‐ number of participants with relapse/exacerbations of psychosis (separated time points) ‐ < 12 months	0 per 1000	0 per 1000 (0 to 0)	RR 2.78 (0.12 to 65.08)	90 (1 RCT)	⊕⊝⊝⊝ Very lowc,d
Leaving the study early ‐ for any reason ‐ overall acceptability (separated time points) ‐ < 12 months	22 per 1000	36 per 1000 (5 to 281)	RR 1.66 (0.21 to 12.94)	90 (2 RCTs)	⊕⊝⊝⊝ Very lowa,d
Functioning ‐ mean endpoint functioning scales (high = good) ‐ < 12 months	‐	SMD 0.03 lower (0.61 lower to 0.55 higher)	‐	46 (2 RCTs)	⊕⊝⊝⊝ Very lowa,b
Adverse events ‐ Participants with at least one AE (separated time points) ‐ < 12 months	591 per 1000	278 per 1000 (124 to 632)	RR 0.47 (0.21 to 1.07)	40 (1 RCT)	⊕⊝⊝⊝ Very lowc,e
*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; MD: mean difference; 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_438842958834426261.

^a Downgraded by two levels for risk of bias: two out of two studies were at overall high risk of bias
^b Downgraded by two levels for imprecision: inadequate information size (< 800 participants for a continuous outcome) and CI includes both better outcome for CBT without antipsychotics and better outcome for standard care
^c Downgraded by two levels for risk of bias: one out of one study was at overall high risk of bias
^d Downgraded by two levels for imprecision: inadequate information size (< 1000 participants for a dichotomous outcome) and CI includes both better outcome for CBT without antipsychotics and better outcome for standard care
^e Downgraded by one level for imprecision: inadequate information size (< 1000 participants for a dichotomous outcome)

Summary of findings 3. Summary of findings table ‐ CBT without antipsychotics compared to CBT plus antipsychotics for schizophrenia.

CBT without antipsychotics compared to CBT plus antipsychotics for schizophrenia
Patient or population: schizophrenia Setting: Intervention: CBT without antipsychotics Comparison: CBT plus antipsychotics
Outcomes	Anticipated absolute effects* (95% CI)		Relative effect (95% CI)	№ of participants (studies)	Certainty of the evidence (GRADE)	Comments
	Risk with CBT plus antipsychotics	Risk with CBT without antipsychotics
Mental state ‐ general: mean endpoint overall symptoms scales (PANSS/BPRS) (high = poor) (separated time points) ‐ < 12 months	‐	SMD 0.3 higher (0.06 lower to 0.65 higher)	‐	125 (3 RCTs)	⊕⊝⊝⊝ Very lowa,b
Global state ‐ number of participants with relapse/exacerbations of psychosis (separated time points) ‐ < 12 months	0 per 1000	0 per 1000 (0 to 0)	RR 2.89 (0.12 to 67.75)	90 (1 RCT)	⊕⊝⊝⊝ Very lowc,d
Leaving the study early ‐ for any reason ‐ overall acceptability (separated time points) ‐ < 12 months	65 per 1000	46 per 1000 (8 to 268)	RR 0.70 (0.12 to 4.11)	90 (2 RCTs)	⊕⊝⊝⊝ Very lowe,f
Functioning ‐ mean endpoint functioning scales (high = good) ‐ < 12 months	‐	SMD 0.13 higher (0.25 lower to 0.5 higher)	‐	110 (3 RCTs)	⊕⊝⊝⊝ Very lowa,g
Adverse events ‐ Participants with at least one AE (separated time points) ‐ < 12 months	762 per 1000	274 per 1000 (130 to 610)	RR 0.36 (0.17 to 0.80)	39 (1 RCT)	⊕⊕⊝⊝ Lowf,h
*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_438842947953353100.

^a Downgraded by two levels for risk of bias: two out of three studies were at overall high risk of bias
^b Downgraded by two levels for imprecision: inadequate information size (< 800 participants for a continuous outcome) and CI includes both no difference and better outcome for CBT plus antipsychotics
^c Downgraded by two levels for risk of bias: one out of one study was at overall high risk of bias
^d Downgraded by two levels for imprecision: inadequate information size (< 1000 participants for a dichotomous outcome) and CI includes both no difference and better outcome for CBT plus antipsychotics
^e Downgraded by two levels for risk of bias: two out of two studies were at overall high risk of bias
^f Downgraded by one level for imprecision: inadequate information size (< 1000 participants for a dichotomous outcome)
^g Downgraded by two levels for imprecision: inadequate information size (< 800 participants for a continuous outcome) and CI ranges from better outcome with CBT without antipsychotics to better outcome with CBT plus antipsychotics
^h Downgraded by one level for risk of bias: the study is at overall high risk of bias

Background

Schizophrenia is a chronic and severe psychiatric disorder seen in approximately 1% of the population and which causes significant functional loss (McGrath 2008).

Antipsychotic drugs are effective for the acute treatment and relapse prevention of the disorder (Huhn 2019; Schneider‐Thoma 2022), but are associated with burdensome side effects that are likely to contribute to excess mortality associated with the disorder (Hjorthøj 2017).

Psychotherapeutic interventions for schizophrenia have been developed to address the various aspects of the disorder and are widely recommended in accordance with the guidelines of the National Institute for Health and Care Excellence (NICE) (NICE 2014). These interventions are usually offered in combination with antipsychotic medications, so that the separate roles of psychological and of pharmacological treatment cannot be ascertained so far (Bighelli 2020).

There exist studies that have attempted to provide psychotherapy with cognitive behavioural therapy (CBT) to participants that were not receiving antipsychotic medication (Morrison 2012a; Morrison 2014). These studies could be an opportunity to determine the role of psychotherapy alone in improving the symptoms of schizophrenia.

We planned in this Cochrane review to summarise all randomised controlled trials that investigated CBT provided without concomitant antipsychotic medication compared to the combination of CBT and medication, medication alone, or no specific treatment.

Description of the condition

Schizophrenia is a chronic and severe psychiatric disorder affecting approximately 1% of the population worldwide (McGrath 2008). Onset typically occurs in the second decade of life (Kirkbride 2017), and usually has a significant impact in terms of quality of life, functioning, social inclusion, and disability. According to the Global Burden of Disease Study 2019, schizophrenia is ranked 9th in terms of years lived with disability (YLDs) in the age range 15 to 49 years (GBD 2019).

Typical symptom manifestations of the condition include severe thought disturbances, which may lead to delusions (beliefs that are not based on reality) and hallucinations, for example hearing or seeing things that are not there. These manifestations are usually defined as 'positive' symptoms, because they are unusual by their presence. By contrast, 'negative' symptoms are unusual by their absence, and can include a restricted range and intensity of emotional expression and a reduced ability to experience pleasure.

Difficulties with concentration, attention, and motivation may also lead to reduced participation in or abandoning education activities and reduced employment; it is estimated that up to 80% of people with schizophrenia are not employed (Marwaha 2004).

Different stages can be identified in the course of the disease. In the onset or prodromal phase, subtle modifications in the person's behaviour, cognition, and feelings can be identified, which then develop into clear psychotic symptoms during the acute phase. After the acute episode, florid symptoms recede during the remission phase (Andreasen 2005). In this stage, most individuals still require maintenance treatment in order to prevent symptoms recrudescence (relapse).

Description of the intervention

Antipsychotic medication is the current first‐line treatment for schizophrenia. Due to the chronic nature of the disease, long‐term treatment with antipsychotics is usually needed to prevent the risk of relapse (Ceraso 2020; Schneider‐Thoma 2022). Unfortunately, these medications have many adverse events that make their use complicated, including movement disorders, weight gain, metabolic problems, and sexual dysfunction (Leucht 2013; Schneider‐Thoma 2019). Also, for this reason, people with schizophrenia are often ambivalent about taking medications, and rates of medication non‐compliance are high (Lacro 2002; Moncrieff 2009). Moreover, recent guidelines tend to shift their recommendations from not recommending to partially recommending antipsychotic discontinuation, both for schizophrenia in general and for first‐episode schizophrenia (Shimomura 2020).

Some authors have argued that the efficacy of antipsychotics may have been overestimated, and their adverse effects underestimated, suggesting that clinicians reconsider “whether everyone who meets the criteria for a schizophrenia spectrum diagnosis requires antipsychotics in order to recover" (Morrison 2012a).

Cognitive behavioural therapy (CBT) is a psychological intervention developed by Aaron Beck in 1979 to be used in the treatment of depression (Beck 1979). The aims of CBT are to establish links between the person's thoughts, feelings, or actions with respect to current or past symptoms and/or functioning; individuals are encouraged to re‐evaluate their perceptions, beliefs, or reasoning in relation to the target symptoms. Alternative ways of coping with the target symptom can be promoted, with the aim of reducing stress and improving functioning (NICE 2014).

CBT began to be used for schizophrenia in the late 1980s and early 1990s (Kingdon 1998). CBT for psychosis (CBTp) is a problem‐centred, structured type of therapy based on patient‐clinician collaboration. It was developed with the primary aim of reducing symptom severity, and there is evidence that it may be effective in this regard when used in addition to antipsychotic medication (Bighelli 2018; Jauhar 2014; Jones 2018a; Jones 2018b; Wykes 2008; Zimmermann 2005). There is evidence that CBT could also be beneficial for improving other outcomes in people with schizophrenia, such as insight, well‐being, and functioning (Jones 2018a; Laws 2018). Based on this literature, current guidelines recommend offering CBT in addition to pharmacological treatment with antipsychotics (NICE 2014).

However, in all existing meta‐analyses, CBT was investigated as provided in combination with antipsychotics (Jauhar 2014; Jones 2018a; Jones 2018b; Wykes 2008; Zimmermann 2005). As a result, the role of CBT without concomitant antipsychotic treatment is unclear.

NICE guidelines for children and adolescents with a first episode of psychosis include the possibility of offering family intervention with individual CBT to young people with schizophrenia and their carers, if they wish to try psychological interventions alone (NICE 2013). As emphasised in the guidelines, such an approach requires strict and regular monitoring of the symptoms, distress, and functioning.

In recent years, there have been attempts to deliver CBT alone to people with schizophrenia (Morrison 2012a; Morrison 2014). These attempts have raised criticism and ethical concerns (Mustafa 2018), but also opened the way to new research and clinical perspectives.

How the intervention might work

CBT encourages the person to establish links between their thoughts, feelings, or behaviours with respect to their symptoms. The aim of creating these connections is to change the way in which the individual interprets and evaluates their experiences and attributes meaning to them.

By helping people recognise their basic feelings, thoughts, and behaviours about the disease, CBT can improve the ability of people with schizophrenia to cope with their psychotic symptoms and can be an important and helpful tool (Morrison 2010). Being able to recognise inappropriate thoughts, feelings, and behaviours can reduce people's stress burden, improve their quality of life, and help recovery (Morrison 2010).

Individuals are stimulated to identify and challenge biased interpretations of their experiences that may have a role in the continuation of symptoms.

Why it is important to do this review

A few attempts and clinical trials have been conducted on the administration of CBT without concomitant antipsychotic medication (Morrison 2012a; Morrison 2014). However, this literature has never been systematically appraised, nor the results synthesised with a meta‐analysis. As a result, no evidence in the form of a systematic review exists on the effects of CBT when administered alone in people with schizophrenia.

Evidence on this issue would be highly relevant for clinical practice.

On the one hand, antipsychotics have burdensome side effects, therefore if it was proven that CBT is effective also without pharmacological treatment, this may reduce patient burden and costs related to the intervention. Nevertheless, the feasibility of such an option should be carefully examined, because not offering an antipsychotic medication goes against current clinical guidelines, and may be considered a critical ethical issue (Mustafa 2018). On the other hand, the combination of the two interventions may increase the efficacy of the treatment in comparison to their stand‐alone use.

The aim of this Cochrane review was to provide a comprehensive examination of the existing evidence on the feasibility and effectiveness of CBT offered without concomitant pharmacological therapy in people with schizophrenia, and help shed light on this challenging matter.

Objectives

To investigate the effects of CBT for schizophrenia when administered without concomitant pharmacological treatment with antipsychotics.

Methods

Criteria for considering studies for this review

Types of studies

We considered all relevant randomised controlled trials (RCTs). We included RCTs meeting our inclusion criteria and reporting useable data either for the qualitative or quantitative synthesis. Both open‐label and rater‐blind studies were eligible for inclusion in the review. We excluded quasi‐randomised studies, such as those that allocated intervention by alternate days of the week. Where people were given treatments in addition to CBT, we only included data if the adjunct treatment was evenly distributed between groups, and it was only the CBT that was randomised.

Types of participants

People with schizophrenia, as defined in the included studies, or related disorders such as schizophreniform disorder, schizoaffective disorder and delusional disorder, by any means of diagnosis. We did not include trials where the majority of participants had such disorders as bipolar affective disorder or substance‐induced psychosis.

If a study included participants with other diagnoses, we included the study only if participants with a diagnosis of schizophrenia or related disorders constituted at least 50% of the participants.

We excluded trials where participants had organic psychoses, as well as studies that recruited participants with prodromal symptoms or at‐risk mental states.

We wanted to ensure that the information was as relevant as possible to the current care of people with schizophrenia, and so aimed to highlight the current clinical state clearly (acute, early post‐acute, partial remission, remission) as well as the stage (first episode, early illness, chronic) and whether the studies focused primarily on people with particular problems (e.g. negative symptoms, treatment‐resistant illnesses). We planned to investigate the effect of the intervention on chronic patients and on people with a first episode of psychosis in a subgroup analysis (see Subgroup analysis and investigation of heterogeneity).

Where a study recruited participants from early intervention services without providing additional details about inclusion criteria, we evaluated on a case‐by‐case basis if the participants matched the characteristics of the population of the current review as described above, requesting additional details from the authors as necessary.

Types of interventions

1. CBT without concomitant medication for schizophrenia

'Cognitive behavioural therapy' has been used as a broad label that can include a variety of interventions. It typically includes elements of cognitive restructuring or cognitive therapy (CT) and elements of behavioural therapy (BT) that are delivered together. However, CT and BT elements could also be approached separately in the therapeutic process. In this review, we considered any intervention that includes CT or BT elements alone, as well as CBT interventions that include both components, as eligible under the term 'CBT'. In this review, CBT was evaluated in people with schizophrenia who were not taking concurrent medication.

The group that received structured CBT and did not receive antipsychotic treatment was used as the intervention group. For simplicity, we referred to this condition as 'CBT'.

2. Control

We included any control condition applied in studies that aimed to investigate the effects of CBT without concomitant medication in people with schizophrenia.

We classified the control condition as follows.

2.1 No specific treatment

Participants are randomised to receive standard care, explicitly excluding the prescription of an antipsychotic medication. The possibility of receiving rescue medication, if needed, was not excluded. For simplicity, we referred to this control condition as 'No specific treatment'.

2.2 Antipsychotics

Participants are randomised to standard care, which normally includes antipsychotics; participants in this group may have received some form of psychosocial support, but not CBT. For simplicity, we referred to this control condition as 'Antipsychotics', abbreviated 'AP' in the forest plots.

2.3 CBT in combination with antipsychotics

Participants are randomised to receive CBT in addition to standard care, which usually includes antipsychotics. No distinction will be made between antipsychotics (e.g. first or second generation), doses, or administration forms (oral, injection), and all will be included. For simplicity, we referred to this control condition as 'CBT plus antipsychotics', abbreviated 'CBT + AP' in the forest plots.

Types of outcome measures

We aimed to divide all outcomes into short term (up to 6 months since the onset of therapy), medium term (up to 12 months), and long term (over 12 months). This grouping reflects NICE recommendations that CBT should be delivered for at least 16 sessions (NICE 2014).

For outcomes such as 'clinically important change', 'any change', and 'relapse', we used the definitions reported in the included trials.

For valid scales, 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/subscale addressing positive symptoms (e.g. PANSS positive, PANSS positive factor, 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/subscale addressing negative symptoms (e.g. PANSS negative, PANSS negative factor, 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/subscale addressing depressive symptoms (e.g. Montgomery‐Asberg Depression Rating Scale (MADRS), Beck Depression Inventory (BDI))

1.2.7 Clinically important change in anxiety symptoms, as defined by individual studies

1.2.8 Average endpoint or change score on a published scale/subscale addressing anxiety (e.g. Hamilton Anxiety Rating scale (HAM‐A) or Beck Anxiety Inventory (BAI))

2. Global state

2.1 Relapse

In case more measures of relapse were reported, we gave priority to: (i) relapse defined with operationalised criteria (e.g. worsening on a symptom scale); (ii) clinical judgement.

2.2 Recovery

Recovery is intended as "the ability to function in the community, socially and vocationally, as well as being relatively free of disease‐related psychopathology" (Andreasen 2005). We extracted the number of participants in recovery, based on this definition or on similar definitions provided in the studies.

2.3 Remission

Symptomatic remission according to Andreasen criteria (Andreasen 2005).

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 Clinical Global Impression (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

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 functioning scale

We accepted 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 cognitive functioning scale

We accepted any published cognitive functioning scale (e.g. the MATRICS 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 quality of life scale

We accepted 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 accepted 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/effects

10.1 At least one adverse event/effect

10.2 Specific adverse events

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

10.3 Incidence of various specific adverse effects

10.4 Average endpoint or change score on adverse effects scale

11. Mortality

11.1 Overall mortality

11.2 Mortality due to natural causes

11.3 Mortality due to suicide

12. Economic outcomes

12.1 Costs due to treatment

12.2 Total direct and indirect costs

12.3 Average change in total cost of medical and mental health care

Search methods for identification of studies

Electronic searches

The Information Specialist searched the register of the Cochrane Schizophrenia Group up to 6 March 2022 using the following search strategy:

*Cognit* in Intervention Field of STUDY

In such a study‐based register, searching the major concept retrieves 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 allows rapid and accurate searches that reduce waste in the next steps of systematic reviewing (Shokraneh 2019).

Following Cochrane methods (Lefebvre 2021), the Information Specialist compiles this register from systematic searches of major resources and their monthly updates (unless otherwise specified).

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

• MEDLINE

• Embase

• Allied and Complementary Medicine Database (AMED)

• BIOSIS

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

• PsycINFO

• PubMed

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

• World Health Organization International Clinical Trials Registry Platform (WHO ICTRP) (www.who.int/clinical-trials-registry-platform)

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

The register of the Cochrane Schizophrenia Group also includes handsearches and conference proceedings (see the Group's website). It does not place any limitations on language, date, document type, or publication status.

Searching other resources

1. Reference searching

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

2. Personal contact

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

Data collection and analysis

Selection of studies

At least two review authors (IB, OC and NHS) independently inspected citations from the searches and identified potentially relevant abstracts. Where disputes arose, we acquired the full report for more detailed scrutiny. We then obtained the full reports of the abstracts or reports deemed potentially relevant, and at least two review authors (IB, OC and NHS) independently inspected the full reports of the abstracts or reports for inclusion in the review. Where it was not possible to resolve a disagreement by discussion, we consulted the senior author of the team. If disagreement persisted, we attempted to contact the study authors for clarification. We documented all decisions made.

Data extraction and management

1. Extraction

At least two review authors (IB and OC) independently extracted data from the included studies. Where possible, we attempted to extract data presented only in graphs and figures, but included these data only if two review authors independently obtained the same result. Any disagreements were discussed. Where it was not possible to resolve disagreements by discussion, we consulted the senior review author. We documented all decisions. If necessary, we attempted to contact authors through an open‐ended request to obtain missing information or for clarification. Review author SL helped clarify issues regarding any remaining problems, and we documented these final decisions.

2. Management

2.1 Forms

We extracted data using pre‐designed forms developed specifically for this review.

2.2 Scale‐derived data

We included continuous data from rating scales only if:

• the psychometric properties of the measuring instrument have been described in a peer‐reviewed journal (Marshall 2000);

• the measuring instrument has not been written or modified by one of the trialists for that particular trial; and

• the instrument should be a global assessment of an area of functioning and not subscores which are not in themselves validated or shown to be reliable. However, we will include subscores of scales if these were validated, or if these were predefined in a scale such as the positive symptom, negative symptom, and general symptom scores of the PANSS (Kay 1986).

Ideally, the measurement instrument should either be a self‐report or be completed by an independent rater or relative (not the therapist). We realise that this is not often reported clearly.

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. We have decided to primarily use endpoint data, and only use change data if the former were not available. If necessary, we combined endpoint and change data in the analysis. This procedure is possible when using mean differences (MDs) (Deeks 2011), and also when using standardised mean differences (SMDs). 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 be expected (Da Costa 2013). We planned to analyse endpoint and change data separately in a sensitivity analysis for the primary outcome mental state (Sensitivity analysis).

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 inclusion.

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 possible lowest values 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. 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 lowest possible score different from 0, and whether the adjustment described above was needed.

If the ratio obtained is lower than one, it strongly suggests that data are skewed. If it is higher than one but less than two, there is the suggestion that data are skewed; if the ratio is larger than two, we included these data, because it is less likely that they were skewed (Altman 1996).

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

These skewed results would nevertheless be reported in 'Additional tables'.

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

2.5 Common measurement

To facilitate comparison between trials, we aimed, where relevant, to convert 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).

2.6 Conversion of continuous to binary

Where possible, we attempted to convert outcome measures to dichotomous data. This can be done by identifying cut‐off points on rating scales and dividing 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 CGI, Guy 1976, of raters, this could be considered a clinically significant response (Leucht 2005a; Leucht 2005b), in particular for acutely ill participants. If data based on these thresholds were not available, we used the primary cut‐off presented by the original authors, because the exact cut‐off is not so important in a meta‐analysis using risk ratios or odds ratios as effect sizes (Furukawa 2010).

2.7 Direction of graphs

Where possible, we entered data such that the area to the left of the line of no effect indicated a favourable outcome for CBT. Where keeping to this made it impossible to avoid outcome titles with clumsy double‐negatives (e.g. 'not un‐improved'), we reported data where the left of the line indicated an unfavourable outcome and noted this in the relevant graphs.

Assessment of risk of bias in included studies

Two review authors (OC and IB) independently assessed risk of bias using the Cochrane RoB 2 tool, referring to the criteria described in Chapter 8 of the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2020; Sterne 2019).

The risk of bias assessment is based on the following criteria:

• bias arising from the randomisation process;

• bias due to deviations from intended interventions;

• bias due to missing outcome data;

• bias in measurement of the outcome; and

• bias in selection of the reported result.

For each domain, we rated the available 'signalling questions' to reach a judgement of low risk of bias, some concerns, or high risk of bias, following the tool algorithms implemented in the RoB 2 Excel tool (available at www.riskofbias.info).

RoB 2 generally allows assessors to address studies from two angles: 1) effect of assignment to the interventions at baseline, regardless of whether the interventions were received as intended ('intention‐to‐treat effect'); 2) adherence to the interventions ('per‐protocol effect') (Chapter 8.2.2, Higgins 2020). For the purposes of this review, we aimed to assess the intention‐to‐treat effect.

We performed an evaluation with the RoB 2 tool for the following outcomes.

• Mental state: average endpoint or change score on a general mental state scale (e.g. PANSS, BPRS) (primary outcome) – medium term

• Global state: relapse – long term

• Leaving the study early: leaving the study early for any reason – medium term

• Functioning: average endpoint or change score on functioning scale – medium term

• At least one adverse event – long term

For cluster‐randomised trials, we planned to use RoB 2 for cluster‐randomised trials, provided at sites.google.com/site/riskofbiastool/welcome/rob-2-0-tool.

For cross‐over trials, as we intended to only use data from the first phase (see Measures of treatment effect), we planned to use the standard version of RoB 2.

In the case of disagreement, we made the final rating by consensus. Where inadequate details of randomisation and other characteristics of trials were provided, we attempted to contact the study authors to obtain further information. We reported non‐concurrence in quality assessment, but if disputes arose regarding the category to which a trial was to be allocated, we resolved this by discussion.

We noted the level of 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 that 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 illustrative comparative risks where possible.

2. Continuous data

For continuous outcomes, we estimated MD between groups, in particular when natural measures (such as days, kilograms, etc.) were used. We prefer not to calculate standardised effect size measures (SMD). However, if scales of very considerable similarity were used, we presumed there was a small difference in measurement, and we calculated SMD. It should be noted that SMD can be transformed to MD by using the formula MD = SMD x standard deviation of the scale of interest (Higgins 2020).

Unit of analysis issues

1. Cluster trials

Studies increasingly employ 'cluster randomisation' (such as randomisation by clinician or practice), but the analysis and pooling of clustered data pose problems. Authors often fail to account for intraclass 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).

We planned that where clustering was incorporated into the analysis of primary studies, we would present these data as if from a non‐cluster randomised study, but adjust for the clustering effect.

Where clustering was not accounted for in primary studies, we presented data in a table, with a (*) symbol to indicate the presence of a probable unit of analysis error. We planned to attempt to contact the first authors of studies to obtain intraclass correlation coefficients (ICCs) for their clustered data and to adjust for this 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 is 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 would assume it to be 0.1 (Ukoumunne 1999).

If cluster studies were appropriately analysed and had taken ICCs and relevant data documented in the report into account, synthesis with other studies would have been possible using the generic inverse variance technique.

2. Cross‐over trials

A major concern of cross‐over trials is the carry‐over effect. This occurs if an effect (e.g. pharmacological, physiological, or psychological) of the treatment in the first phase is carried over to the second phase. As a result, 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 are not appropriate if the condition of interest is unstable (Elbourne 2002). As both carry‐over and unstable conditions are very likely in severe mental illness, we planned to only use data from the first phase of cross‐over studies.

3. Studies with multiple treatment groups

Where a study involved more than two relevant treatment arms, we presented each different treatment arm in the relevant comparison.

Dealing with missing data

1. Overall loss of credibility

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

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

2. Binary

We presented data using an intention‐to‐treat (ITT) analysis. If studies presented data only on completers, we still used these studies and planned to undertake a sensitivity analysis excluding studies using completer analyses (Sensitivity analysis).

3. Continuous

3.1 Standard deviations

If standard deviations (SDs) were not reported, we would attempt to obtain the missing values from the study authors. If these were not available, where there were missing measures of variance for continuous data, but an exact standard error (SE) and CIs were available for group means, and either P value or t value available for differences in the mean, we would calculate SDs according to the rules described in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2020). When only the SE is reported, SDs are calculated by 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 would calculate the SDs according to a validated imputation method that is based on the SDs of the other included studies (Furukawa 2006). Although some of these imputation strategies can introduce error, the alternative would be to exclude a given study’s outcome and thus to lose information. Nevertheless, we would have examined the validity of the imputations in a sensitivity analysis that excludes imputed values.

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

Various methods are available to account for participants who left the trials early or who were lost to follow‐up. Some trials simply present the results of study completers; others use 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 seem to be somewhat better than LOCF (Leon 2006), we feel that the high percentage of participants leaving the studies early and differences between groups in their reasons for doing so is often the core problem in randomised schizophrenia trials, therefore we did not exclude studies based on the statistical approach used. However, by preference, we used the more sophisticated approaches, that is we preferred to use MMRM or multiple‐imputation to LOCF, and only presented completer analyses if some kind of ITT data were not available at all. Moreover, we addressed this issue in the 'missing outcome data' item of the RoB 2 tool.

Assessment of heterogeneity

1. Clinical heterogeneity

We considered all included studies initially, without seeing comparison data, to judge clinical heterogeneity. We simply inspected all studies for participants who were clearly outliers or situations that arose that had not been predicted, and, where found, discussed with other review authors such situations or participant groups.

2. Methodological heterogeneity

We considered all included studies initially, without seeing comparison data, to judge methodological heterogeneity. We simply inspected all studies for clearly outlying methods that we had not predicted would arise and discussed with other review authors any such methodological outliers.

3. Statistical heterogeneity

3.1 Visual inspection

We inspected graphs visually to investigate the possibility of statistical heterogeneity.

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 CI for I²). We interpreted the I² statistic as follows (Chapter 10, Higgins 2020):

• 0% to 40%: might not be important;

• 30% to 60%: may represent moderate heterogeneity;

• 50% to 90%: may represent substantial heterogeneity;

• 75% to 100%: considerable heterogeneity.

We planned that if we found substantial levels of heterogeneity in the primary outcome, we would explore reasons for the 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 for Systematic Reviews of Interventions (Higgins 2020).

1. Protocol versus full study

We attempted to locate protocols of the included trials. If the protocol was available, we compared the outcomes in the protocol with those in the published report. If the protocol was not available, we compared the outcomes listed in the methods section of the trial report with the reported results. If details from ClinicalTrials.gov and WHO ICTRP were available, they would have been included in the search results, and could be used to compare the differences between planned methods and published results.

2. Funnel plot

We are aware that 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 where there were fewer than 10 studies, or where all studies were of similar size. Where funnel plots were possible, we planned to seek statistical advice on their interpretation. We planned to additionally apply the Egger's test for funnel plot asymmetry to support the visual inspection of the forest plot with a statistical test. We planned to perform Egger's test with R.

Data synthesis

We understand that there is no closed argument regarding preference for the use of fixed‐effect or random‐effects models. The random‐effects method incorporates an assumption that the different studies are estimating different, yet related, intervention effects. This often seems to be true to us, and the random‐effects model takes into account differences between studies, even if there is no statistically significant heterogeneity. However, there is a disadvantage to the random‐effects model in that it puts added weight onto small studies, which are often the most biased studies. Depending on the direction of effect, these studies can either inflate or deflate the effect size. We chose to use a random‐effects model for analyses. We planned to apply the fixed‐effect model in a sensitivity analysis of the primary outcome.

Subgroup analysis and investigation of heterogeneity

1. Subgroup analyses

We planned to conduct the following subgroup analysis for the primary outcome if at least 10 studies were included in the analysis (Higgins 2020). Since no analyses included at least 10 studies, it was not possible to conduct any subgroup analyses.

1.1 Stage of the illness

We planned to separately investigate studies that enrolled chronic participants and studies that enrolled participants at their first episode of psychosis.

1.2 Special populations

We planned to investigate the effect of the intervention in different subpopulations if data were provided separately (e.g. participants with schizophrenia versus participants with other psychotic disorders such as schizophreniform disorder, schizoaffective disorder, and delusional disorder).

1.3 Individual versus group

We planned to separately investigate studies that provided CBT in an individual versus in a group setting.

1.4 Clinical staff training

We planned to separately investigate studies in which CBT was delivered by different professionals (e.g. therapists expert in CBT, therapists in training, nurses).

2. Investigation of heterogeneity

We reported if inconsistency was substantial. Firstly, we investigated whether data had been entered correctly. If they had, we would consider the following strategies:

• pool data despite the heterogeneity. An example where this strategy may be appropriate is when the effects of all studies are in the same direction. In other words, the heterogeneity reflects the degree of an effect rather than its direction, which is less problematic. Another example is when heterogeneity can be explained by appropriate subgroup analyses;

• exclude outlying studies. This strategy may apply if re‐inspection of such studies revealed methodological or clinical differences that were previously overlooked;

• not pool the studies.

All decisions in this regard have been described and discussed.

Sensitivity analysis

Where possible, we planned to perform sensitivity analyses for the primary outcome to explore the influence of the factors listed below on effect size. We planned that where there were substantial differences in the direction or precision of effect estimates in any of these sensitivity analyses, we would discuss them in the Discussion. However, since no comparison included more than three studies, we did not perform any sensitivity analyses.

1. Blinding of outcome assessor

We planned to exclude studies that did not apply a blind outcome assessor.

2. Assumptions for missing data

We planned to exclude studies using completer analyses only (see Dealing with missing data).

3. Loss to follow‐up

We planned to exclude studies where the overall loss of data was greater than 50%.

4. Risk of bias

We planned to analyse the effects of excluding trials at overall high risk of bias (see Assessment of risk of bias in included studies).

5. Imputed values

We planned to undertake a sensitivity analysis excluding trials where imputed values were used for ICC in calculating the design effect in cluster‐randomised trials or where SDs were imputed.

6. Fixed‐effect

We synthesised data using the random‐effects model; however, we also planned to synthesise data for the primary outcome using a fixed‐effect model to evaluate whether this altered the significance of the results.

7. Separating endpoint and change data

We planned to analyse studies providing data in the form of endpoint scores and change scores separately for the continuous primary outcome mental state.

8. Skewed data

We planned to perform a sensitivity analysis excluding studies for which there was the suggestion of skewedness (mean/SD ratio lower than 2; see Data extraction and management). We planned that if this changed the results in comparison with the main analysis (from significantly favouring the intervention to significantly favouring the control, or vice versa), we would exclude these studies from the main analysis and present their data in 'Additional tables'.

Summary of findings and assessment of the certainty of the evidence

Summary of findings table

We used the GRADE approach to interpret findings (Schünemann 2017), and employed GRADEpro GDT software to export data from the review to create summary of findings tables (GRADEpro GDT; RevMan Web 2022). These tables provide outcome‐specific information concerning the overall certainty of evidence from each study included in the comparison, the magnitude of effect of the interventions examined, and the sum of available data on all outcomes we rate as important to patient care and decision‐making. The overall RoB 2 judgements were used to feed into the GRADE assessment. We selected the following main outcomes for inclusion in the summary of findings tables.

• Mental state: average endpoint or change score on a general mental state scale (e.g. PANSS, BPRS) (primary outcome) – medium term

• Global state: relapse – long term

• Leaving the study early: leaving the study early for any reason – medium term

• Functioning: average endpoint or change score on functioning scale – medium term

• At least one adverse event – long term

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

Results

Description of studies

For details, see Included studies and Excluded studies.

Results of the search

Our search of the Cochrane Schizophrenia Group's Study‐Based Register of Trials identified 1387 eligible records; 1 additional record was retrieved by handsearching. We excluded 1349 records based on title and abstract. We assessed the remaining 38 full‐text articles and included 4 studies (29 full‐text articles) (Figure 1).

1.

Study flow diagram.

Included studies

We included 4 studies (300 participants); all trials contributed to the meta‐analysis.

1. Design and duration

All studies were RCTs. The studies had an intervention period of 26 to 39 weeks and a follow‐up period of 26 to 104 weeks (Francey 2020 – STAGES; Morrison 2014 – ACTION; Morrison 2018 – COMPARE; Morrison 2020 – MAPS). One study was triple‐blind (Francey 2020 – STAGES); the other three studies employed a blind rater.

2. Participants

Participants in one study were diagnosed using Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition (DSM‐IV) criteria (Francey 2020 – STAGES); the remaining three studies used the International Classification of Diseases 10th Revision (ICD‐10) (Morrison 2014 – ACTION; Morrison 2018 – COMPARE; Morrison 2020 – MAPS). Two studies also accepted participants who met entry criteria for an early intervention for psychosis service (Morrison 2014 – ACTION; Morrison 2018 – COMPARE), while the other two studies specifically recruited first episode patients, also including children and adolescents (Francey 2020 – STAGES; Morrison 2020 – MAPS). The average age of participants was 21.94 years.

3. Size

The mean number of participants was 75. The smallest study included 61 participants (Morrison 2020 – MAPS), and the largest study included 90 participants (Francey 2020 – STAGES).

4. Setting

Participants were recruited in oupatient service in two studies (Francey 2020 – STAGES; Morrison 2014 – ACTION), and both inpatient‐outpatient settings in one study (Morrison 2018 – COMPARE); this information was unclear in the remaining study (Morrison 2020 – MAPS). Three studies were conducted in the UK (Morrison 2014 – ACTION; Morrison 2018 – COMPARE; Morrison 2020 – MAPS), and one study was conducted in Australia (Francey 2020 – STAGES).

5. Interventions

All studies had an intervention group using CBT techniques, and the intervention was manualised. Two studies planned 26 CBT sessions (Morrison 2014 – ACTION; Morrison 2018 – COMPARE); one study planned 24 CBT sessions (Francey 2020 – STAGES); and this information was not specified in the remaining study (Morrison 2020 – MAPS). Therapy was provided face‐to‐face and individually in all studies. Pharmacotherapy was allowed in all studies.

6. Outcomes

Scales used to assess symptoms and other conditions varied. Study reporting was in some cases incomplete, and authors were approached for missing outcome data and clarifications by email. The supplementary materials and appendices were also inspected and used as a source of data.

6.1. Outcome scales

6.1.1 Mental state

Brief Psychiatric Rating Scale (BPRS) (Overall 1962)

BPRS is a clinician‐rated scale used to quantify the severity of psychiatric symptoms, including psychotic symptoms. The most commonly used form of the scale has 18 items that include positive, negative, and affective symptoms. Each item is graded on a 7‐point Likert scale ranging from 1 to 7 (1 "not present" and 7 "very severe"). A total score can be obtained by summing the scores of all items as a measure of overall schizophrenia symptoms (range from 18 to 126, a higher number showing higher severity of symptoms). One study used the BPRS (Francey 2020 – STAGES).

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

Created using the BPRS (see above), the PANSS is a 30‐item clinician‐rated assessment that includes positive, negative, and general psychopathology symptoms of schizophrenia. Each item is graded on a 7‐point Likert scale (1 "absent" and 7 "severe"). A total score can be derived by adding the scores of all items as a measure of overall schizophrenia symptoms (ranging from 30 to 210, a higher score showing higher severity of the condition). PANSS comprises three subscales: 1) positive syndromes, 2) negative syndromes, and 3) general psychopathology. Three studies used the PANSS (Morrison 2014 – ACTION; Morrison 2018 – COMPARE; Morrison 2020 – MAPS).

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

The first instrument developed to provide a comprehensive assessment of negative symptoms in schizophrenia was the Scale for the Assessment of Negative Symptoms (SANS) (Andreasen 1984), which consists of five scales that assess five different aspects of negative symptoms: affective blunting, alogia, avolition‐apathy, anhedonia‐asociality, and attentional impairment. Each of these negative symptoms can be rated globally, but in addition detailed observations are made to achieve the global rating.

Scale for the Assessment of Positive Symptoms (SAPS)

The SAPS permits a detailed evaluation and global ratings of hallucinations, delusions, positive formal thought disorder, and bizarre behaviour (Andreasen 1984).

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

HADS was developed by Zigmond and Snaith in 1983 and is a self‐assessment scale used for screening purposes to determine anxiety and depression, not for diagnosis. The 14‐item HADS includes the anxiety subscale (HADS‐A) and the depression subscale (HADS‐D), each consisting of 7 items. A score between 0 and 3 is obtained for each item by choosing from among four options. By adding the subscale scores, a total of 0 to 21 points can be obtained from each of the depression and anxiety subscales. Zero 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 the HADS (Morrison 2018 – COMPARE; Morrison 2020 – MAPS).

Beck Depression Inventory (BDI) (Beck 1961)

BDI is a 21‐item self‐reporting questionnaire for evaluating the severity of depression in normal and psychiatric populations. Developed by Beck and colleagues in 1961, it relied on the theory of negative cognitive distortions as central to depression. A total score is obtained by scoring each item between 0 and 3 points. Each of the 21 questions presents four different statements and asks respondents to select the option that best represents them. Statements refer to depressive states in varying degrees of severity (from “I do not feel sad” to “I am so sad or unhappy that I can’t stand it”). Only one study used the BDI (Morrison 2014 – ACTION).

Social Interaction Anxiety Scale (SIAS)

The Social Interaction Anxiety Scale (SIAS) and its companion scale, the Social Phobia Scale (SPS), were created in response to a need for tools that assess various commonly feared social situations (Mattick 1998). The SIAS was developed on the premise that social anxiety manifests itself in two types of situations: those involving social interaction with others (e.g. initiating and maintaining conversation) and those involving being observed or scrutinised by others (e.g. giving a speech or eating in public) (Mattick 1998). The SIAS is an attempt to quantify the first of the two concepts, social interaction anxiety, and includes 20 questions rated on a 5‐point Likert scale ranging from "not at all typical of me" to "extremely typical of me". Items are self‐statements that describe how people react to interpersonal contacts in dyads or groups. Summing the ratings after opposite scoring three positively worded items yields a total SIAS score. Only one study used the SIAS (Morrison 2014 – ACTION).

6.1.2 Global state

Clinical Global Impression (CGI) (Guy 1976)

CGI is a clinically graded 7‐step scale consisting of two scales that measure overall disease severity (CGI‐Severity, or CGI‐S) and overall clinical improvement (CGI‐Improvement, or CGI‐I). Lower scores represent greater improvement in the disease (or less deterioration). A CGI‐I score of 1 "very much improved" or 2 "much improved" corresponds to a clinically meaningful improvement. One study used the CGI (Morrison 2018 – COMPARE).

6.1.3 Functioning

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

PSP is a short scale with a 100‐point single‐item, with scores of 1 to 10 reflecting a lack of basic functional autonomy, and scores of 91 to 100 reflecting excellent functioning. Ratings are based on the evaluation of four objective indicators: "(a) socially useful activities, including work and study; (b) personal and social relationships, (c) self‐care, (d) disruptive and aggressive behavior". These are rated on a 6‐point severity scale (absent to extremely severe), following specific operational definitions (Morosini 2000). Two studies used the PSP (Morrison 2014 – ACTION; Morrison 2018 – COMPARE).

Social and Occupational Functioning Assessment Scale (SOFAS) (APA 1994)

SOFAS was developed to assess the level of social and occupational functioning of individuals not directly affected by the overall severity of their psychiatric symptoms. The scale also takes into account the impact on a person's general health when assessing social and professional functioning. This scale, found in DSM‐IV, is used in particular to determine 5th axis features and to measure functionality. It ranges from 0 to 100 points, with higher scores indicating good functioning. One study used the SOFAS (Francey 2020 – STAGES).

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

FESFS was developed in 2014 by Lecomte and colleagues and assesses nine functional areas: "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”. The participant is asked to rate his or her perceived competencies and frequency of involvement in these areas on a scale from 1 to 4 (e.g. "strongly agree" to "strongly disagree" or "strongly disagree" to "always agree"). Higher scores indicate better social and occupational functioning. One study used the FESFS (Morrison 2020 – MAPS).

6.1.4 Cognitive functioning

No studies used a scale measuring cognitive functioning.

6.1.5 Quality of life

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

WHOQOL‐BREF is a self‐reported scale used to assess quality of life. It contains 26 questions about recent health satisfaction, psychological functioning, social relationships, and environmental opportunities. Each question is rated from 1 to 5 on a 5‐point Likert scale. An overall score of 26 to 130 is determined by summing the scores of all items, with higher scores indicating better quality of life. One study used the WHOQOL‐BREF (Morrison 2018 – COMPARE).

EQ‐5D‐5L (The EuroQol Group 1990)

EQ‐5D‐5L is a self‐report scale that measures quality of life. The five dimensions of the descriptive system are mobility, self‐care, usual activities, pain/discomfort, and anxiety/depression. There are five levels for each dimension: no problems, minor problems, moderate problems, severe problems, and extreme problems. The participant is asked to rate his or her health by checking the box beside the most suitable statement within each of the five dimensions. This decision yields a 1‐digit number expressing the level chosen for that dimension. The numbers for the five dimensions can be combined to generate a 5‐digit number that defines the person's health status. One study used the EQ‐5D‐5L (Morrison 2020 – MAPS).

Quality of Life Scale (QLS) (Heinrichs 1984)

QLS is a 21‐item scale based on a semi‐structured interview designed to assess deficit symptoms in schizophrenia. It is made up of 21 items divided into 4 categories: 1) interpersonal relations, 2) instrumental role, 3) intrapsychic foundations, and 4) common objects and activities. Each item is rated on a 7‐point Likert scale ranging from 0 to 6, with a higher score representing less impairment in the past four weeks. One study used the QLS (Francey 2020 – STAGES).

7. Funding sources

All studies reported receiving public funding.

Excluded studies

We excluded three studies based on full‐text assessment. Reasons for exclusion were non‐randomised study design (Morrison 2012), and lack of a CBT without antipsychotic arm (Morrison 2016; Morrison 2018 – FOCUS).

We identified no ongoing studies or studies awaiting classification.

Risk of bias in included studies

The risk of bias assessment for each of the predefined outcomes is located in the Risk of bias (tables), 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.

Regarding overall risk of bias, one study had a low risk of bias (Francey 2020 – STAGES); one study had some concerns (Morrison 2014 – ACTION); and two studies had a high risk of bias (Morrison 2018 – COMPARE; Morrison 2020 – MAPS).

The main problem in the included studies was represented by domain deviations from intended interventions. The included studies tried to administer CBT without antipsychotics medication to at least one group of participants, which can be challenging. Indeed, during the trial there were deviations from the randomised intervention.

In Morrison 2014 – ACTION, some participants started to receive antipsychotic medication during the trial against randomisation; however, the numbers were the same in both groups. We thus assessed this study as some concerns for the domain deviations from intended interventions.

In Morrison 2018 – COMPARE and Morrison 2020 – MAPS, many participants randomised to the CBT without antipsychotics group received antipsychotics (30.8% in COMPARE and 33.3% in MAPS). This deviation from intended intervention was not balanced between groups, because fewer participants randomised to receive antipsychotics started to receive CBT during the trial (4% in COMPARE and 11% in MAPS). We thus assessed this study as high risk of bias for this domain.

Francey 2020 – STAGES was triple‐blind (participants, clinicians, and outcome assessors), therefore according to the RoB 2 algorithm deviations from intended interventions do not represent a problem, and we assessed this domain as low risk of bias.

Effects of interventions

See: Table 1; Table 2; Table 3

See Table 1, Table 2, Table 3, and forest plots for detailed results.

Comparison 1: CBT versus no specific treatment

For a correct interpretation of the results, it must be considered that about 27% of participants in the CBT arm of the study included in this comparison also received antipsychotics during the study, against randomisation. The same percentage of participants received antipsychotics in the control arm, against randomisation. For details, see Table 4.

1. Numbers of participants receiving antipsychotic medication during the trial.

	CBT	CBT plus antipsychotics	Antipsychotics	No specific treatment
Francey 2020 – STAGES	19/41 (46.3%)	‐	‐	‐
Morrison 2018 – COMPARE	8/26 (30.8%)	18/25 (72%)	16/24 (66.7%)	‐
Morrison 2020 – MAPS	6/18 (33.3%)	12/21 (57.1%)	16/22 (72.7%)	‐
Morrison 2014 – ACTION	10/37 (27%)	‐	‐	10/37 (27%)

CBT: cognitive behavioural therapy

Primary outcome

1. Mental state

1.1 General

1.1.1 Mean endpoint or change score on general mental state scale

One study provided data on this outcome. Results did not show a difference between CBT and no specific treatment on mental state measured with PANSS at up to 6 months and after 1 year (up to 6 months: mean difference (MD) −6.99, 95% confidence interval (CI) −14.91 to 0.93, 1 RCT, n = 42; up to 12 months: MD −9.77, 95% CI −20.59 to 1.05, 1 RCT, n = 39, low−certainty evidence). At over one year, CBT without antipsychotics was associated with a reduction in PANSS scores compared to no specific treatment (MD −14.77, 95% CI −27.75 to −1.79, 1 RCT, n = 34). No difference emerged between time points (test for subgroup differences P = 0.60) (Analysis 1.1).

1.1. Analysis.

Comparison 1: CBT versus no specific treatment, Outcome 1: Mental state – general: mean endpoint PANSS total (high = poor)

Secondary outcomes

1. Mental state

1.1 General

1.1.1 Number of participants with clinically important change in general mental state

One study provided data on this outcome. Results did not show a difference between CBT and no specific treatment on number of participants with clinically important change in general mental state (up to 12 months: risk ratio (RR) 2.33, 95% CI 0.65 to 8.34, 1 RCT, n = 74; > 1 year: RR 2.33, 95% CI 0.65 to 8.34, 1 RCT, n = 74) (Analysis 1.2).

1.2. Analysis.

Comparison 1: CBT versus no specific treatment, Outcome 2: Number of participants with clinically important change in general mental state (separated time points)

1.2 Specific

1.2.1 Mean endpoint or change score on positive symptoms scale

Based on one study, results did not show a difference between CBT and no specific treatment on positive symptoms measured with PANSS positive (up to 6 months: MD −1.28, 95% CI −4.23 to 1.67, 1 RCT, n = 42; up to 12 months: MD −2.30, 95% CI −6.52 to 1.92, 1 RCT, n = 40; > 1 year: MD −4.20, 95% CI −8.56 to 0.16, n = 37). No difference emerged between time points (test for subgroup differences P = 0.55) (Analysis 1.3).

1.3. Analysis.

Comparison 1: CBT versus no specific treatment, Outcome 3: Mental state – specific: mean endpoint PANSS positive (high = poor)

1.2.2 Mean endpoint or change score on negative symptoms scale

One study provided data on this outcome. Results did not show a difference between CBT and no specific treatment on negative symptoms measured with PANSS negative up to 6 months (up to 6 months: MD −1.47, 95% CI −3.72 to 0.78, 1 RCT, n = 42). At later time points, CBT without antipsychotics was associated with a reduction in negative symptoms (up to 12 months: MD −3.34, 95% CI −6.53 to −0.15, 1 RCT, n = 39; > 1 year: MD −4.06, 95% CI −7.50 to −0.62, n = 34). Test for subgroup differences did not show a difference between time points (P = 0.39) (Analysis 1.4).

1.4. Analysis.

Comparison 1: CBT versus no specific treatment, Outcome 4: Mental state – specific: mean endpoint PANSS negative (high = poor)

1.2.3 Mean endpoint or change score on depressive symptoms scale

Based on one study, results did not show a difference between CBT and no specific treatment on depressive symptoms measured with BDI at any of the analysed time points (up to 6 months: MD 0.20, 95% CI −2.80 to 3.20, 1 RCT, n = 40; up to 12 months: MD 0.44, 95% CI −2.94 to 3.82, 1 RCT, n = 35; > 1 year: MD −1.88, 95% CI −5.62 to 1.86, n = 32). Test for subgroup differences did not show a difference between time points (P = 0.32) (Analysis 1.5).

1.5. Analysis.

Comparison 1: CBT versus no specific treatment, Outcome 5: Mental state – specific: mean endpoint BDI (high = poor)

1.2.4 Mean endpoint or change score on anxiety symptoms scale

One study provided data on this outcome. Results did not show a difference between CBT and no specific treatment on anxiety symptoms measured with BAI (Beck Anxiety Inventory) at any of the analysed time points (up to 6 months: MD −3.15, 95% CI −13.34 to 7.04, 1 RCT, n = 37; up to 12 months: MD −11.86, 95% CI −25.61 to 1.89, 1 RCT, n = 29; > 1 year: MD −12.75, 95% CI −26.32 to 0.82, n = 32). Test for subgroup differences did not show a difference between time points (P = 0.44) (Analysis 1.6).

1.6. Analysis.

Comparison 1: CBT versus no specific treatment, Outcome 6: Mental state – specific: mean endpoint BAI (high = poor)

2. Global state

2.1 Relapse/exacerbations of psychosis

One study provided data on this outcome. Results did not show a difference between CBT and no specific treatment on relapse/exacerbations of psychosis at any of the analysed time points (up to 12 months: RR 3.00, 95% CI 0.13 to 71.34, 1 RCT, n = 74; > 1 year: RR 0.5, 95% CI 0.05 to 5.28, n = 74, low‐certainty evidence). Test for subgroup differences did not show a difference between time points (P = 0.37) (Analysis 1.7).

1.7. Analysis.

Comparison 1: CBT versus no specific treatment, Outcome 7: Global state – number of participants with relapse/exacerbations of psychosis (separated time points)

3. Service use

3.1 Readmission to hospital

Based on one study, results did not show a significant difference between CBT and no specific treatment on readmission to hospital up to 12 months (RR 1.00, 95% CI 0.27 to 3.70, 1 RCT, n = 74) (Analysis 1.8).

1.8. Analysis.

Comparison 1: CBT versus no specific treatment, Outcome 8: Service use – readmission to hospital

3.2 Number of days in hospital

Based on data from one study, CBT is associated with a lower number of days in hospital compared to no specific treatment (up to 12 months: MD −22.45, 95% CI −28.82 to −16.08, 1 RCT, n = 74) (Analysis 1.9).

1.9. Analysis.

Comparison 1: CBT versus no specific treatment, Outcome 9: Service use – mean number of days in hospital

4. Leaving the study early

4.1 For any reason – overall acceptability

One study provided data on this outcome. There was no evidence of a difference between CBT and no specific treatment in leaving the study early for any reason at any of the analysed time points (up to 6 months: RR 1.33, 95% CI 0.32 to 5.55, 1 RCT, n = 74; up to 12 months: RR 0.80, 95% CI 0.23 to 2.75, 1 RCT, n = 74, low‐certainty evidence; > 1 year: RR 2.50, 95% CI 0.52 to 12.08). No differences emerged between time points (P = 0.53) (Analysis 1.10).

1.10. Analysis.

Comparison 1: CBT versus no specific treatment, Outcome 10: Leaving the study early – for any reason – overall acceptability (separated time points)

5. Functioning

5.1 Mean endpoint or change score on functioning scale

One study provided data on this outcome. Results did not show a difference between CBT and no specific treatment in functioning measured with PSP at up to six months (MD −7.85, 95% CI −18.13 to 2.43, 1 RCT, n = 42) and after more than one year (MD −8.80, 95% CI −21.86 to 4.26, 1 RCT, n = 37). CBT without antipsychotics was associated with better functioning at up to 12 months (MD −12.42, 95% CI −22.75 to −2.09, 1 RCT, n = 40, low‐certainty evidence). No differences emerged between the different time points (P = 0.82) (Analysis 1.11).

1.11. Analysis.

Comparison 1: CBT versus no specific treatment, Outcome 11: Functioning – mean endpoint PSP (high = good)

6. Cognitive functioning

No study provided data for this outcome.

7. Quality of life

No study provided data for this outcome.

8. Satisfaction with care

No study provided data for this outcome.

9. Behaviour

9.1 Suicide attempts

Based on one study, results did not show a difference between CBT and no specific treatment in suicide attempts at up to 12 months (RR 1.00, 95% CI 0.06 to 15.40, 1 RCT, n = 74) (Analysis 1.12).

1.12. Analysis.

Comparison 1: CBT versus no specific treatment, Outcome 12: Behaviour – participants with suicide attempts

10. Adverse events

10.1 Deterioration of symptoms

One study provided data on this outcome. Results did not show a difference between CBT and no specific treatment in deterioration of symptoms at any time point (up to 12 months: RR 1.00, 95% CI 0.15 to 6.73, 1 RCT, n = 74; > 1 year: RR 0.50, 95% CI 0.10 to 2.56, 1 RCT, n = 74). There was no difference between time points (P = 0.59) (Analysis 1.13).

1.13. Analysis.

Comparison 1: CBT versus no specific treatment, Outcome 13: Adverse events – deterioration of symptoms

10.2 Participants with serious adverse event

Based on one study, results did not show a difference between CBT and no specific treatment in serious adverse events (up to 12 months: RR 0.33, 95% CI 0.07 to 1.54, 1 RCT, n = 74) (Analysis 1.14).

1.14. Analysis.

Comparison 1: CBT versus no specific treatment, Outcome 14: Adverse events – participants with serious adverse events

11. Mortality

11.1 Overall mortality

One study provided data on this outcome. There was no evidence of a difference between CBT and no specific treatment in overall mortality at any time point (up to 12 months: RR 0.33, 95% CI 0.01 to 7.93, 1 RCT, n = 74; > 1 year: RR 0.20, 95% CI 0.01 to 4.03, n = 74). There was no difference between time points (P = 0.82) (Analysis 1.15).

1.15. Analysis.

Comparison 1: CBT versus no specific treatment, Outcome 15: Mortality – overall mortality

11.2 Mortality due to natural causes

One study provided data on this outcome. There was no evidence of a difference between CBT and no specific treatment in mortality due to natural causes at any time point (up to 12 months: RR 0.20, 95% CI 0.01 to 4.03, 1 RCT, n = 74; > 1 year: RR 0.11, 95% CI 0.01 to 1.99, n = 74). There was no difference between time points (P = 0.78) (Analysis 1.16).

1.16. Analysis.

Comparison 1: CBT versus no specific treatment, Outcome 16: Mortality – mortality due to natural causes

12. Economic outcomes

No study provided data for this outcome.

Comparison 2: CBT versus antipsychotics

For a correct interpretation of the results, it must be considered that about 32% participants in the CBT arms of the studies included in this comparison also received antipsychotics during the studies, against randomisation. For details, see Table 4.

Primary outcome

1. Mental state

1.1 General

1.1.1 Mean endpoint or change score on general mental state scale

Two studies provided data on this outcome. Results did not show a difference between CBT and antipsychotics on mental state measured with PANSS at any of the predefined examined time points (up to 6 months: MD −3.75, 95% CI −11.61 to 4.12, I² = 39% (not important to moderate heterogeneity), 2 RCTs, n = 78; up to 12 months: MD 3.38, 95% CI −2.38 to 9.14, I² = 0%, 2 RCTs, n = 63, very low‐certainty evidence). No differences emerged between time points (test for subgroup differences P = 0.15) (Analysis 2.1).

2.1. Analysis.

Comparison 2: CBT versus antipsychotics, Outcome 1: Mental state – general: mean endpoint PANSS total (high = poor)

Secondary outcomes

1. Mental state

1.1 General

1.1.1 Clinically important change in general mental state

Two studies provided data on this outcome. There was no evidence of a difference between CBT and antipsychotics in number of participants with a clinically important change in general mental state at any of the analysed time points (up to 6 months: RR 1.15, 95% CI 0.44 to 2.98, I² = 0%, 2 RCTs, n = 90; up to 12 months: RR 0.90, 95% CI 0.36 to 2.28, I² = 0%, 2 RCTs, n = 90). No differences emerged between time points (test for subgroup differences P = 0.72) (Analysis 2.2).

2.2. Analysis.

Comparison 2: CBT versus antipsychotics, Outcome 2: Number of participants with clinically important change in general mental state (separated time points)

1.2 Specific

1.2.1 Mean endpoint or change score on positive symptoms scale

Two studies provided data on this outcome. There was no evidence of a difference between CBT and antipsychotics in positive symptoms measured with PANSS positive at any of the analysed time points (up to 6 months: MD 0.25, 95% CI −3.08 to 3.58, I² = 0%, 2 RCTs, n = 77; up to 12 months: MD 1.96, 95% CI −2.80 to 6.73, I² = 62% (substantial heterogeneity), 2 RCTs, n = 63). No differences emerged between time points (test for subgroup differences P = 0.56) (Analysis 2.3).

2.3. Analysis.

Comparison 2: CBT versus antipsychotics, Outcome 3: Mental state – specific: mean endpoint PANSS positive (high = poor)

1.2.2 Mean endpoint or change score on negative symptoms scale

Two studies provided data on this outcome. There was no evidence of a difference between CBT and antipsychotics in negative symptoms measured with PANSS negative at any of the analysed time points (up to 6 months: MD −1.19, 95% CI −5.27 to 2.89, I² = 67% (substantial heterogeneity), 2 RCTs, n = 78; up to 12 months: MD 0.54, 95% CI −2.63 to 3.71, I² = 46% (moderate heterogeneity), 2 RCTs, n = 63). No differences emerged between time points (test for subgroup differences P = 0.51) (Analysis 2.4).

2.4. Analysis.

Comparison 2: CBT versus antipsychotics, Outcome 4: Mental state – specific: mean endpoint PANSS negative (high = poor)

1.2.3 Mean endpoint or change score on depressive symptoms scale

Two studies provided data on this outcome. Results did not show a difference between CBT and antipsychotics in depressive symptoms measured with HADS depression at any of the analysed time points (up to 6 months: MD −1.75, 95% CI −4.27 to 0.77, I² = 0%, 2 RCTs, n = 59; up to 12 months: MD −0.68, 95% CI −3.08 to 1.73, I² = 0%, 2 RCTs, n = 54). No differences emerged between time points (test for subgroup differences P = 0.55) (Analysis 2.5).

2.5. Analysis.

Comparison 2: CBT versus antipsychotics, Outcome 5: Mental state – specific: mean endpoint HADS Depression (high = poor)

1.2.4 Mean endpoint or change score on anxiety symptoms scale

Two studies provided data on this outcome. There was no evidence of a difference between CBT and antipsychotics in symptoms of anxiety measured with HADS anxiety at any of the analysed time points (up to 6 months: MD −1.30, 95% CI −4.02 to 1.43, I² = 0%, 2 RCTs, n = 60; up to 12 months: MD −1.34, 95% CI −3.77 to 1.09, I² = 0%, 2 RCTs, n = 54). No differences emerged between time points (test for subgroup differences P = 0.98) (Analysis 2.6).

2.6. Analysis.

Comparison 2: CBT versus antipsychotics, Outcome 6: Mental state – specific: mean endpoint HADS Anxiety (high = poor)

2. Global state

2.1 Relapse/exacerbations of psychosis

Two studies provided data on this outcome. There was no evidence of a difference between CBT and antipsychotics in relapse or exacerbations of psychosis at any of the analysed time points (up to 6 months: RR 2.35, 95% CI 0.36 to 15.31, I² = 0%, 2 RCTs, n = 90; up to 12 months: RR 2.78, 95% CI 0.12 to 65.08, I² = 0%, 2 RCTs, n = 90, very low‐certainty evidence). No differences emerged between time points (test for subgroup differences P = 0.93) (Analysis 2.7).

2.7. Analysis.

Comparison 2: CBT versus antipsychotics, Outcome 7: Global state – number of participants with relapse/exacerbations of psychosis (separated time points)

2.2 Mean endpoint or change score on global state scale

One study provided data on this outcome. No difference was shown between CBT and antipsychotics in CGI‐Clinician endpoint scores at any of the analysed time points (up to 6 months: MD 0.13, 95% CI −0.49 to 0.75, 1 RCT, n = 44; up to 12 months: MD 0.15, 95% CI −0.50 to 0.80, 1 RCT, n = 43). No differences emerged between time points (test for subgroup differences P = 0.97). Similar results were obtained by CGI‐Improvement (up to 6 months: MD −0.17, 95% CI −0.89 to 0.55, 1 RCT, n = 44; up to 12 months: MD 0.05, 95% CI −0.57 to 0.67, 1 RCT, n = 42). Test for subgroup differences did not reveal a difference between time points (P = 0.65) (Analysis 2.8; Analysis 2.9).

2.8. Analysis.

Comparison 2: CBT versus antipsychotics, Outcome 8: Global state – mean endpoint CGI‐Clinician (high = poor)

2.9. Analysis.

Comparison 2: CBT versus antipsychotics, Outcome 9: Global state – mean endpoint CGI‐Clinician – Improvement (high = poor)

3. Service use

3.1 Readmission to hospital

Two studies provided data on this outcome. In one study, no participants were readmitted to the hospital at up to six months. In the other study, three participants in the CBT group and zero participants in the antipsychotics group were readmitted to the hospital (up to 12 months: RR 6.48, 95% CI 0.35 to 119.32, 1 RCT, n = 50) (Analysis 2.10).

2.10. Analysis.

Comparison 2: CBT versus antipsychotics, Outcome 10: Service use – readmission to hospital

3.2 Number of days in hospital

One study provided data on this outcome. Mean number of days in hospital was 50 in the CBT arm and 0 in the standard care (including antipsychotics) arm. It was not possible to estimate a mean difference (Analysis 2.11).

2.11. Analysis.

Comparison 2: CBT versus antipsychotics, Outcome 11: Service use – number of days in hospital (separated time points)

4. Leaving the study early

4.1 For any reason – overall acceptability

Two studies provided data on this outcome. There was no evidence of a difference between CBT and antipsychotics in leaving study early for any reason at any of the analysed time points (up to 6 months: RR 1.66, 95% CI 0.21 to 12.94, I² = 0%, 2 RCTs, n = 90; up to 12 months: RR 1.66, 95% CI 0.21 to 12.94, I² = 0%, 2 RCTs, n = 90, very low‐certainty evidence). No differences emerged between time points (test for subgroup differences P = 1.00) (Analysis 2.12).

2.12. Analysis.

Comparison 2: CBT versus antipsychotics, Outcome 12: Leaving the study early – for any reason – overall acceptability (separated time points)

5. Functioning

5.1 Mean endpoint or change score on functioning scale

Two studies provided data on this outcome. Results did not show a difference between CBT and antipsychotics in functioning (measured with PSP and FESFS scales) (up to 6 months: standardised mean difference (SMD) 0.18, 95% CI −0.37 to 0.72, I² = 0%, 2 RCTs, n = 53; up to 12 months: SMD −0.03, 95% CI −0.61 to 0.55, I² = 0%, 2 RCTs, n = 46, very low‐certainty evidence). No differences emerged between time points (test for subgroup differences P = 0.92) (Analysis 2.15).

2.15. Analysis.

Comparison 2: CBT versus antipsychotics, Outcome 15: Functioning – mean endpoint functioning scales (high = good)

6. Cognitive functioning

No study reported data on this outcome.

7. Quality of life

7.1 Mean endpoint or change score on quality of life scale

Based on two studies, results did not show a difference between CBT and antipsychotics in quality of life (measured with WHOQOL and EQ‐5D‐5L at any of the predefined time points (up to 6 months: SMD 0.11, 95% CI −0.42 to 0.64, I² = 0%, 2 RCTs, n = 56; up to 12 months: SMD −0.09, 95% CI −0.64 to 0.46, I² = 0%, 2 RCTs, n = 51). No differences emerged between time points (test for subgroup differences P = 0.61) (Analysis 2.18).

2.18. Analysis.

Comparison 2: CBT versus antipsychotics, Outcome 18: Quality of life – mean endpoint quality of life scales (high = good)

8. Satisfaction with care

No study reported data on this outcome.

9. Behaviour

9.1 Violent incidents

Two studies provided data on this outcome. No difference was found between CBT and antipsychotics in violent incidents (up to 12 months: RR 1.22, 95% CI 0.19 to 7.84, 2 RCTs, n = 90) (Analysis 2.19).

2.19. Analysis.

Comparison 2: CBT versus antipsychotics, Outcome 19: Behaviour – violent incidents (separated time points)

9.2 Self‐injury

Two studies provided data on this outcome. No difference was found between CBT and antipsychotics in self‐injury (up to 12 months: RR 1.05, 95% CI 0.43 to 2.56, 2 RCTs, n = 90) (Analysis 2.20).

2.20. Analysis.

Comparison 2: CBT versus antipsychotics, Outcome 20: Behaviour – self‐injury (separated time points)

9.3 Suicide attempts

Based on one study, results did not show a difference between CBT and antipsychotics in suicide attempts (up to 12 months: RR 3.63, 95% CI 0.16 to 84.11, 1 RCT, n = 40) (Analysis 2.21).

2.21. Analysis.

Comparison 2: CBT versus antipsychotics, Outcome 21: Behaviour – suicide attempts (separated time points)

10. Adverse events

10.1 Participants with at least one adverse event

Based on one study, results did not show a difference between CBT and antipsychotics in number of participants experiencing at least one adverse event (up to 12 months: RR 0.47, 95% CI 0.21 to 1.07, 1 RCT, n = 40, very low‐certainty evidence) (Analysis 2.22).

2.22. Analysis.

Comparison 2: CBT versus antipsychotics, Outcome 22: Adverse events – participants with at least 1 adverse event

10.2 Participants with at least one serious adverse event

Two studies provided data on this outcome. There was no evidence of a difference between CBT and antipsychotics in number of participants experiencing at least one serious adverse event (up to 12 months: RR 2.83, 95% CI 0.80 to 10.05, I² = 0%, 2 RCTs, n = 90) (Analysis 2.23).

2.23. Analysis.

Comparison 2: CBT versus antipsychotics, Outcome 23: Adverse events – participants with serious adverse events

10.3 Deterioration of symptoms

Two studies provided data on this outcome. Results did not show a difference between CBT and antipsychotics in deterioration of symptoms at any of the analysed time points (up to 6 months: RR 2.13, 95% CI 0.41 to 11.02, I² = 0%, 2 RCTs, n = 90; up to 12 months: RR 0.46, 95% CI 0.04 to 4.77, 2 RCTs, n = 90). Test for subgroup differences did not show a difference between time points (P = 0.29) (Analysis 2.24).

2.24. Analysis.

Comparison 2: CBT versus antipsychotics, Outcome 24: Adverse events – deterioration of symptoms

10.4 Mean endpoint or change score on adverse event scale

Two studies provided data on this outcome. CBT was associated with fewer adverse events in ANNSERS (Antipsychotic Non‐Neurological Side Effects Rating Scale) endpoint scores compared to antipsychotics at all of the analysed time points (up to 6 months: MD −4.94, 95% CI −8.60 to −1.28, I² = 0%, 2 RCTs, n = 48; up to 12 months: MD −6.96, 95% CI −11.55 to −2.37, I² = 0%, 2 RCTs, n = 42) (Analysis 2.25).

2.25. Analysis.

Comparison 2: CBT versus antipsychotics, Outcome 25: Adverse events – mean endpoint ANNSERS (high = poor)

11. Mortality

No study reported data on this outcome.

12. Economic outcomes

No study reported data on this outcome.

Comparison 3: CBT versus CBT plus antipsychotics

For a correct interpretation of the results, it must be considered that, on average, about 39% of the participants in the CBT arms of the studies included in this comparison also received antipsychotics during the studies, against randomisation. For details, see Table 4.

Primary outcome

1. Mental state

1.1 General

1.1.1 Mean endpoint or change score on general mental state scale

Three studies provided data on this outcome. Results did not show a difference between CBT and CBT plus antipsychotics on mental state measured with PANSS or BPRS at any of the prespecified time points (up to 6 months: SMD 0.20, 95% CI −0.21 to 0.60, I² = 31% (not important heterogeneity), 3 RCTs, n = 143; up to 12 months: SMD 0.30, 95% CI −0.06 to 0.65, I² = 0%, 3 RCTs, n = 125, very low‐certainty evidence; > 1 year: SMD 0.22, 95% CI −0.32 to 0.76, 1 RCT, n = 55) (Analysis 3.3). No differences emerged between time points (test for subgroup differences P = 0.93).

3.3. Analysis.

Comparison 3: CBT versus CBT plus antipsychotics, Outcome 3: Mental state – general: mean endpoint overall symptoms scales (PANSS/BPRS) (high = poor) (separated time points)

Secondary outcomes

1. Mental state

1.1 General

1.1.1 Clinically important change in general mental state

Two studies provided data on this outcome. No difference was found between CBT and CBT plus antipsychotics in number of participants with a clinically important change in general mental state at any of the analysed time points (up to 6 months: RR 0.62, 95% CI 0.15 to 2.53, I² = 59% (moderate heterogeneity), 2 RCTs, n = 90; up to 12 months: RR 0.80, 95% CI 0.33 to 1.98, I² = 0%, 2 RCTs, n = 90) (Analysis 3.4). No differences emerged between time points (test for subgroup differences P = 0.76).

3.4. Analysis.

Comparison 3: CBT versus CBT plus antipsychotics, Outcome 4: Mental state – general: number of participants with clinically important change in general mental state (separated time points)

1.2 Specific

1.2.1 Mean endpoint or change score on positive symptoms scale

Three studies provided data on this outcome. CBT plus antipsychotics was associated with a greater reduction in positive symptoms (PANSS and BPRS scales) up to 12 months, but not at up to 6 months (up to 6 months: SMD 0.22, 95% CI −0.19 to 0.64, I² = 35% (moderate heterogeneity), 3 RCTs, n = 146; up to 12 months: SMD 0.40, 95% CI 0.05 to 0.76, I² = 0%, 3 RCTs, n = 126). Test for subgroup differences confirmed a difference between time points (P = 0.03) (Analysis 3.7).

3.7. Analysis.

Comparison 3: CBT versus CBT plus antipsychotics, Outcome 7: Mental state – specific: mean endpoint positive symptoms scales (high = poor) (separated time points)

1.2.2 Mean endpoint or change score on negative symptoms scale

Three studies provided data on this outcome. Results showed a trend in favour of CBT plus antipsychotics in terms of reduction of negative symptoms (PANSS and SANS scales) at up to 12 months, but the CI does not exclude the possibility of no difference with CBT (up to 6 months: SMD 0.10, 95% CI −0.36 to 0.55, I² = 46% (moderate heterogeneity), 3 RCTs, n = 143; up to 12 months: SMD 0.40, 95% CI −0.01 to 0.81, I² = 20% (not important heterogeneity), 3 RCTs, n = 123). No differences emerged between time points (test for subgroup differences P = 0.33) (Analysis 3.10).

3.10. Analysis.

Comparison 3: CBT versus CBT plus antipsychotics, Outcome 10: Mental state – specific: mean endpoint negative symptoms scales (high = poor)

1.2.3 Mean endpoint or change score on depressive symptoms scale

Three studies provided data on this outcome. There was no evidence of a difference between CBT and CBT plus antipsychotics in depressive symptoms (HADS depression) in mental state at any time points (up to 6 months: SMD 0.18, 95% CI −0.18 to 0.54, I² = 0%, 3 RCTs, n = 123; up to 12 months: SMD 0.16, 95% CI −0.22 to 0.54, I² = 0%, 3 RCTs, n = 108). Test for subgroup differences did not show a difference between time points (P = 0.94) (Analysis 3.13).

3.13. Analysis.

Comparison 3: CBT versus CBT plus antipsychotics, Outcome 13: Mental state – specific: mean endpoint depression scales (high = poor)

1.2.4 Mean endpoint or change score on anxiety symptoms scale

Three studies provided data on this outcome. No difference was found between CBT and CBT plus antipsychotics in anxiety symptoms (HADS anxiety) at any time points (up to 6 months: SMD 0.03, 95% CI −0.37 to 0.43, I² = 17% (not important heterogeneity), 3 RCTs, n = 125; up to 12 months: SMD 0.21, 95% CI −0.17 to 0.59, I² = 0%, 3 RCTs, n = 109; > 1 year: SMD 0.23, 95% CI −0.31 to 0.77, 1 RCT, n = 55). Test for subgroup differences did not show a difference between time points (P = 0.76) (Analysis 3.16).

3.16. Analysis.

Comparison 3: CBT versus CBT plus antipsychotics, Outcome 16: Mental state – specific: mean endpoint anxiety scales (high = poor)

2. Global state

2.1 Relapse/exacerbations of psychosis

Two studies provided data on this outcome. There was no evidence of a difference between CBT and CBT plus antipsychotics in relapse or exacerbations of psychosis at any of the analysed time points (up to 6 months: RR 2.21, 95% CI 0.30 to 16.35, I² = 0%, 2 RCTs, n = 90; up to 12 months: RR 2.89, 95% CI 0.12 to 67.75, 1 RCT, n = 90, very low‐certainty evidence). No differences emerged between time points (test for subgroup differences P = 0.89) (Analysis 3.17).

3.17. Analysis.

Comparison 3: CBT versus CBT plus antipsychotics, Outcome 17: Global state – number of participants with relapse/exacerbations of psychosis (separated time points)

2.2 Mean endpoint or change score on global state scale

Only one study used CGI and CGI‐Improvement scales. Results showed a trend in favour of CBT plus antipsychotics in mean endpoint scores on global state scales at up to six months, although the CI does not exclude the possibility of no difference (CGI‐Clinician: up to 6 months: MD 0.59, 95% CI −0.00 to 1.18, 1 RCT, n = 43; up to 12 months: MD 0.38, 95% CI −0.28 to 1.04, 1 RCT, n = 41). No differences emerged between time points (test for subgroup differences P = 0.64) (Analysis 3.18). Results of CGI‐Improvement showed a similar pattern (up to 6 months: MD 0.64, 95% CI −0.00 to 1.28, 1 RCT, n = 43; up to 12 months: MD 0.25, 95% CI −0.41 to 0.91, 1 RCT, n = 40). No differences emerged between time points (test for subgroup differences P = 0.41) (Analysis 3.19).

3.18. Analysis.

Comparison 3: CBT versus CBT plus antipsychotics, Outcome 18: Global state – mean endpoint CGI‐Clinician (high = poor)

3.19. Analysis.

Comparison 3: CBT versus CBT plus antipsychotics, Outcome 19: Global state – mean endpoint CGI‐Clinician – Improvement (high = poor)

3. Service use

3.1 Readmission to hospital

One study provided data on this outcome. There was no evidence of a difference between CBT and CBT plus antipsychotics in readmission to hospital (up to 6 months: RR 0.39, 95% CI 0.02 to 8.93, 1 RCT, n = 39; up to 12 months: RR 0.64, 95% CI 0.12 to 3.52, 1 RCT, n = 51). No differences emerged between time points (test for subgroup differences P = 0.78) (Analysis 3.20).

3.20. Analysis.

Comparison 3: CBT versus CBT plus antipsychotics, Outcome 20: Service use – readmission to hospital (separated time points)

3.2 Number of days in hospital

One study provided data on this outcome. No difference was shown between CBT and CBT plus antipsychotics in number of days in hospital (up to 12 months: RR 7.75, 95% CI −26.53 to 42.03, 1 RCT, n = 41) (Analysis 3.21).

3.21. Analysis.

Comparison 3: CBT versus CBT plus antipsychotics, Outcome 21: Service use – number of days in hospital (separated time points)

4. Leaving the study early

4.1 For any reason – overall acceptability

Three studies provided data on this outcome. No difference was found between CBT and CBT plus antipsychotics in leaving the study early for any reason at any of the analysed time points (up to 6 months: RR 0.82, 95% CI 0.59 to 1.14, I² = 0%, 3 RCTs, n = 180; up to 12 months: RR 0.70, 95% CI 0.12 to 4.11, I² = 0%, 2 RCTs, n = 90, very low‐certainty evidence). Test for subgroup differences did not show a difference between time points (P = 0.87) (Analysis 3.22).

3.22. Analysis.

Comparison 3: CBT versus CBT plus antipsychotics, Outcome 22: Leaving the study early – for any reason – overall acceptability (separated time points)

4.2 Due to adverse events

One study provided data on this outcome. There was no evidence of a difference between CBT and CBT plus antipsychotics in leaving the study early due to adverse events up to six months (RR 0.19, 95% CI 0.01 to 3.88, 1 RCT, n = 100) (Analysis 3.23).

3.23. Analysis.

Comparison 3: CBT versus CBT plus antipsychotics, Outcome 23: Leaving the study early due to adverse effects (separated time points)

5. Functioning

5.1 Mean endpoint or change score on functioning scale

Three studies provided data on this outcome. No difference was found between CBT and CBT plus antipsychotics in functioning (PSP, FESFS, and SOFAS scales) at any of the analysed time points (up to 6 months: SMD 0.06, 95% CI −0.30 to 0.42, I² = 0%, 3 RCTs, n = 119; up to 12 months: SMD 0.13, 95% CI −0.25 to 0.50, I² = 0%, 3 RCTs, n = 110, very low‐certainty evidence; > 1 year: SMD 0.27, 95% CI −0.27 to 0.81, 1 RCT, n = 56). Test for subgroup differences did not show a difference between time points (P = 0.81) (Analysis 3.27).

3.27. Analysis.

Comparison 3: CBT versus CBT plus antipsychotics, Outcome 27: Functioning – mean endpoint functioning scales (high = good)

6. Cognitive functioning

No studies reported data on this outcome.

7. Quality of life

7.1 Mean endpoint or change score on quality of life scale

Three studies provided data on this outcome. No difference was found between CBT and CBT plus antipsychotics in quality of life (WHOQOL, EQ‐5D‐5L, and QLS) at any of the analysed time points (up to 6 months: SMD 0.37, 95% CI −0.11 to 0.85, I² = 32% (not important to moderate heterogeneity), 3 RCTs, n = 116; up to 12 months: SMD 0.17, 95% CI −0.21 to 0.56, I² = 0%, 3 RCTs, n = 106; > 1 year: SMD 0.15, 95% CI −0.40 to 0.70, 1 RCT, n = 52). No differences emerged between time points (test for subgroup differences P = 0.79) (Analysis 3.31).

3.31. Analysis.

Comparison 3: CBT versus CBT plus antipsychotics, Outcome 31: Quality of life – mean endpoint quality of life scales (high = good)

8. Satisfaction with care

No studies reported data on this outcome.

9. Behaviour

9.1 Violent incidents

Two studies provided data on this outcome. No difference was found between CBT and CBT plus antipsychotics in violent incidents at up to 12 months (RR 1.16, 95% CI 0.18 to 7.54, I² = 0%, 2 RCTs, n = 90) (Analysis 3.32).

3.32. Analysis.

Comparison 3: CBT versus CBT plus antipsychotics, Outcome 32: Behaviour – violent incidents (separated time points)

9.2 Self‐injury

Three studies provided data on this outcome. Results did not show a difference between CBT and CBT plus antipsychotics in self‐injury at any of the analysed time points (up to 6 months: RR 0.82, 95% CI 0.30 to 2.25, 1 RCT, n = 90; up to 12 months: RR 0.55, 95% CI 0.26 to 1.13, I² = 0%, 2 RCTs, n = 90). No differences emerged between time points (test for subgroup differences P = 0.52) (Analysis 3.33).

3.33. Analysis.

Comparison 3: CBT versus CBT plus antipsychotics, Outcome 33: Behaviour – self‐injury (separated time points)

9.3 Suicide attempts

One study provided data on this outcome. No difference was found between CBT and CBT plus antipsychotics in suicide attempts at up to 12 months (RR 1.17, 95% CI 0.08 to 17.35, 1 RCT, n = 39) (Analysis 3.34, Analysis 3.35).

3.34. Analysis.

Comparison 3: CBT versus CBT plus antipsychotics, Outcome 34: Behaviour – suicide attempts (separated time points)

3.35. Analysis.

Comparison 3: CBT versus CBT plus antipsychotics, Outcome 35: Adverse events – participants with at least 1 adverse event (pooled time points)

10. Adverse events

10.1 Participants with at least one adverse event

Two studies provided data on this outcome. No difference was found between CBT and CBT plus antipsychotics in participants with at least one adverse event at up to six months (RR 1.06, 95% CI 0.78 to 1.43, 1 RCT, n = 90). Results showed that fewer participants had at least one adverse event with CBT without antipsychotics at up to 12 months (RR 0.36, 95% CI 0.17 to 0.80, 1 RCT, n = 39, low‐certainty evidence). Test for subgroup differences confirmed a difference between time points (P = 0.01) (Analysis 3.36).

3.36. Analysis.

Comparison 3: CBT versus CBT plus antipsychotics, Outcome 36: Adverse events – participants with at least 1 adverse event (separated time points)

10.2 Participants with at least one serious adverse event

Three studies provided data on this outcome. There was no evidence of a difference between CBT and CBT plus antipsychotics in participants with serious adverse events at any of the analysed time points (up to 6 months: RR 0.96, 95% CI 0.37 to 2.50, 1 RCT, n = 90; up to 12 months: RR 0.71, 95% CI 0.32 to 1.57, I² = 0%, 2 RCTs, n = 100). No differences emerged between time points (test for subgroup differences P = 0.64) (Analysis 3.37).

3.37. Analysis.

Comparison 3: CBT versus CBT plus antipsychotics, Outcome 37: Adverse events – participants with serious adverse events (separated time points)

10.3 Deterioration of symptoms

Three studies provided data on this outcome. No difference was found between CBT and CBT plus antipsychotics in deterioration of symptoms at any of the analysed time points (up to 6 months: RR 1.07, 95% CI 0.37 to 3.10, I² = 0%, 3 RCTs, n = 180; up to 12 months: RR 1.05, 95% CI 0.11 to 9.74, I² = 0%, 2 RCTs, n = 90). No differences emerged between time points (test for subgroup differences P = 0.99) (Analysis 3.38).

3.38. Analysis.

Comparison 3: CBT versus CBT plus antipsychotics, Outcome 38: Adverse events – deterioration of symptoms (separated time points)

10.4 Lack of improvement

One study provided data on this outcome. No difference was found between CBT and CBT plus antipsychotics in lack of improvement up to six months (RR 0.96, 95% CI 0.33 to 2.74, 1 RCT, n = 90) (Analysis 3.39).

3.39. Analysis.

Comparison 3: CBT versus CBT plus antipsychotics, Outcome 39: Adverse events – lack of improvement (separated time points)

10.5 Mean endpoint or change score on adverse event scale

Three studies provided data on this outcome. There was no evidence of a difference between CBT and CBT plus antipsychotics in adverse events measured with ANNSERS scale at any of the analysed time points (up to 6 months: MD −3.68, 95% CI −9.66 to 2.31, I² = 73% (substantial heterogeneity), 2 RCTs, n = 52; up to 12 months: MD −3.47, 95% CI −12.06 to 5.11, I² = 82% (substantial heterogeneity), 2 RCTs, n = 49). No differences emerged between time points (test for subgroup differences P = 0.97) (Analysis 3.40).

3.40. Analysis.

Comparison 3: CBT versus CBT plus antipsychotics, Outcome 40: Adverse events – mean endpoint ANNSERS (high = poor)

Results of UKU scale (UKU Side Effect Rating Scale) did not show a difference between CBT and CBT plus antipsychotics at any of the analysed time points (up to 6 months: MD −1.86, 95% CI −6.37 to 2.66, I² = 80% (substantial heterogeneity), 3 RCTs, n = 122; up to 12 months: MD −3.47, 95% CI −12.06 to 5.11, I² = 82% (substantial heterogeneity), 2 RCTs, n = 49). No differences emerged between time points (test for subgroup differences P = 0.74) (Analysis 3.41).

3.41. Analysis.

Comparison 3: CBT versus CBT plus antipsychotics, Outcome 41: Adverse events – mean endpoint UKU (high = poor)

11. Mortality

No study reported data on this outcome.

12. Economic outcomes

No study reported data on this outcome.

Discussion

Summary of main results

Four studies were eligible for inclusion in the review. All four studies (300 participants) provided data for the meta‐analysis.

Based on low‐certainty evidence, CBT was associated with a reduction in PANSS total scores (Analysis 1.1) and negative symptoms (Analysis 1.4) at longer than 12 months when compared to no specific treatment. CBT was also associated with a lower duration of hospital stay (Analysis 1.9) and better functioning at up to 12 months (Analysis 1.11).

We found no differences between CBT and antipsychotics for any of the investigated outcomes, with the exception of adverse events measured with ANNSERS (Analysis 2.25).

Based on very low‐certainty evidence, CBT was less effective than CBT plus antipsychotics in reducing positive symptoms at up to 12 months (Analysis 3.7) and negative symptoms at up to 12 months (even if the CI includes the possibility of no difference) (Analysis 3.10). CBT was also associated with worse global state measured with CGI at up to six months (even if the CI includes the possibility of no difference) (Analysis 3.18), and a lower number of participants experiencing at least one adverse event at up to 12 months (Analysis 3.36).

Overall completeness and applicability of evidence

Despite the high number of systematic reviews and meta‐analyses on CBT for schizophrenia, we could not identify any review specifically focusing on CBT delivered without concomitant antipsychotic medication.

This review summarises the available evidence on the provision of CBT without antipsychotics to people with schizophrenia. We followed methodologically robust Cochrane standards in conducting the review, which provides an overall picture of the most relevant outcomes.

Caution is advised in the interpretation of results, due to the low number of included studies and the very low number of studies providing data for each outcome (one to three studies).

The low number of included studies identified reflects the challenge and controversy surrounding the idea of providing CBT without medication to people with schizophrenia.

Indeed, a considerable number of participants in the included studies received antipsychotic treatment against the randomisation plan (about 35% of those in the 'CBT without antipsychotic' study arm) (Table 4), reflecting the potential challenges of such an approach in real‐world clinical practice.

Moreover, it should be noted that all of the identified studies recruited participants from early intervention for psychosis services (Early Psychosis Prevention and Intervention Centre (EPPIC) in Australia and Early Intervention in Psychosis (EIP) service in the UK) or explicitly with first‐episode psychosis. Indeed, the mean age of the people included in this meta‐analysis was around 22 years. It is therefore questionable whether these results can be generalised to the general population of people with schizophrenia.

Additionally, study participants had on average a baseline severity of 80.63 on the PANSS, which corresponds to a moderately ill CGI score (Leucht 2005). Only in Francey 2020 – STAGES were participants markedly ill at baseline. This should also be taken into account when generalising the results.

As a methodological limitation, it should be noted that the aim of the included studies was to investigate the feasibility (Morrison 2018 – COMPARE; Morrison 2020 – MAPS), feasibility and effectiveness (Morrison 2014 – ACTION), or to determine non‐inferiority of CBT without antipsychotic medication to standard care (Francey 2020 – STAGES). Studies designed to investigate the efficacy of this approach would probably require a different sample size calculation.

Quality of the evidence

We assessed the certainty of the evidence as very low to low using the GRADE approach.

For all outcomes evaluated in comparison 1, CBT without antipsychotics compared to standard care without antipsychotics, we evaluated the certainty of the evidence as low due to inadequate information size and imprecise results.

For all outcomes evaluated in comparison 2, CBT without antipsychotics compared to standard care including antipsychotics, we evaluated the certainty of the evidence as very low due to high risk of bias, inadequate information size, and imprecise results. Only for the outcome 'participants with at least one adverse event' were results not imprecise, favouring CBT without antipsychotics, but the certainty of the evidence was still very low due to high risk of bias and the very low number of participants.

For all outcomes evaluated in comparison 3, CBT without antipsychotics compared to CBT in combination with antipsychotics, we evaluated the certainty of the evidence as very low because the majority of the studies contributing to this outcome were at high risk of bias; information size was inadequate; and results were imprecise. For the outcome 'participants with at least one adverse event', there was also substantial heterogeneity between results from the various studies providing data.

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 search is about a year and a half old. However, due to the very special topic investigated and thanks to communications with the authors of the included studies, we doubt that any new eligible studies would have been published in the meantime. On the other hand, it must be noted that the search was very broad, retrieving all the references where the term "cognit" was used, thereby minimising the risk of missing potentially relevant studies.

Secondly, we were unable to conduct any subgroup analyses to investigate potential moderators, since all analyses included a maximum of one to three studies. For the same reason, it was not possible to conduct sensitivity analyses exploring the influence of factors such as risk of bias.

Agreements and disagreements with other studies or reviews

We are not aware of previous systematic reviews that have attempted to synthesise the evidence on this specific topic.

Existing evidence is available about CBT when delivered in addition to standard care, meaning implicitly in combination with antipsychotics. These reviews have shown efficacy for multiple symptoms (Bighelli 2018; Jauhar 2014; Jones 2018a; Jones 2018b).

In randomised trials investigating CBT included in previous reviews, the dose or type of antipsychotic medication provided is rarely specified in detail, so that it is not possible to investigate its role with, for example, meta‐regression analyses.

An exploratory trial by Morrison and colleagues examined 20 participants with schizophrenia receiving cognitive therapy who had not been taking antipsychotic medication for at least six months (Morrison 2012b). Post‐treatment scores showed a reduction in symptoms measured by the PANSS and an improvement in functioning, both at post‐treatment and follow‐up. Only three participants had an increase in PANSS scores, but without reaching a clinically significant deterioration. Despite the limitations connected to the open and non‐randomised design, the authors concluded that cognitive therapy without antipsychotics is acceptable, and conducted a randomised trial to further test this hypothesis (Morrison 2014 – ACTION). Results of the current systematic review also did not find a difference between CBT without antipsychotics and control conditions in terms of number of participants with relapse or deterioration of symptoms. This approach might therefore be safely further investigated in future trials.

Authors' conclusions

Implications for practice.

Based on limited and very low‐certainty evidence, offering cognitive behavioural therapy (CBT) without antipsychotics was not associated with a worsening of symptoms or increased hospital admissions. We also found that the combination of CBT with antipsychotics might be beneficial for some outcomes compared to CBT alone.

An interpretation of the results for the practice is difficult because the results are to some degree conflicting: on the one hand, adding antipsychotics to CBT helps, as shown by results of comparison 3, where CBT was less efficacious than CBT plus antipsychotics. On the other hand, when compared directly in comparison 2, CBT and antipsychotics did not show a difference in efficacy.

Moreover, to date only two working groups have attempted to conduct clinical trials in which a group of participants with schizophrenia was randomised to not receive antipsychotic medication. These studies have investigated selected subpopulations of participants (people refusing to take antipsychotics, people with a first episode; see Overall completeness and applicability of evidence). On average, participants were found to be only moderately ill at baseline, thus the findings may not be generalisable to the more severely ill participants. For all of these reasons, the results should be interpreted cautiously and can hardly be generalised to the general population of people with schizophrenia. It must also be noted that a substantial proportion of participants in the included studies ultimately needed antipsychotics, and this might also be the case in clinical practice.

The information in this review can be used in the context of a shared decision‐making process to make informed decisions about the individual clinical plan.

Implications for research.

As highlighted above, one of the major limitations of the current evidence was the limited number of studies and participants.

A large, final, and high‐quality randomised trial, employing blind raters and adopting an intention‐to‐treat analysis approach, is warranted. This study should be appropriately funded and have resources in order to be adequately powered to detect feasibility and efficacy.

In particular, attempts could be made to recruit participants not only at the early stage of the disorder, but also in the later phases, and eventually also the more severely ill.

The current results show that such investigations might be safe (no worsening of symptoms or relapses detected); however, the high number of participants that received antipsychotic treatment during the trial against randomisation show that feasibility issues exist (Table 4).

It would also be useful if new randomised trials offering CBT in addition to standard care strictly monitor and provide detailed information about the type and dose of antipsychotics received by participants so that the role of the medication can be investigated in the analyses.

Finally, the analysis of data at a patient level with an individual patient data meta‐analysis approach would allow a detailed investigation of the individual characteristics of the participants and determine for which people CBT without antipsychotics may be feasible or efficacious, or both.

History

Protocol first published: Issue 7, 2022

Risk of bias

Risk of bias for analysis 1.1 Mental state – general: mean endpoint PANSS total (high = poor).

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 1.1.1 < 6 months

Morrison 2014 – ACTION

Low risk of bias

Participants were randomised. Randomisation was independent and concealed. No baseline difference to suggest a problem with the randomisation process.

Some concerns

Participants and therapists are not blind; there are deviations from intended interventions, but these are equally distributed in the two groups; ITT analysis was performed.

Low risk of bias

Data on this outcome are available for nearly all randomised participants.

Low risk of bias

Outcome was measured with a valid method by blind outcome assessors

Low risk of bias

The outcome is predefined in the trial registration of this study.

Some concerns

Calculated based on RoB2 algorithm.

Subgroup 1.1.2 < 12 months

Morrison 2014 – ACTION

Low risk of bias

Participants were randomised. Randomisation was independent and concealed. No baseline difference to suggest a problem with the randomisation process.

Some concerns

Participants and therapists are not blind; there are deviations from intended interventions, but these are equally distributed in the two groups; ITT analysis was performed.

Low risk of bias

Data on this outcome are available for nearly all randomised participants.

Low risk of bias

Outcome was measured with a valid method by blind outcome assessors

Low risk of bias

The outcome is predefined in the trial registration of this study.

Some concerns

Calculated based on RoB2 algorithm.

Subgroup 1.1.3 > 1 year

Morrison 2014 – ACTION

Low risk of bias

Participants were randomised. Randomisation was independent and concealed. No baseline difference to suggest a problem with the randomisation process.

Some concerns

Participants and therapists are not blind; there are deviations from intended interventions, but these are equally distributed in the two groups; ITT analysis was performed.

Low risk of bias

Data on this outcome are available for nearly all randomised participants.

Low risk of bias

Outcome was measured with a valid method by blind outcome assessors

Low risk of bias

The outcome is predefined in the trial registration of this study.

Some concerns

Calculated based on RoB2 algorithm.

Risk of bias for analysis 1.7 Global state – number of participants with relapse/exacerbations of psychosis (separated 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

Subgroup 1.7.1 < 12 months

Morrison 2014 – ACTION

Low risk of bias

Participants were randomised. Randomisation was independent and concealed. No baseline difference to suggest a problem with the randomisation process.

Some concerns

Participants and therapists are not blind; there are deviations from intended interventions, but these are equally distributed in the two groups; ITT analysis was performed.

Some concerns

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

Low risk of bias

Outcome was measured with a valid method by blind outcome assessors

Low risk of bias

The outcome is predefined in the trial registration of this study.

Some concerns

Calculated based on RoB2 algorithm.

Subgroup 1.7.2 > 1 year

Morrison 2014 – ACTION

Low risk of bias

Participants were randomised. Randomisation was independent and concealed. No baseline difference to suggest a problem with the randomisation process.

Some concerns

Participants and therapists are not blind; there are deviations from intended interventions, but these are equally distributed in the two groups; ITT analysis was performed.

Some concerns

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

Low risk of bias

Outcome was measured with a valid method by blind outcome assessors

Low risk of bias

The outcome is predefined in the trial registration of this study.

Some concerns

Calculated based on RoB2 algorithm.

Risk of bias for analysis 1.10 Leaving the study early – for any reason – overall acceptability (separated 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

Subgroup 1.10.1 < 6 months

Morrison 2014 – ACTION

Low risk of bias

Participants were randomised. Randomisation was independent and concealed. No baseline difference to suggest a problem with the randomisation process.

Some concerns

Participants and therapists are not blind; there are deviations from intended interventions, but these are equally distributed in the two groups; ITT analysis was performed.

Low risk of bias

Data on this outcome are available for nearly all randomised participants.

Low risk of bias

Outcome was measured with a valid method by blind outcome assessors

Low risk of bias

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

Some concerns

Calculated based on RoB2 algorithm.

Subgroup 1.10.2 < 12 months

Morrison 2014 – ACTION

Low risk of bias

Participants were randomised. Randomisation was independent and concealed. No baseline difference to suggest a problem with the randomisation process.

Some concerns

Participants and therapists are not blind; there are deviations from intended interventions, but these are equally distributed in the two groups; ITT analysis was performed.

Low risk of bias

Data on this outcome are available for nearly all randomised participants.

Low risk of bias

Outcome was measured with a valid method by blind outcome assessors

Low risk of bias

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

Some concerns

Calculated based on RoB2 algorithm.

Subgroup 1.10.3 > 1 year

Morrison 2014 – ACTION

Low risk of bias

Participants were randomised. Randomisation was independent and concealed. No baseline difference to suggest a problem with the randomisation process.

Some concerns

Participants and therapists are not blind; there are deviations from intended interventions, but these are equally distributed in the two groups; ITT analysis was performed.

Low risk of bias

Data on this outcome are available for nearly all randomised participants.

Low risk of bias

Outcome was measured with a valid method by blind outcome assessors

Low risk of bias

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

Some concerns

Calculated based on RoB2 algorithm.

Risk of bias for analysis 1.11 Functioning – mean endpoint PSP (high = good).

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 1.11.1 < 6 months

Morrison 2014 – ACTION

Low risk of bias

Participants were randomised. Randomisation was independent and concealed. No baseline difference to suggest a problem with the randomisation process.

Some concerns

Participants and therapists are not blind; there are deviations from intended interventions, but these are equally distributed in the two groups; ITT analysis was performed.

Low risk of bias

Data on this outcome are available for nearly all randomised participants.

Low risk of bias

Outcome was measured with a valid method by blind outcome assessors

Some concerns

The outcome is not mentioned in the trial registration.

Some concerns

Calculated based on RoB2 algorithm.

Subgroup 1.11.2 < 12 months

Morrison 2014 – ACTION

Low risk of bias

Participants were randomised. Randomisation was independent and concealed. No baseline difference to suggest a problem with the randomisation process.

Some concerns

Participants and therapists are not blind; there are deviations from intended interventions, but these are equally distributed in the two groups; ITT analysis was performed.

Low risk of bias

Data on this outcome are available for nearly all randomised participants.

Low risk of bias

Outcome was measured with a valid method by blind outcome assessors

Some concerns

The outcome is not mentioned in the trial registration.

Some concerns

Calculated based on RoB2 algorithm.

Subgroup 1.11.3 > 1 year

Morrison 2014 – ACTION

Low risk of bias

Participants were randomised. Randomisation was independent and concealed. No baseline difference to suggest a problem with the randomisation process.

Some concerns

Participants and therapists are not blind; there are deviations from intended interventions, but these are equally distributed in the two groups; ITT analysis was performed.

Low risk of bias

Data on this outcome are available for nearly all randomised participants.

Low risk of bias

Outcome was measured with a valid method by blind outcome assessors

Some concerns

The outcome is not mentioned in the trial registration.

Some concerns

Calculated based on RoB2 algorithm.

Risk of bias for analysis 2.1 Mental state – general: mean endpoint PANSS total (high = poor).

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 2.1.1 < 6 months

Morrison 2018 – COMPARE

Low risk of bias

Participants were randomised. Randomisation was independent and concealed. No baseline difference to suggest a problem with the randomisation process.

High risk of bias

Participants and therapists are not blind; there are deviations from intended interventions not balanced between the groups; ITT analysis was performed.

Low risk of bias

Data on this outcome are available for nearly all randomised participants.

Low risk of bias

Outcome was measured with a valid method by blind outcome assessors

Low risk of bias

The outcome is predefined in the trial registration of this study.

High risk of bias

Calculated based on RoB2 algorithm.

Morrison 2020 – MAPS

Low risk of bias

Participants were randomised. Randomisation was independent and concealed. No baseline difference to suggest a problem with the randomisation process

High risk of bias

Participants and therapists are not blind; there are deviations from intended interventions not balanced between the groups; ITT analysis was performed.

Some concerns

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

Low risk of bias

Outcome was measured with a valid method by blind outcome assessors

Low risk of bias

The outcome is predefined in the trial registration of this study.

High risk of bias

Calculated based on RoB2 algorithm.

Subgroup 2.1.2 < 12 months

Morrison 2018 – COMPARE

Low risk of bias

Participants were randomised. Randomisation was independent and concealed. No baseline difference to suggest a problem with the randomisation process.

High risk of bias

Participants and therapists are not blind; there are deviations from intended interventions not balanced between the groups; ITT analysis was performed.

Low risk of bias

Data on this outcome are available for nearly all randomised participants.

Low risk of bias

Outcome was measured with a valid method by blind outcome assessors

Low risk of bias

The outcome is predefined in the trial registration of this study.

High risk of bias

Calculated based on RoB2 algorithm.

Morrison 2020 – MAPS

Low risk of bias

Participants were randomised. Randomisation was independent and concealed. No baseline difference to suggest a problem with the randomisation process

High risk of bias

Participants and therapists are not blind; there are deviations from intended interventions not balanced between the groups; ITT analysis was performed.

Some concerns

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

Low risk of bias

Outcome was measured with a valid method by blind outcome assessors

Low risk of bias

The outcome is predefined in the trial registration of this study.

High risk of bias

Calculated based on RoB2 algorithm.

Risk of bias for analysis 2.7 Global state – number of participants with relapse/exacerbations of psychosis (separated 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

Subgroup 2.7.1 < 6 months

Morrison 2018 – COMPARE

Low risk of bias

Participants were randomised. Randomisation was independent and concealed. No baseline difference to suggest a problem with the randomisation process.

High risk of bias

Participants and therapists are not blind; there are deviations from intended interventions not balanced between the groups; ITT analysis was performed.

Low risk of bias

Data on this outcome are available for nearly all randomised participants.

Low risk of bias

Outcome was measured with a valid method by blind outcome assessors

Low risk of bias

The outcome is predefined in the trial registration of this study.

High risk of bias

Calculated based on RoB2 algorithm.

Morrison 2020 – MAPS

Low risk of bias

Participants were randomised. Randomisation was independent and concealed. No baseline difference to suggest a problem with the randomisation process

High risk of bias

Participants and therapists are not blind; there are deviations from intended interventions not balanced between the groups; ITT analysis was performed.

Some concerns

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

Low risk of bias

Outcome was measured with a valid method by blind outcome assessors

Low risk of bias

The outcome is predefined in the trial registration of this study.

High risk of bias

Calculated based on RoB2 algorithm.

Subgroup 2.7.2 < 12 months

Morrison 2018 – COMPARE

Low risk of bias

Participants were randomised. Randomisation was independent and concealed. No baseline difference to suggest a problem with the randomisation process.

High risk of bias

Participants and therapists are not blind; there are deviations from intended interventions not balanced between the groups; ITT analysis was performed.

Low risk of bias

Data on this outcome are available for nearly all randomised participants.

Low risk of bias

Outcome was measured with a valid method by blind outcome assessors

Low risk of bias

The outcome is predefined in the trial registration of this study.

High risk of bias

Calculated based on RoB2 algorithm.

Morrison 2020 – MAPS

Low risk of bias

Participants were randomised. Randomisation was independent and concealed. No baseline difference to suggest a problem with the randomisation process

High risk of bias

Participants and therapists are not blind; there are deviations from intended interventions not balanced between the groups; ITT analysis was performed.

Some concerns

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

Low risk of bias

Outcome was measured with a valid method by blind outcome assessors

Low risk of bias

The outcome is predefined in the trial registration of this study.

High risk of bias

Calculated based on RoB2 algorithm.

Risk of bias for analysis 2.12 Leaving the study early – for any reason – overall acceptability (separated 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

Subgroup 2.12.1 < 6 months

Morrison 2018 – COMPARE

Low risk of bias

Participants were randomised. Randomisation was independent and concealed. No baseline difference to suggest a problem with the randomisation process.

High risk of bias

Participants and therapists are not blind; there are deviations from intended interventions not balanced between the groups; ITT analysis was performed.

Low risk of bias

Data on this outcome are available for nearly all randomised participants.

Low risk of bias

Outcome was measured with a valid method by blind outcome assessors

Low risk of bias

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

High risk of bias

Calculated based on RoB2 algorithm.

Morrison 2020 – MAPS

Low risk of bias

Participants were randomised. Randomisation was independent and concealed. No baseline difference to suggest a problem with the randomisation process

High risk of bias

Participants and therapists are not blind; there are deviations from intended interventions not balanced between the groups; ITT analysis was performed.

Low risk of bias

Data on this outcome are available for nearly all randomised participants.

Low risk of bias

Outcome was measured with a valid method by blind outcome assessors

Low risk of bias

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

High risk of bias

Calculated based on RoB2 algorithm.

Subgroup 2.12.2 < 12 months

Morrison 2018 – COMPARE

Low risk of bias

Participants were randomised. Randomisation was independent and concealed. No baseline difference to suggest a problem with the randomisation process.

High risk of bias

Participants and therapists are not blind; there are deviations from intended interventions not balanced between the groups; ITT analysis was performed.

Low risk of bias

Data on this outcome are available for nearly all randomised participants.

Low risk of bias

Outcome was measured with a valid method by blind outcome assessors

Low risk of bias

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

High risk of bias

Calculated based on RoB2 algorithm.

Morrison 2020 – MAPS

Low risk of bias

Participants were randomised. Randomisation was independent and concealed. No baseline difference to suggest a problem with the randomisation process

High risk of bias

Participants and therapists are not blind; there are deviations from intended interventions not balanced between the groups; ITT analysis was performed.

Low risk of bias

Data on this outcome are available for nearly all randomised participants.

Low risk of bias

Outcome was measured with a valid method by blind outcome assessors

Low risk of bias

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

High risk of bias

Calculated based on RoB2 algorithm.

Risk of bias for analysis 2.15 Functioning – mean endpoint functioning scales (high = good).

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 2.15.1 < 6 months

Morrison 2018 – COMPARE

Low risk of bias

Participants were randomised. Randomisation was independent and concealed. No baseline difference to suggest a problem with the randomisation process.

High risk of bias

Participants and therapists are not blind; there are deviations from intended interventions not balanced between the groups; ITT analysis was performed.

Low risk of bias

Data on this outcome are available for nearly all randomised participants.

Low risk of bias

Outcome was measured with a valid method by blind outcome assessors

Low risk of bias

The outcome is predefined in the trial registration of this study.

High risk of bias

Calculated based on RoB2 algorithm.

Morrison 2020 – MAPS

Low risk of bias

Participants were randomised. Randomisation was independent and concealed. No baseline difference to suggest a problem with the randomisation process

High risk of bias

Participants and therapists are not blind; there are deviations from intended interventions not balanced between the groups; ITT analysis was performed.

Some concerns

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

Low risk of bias

Outcome was measured with a valid method by blind outcome assessors

Low risk of bias

The outcome is predefined in the trial registration of this study.

High risk of bias

Calculated based on RoB2 algorithm.

Subgroup 2.15.2 < 12 months

Morrison 2018 – COMPARE

Low risk of bias

Participants were randomised. Randomisation was independent and concealed. No baseline difference to suggest a problem with the randomisation process.

High risk of bias

Participants and therapists are not blind; there are deviations from intended interventions not balanced between the groups; ITT analysis was performed.

Low risk of bias

Data on this outcome are available for nearly all randomised participants.

Low risk of bias

Outcome was measured with a valid method by blind outcome assessors

Low risk of bias

The outcome is predefined in the trial registration of this study.

High risk of bias

Calculated based on RoB2 algorithm.

Morrison 2020 – MAPS

Low risk of bias

Participants were randomised. Randomisation was independent and concealed. No baseline difference to suggest a problem with the randomisation process

High risk of bias

Participants and therapists are not blind; there are deviations from intended interventions not balanced between the groups; ITT analysis was performed.

Some concerns

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

Low risk of bias

Outcome was measured with a valid method by blind outcome assessors

Low risk of bias

The outcome is predefined in the trial registration of this study.

High risk of bias

Calculated based on RoB2 algorithm.

Risk of bias for analysis 2.22 Adverse events – participants with at least 1 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

Subgroup 2.22.1 < 12 months

Morrison 2020 – MAPS

Low risk of bias

Participants were randomised. Randomisation was independent and concealed. No baseline difference to suggest a problem with the randomisation process

High risk of bias

Participants and therapists are not blind; there are deviations from intended interventions not balanced between the groups; ITT analysis was performed.

Low risk of bias

Data on this outcome are available for nearly all randomised participants.

Low risk of bias

Outcome was measured with a valid method by blind outcome assessors

Low risk of bias

The outcome is predefined in the trial registration of this study.

High risk of bias

Calculated based on RoB2 algorithm.

Risk of bias for analysis 3.3 Mental state – general: mean endpoint overall symptoms scales (PANSS/BPRS) (high = poor) (separated 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

Subgroup 3.3.1 < 6 months

Francey 2020 – STAGES

Low risk of bias

Participants were randomised. Randomisation was independent and concealed. No baseline difference to suggest a problem with the randomisation process.

Low risk of bias

Clinicians, research staff and participants are blind; intention to treat analysis was performed.

Low risk of bias

Data on this outcome are available for nearly all randomised participants.

Low risk of bias

Outcome was measured with a valid method by blind outcome assessors

Low risk of bias

The outcome is mentioned in the trial registration of this study.

Low risk of bias

Calculated based on RoB2 algorithm.

Morrison 2018 – COMPARE

Low risk of bias

Participants were randomised. Randomisation was independent and concealed. No baseline difference to suggest a problem with the randomisation process.

High risk of bias

Participants and therapists are not blind; there are deviations from intended interventions not balanced between the groups; ITT analysis was performed.

Low risk of bias

Data on this outcome are available for nearly all randomised participants.

Low risk of bias

Outcome was measured with a valid method by blind outcome assessors

Low risk of bias

The outcome is predefined in the trial registration of this study.

High risk of bias

Calculated based on RoB2 algorithm.

Morrison 2020 – MAPS

Low risk of bias

Participants were randomised. Randomisation was independent and concealed. No baseline difference to suggest a problem with the randomisation process

High risk of bias

Participants and therapists are not blind; there are deviations from intended interventions not balanced between the groups; ITT analysis was performed.

Some concerns

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

Low risk of bias

Outcome was measured with a valid method by blind outcome assessors

Low risk of bias

The outcome is predefined in the trial registration of this study.

High risk of bias

Calculated based on RoB2 algorithm.

Subgroup 3.3.2 < 12 months

Francey 2020 – STAGES

Low risk of bias

Participants were randomised. Randomisation was independent and concealed. No baseline difference to suggest a problem with the randomisation process.

Low risk of bias

Clinicians, research staff and participants are blind; intention to treat analysis was performed.

Low risk of bias

Data on this outcome are available for nearly all randomised participants.

Low risk of bias

Outcome was measured with a valid method by blind outcome assessors

Low risk of bias

The outcome is mentioned in the trial registration of this study.

Low risk of bias

Calculated based on RoB2 algorithm.

Morrison 2018 – COMPARE

Low risk of bias

Participants were randomised. Randomisation was independent and concealed. No baseline difference to suggest a problem with the randomisation process.

High risk of bias

Participants and therapists are not blind; there are deviations from intended interventions not balanced between the groups; ITT analysis was performed.

Low risk of bias

Data on this outcome are available for nearly all randomised participants.

Low risk of bias

Outcome was measured with a valid method by blind outcome assessors

Low risk of bias

The outcome is predefined in the trial registration of this study.

High risk of bias

Calculated based on RoB2 algorithm.

Morrison 2020 – MAPS

Low risk of bias

Participants were randomised. Randomisation was independent and concealed. No baseline difference to suggest a problem with the randomisation process

High risk of bias

Participants and therapists are not blind; there are deviations from intended interventions not balanced between the groups; ITT analysis was performed.

Some concerns

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

Low risk of bias

Outcome was measured with a valid method by blind outcome assessors

Low risk of bias

The outcome is predefined in the trial registration of this study.

High risk of bias

Calculated based on RoB2 algorithm.

Subgroup 3.3.3 > 1 year

Francey 2020 – STAGES

Low risk of bias

Participants were randomised. Randomisation was independent and concealed. No baseline difference to suggest a problem with the randomisation process.

Low risk of bias

Clinicians, research staff and participants are blind; intention to treat analysis was performed.

Low risk of bias

Data on this outcome are available for nearly all randomised participants.

Low risk of bias

Outcome was measured with a valid method by blind outcome assessors

Low risk of bias

The outcome is mentioned in the trial registration of this study.

Low risk of bias

Calculated based on RoB2 algorithm.

Risk of bias for analysis 3.17 Global state – number of participants with relapse/exacerbations of psychosis (separated 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

Subgroup 3.17.1 < 6 months

Morrison 2018 – COMPARE

Low risk of bias

Participants were randomised. Randomisation was independent and concealed. No baseline difference to suggest a problem with the randomisation process.

High risk of bias

Participants and therapists are not blind; there are deviations from intended interventions not balanced between the groups; ITT analysis was performed.

Low risk of bias

Data on this outcome are available for nearly all randomised participants.

Low risk of bias

Outcome was measured with a valid method by blind outcome assessors

Low risk of bias

The outcome is predefined in the trial registration of this study.

High risk of bias

Calculated based on RoB2 algorithm.

Morrison 2020 – MAPS

Low risk of bias

Participants were randomised. Randomisation was independent and concealed. No baseline difference to suggest a problem with the randomisation process

High risk of bias

Participants and therapists are not blind; there are deviations from intended interventions not balanced between the groups; ITT analysis was performed.

Some concerns

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

Low risk of bias

Outcome was measured with a valid method by blind outcome assessors

Low risk of bias

The outcome is predefined in the trial registration of this study.

High risk of bias

Calculated based on RoB2 algorithm.

Subgroup 3.17.2 < 12 months

Morrison 2018 – COMPARE

Low risk of bias

Participants were randomised. Randomisation was independent and concealed. No baseline difference to suggest a problem with the randomisation process.

High risk of bias

Participants and therapists are not blind; there are deviations from intended interventions not balanced between the groups; ITT analysis was performed.

Low risk of bias

Data on this outcome are available for nearly all randomised participants.

Low risk of bias

Outcome was measured with a valid method by blind outcome assessors

Low risk of bias

The outcome is predefined in the trial registration of this study.

High risk of bias

Calculated based on RoB2 algorithm.

Morrison 2020 – MAPS

Low risk of bias

Participants were randomised. Randomisation was independent and concealed. No baseline difference to suggest a problem with the randomisation process

High risk of bias

Participants and therapists are not blind; there are deviations from intended interventions not balanced between the groups; ITT analysis was performed.

Some concerns

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

Low risk of bias

Outcome was measured with a valid method by blind outcome assessors

Low risk of bias

The outcome is predefined in the trial registration of this study.

High risk of bias

Calculated based on RoB2 algorithm.

Risk of bias for analysis 3.22 Leaving the study early – for any reason – overall acceptability (separated 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

Subgroup 3.22.1 < 6 months

Francey 2020 – STAGES

Low risk of bias

Participants were randomised. Randomisation was independent and concealed. No baseline difference to suggest a problem with the randomisation process.

Low risk of bias

Clinicians, research staff and participants are blind; intention to treat analysis was performed.

Low risk of bias

Data on this outcome are available for nearly all randomised participants.

Low risk of bias

Outcome was measured with a valid method by blind outcome assessors

Low risk of bias

The outcome is not mentioned in the trial registration, 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. Randomisation was independent and concealed. No baseline difference to suggest a problem with the randomisation process.

High risk of bias

Participants and therapists are not blind; there are deviations from intended interventions not balanced between the groups; ITT analysis was performed.

Low risk of bias

Data on this outcome are available for nearly all randomised participants.

Low risk of bias

Outcome was measured with a valid method by blind outcome assessors

Low risk of bias

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

High risk of bias

Calculated based on RoB2 algorithm.

Morrison 2020 – MAPS

Low risk of bias

Participants were randomised. Randomisation was independent and concealed. No baseline difference to suggest a problem with the randomisation process

High risk of bias

Participants and therapists are not blind; there are deviations from intended interventions not balanced between the groups; ITT analysis was performed.

Low risk of bias

Data on this outcome are available for nearly all randomised participants.

Low risk of bias

Outcome was measured with a valid method by blind outcome assessors

Low risk of bias

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

High risk of bias

Calculated based on RoB2 algorithm.

Subgroup 3.22.2 < 12 months

Morrison 2018 – COMPARE

Low risk of bias

Participants were randomised. Randomisation was independent and concealed. No baseline difference to suggest a problem with the randomisation process.

High risk of bias

Participants and therapists are not blind; there are deviations from intended interventions not balanced between the groups; ITT analysis was performed.

Low risk of bias

Data on this outcome are available for nearly all randomised participants.

Low risk of bias

Outcome was measured with a valid method by blind outcome assessors

Low risk of bias

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

High risk of bias

Calculated based on RoB2 algorithm.

Morrison 2020 – MAPS

Low risk of bias

Participants were randomised. Randomisation was independent and concealed. No baseline difference to suggest a problem with the randomisation process

High risk of bias

Participants and therapists are not blind; there are deviations from intended interventions not balanced between the groups; ITT analysis was performed.

Low risk of bias

Data on this outcome are available for nearly all randomised participants.

Low risk of bias

Outcome was measured with a valid method by blind outcome assessors

Low risk of bias

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

High risk of bias

Calculated based on RoB2 algorithm.

Risk of bias for analysis 3.27 Functioning – mean endpoint functioning scales (high = good).

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 3.27.1 < 6 months

Francey 2020 – STAGES

Low risk of bias

Participants were randomised. Randomisation was independent and concealed. No baseline difference to suggest a problem with the randomisation process.

Low risk of bias

Clinicians, research staff and participants are blind; intention to treat analysis was performed.

Low risk of bias

Data on this outcome are available for nearly all randomised participants.

Low risk of bias

Outcome was measured with a valid method by blind outcome assessors

Low risk of bias

The outcome is predefined in the trial registration of this study.

Low risk of bias

Calculated based on RoB2 algorithm.

Morrison 2018 – COMPARE

Low risk of bias

Participants were randomised. Randomisation was independent and concealed. No baseline difference to suggest a problem with the randomisation process.

High risk of bias

Participants and therapists are not blind; there are deviations from intended interventions not balanced between the groups; ITT analysis was performed.

Low risk of bias

Data on this outcome are available for nearly all randomised participants.

Low risk of bias

Outcome was measured with a valid method by blind outcome assessors

Low risk of bias

The outcome is predefined in the trial registration of this study.

High risk of bias

Calculated based on RoB2 algorithm.

Morrison 2020 – MAPS

Low risk of bias

Participants were randomised. Randomisation was independent and concealed. No baseline difference to suggest a problem with the randomisation process

High risk of bias

Participants and therapists are not blind; there are deviations from intended interventions not balanced between the groups; ITT analysis was performed.

Some concerns

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

Low risk of bias

Outcome was measured with a valid method by blind outcome assessors

Low risk of bias

The outcome is predefined in the trial registration of this study.

High risk of bias

Calculated based on RoB2 algorithm.

Subgroup 3.27.2 < 12 months

Francey 2020 – STAGES

Low risk of bias

Participants were randomised. Randomisation was independent and concealed. No baseline difference to suggest a problem with the randomisation process.

Low risk of bias

Clinicians, research staff and participants are blind; intention to treat analysis was performed.

Low risk of bias

Data on this outcome are available for nearly all randomised participants.

Low risk of bias

Outcome was measured with a valid method by blind outcome assessors

Low risk of bias

The outcome is predefined in the trial registration of this study.

Low risk of bias

Calculated based on RoB2 algorithm.

Morrison 2018 – COMPARE

Low risk of bias

Participants were randomised. Randomisation was independent and concealed. No baseline difference to suggest a problem with the randomisation process.

High risk of bias

Participants and therapists are not blind; there are deviations from intended interventions not balanced between the groups; ITT analysis was performed.

Low risk of bias

Data on this outcome are available for nearly all randomised participants.

Low risk of bias

Outcome was measured with a valid method by blind outcome assessors

Low risk of bias

The outcome is predefined in the trial registration of this study.

High risk of bias

Calculated based on RoB2 algorithm.

Morrison 2020 – MAPS

Low risk of bias

Participants were randomised. Randomisation was independent and concealed. No baseline difference to suggest a problem with the randomisation process

High risk of bias

Participants and therapists are not blind; there are deviations from intended interventions not balanced between the groups; ITT analysis was performed.

Some concerns

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

Low risk of bias

Outcome was measured with a valid method by blind outcome assessors

Low risk of bias

The outcome is predefined in the trial registration of this study.

High risk of bias

Calculated based on RoB2 algorithm.

Subgroup 3.27.3 > 1 year

Francey 2020 – STAGES

Low risk of bias

Participants were randomised. Randomisation was independent and concealed. No baseline difference to suggest a problem with the randomisation process.

Low risk of bias

Clinicians, research staff and participants are blind; intention to treat analysis was performed.

Low risk of bias

Data on this outcome are available for nearly all randomised participants.

Low risk of bias

Outcome was measured with a valid method by blind outcome assessors

Low risk of bias

The outcome is predefined in the trial registration of this study.

Low risk of bias

Calculated based on RoB2 algorithm.

Risk of bias for analysis 3.35 Adverse events – participants with at least 1 adverse event (pooled 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

Francey 2020 – STAGES

Low risk of bias

Participants were randomised. Randomisation was independent and concealed. No baseline difference to suggest a problem with the randomisation process.

Low risk of bias

Clinicians, research staff and participants are blind; intention to treat analysis was performed.

Low risk of bias

Data on this outcome are available for nearly all randomised participants.

Low risk of bias

Outcome was measured with a valid method by blind outcome assessors

Low risk of bias

The outcome is predefined in the protocol of this study.

Low risk of bias

Calculated based on RoB2 algorithm.

Morrison 2020 – MAPS

Low risk of bias

Participants were randomised. Randomisation was independent and concealed. No baseline difference to suggest a problem with the randomisation process

High risk of bias

Participants and therapists are not blind; there are deviations from intended interventions not balanced between the groups; ITT analysis was performed.

Low risk of bias

Data on this outcome are available for nearly all randomised participants.

Low risk of bias

Outcome was measured with a valid method by blind outcome assessors

Low risk of bias

The outcome is predefined in the trial registration of this study.

High risk of bias

Calculated based on RoB2 algorithm.

Data and analyses

Comparison 1. CBT versus no specific treatment.

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)	1		Mean Difference (IV, Random, 95% CI)	Subtotals only
1.1.1 < 6 months	1	42	Mean Difference (IV, Random, 95% CI)	‐6.99 [‐14.91, 0.93]
1.1.2 < 12 months	1	39	Mean Difference (IV, Random, 95% CI)	‐9.77 [‐20.59, 1.05]
1.1.3 > 1 year	1	34	Mean Difference (IV, Random, 95% CI)	‐14.77 [‐27.75, ‐1.79]
1.2 Number of participants with clinically important change in general mental state (separated time points)	1		Risk Ratio (IV, Random, 95% CI)	Subtotals only
1.2.1 < 12 months	1	74	Risk Ratio (IV, Random, 95% CI)	2.33 [0.65, 8.34]
1.2.2 > 1 year	1	74	Risk Ratio (IV, Random, 95% CI)	2.33 [0.65, 8.34]
1.3 Mental state – specific: mean endpoint PANSS positive (high = poor)	1		Mean Difference (IV, Random, 95% CI)	Subtotals only
1.3.1 < 6 m	1	42	Mean Difference (IV, Random, 95% CI)	‐1.28 [‐4.23, 1.67]
1.3.2 < 12 m	1	40	Mean Difference (IV, Random, 95% CI)	‐2.30 [‐6.52, 1.92]
1.3.3 > 1 year	1	37	Mean Difference (IV, Random, 95% CI)	‐4.20 [‐8.56, 0.16]
1.4 Mental state – specific: mean endpoint PANSS negative (high = poor)	1		Mean Difference (IV, Random, 95% CI)	Subtotals only
1.4.1 < 6 m	1	42	Mean Difference (IV, Random, 95% CI)	‐1.47 [‐3.72, 0.78]
1.4.2 < 12 m	1	39	Mean Difference (IV, Random, 95% CI)	‐3.34 [‐6.53, ‐0.15]
1.4.3 > 1 year	1	34	Mean Difference (IV, Random, 95% CI)	‐4.06 [‐7.50, ‐0.62]
1.5 Mental state – specific: mean endpoint BDI (high = poor)	1		Mean Difference (IV, Random, 95% CI)	Subtotals only
1.5.1 < 6 m	1	40	Mean Difference (IV, Random, 95% CI)	0.20 [‐2.80, 3.20]
1.5.2 < 12 m	1	35	Mean Difference (IV, Random, 95% CI)	0.44 [‐2.94, 3.82]
1.5.3 > 1 year	1	32	Mean Difference (IV, Random, 95% CI)	‐1.88 [‐5.62, 1.86]
1.6 Mental state – specific: mean endpoint BAI (high = poor)	1		Mean Difference (IV, Random, 95% CI)	Subtotals only
1.6.1 < 6 m	1	37	Mean Difference (IV, Random, 95% CI)	‐3.15 [‐13.34, 7.04]
1.6.2 < 12 m	1	29	Mean Difference (IV, Random, 95% CI)	‐11.86 [‐25.61, 1.89]
1.6.3 > 1 year	1	32	Mean Difference (IV, Random, 95% CI)	‐12.75 [‐26.32, 0.82]
1.7 Global state – number of participants with relapse/exacerbations of psychosis (separated time points)	1		Risk Ratio (IV, Random, 95% CI)	Subtotals only
1.7.1 < 12 months	1	74	Risk Ratio (IV, Random, 95% CI)	3.00 [0.13, 71.34]
1.7.2 > 1 year	1	74	Risk Ratio (IV, Random, 95% CI)	0.50 [0.05, 5.28]
1.8 Service use – readmission to hospital	1		Risk Ratio (IV, Random, 95% CI)	Subtotals only
1.8.1 < 12 months	1	74	Risk Ratio (IV, Random, 95% CI)	1.00 [0.27, 3.70]
1.9 Service use – mean number of days in hospital	1		Mean Difference (IV, Random, 95% CI)	Subtotals only
1.9.1 < 12 months	1	74	Mean Difference (IV, Random, 95% CI)	‐22.45 [‐28.82, ‐16.08]
1.10 Leaving the study early – for any reason – overall acceptability (separated time points)	1		Risk Ratio (IV, Random, 95% CI)	Subtotals only
1.10.1 < 6 months	1	74	Risk Ratio (IV, Random, 95% CI)	1.33 [0.32, 5.55]
1.10.2 < 12 months	1	74	Risk Ratio (IV, Random, 95% CI)	0.80 [0.23, 2.75]
1.10.3 > 1 year	1	74	Risk Ratio (IV, Random, 95% CI)	2.50 [0.52, 12.08]
1.11 Functioning – mean endpoint PSP (high = good)	1		Mean Difference (IV, Random, 95% CI)	Subtotals only
1.11.1 < 6 months	1	42	Mean Difference (IV, Random, 95% CI)	‐7.85 [‐18.13, 2.43]
1.11.2 < 12 months	1	40	Mean Difference (IV, Random, 95% CI)	‐12.42 [‐22.75, ‐2.09]
1.11.3 > 1 year	1	37	Mean Difference (IV, Random, 95% CI)	‐8.80 [‐21.86, 4.26]
1.12 Behaviour – participants with suicide attempts	1		Risk Ratio (IV, Random, 95% CI)	Subtotals only
1.12.1 < 12 months	1	74	Risk Ratio (IV, Random, 95% CI)	1.00 [0.06, 15.40]
1.13 Adverse events – deterioration of symptoms	1		Risk Ratio (IV, Random, 95% CI)	Subtotals only
1.13.1 < 12 months	1	74	Risk Ratio (IV, Random, 95% CI)	1.00 [0.15, 6.73]
1.13.2 > 1 year	1	74	Risk Ratio (IV, Random, 95% CI)	0.50 [0.10, 2.56]
1.14 Adverse events – participants with serious adverse events	1		Risk Ratio (IV, Random, 95% CI)	Subtotals only
1.14.1 < 12 months	1	74	Risk Ratio (IV, Random, 95% CI)	0.33 [0.07, 1.55]
1.15 Mortality – overall mortality	1		Risk Ratio (IV, Random, 95% CI)	Subtotals only
1.15.1 < 6 months	1	74	Risk Ratio (IV, Random, 95% CI)	Not estimable
1.15.2 < 12 months	1	74	Risk Ratio (IV, Random, 95% CI)	0.33 [0.01, 7.93]
1.15.3 > 1 year	1	74	Risk Ratio (IV, Random, 95% CI)	0.20 [0.01, 4.03]
1.16 Mortality – mortality due to natural causes	1		Risk Ratio (IV, Random, 95% CI)	Subtotals only
1.16.1 < 6 months	1	74	Risk Ratio (IV, Random, 95% CI)	Not estimable
1.16.2 < 12 months	1	74	Risk Ratio (IV, Random, 95% CI)	0.20 [0.01, 4.03]
1.16.3 > 1 year	1	74	Risk Ratio (IV, Random, 95% CI)	0.11 [0.01, 1.99]

Comparison 2. CBT versus antipsychotics.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
2.1 Mental state – general: mean endpoint PANSS total (high = poor)	2		Mean Difference (IV, Random, 95% CI)	Subtotals only
2.1.1 < 6 months	2	78	Mean Difference (IV, Random, 95% CI)	‐3.75 [‐11.61, 4.12]
2.1.2 < 12 months	2	63	Mean Difference (IV, Random, 95% CI)	3.38 [‐2.38, 9.14]
2.2 Number of participants with clinically important change in general mental state (separated time points)	2		Risk Ratio (IV, Random, 95% CI)	Subtotals only
2.2.1 < 6 months	2	90	Risk Ratio (IV, Random, 95% CI)	1.15 [0.44, 2.98]
2.2.2 < 12 months	2	90	Risk Ratio (IV, Random, 95% CI)	0.90 [0.36, 2.28]
2.3 Mental state – specific: mean endpoint PANSS positive (high = poor)	2		Mean Difference (IV, Random, 95% CI)	Subtotals only
2.3.1 < 6 m	2	77	Mean Difference (IV, Random, 95% CI)	0.25 [‐3.08, 3.58]
2.3.2 < 12 m	2	63	Mean Difference (IV, Random, 95% CI)	1.96 [‐2.80, 6.73]
2.4 Mental state – specific: mean endpoint PANSS negative (high = poor)	2		Mean Difference (IV, Random, 95% CI)	Subtotals only
2.4.1 < 6 m	2	78	Mean Difference (IV, Random, 95% CI)	‐1.19 [‐5.27, 2.89]
2.4.2 < 12 m	2	63	Mean Difference (IV, Random, 95% CI)	0.54 [‐2.63, 3.71]
2.5 Mental state – specific: mean endpoint HADS Depression (high = poor)	2		Mean Difference (IV, Random, 95% CI)	Subtotals only
2.5.1 < 6 m	2	59	Mean Difference (IV, Random, 95% CI)	‐1.75 [‐4.27, 0.77]
2.5.2 < 12 m	2	54	Mean Difference (IV, Random, 95% CI)	‐0.68 [‐3.08, 1.73]
2.6 Mental state – specific: mean endpoint HADS Anxiety (high = poor)	2		Mean Difference (IV, Random, 95% CI)	Subtotals only
2.6.1 < 6 months	2	60	Mean Difference (IV, Random, 95% CI)	‐1.30 [‐4.02, 1.43]
2.6.2 < 12 months	2	54	Mean Difference (IV, Random, 95% CI)	‐1.34 [‐3.77, 1.09]
2.7 Global state – number of participants with relapse/exacerbations of psychosis (separated time points)	2		Risk Ratio (IV, Random, 95% CI)	Subtotals only
2.7.1 < 6 months	2	90	Risk Ratio (IV, Random, 95% CI)	2.35 [0.36, 15.31]
2.7.2 < 12 months	2	90	Risk Ratio (IV, Random, 95% CI)	2.78 [0.12, 65.08]
2.8 Global state – mean endpoint CGI‐Clinician (high = poor)	1		Mean Difference (IV, Random, 95% CI)	Subtotals only
2.8.1 < 6 m	1	44	Mean Difference (IV, Random, 95% CI)	0.13 [‐0.49, 0.75]
2.8.2 < 12 m	1	43	Mean Difference (IV, Random, 95% CI)	0.15 [‐0.50, 0.80]
2.9 Global state – mean endpoint CGI‐Clinician – Improvement (high = poor)	1		Mean Difference (IV, Random, 95% CI)	Subtotals only
2.9.1 < 6 m	1	44	Mean Difference (IV, Random, 95% CI)	‐0.17 [‐0.89, 0.55]
2.9.2 < 12 m	1	42	Mean Difference (IV, Random, 95% CI)	0.05 [‐0.57, 0.67]
2.10 Service use – readmission to hospital	2		Risk Ratio (IV, Random, 95% CI)	Subtotals only
2.10.1 < 6 months	1	40	Risk Ratio (IV, Random, 95% CI)	Not estimable
2.10.2 < 12 months	1	50	Risk Ratio (IV, Random, 95% CI)	6.48 [0.35, 119.32]
2.11 Service use – number of days in hospital (separated time points)	1		Mean Difference (IV, Random, 95% CI)	Subtotals only
2.11.1 < 12 months	1	41	Mean Difference (IV, Random, 95% CI)	Not estimable
2.12 Leaving the study early – for any reason – overall acceptability (separated time points)	2		Risk Ratio (IV, Random, 95% CI)	Subtotals only
2.12.1 < 6 months	2	90	Risk Ratio (IV, Random, 95% CI)	1.66 [0.21, 12.94]
2.12.2 < 12 months	2	90	Risk Ratio (IV, Random, 95% CI)	1.66 [0.21, 12.94]
2.13 Functioning – mean endpoint PSP (high = good)	1		Mean Difference (IV, Random, 95% CI)	Subtotals only
2.13.1 < 6 months	1	44	Mean Difference (IV, Random, 95% CI)	3.90 [‐5.00, 12.80]
2.13.2 < 12 months	1	43	Mean Difference (IV, Random, 95% CI)	‐0.50 [‐9.70, 8.70]
2.14 Functioning – mean endpoint FESFS (high = good)	1		Mean Difference (IV, Random, 95% CI)	Subtotals only
2.14.1 < 6 months	1	9	Mean Difference (IV, Random, 95% CI)	‐0.84 [‐6.03, 4.35]
2.14.2 < 12 months	1	3	Mean Difference (IV, Random, 95% CI)	‐0.67 [‐8.46, 7.12]
2.15 Functioning – mean endpoint functioning scales (high = good)	2		Std. Mean Difference (IV, Random, 95% CI)	Subtotals only
2.15.1 < 6 months	2	53	Std. Mean Difference (IV, Random, 95% CI)	0.18 [‐0.37, 0.72]
2.15.2 < 12 months	2	46	Std. Mean Difference (IV, Random, 95% CI)	‐0.03 [‐0.61, 0.55]
2.16 Quality of life – mean endpoint WHOQOL (high = good)	1		Mean Difference (IV, Random, 95% CI)	Subtotals only
2.16.1 < 6 months	1	41	Mean Difference (IV, Random, 95% CI)	‐0.01 [‐10.98, 10.96]
2.16.2 < 12 months	1	43	Mean Difference (IV, Random, 95% CI)	‐2.45 [‐13.05, 8.15]
2.17 Quality of life – mean endpoint EQ‐5D‐5L (high = good)	1		Mean Difference (IV, Random, 95% CI)	Subtotals only
2.17.1 < 6 months	1	15	Mean Difference (IV, Random, 95% CI)	0.13 [‐0.18, 0.43]
2.17.2 < 12 months	1	8	Mean Difference (IV, Random, 95% CI)	0.04 [‐0.24, 0.32]
2.18 Quality of life – mean endpoint quality of life scales (high = good)	2		Std. Mean Difference (IV, Random, 95% CI)	Subtotals only
2.18.1 < 6 months	2	56	Std. Mean Difference (IV, Random, 95% CI)	0.11 [‐0.42, 0.64]
2.18.2 < 12 months	2	51	Std. Mean Difference (IV, Random, 95% CI)	‐0.09 [‐0.64, 0.46]
2.19 Behaviour – violent incidents (separated time points)	2		Risk Ratio (IV, Random, 95% CI)	Subtotals only
2.19.1 < 12 months	2	90	Risk Ratio (IV, Random, 95% CI)	1.22 [0.19, 7.84]
2.20 Behaviour – self‐injury (separated time points)	2		Risk Ratio (IV, Random, 95% CI)	Subtotals only
2.20.1 < 12 months	2	90	Risk Ratio (IV, Random, 95% CI)	1.05 [0.43, 2.56]
2.21 Behaviour – suicide attempts (separated time points)	1		Risk Ratio (IV, Random, 95% CI)	Subtotals only
2.21.1 < 12 months	1	40	Risk Ratio (IV, Random, 95% CI)	3.63 [0.16, 84.11]
2.22 Adverse events – participants with at least 1 adverse event	1		Risk Ratio (IV, Random, 95% CI)	Subtotals only
2.22.1 < 12 months	1	40	Risk Ratio (IV, Random, 95% CI)	0.47 [0.21, 1.07]
2.23 Adverse events – participants with serious adverse events	2		Risk Ratio (IV, Random, 95% CI)	Subtotals only
2.23.1 < 12 months	2	90	Risk Ratio (IV, Random, 95% CI)	2.83 [0.80, 10.05]
2.24 Adverse events – deterioration of symptoms	2		Risk Ratio (IV, Random, 95% CI)	Subtotals only
2.24.1 < 6 months	2	90	Risk Ratio (IV, Random, 95% CI)	2.13 [0.41, 11.02]
2.24.2 < 12 months	2	90	Risk Ratio (IV, Random, 95% CI)	0.46 [0.04, 4.77]
2.25 Adverse events – mean endpoint ANNSERS (high = poor)	2		Mean Difference (IV, Random, 95% CI)	Subtotals only
2.25.1 < 6 months	2	48	Mean Difference (IV, Random, 95% CI)	‐4.94 [‐8.60, ‐1.28]
2.25.2 < 12 months	2	42	Mean Difference (IV, Random, 95% CI)	‐6.96 [‐11.55, ‐2.37]

2.13. Analysis.

Comparison 2: CBT versus antipsychotics, Outcome 13: Functioning – mean endpoint PSP (high = good)

2.14. Analysis.

Comparison 2: CBT versus antipsychotics, Outcome 14: Functioning – mean endpoint FESFS (high = good)

2.16. Analysis.

Comparison 2: CBT versus antipsychotics, Outcome 16: Quality of life – mean endpoint WHOQOL (high = good)

2.17. Analysis.

Comparison 2: CBT versus antipsychotics, Outcome 17: Quality of life – mean endpoint EQ‐5D‐5L (high = good)

Comparison 3. CBT versus CBT plus antipsychotics.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
3.1 Mental state – general: mean endpoint PANSS total (high = poor)	2		Mean Difference (IV, Random, 95% CI)	Subtotals only
3.1.1 < 6 months	2	77	Mean Difference (IV, Random, 95% CI)	3.16 [‐5.41, 11.73]
3.1.2 < 12 months	2	63	Mean Difference (IV, Random, 95% CI)	2.32 [‐3.89, 8.52]
3.2 Mental state – general: mean endpoint BPRS total (high = poor)	1		Mean Difference (IV, Random, 95% CI)	Subtotals only
3.2.1 < 6 months	1	66	Mean Difference (IV, Random, 95% CI)	1.40 [‐5.10, 7.90]
3.2.2 < 12 months	1	62	Mean Difference (IV, Random, 95% CI)	4.40 [‐0.82, 9.62]
3.2.3 > 1 year	1	55	Mean Difference (IV, Random, 95% CI)	2.40 [‐3.35, 8.15]
3.3 Mental state – general: mean endpoint overall symptoms scales (PANSS/BPRS) (high = poor) (separated time points)	3		Std. Mean Difference (IV, Random, 95% CI)	Subtotals only
3.3.1 < 6 months	3	143	Std. Mean Difference (IV, Random, 95% CI)	0.20 [‐0.21, 0.60]
3.3.2 < 12 months	3	125	Std. Mean Difference (IV, Random, 95% CI)	0.30 [‐0.06, 0.65]
3.3.3 > 1 year	1	55	Std. Mean Difference (IV, Random, 95% CI)	0.22 [‐0.32, 0.76]
3.4 Mental state – general: number of participants with clinically important change in general mental state (separated time points)	2		Risk Ratio (IV, Random, 95% CI)	Subtotals only
3.4.1 < 6 months	2	90	Risk Ratio (IV, Random, 95% CI)	0.62 [0.15, 2.53]
3.4.2 < 12 months	2	90	Risk Ratio (IV, Random, 95% CI)	0.80 [0.33, 1.98]
3.5 Mental state – specific: mean endpoint PANSS positive (high = poor)	2		Mean Difference (IV, Random, 95% CI)	Subtotals only
3.5.1 < 6 m	2	77	Mean Difference (IV, Random, 95% CI)	2.68 [0.21, 5.15]
3.5.2 < 12 m	2	63	Mean Difference (IV, Random, 95% CI)	2.80 [‐1.16, 6.76]
3.6 Mental state – specific: mean endpoint BPRS positive (high = poor)	1		Mean Difference (IV, Random, 95% CI)	Subtotals only
3.6.1 < 6 m	1	69	Mean Difference (IV, Random, 95% CI)	‐0.40 [‐2.38, 1.58]
3.6.2 < 12 m	1	63	Mean Difference (IV, Random, 95% CI)	1.50 [‐0.28, 3.28]
3.6.3 > 1 year	1	55	Mean Difference (IV, Random, 95% CI)	0.70 [‐1.19, 2.59]
3.7 Mental state – specific: mean endpoint positive symptoms scales (high = poor) (separated time points)	3		Std. Mean Difference (IV, Random, 95% CI)	Subtotals only
3.7.1 < 6 m	3	146	Std. Mean Difference (IV, Random, 95% CI)	0.22 [‐0.19, 0.64]
3.7.2 < 12 m	3	126	Std. Mean Difference (IV, Random, 95% CI)	0.40 [0.05, 0.76]
3.8 Mental state – specific: mean endpoint PANSS negative (high = poor)	2		Mean Difference (IV, Random, 95% CI)	Subtotals only
3.8.1 < 6 m	2	78	Mean Difference (IV, Random, 95% CI)	0.34 [‐4.42, 5.11]
3.8.2 < 12 m	2	63	Mean Difference (IV, Random, 95% CI)	1.01 [‐1.36, 3.38]
3.9 Mental state – specific: mean endpoint SANS negative (high = poor)	1		Mean Difference (IV, Random, 95% CI)	Subtotals only
3.9.1 < 6 m	1	65	Mean Difference (IV, Random, 95% CI)	1.00 [‐6.32, 8.32]
3.9.2 < 12 m	1	60	Mean Difference (IV, Random, 95% CI)	7.80 [1.98, 13.62]
3.9.3 > 1 year	1	55	Mean Difference (IV, Random, 95% CI)	2.60 [‐3.55, 8.75]
3.10 Mental state – specific: mean endpoint negative symptoms scales (high = poor)	3		Std. Mean Difference (IV, Random, 95% CI)	Subtotals only
3.10.1 < 6 m	3	143	Std. Mean Difference (IV, Random, 95% CI)	0.10 [‐0.36, 0.55]
3.10.2 < 12 m	3	123	Std. Mean Difference (IV, Random, 95% CI)	0.40 [‐0.01, 0.81]
3.11 Mental state – specific: mean endpoint HADS Depression (high = poor)	2		Mean Difference (IV, Random, 95% CI)	Subtotals only
3.11.1 < 6 m	2	58	Mean Difference (IV, Random, 95% CI)	0.87 [‐1.55, 3.29]
3.11.2 < 12 m	2	49	Mean Difference (IV, Random, 95% CI)	‐0.45 [‐2.86, 1.96]
3.12 Mental state – specific: mean endpoint HAM‐D (high = poor)	1		Mean Difference (IV, Random, 95% CI)	Subtotals only
3.12.1 < 6 m	1	65	Mean Difference (IV, Random, 95% CI)	1.30 [‐2.82, 5.42]
3.12.2 < 12 m	1	59	Mean Difference (IV, Random, 95% CI)	2.90 [‐0.77, 6.57]
3.12.3 > 1 year	1	55	Mean Difference (IV, Random, 95% CI)	2.60 [‐1.53, 6.73]
3.13 Mental state – specific: mean endpoint depression scales (high = poor)	3		Std. Mean Difference (IV, Random, 95% CI)	Subtotals only
3.13.1 < 6 m	3	123	Std. Mean Difference (IV, Random, 95% CI)	0.18 [‐0.18, 0.54]
3.13.2 < 12 m	3	108	Std. Mean Difference (IV, Random, 95% CI)	0.16 [‐0.22, 0.54]
3.14 Mental state – specific: mean endpoint HADS Anxiety (high = poor)	2		Mean Difference (IV, Random, 95% CI)	Subtotals only
3.14.1 < 6 months	2	59	Mean Difference (IV, Random, 95% CI)	0.53 [‐4.54, 5.60]
3.14.2 < 12 months	2	50	Mean Difference (IV, Random, 95% CI)	1.13 [‐1.44, 3.69]
3.15 Mental state – specific: mean endpoint HAM‐A (high = poor)	1		Mean Difference (IV, Random, 95% CI)	Subtotals only
3.15.1 < 6 months	1	66	Mean Difference (IV, Random, 95% CI)	‐0.90 [‐5.93, 4.13]
3.15.2 < 12 months	1	59	Mean Difference (IV, Random, 95% CI)	1.50 [‐2.78, 5.78]
3.15.3 > 1 year	1	55	Mean Difference (IV, Random, 95% CI)	2.20 [‐2.93, 7.33]
3.16 Mental state – specific: mean endpoint anxiety scales (high = poor)	3		Std. Mean Difference (IV, Random, 95% CI)	Subtotals only
3.16.1 < 6 months	3	125	Std. Mean Difference (IV, Random, 95% CI)	0.03 [‐0.37, 0.43]
3.16.2 < 12 months	3	109	Std. Mean Difference (IV, Random, 95% CI)	0.21 [‐0.17, 0.59]
3.16.3 > 1 year	1	55	Std. Mean Difference (IV, Random, 95% CI)	0.23 [‐0.31, 0.77]
3.17 Global state – number of participants with relapse/exacerbations of psychosis (separated time points)	2		Risk Ratio (IV, Random, 95% CI)	Subtotals only
3.17.1 < 6 months	2	90	Risk Ratio (IV, Random, 95% CI)	2.21 [0.30, 16.35]
3.17.2 < 12 months	2	90	Risk Ratio (IV, Random, 95% CI)	2.89 [0.12, 67.75]
3.18 Global state – mean endpoint CGI‐Clinician (high = poor)	1		Mean Difference (IV, Random, 95% CI)	Subtotals only
3.18.1 < 6 m	1	43	Mean Difference (IV, Random, 95% CI)	0.59 [‐0.00, 1.18]
3.18.2 < 12 m	1	41	Mean Difference (IV, Random, 95% CI)	0.38 [‐0.28, 1.04]
3.19 Global state – mean endpoint CGI‐Clinician – Improvement (high = poor)	1		Mean Difference (IV, Random, 95% CI)	Subtotals only
3.19.1 < 6 m	1	43	Mean Difference (IV, Random, 95% CI)	0.64 [‐0.00, 1.28]
3.19.2 < 12 m	1	40	Mean Difference (IV, Random, 95% CI)	0.25 [‐0.41, 0.91]
3.20 Service use – readmission to hospital (separated time points)	2		Risk Ratio (IV, Random, 95% CI)	Subtotals only
3.20.1 < 6 months	1	39	Risk Ratio (IV, Random, 95% CI)	0.39 [0.02, 8.93]
3.20.2 < 12 months	1	51	Risk Ratio (IV, Random, 95% CI)	0.64 [0.12, 3.52]
3.21 Service use – number of days in hospital (separated time points)	1		Mean Difference (IV, Random, 95% CI)	Subtotals only
3.21.1 < 12 months	1	41	Mean Difference (IV, Random, 95% CI)	7.75 [‐26.53, 42.03]
3.22 Leaving the study early – for any reason – overall acceptability (separated time points)	3		Risk Ratio (IV, Random, 95% CI)	Subtotals only
3.22.1 < 6 months	3	180	Risk Ratio (IV, Random, 95% CI)	0.82 [0.59, 1.14]
3.22.2 < 12 months	2	90	Risk Ratio (IV, Random, 95% CI)	0.70 [0.12, 4.11]
3.23 Leaving the study early due to adverse effects (separated time points)	1		Risk Ratio (IV, Random, 95% CI)	Subtotals only
3.23.1 < 6 months	1	90	Risk Ratio (IV, Random, 95% CI)	0.19 [0.01, 3.88]
3.24 Functioning – mean endpoint PSP (high = good)	1		Mean Difference (IV, Random, 95% CI)	Subtotals only
3.24.1 < 6 months	1	43	Mean Difference (IV, Random, 95% CI)	2.43 [‐6.24, 11.10]
3.24.2 < 12 months	1	41	Mean Difference (IV, Random, 95% CI)	0.05 [‐9.04, 9.14]
3.25 Functioning – mean endpoint FESFS (high = good)	1		Mean Difference (IV, Random, 95% CI)	Subtotals only
3.25.1 < 6 months	1	6	Mean Difference (IV, Random, 95% CI)	0.31 [‐5.38, 6.00]
3.25.2 < 12 months	1	4	Mean Difference (IV, Random, 95% CI)	‐0.94 [‐7.87, 5.99]
3.26 Functioning – mean endpoint SOFAS (high = good)	1		Mean Difference (IV, Random, 95% CI)	Subtotals only
3.26.1 < 6 months	1	70	Mean Difference (IV, Random, 95% CI)	‐0.20 [‐7.29, 6.89]
3.26.2 < 12 months	1	65	Mean Difference (IV, Random, 95% CI)	3.50 [‐4.04, 11.04]
3.26.3 > 1 year	1	56	Mean Difference (IV, Random, 95% CI)	4.00 [‐4.08, 12.08]
3.27 Functioning – mean endpoint functioning scales (high = good)	3		Std. Mean Difference (IV, Random, 95% CI)	Subtotals only
3.27.1 < 6 months	3	119	Std. Mean Difference (IV, Random, 95% CI)	0.06 [‐0.30, 0.42]
3.27.2 < 12 months	3	110	Std. Mean Difference (IV, Random, 95% CI)	0.13 [‐0.25, 0.50]
3.27.3 > 1 year	1	56	Std. Mean Difference (IV, Random, 95% CI)	0.27 [‐0.27, 0.81]
3.28 Quality of life – mean endpoint WHOQOL (high = good)	1		Mean Difference (IV, Random, 95% CI)	Subtotals only
3.28.1 < 6 months	1	39	Mean Difference (IV, Random, 95% CI)	9.90 [‐0.68, 20.48]
3.28.2 < 12 months	1	36	Mean Difference (IV, Random, 95% CI)	‐0.88 [‐12.57, 10.81]
3.29 Quality of life – mean endpoint EQ‐5D‐5L (high = good)	1		Mean Difference (IV, Random, 95% CI)	Subtotals only
3.29.1 < 6 months	1	13	Mean Difference (IV, Random, 95% CI)	0.23 [‐0.04, 0.50]
3.29.2 < 12 months	1	10	Mean Difference (IV, Random, 95% CI)	0.05 [‐0.27, 0.37]
3.30 Quality of life – mean endpoint Quality of Life Scale (high = good)	1		Mean Difference (IV, Random, 95% CI)	Subtotals only
3.30.1 < 6 months	1	64	Mean Difference (IV, Random, 95% CI)	1.10 [‐11.28, 13.48]
3.30.2 < 12 months	1	60	Mean Difference (IV, Random, 95% CI)	7.50 [‐4.65, 19.65]
3.30.3 > 1 year	1	52	Mean Difference (IV, Random, 95% CI)	3.30 [‐8.67, 15.27]
3.31 Quality of life – mean endpoint quality of life scales (high = good)	3		Std. Mean Difference (IV, Random, 95% CI)	Subtotals only
3.31.1 < 6 months	3	116	Std. Mean Difference (IV, Random, 95% CI)	0.37 [‐0.11, 0.85]
3.31.2 < 12 months	3	106	Std. Mean Difference (IV, Random, 95% CI)	0.17 [‐0.21, 0.56]
3.31.3 > 1 year	1	52	Std. Mean Difference (IV, Random, 95% CI)	0.15 [‐0.40, 0.70]
3.32 Behaviour – violent incidents (separated time points)	2		Risk Ratio (IV, Random, 95% CI)	Subtotals only
3.32.1 < 12 months	2	90	Risk Ratio (IV, Random, 95% CI)	1.16 [0.18, 7.54]
3.33 Behaviour – self‐injury (separated time points)	3		Risk Ratio (IV, Random, 95% CI)	Subtotals only
3.33.1 < 6 months	1	90	Risk Ratio (IV, Random, 95% CI)	0.82 [0.30, 2.25]
3.33.2 < 12 months	2	90	Risk Ratio (IV, Random, 95% CI)	0.55 [0.26, 1.13]
3.34 Behaviour – suicide attempts (separated time points)	1		Risk Ratio (IV, Random, 95% CI)	Subtotals only
3.34.1 < 12 months	1	39	Risk Ratio (IV, Random, 95% CI)	1.17 [0.08, 17.35]
3.35 Adverse events – participants with at least 1 adverse event (pooled time points)	2	129	Risk Ratio (IV, Random, 95% CI)	0.66 [0.23, 1.87]
3.36 Adverse events – participants with at least 1 adverse event (separated time points)	2		Risk Ratio (IV, Random, 95% CI)	Subtotals only
3.36.1 < 6 months	1	90	Risk Ratio (IV, Random, 95% CI)	1.06 [0.78, 1.43]
3.36.2 < 12 months	1	39	Risk Ratio (IV, Random, 95% CI)	0.36 [0.17, 0.80]
3.37 Adverse events – participants with serious adverse events (separated time points)	3		Risk Ratio (IV, Random, 95% CI)	Subtotals only
3.37.1 < 6 months	1	90	Risk Ratio (IV, Random, 95% CI)	0.96 [0.37, 2.50]
3.37.2 < 12 months	2	90	Risk Ratio (IV, Random, 95% CI)	0.71 [0.32, 1.57]
3.38 Adverse events – deterioration of symptoms (separated time points)	3		Risk Ratio (IV, Random, 95% CI)	Subtotals only
3.38.1 < 6 months	3	180	Risk Ratio (IV, Random, 95% CI)	1.07 [0.37, 3.10]
3.38.2 < 12 months	2	90	Risk Ratio (IV, Random, 95% CI)	1.05 [0.11, 9.74]
3.39 Adverse events – lack of improvement (separated time points)	1		Risk Ratio (IV, Random, 95% CI)	Subtotals only
3.39.1 < 6 months	1	90	Risk Ratio (IV, Random, 95% CI)	0.96 [0.33, 2.74]
3.40 Adverse events – mean endpoint ANNSERS (high = poor)	2		Mean Difference (IV, Random, 95% CI)	Subtotals only
3.40.1 < 6 months	2	52	Mean Difference (IV, Random, 95% CI)	‐3.68 [‐9.66, 2.31]
3.40.2 < 12 months	2	49	Mean Difference (IV, Random, 95% CI)	‐3.47 [‐12.06, 5.11]
3.41 Adverse events – mean endpoint UKU (high = poor)	3		Mean Difference (IV, Random, 95% CI)	Subtotals only
3.41.1 < 6 months	3	122	Mean Difference (IV, Random, 95% CI)	‐1.86 [‐6.37, 2.66]
3.41.2 < 12 months	2	49	Mean Difference (IV, Random, 95% CI)	‐3.47 [‐12.06, 5.11]

3.1. Analysis.

Comparison 3: CBT versus CBT plus antipsychotics, Outcome 1: Mental state – general: mean endpoint PANSS total (high = poor)

3.2. Analysis.

Comparison 3: CBT versus CBT plus antipsychotics, Outcome 2: Mental state – general: mean endpoint BPRS total (high = poor)

3.5. Analysis.

Comparison 3: CBT versus CBT plus antipsychotics, Outcome 5: Mental state – specific: mean endpoint PANSS positive (high = poor)

3.6. Analysis.

Comparison 3: CBT versus CBT plus antipsychotics, Outcome 6: Mental state – specific: mean endpoint BPRS positive (high = poor)

3.8. Analysis.

Comparison 3: CBT versus CBT plus antipsychotics, Outcome 8: Mental state – specific: mean endpoint PANSS negative (high = poor)

3.9. Analysis.

Comparison 3: CBT versus CBT plus antipsychotics, Outcome 9: Mental state – specific: mean endpoint SANS negative (high = poor)

3.11. Analysis.

Comparison 3: CBT versus CBT plus antipsychotics, Outcome 11: Mental state – specific: mean endpoint HADS Depression (high = poor)

3.12. Analysis.

Comparison 3: CBT versus CBT plus antipsychotics, Outcome 12: Mental state – specific: mean endpoint HAM‐D (high = poor)

3.14. Analysis.

Comparison 3: CBT versus CBT plus antipsychotics, Outcome 14: Mental state – specific: mean endpoint HADS Anxiety (high = poor)

3.15. Analysis.

Comparison 3: CBT versus CBT plus antipsychotics, Outcome 15: Mental state – specific: mean endpoint HAM‐A (high = poor)

3.24. Analysis.

Comparison 3: CBT versus CBT plus antipsychotics, Outcome 24: Functioning – mean endpoint PSP (high = good)

3.25. Analysis.

Comparison 3: CBT versus CBT plus antipsychotics, Outcome 25: Functioning – mean endpoint FESFS (high = good)

3.26. Analysis.

Comparison 3: CBT versus CBT plus antipsychotics, Outcome 26: Functioning – mean endpoint SOFAS (high = good)

3.28. Analysis.

Comparison 3: CBT versus CBT plus antipsychotics, Outcome 28: Quality of life – mean endpoint WHOQOL (high = good)

3.29. Analysis.

Comparison 3: CBT versus CBT plus antipsychotics, Outcome 29: Quality of life – mean endpoint EQ‐5D‐5L (high = good)

3.30. Analysis.

Comparison 3: CBT versus CBT plus antipsychotics, Outcome 30: Quality of life – mean endpoint Quality of Life Scale (high = good)

Characteristics of studies

Characteristics of included studies [ordered by study ID]

Francey 2020 – STAGES.

Study characteristics
Methods	Study design: randomised controlled trial Study grouping: parallel group Antipsychotic medication allowed: no for CBT; yes for control Duration in weeks: 36 weeks for intervention, 52 and 104 weeks follow‐up Number of study arms: 2 Type of blinding: triple‐blind Type of data analysis for overall efficacy: ITT (not stated clearly, but more participants than the observed cases/completers are included in the analysis) Number of sites: NA Number of countries: 1 Country: Australia Sponsorship source: public Trial registration ID: ACTRN12607000608460 Publication year: 2020
Participants	Diagnosis: DSM‐IVpsychotic disorder, including schizophrenia, schizophreniform disorder, delusional disorder, brief psychotic disorder, major depressive disorder with psychotic symptoms, substance‐induced psychotic disorder, or psychosis not otherwise specified (NOS); diagnostic criteria: DSM‐IV Inclusion criteria:"Participants were aged 15‐25 years, presenting with FEP to a specialist early psychosis service (Early Psychosis Prevention and Intervention Centre, EPPIC). EPPIC is part of Orygen Youth Health, a public youth mental health service that serves a catchment area of approximately 1 million in the Western region of metropolitan Melbourne, Australia. Eligibility was defined as fulfilling criteria for a DSM‐IV psychotic disorder, including schizophrenia, schizophreniform disorder, delusional disorder, brief psychotic disorder, major depressive disorder with psychotic symptoms, substance induced psychotic disorder, or psychosis not otherwise specified (NOS). Potential participants were also required to meet strict inclusion criteria to minimize risk: ability to provide informed consent; comprehension of the English language; DUP of less than 6 months; living in stable accommodation; low risk to self or others (score of <5 on the Brief Psychiatric Rating Scale version 4 [BPRS‐4 34] Suicidality and Hostility subscales); low previous exposure to antipsychotic medication (less than 7 days or lifetime maximum 1750 mg chlorpromazine equivalent)." Exclusion criteria: "At high risk for suicide, defined as a score of 5 or greater on the Brief Psychiatric Rating Scale (BPRS) (Overall & Gorham 1962) Suicidality subscale in the 2 weeks prior to presentation. This level of suicidality corresponds to suicidal ideation with intent or plan or an impulsive suicide attempt using non‐lethal methods or in full view of potential saviours. At high risk for aggression, defined as a score of 5 or greater on the BPRS Hostility subscale in the 2 weeks prior to presentation, which corresponds to making threats or having thrown things. Previous treatment with antipsychotic medication defined as having exceeded 7 days of antipsychotic treatment or a lifetime dose greater than 1750 mg chlorpromazine equivalents. Previous treatment with lithium or anticonvulsant medication for a manic episode. Current pregnancy." Setting: probably outpatient Special subgroup: first‐episode patients N: 90 Gender: 36 men, 45 women Age: mean 18.54 years CBT without antipsychotic arm: participants total: 46, participants male: 18, participants female: 23, age: mean 18.2 years (SD = 2.6), age range: 15 to 25 years, duration of untreated psychosis: < 6 months, % taking antipsychotics: NA for CBT, 100% for control, % diagnosis schizophrenia: 17.10%, % diagnosis schizoaffective disorder: NA, % delusional disorder: 2.40%, % psychosis NOS: 29.30%, % schizophreniform disorder: 17.10%, % diagnosis affective psychosis: 19.50%, baseline severity BPRS total M = 58.5, SD = 8.9, N = 41 Control arm: participants total: 44, age: mean 18.9 (SD = 2.9), age range: 15 to 25 years, duration of untreated psychosis: < 6 months, % taking antipsychotics: 100%, % diagnosis schizophrenia: 12.50%, % diagnosis schizoaffective disorder: NA, % delusional disorder: 10%, % psychosis NOS: 20%, % schizophreniform disorder: 22.50%, % diagnosis affective psychosis: 20%, baseline severity BPRS total M = 57.3, SD = 9.8, N = 39
Interventions	1. Intensive psychosocial intervention + placebo (CBT), N = 46 Description of treatment: "All participants received CBCM, a comprehensive manualized intervention developed specifically for early psychosis, which has a strong focus on therapeutic engagement. CBCM provides formulation driven CBT and psychoeducation delivered within a therapeutic case management framework. 41 CBCM incorporates CBT for positive and negative psychotic symptoms, comorbidities, enhancement of coping strategies, and relapse prevention. Typical CBT strategies were employed, including symptom monitoring, activity scheduling, and behavioral experiments. Case managers met with young people weekly, usually at the outpatient clinic, and offered support, problem‐solving, links to group programs, and advocacy in addition to formulation‐driven CBT. Therapists were tertiary‐trained mental health professionals, mostly clinical psychologists who received both group and individual supervision in CBCM. They completed a psychological interventions checklist after each therapy session to record the interventions used. CBCM was enhanced by close monitoring of mental state and risks, family work and 24‐h crisis response and home‐visits when required. Participants were also seen regularly by the treating psychiatric registrar and consultant psychiatrist." Number of sessions: 24 planned, 14 mean, 8.8 specifically delivered CBT Sessions frequency and duration: 4 planned, 2.3 actually, duration NA Therapist: tertiary‐trained mental health professionals, mostly clinical psychologists Therapist experience: expert Supervision: yes Modality: individual Delivery: face‐to‐face Treatment manualised: yes 2. Control: intensive psychosocial intervention + low‐dose antipsychotics (CBT plus antipsychotics), N = 44 Description of treatment: participants allocated to the medication group received risperidone (1 mg) or paliperidone (3 mg), depending on when they were enrolled in the study. The study commenced recruitment in 2008 using risperidone 1 mg tablets and matched placebo and then paused in August 2009 when this matched active and placebo trial medication became unavailable. It recommenced in June 2012 using paliperidone 3 mg capsules and matched placebo and paused again in March 2013 when the paliperidone and matched placebo was no longer available. The final recruitment phase from September 2013 to December 2016 used 1 mg risperidone and matched placebo manufactured specifically for the study. The total recruitment period was 4 years and 11 months. Dosing was increased by prescription of additional tablets/capsules after medical review appointment with treating psychiatrist. At each phase of the study, placebo group participants received placebo tablets that were identical in appearance, taste, and packaging to the active medication. Other medications, excluding other antipsychotic medication and mood stabilisers, were permitted during the trial. These were recorded. All participants received CBCM, a comprehensive manualised intervention developed specifically for early psychosis, which has a strong focus on therapeutic engagement. CBCM provides formulation‐driven CBT and psychoeducation delivered within a therapeutic case management framework. CBCM incorporates CBT for positive and negative psychotic symptoms, comorbidities, enhancement of coping strategies, and relapse prevention. Typical CBT strategies were employed, including symptom monitoring, activity scheduling, and behavioural experiments. Case managers met with young people weekly, usually at the outpatient clinic, and offered support, problem‐solving, links to group programmes, and advocacy in addition to formulation‐driven CBT. Therapists were tertiary‐trained mental health professionals, mostly clinical psychologists who received both group and individual supervision in CBCM. They completed a psychological interventions checklist after each therapy session to record the interventions used. CBCM was enhanced by close monitoring of mental state and risks, family work, and 24‐hour crisis response and home visits when required. Participants were also seen regularly by the treating psychiatric registrar and consultant psychiatrist. Number of sessions: 4*6 = 24 planned, 14 mean, 7.9 specifically delivering CBT; weekly psychiatrist appointments for the first 2 months Sessions frequency and duration: 4 planned, 2.3 actually Therapist: experts Therapist experience: tertiary‐trained mental health professionals, mostly clinical psychologists + psychiatrists Supervision: yes Modality: individual Delivery: face‐to‐face Treatment manualised: yes
Outcomes	Mental state – general: mean endpoint BPRS total (< 6 months, < 1 year, > 1 year) Mental state – general: mean endpoint overall symptoms scales (PANSS/BPRS) (< 6 months, < 1 year, > 1 year) Mental state – specific: mean endpoint BPRS positive (< 6 months, < 1 year, > 1 year) Mental state – specific: mean endpoint SANS (< 6 months, < 1 year, > 1 year) Mental state – specific: mean endpoint HAM‐D (< 6 months, < 1 year, > 1 year) Mental state – specific: mean endpoint HAM‐A (< 6 months, < 1 year, > 1 year) Leaving the study early – for any reason – overall acceptability (< 6 months) Leaving the study early – due to adverse effect – overall tolerability (< 6 months) Functioning – mean endpoint SOFAS (< 6 months, < 1 year, > 1 year) Functioning – mean endpoint functioning scales (< 6 months, < 1 year, > 1 year) Quality of life – mean endpoint quality of life scale (< 6 months, < 1 year, > 1 year) Quality of life – mean endpoint quality of life scale (< 6 months, < 1 year, > 1 year) Behaviour – number of participant self‐injury (< 6 months) Adverse events/effects – at least 1 adverse event (< 6 months) Number of participants with serious adverse events (< 6 months) Adverse events – participants with serious adverse events (< 6 months) Adverse events – deterioration of symptoms (< 6 months) Adverse events – lack of improvement (< 6 months) Adverse events – mean endpoint UKU (< 6 months)
Notes	Authors were contacted and provided additional information.

Morrison 2014 – ACTION.

Study characteristics
Methods	Study design: randomised controlled trial Study grouping: parallel group Antipsychotic medication allowed: yes Duration in weeks: 39 weeks for therapy, +39 weeks follow‐up/booster session Number of study arms: 2 Type of blinding: rater blind Type of data analysis for overall efficacy: ITT Number of sites: 2 Number of academic sites: NA Number of countries: 1 Country: England Sponsorship source: public Trial registration ID: ISRCTN29607432 Publication year: 2014
Participants	Diagnosis: ICD‐10 criteria for schizophrenia, schizoaffective disorder, or delusional disorder, or met entry criteria for an early intervention for psychosis service (operationally defined with the PANSS) Inclusion criteria: "Eligible participants aged 16‐65 years were in contact with mental health services, and either met International Classification of Diseases – tenth revision (ICD‐10) criteria for schizophrenia, schizoaffective disorder, or delusional disorder, or met entry criteria for an early intervention for psychosis service (operationally defined with the Positive and Negative Syndrome Scale [PANSS]) to allow for diagnostic uncertainty in early phases of psychosis and the fact that most early‐episode cases in the UK will receive their services from such specialist teams, consistent with NICE guidelines. Participants had also had either at least 6 months without antipsychotic drugs and continuing symptoms or had never received antipsychotics and had chosen not to; all participants scored at least 4 on PANSS delusions or hallucinations, or at least 5 on suspiciousness or persecution, conceptual disorganisation, or grandiosity. All participants were identified via care coordinators and relevant mental health staff within participating mental health trusts at the two study sites." Exclusion criteria: "Exclusion criteria were present receipt of antipsychotic drugs; moderate to severe learning disability; organic impairment; participants not having the capacity to consent to research participation; non‐English‐speaking participants (because their inclusion would prevent the use of standardised assessment techniques); acute inpatient care settings; receipt of cognitive therapy for psychosis or previous cognitive therapy for other disorders in the past 2 years; and a primary diagnosis of substance or alcohol abuse." Setting: outpatients Special subgroup: positive symptoms N: 74 Gender: 39 men, 35 women Age: mean 31.32 years CBT without antipsychotic arm: participants total: 37, participants male: 17, participants female: 20, age: mean 32.95 years (SD = 13.11), age range: 16 to 65 years, duration of untreated psychosis: NA, % taking antipsychotics: 100% (baseline) 27% during trial, baseline PANSS total M = 70.24, SD = 13.75, N = 37 Control arm: participants total: 37, age: mean 29.68 (SD = 11.95), age range: 16 to 65 years, duration of untreated psychosis: NA, % taking antipsychotics: 100% (baseline), 27% (during trial), baseline PANSS total M = 73.27, SD = 13.42, N = 37
Interventions	1. Cognitive therapy + TAU and monitoring (CBT), N = 37 Description of treatment: "Cognitive therapy requires an individualised, problem‐oriented approach and incorporates a manualised process of assessment and formulation. The central features of our approach to treatment of psychosis involve normalisation and evaluation of the appraisals that people make, helping them to test such appraisals with use of behavioural experiments, and helping them to identify and modify unhelpful cognitive and behavioural responses. A more detailed analysis of the treatment strategies can be found in our treatment manuals." Number of sessions: 26 planned, 13.3 (mean) received Sessions frequency and duration: 4 per month, duration 60 minutes Therapist: psychologists, nurse, 1 psychiatrist Therapist experience: expert Supervision: yes Modality: individual Delivery: face‐to‐face Treatment manualised: yes 2. Control: TAU and monitoring only (no specific treatment), N = 37 Description of treatment: "All participants received treatment as usual plus regular monitoring (incorporating a PANSS assessment from a research assistant), which provided benefits over routine care because it aimed to provide warm, empathic, and non‐judgmental face‐to‐face contact, supportive listening, signposting to appropriate local services for unmet needs, and crisis management when needed (usually by referral to a local crisis team, early intervention service, or psychiatric liaison within emergency departments). Treatment as usual was variable across both sites, although both were chosen partly because they had comprehensive early intervention services. In practice, participants within these services received regular care‐coordination and psychosocial interventions, including the offer of family interventions, whereas individuals from other community‐based services often received little more than irregular contact with care coordinators, and many of these participants were discharged by their clinical teams during the trial for non‐attendance or continued reluctance to accept medicine." Number of sessions: NA Sessions frequency and duration: NA Therapist: NA Therapist experience: NA Supervision: NA Modality: NA Delivery: face‐to‐face Treatment manualised: NA
Outcomes	Mental state – general: mean endpoint PANSS total (< 6 months, < 1 year, > 1 year) Mental state – general: mean endpoint overall symptoms scales (PANSS/BPRS) (< 6 months, < 1 year) Mental state – clinically important change in general mental state (< 1 year, > 1 year) Mental state – specific: mean endpoint PANSS positive (< 6 months, < 1 year, > 1 year) Mental state – specific: mean endpoint PANSS negative (< 6 months, < 1 year, > 1 year) Mental state – specific: mean endpoint BDI (< 6 months, < 1 year, > 1 year) Mental state – specific: mean endpoint BAI (< 6 months, < 1 year, > 1 year) Global state – relapse (< 1 year, > 1 year) Service use – admission to hospital (< 1 year, > 1 year) Service use – number of days in hospital (< 1 year, > 1 year) Leaving the study early – for any reason – overall acceptability (< 6 months, < 1 year, > 1 year) Functioning – mean endpoint PSP (< 6 months, < 1 year, > 1 year) Behaviour – suicide attempt (< 1 year) Adverse events/effects – specific adverse events (< 1 year, > 1 year) Number of participants with serious adverse events (< 1 year) Mortality – overall mortality (< 6 months, < 1 year, > 1 year) Mortality – due to natural causes (< 6 months, < 1 year, > 1 year)
Notes	Authors were contacted and provided additional information.

Morrison 2018 – COMPARE.

Study characteristics
Methods	Study design: randomised controlled trial Study grouping: parallel group Antipsychotic medication allowed: no for CBT, yes for control group Duration in weeks: 36 weeks for therapy, +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 academic sites: 0 Number of countries: 1 Country: England Sponsorship source: public Trial registration ID: ISRCTN06022197 Publication year: 2018
Participants	Diagnosis: schizophrenia, schizoaffective disorder, or delusional disorder, or met the entry criteria for an early intervention for psychosis service (ICD‐10) Inclusion criteria: "Eligible participants were aged 16 years or older; met ICD‐­10 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 help­seeking." 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: inpatients and outpatients Special subgroup: first‐episode patients, positive symptoms N: 75 Gender: 43 men, 32 women Age: mean 23.61 years CBT without antipsychotic 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, % taking antipsychotics: not known at baseline, 30.7% during the trial, baseline PANSS total M = 70.35, SD = 8.03, N = 26 CBT + antipsychotic 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, % taking antipsychotics: not known at baseline, 72% during the trial, baseline PANSS total M = 70.76, SD = 8.46, N = 25 Antipsychotics arm: participants total: 24, participants male: 13, participants female: 11, age: mean 23.21, age range: > 16 years, duration of untreated psychosis: 37.33, % taking antipsychotics: not known at baseline, 66.66% during the trial, baseline PANSS total M = 70.13, SD = 10.11, N = 24
Interventions	1. CBT, N = 26 Description of treatment: "26 sessions of therapy based on a specific cognitive model 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. 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." Number of sessions: 26 planned, 14.39 (mean) received Sessions frequency and duration: 4 per month, duration 60 minutes Therapist: appropriately qualified psychological therapist Therapist experience: NA Supervision: yes Modality: individual Delivery: face‐to‐face Treatment manualised: yes Pharmacotherapy allowed: no, only if deteriorated 2. CBT plus antipsychotics, N = 25 Description of treatment: "Participants allocated to the combined treatment group were offered CBT and antipsychotic medications as described for the monotherapy groups" Number of sessions: 26, 14.39 (mean) received Sessions frequency and duration: 4 per month, duration 60 minutes Therapist: appropriately qualified psychological therapist + psychiatrist Therapist experience: NA Supervision: yes Modality: individual Delivery: face‐to‐face Treatment manualised: yes Pharmacotherapy allowed: yes 3. Antipsychotics, 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 Pharmacotherapy allowed: yes
Outcomes	Mental state – general: mean endpoint PANSS total (< 6 months, < 1 year) Mental state – general: mean endpoint overall symptoms scales (PANSS/BPRS) (< 6 months, < 1 year) Mental state – clinically important change in general mental state (< 6 months, < 1 year) Mental state – specific: mean endpoint PANSS positive (< 6 months, < 1 year) Mental state – specific: mean endpoint positive symptoms scales (< 6 months, < 1 year) Mental state – specific: mean endpoint PANSS negative (< 6 months, < 1 year) Mental state – specific: mean endpoint negative symptoms scales (< 6 months, < 1 year) Mental state – specific: mean endpoint HADS Depression (< 6 months, < 1 year) Mental state – specific: mean endpoint depression scales (< 6 months, < 1 year) Mental state – specific: mean endpoint HADS Anxiety (< 6 months, < 1 year) Mental state – specific: mean endpoint anxiety scales (< 6 months, < 1 year) Global state – relapse (< 6 months, < 1 year) Global state – average endpoint or change score on a global state scale – CGI endpoint (< 6 months, < 1 year) Global state – average endpoint or change score on a global state scale – CGI‐Clinician change (< 6 months, < 1 year) Service use – admission to hospital (< 1 year) Service use – number of days in hospital (< 1 year) Leaving the study early – for any reason (< 6 months, < 1 year) Functioning – mean endpoint PSP (< 6 months, < 1 year) Functioning – mean endpoint functioning scales (< 6 months, < 1 year) Quality of life – average endpoint or change score on quality of life scale – WHOQOL‐BREF endpoint (< 6 months, < 1 year) Quality of life – mean endpoint quality of life scales (< 6 months, < 1 year, > 1 year) Behaviour – occurrence of violent incidents (< 1 year) Behaviour – self‐injury (< 1 year) Behaviour – suicide attempt (< 1 year) Adverse events/effects – specific adverse events (< 6 months, < 1 year) Number of participants with serious adverse events (< 6 months, < 1 year, > 1 year) Scales for side effects – ANNSERS endpoint (< 6 months, < 1 year)
Notes	Authors were contacted and provided additional information.

Morrison 2020 – MAPS.

Study characteristics
Methods	Study design: randomised controlled trial Study grouping: parallel group Antipsychotic medication allowed: yes Duration in weeks: 26 weeks for therapy, +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 academic sites: NA Number of countries: 1 Country: England Sponsorship source: public Trial registration ID: ISRCTN80567433 Publication year: 2020
Participants	Diagnosis: schizophrenia, schizoaffective disorder, or delusional disorder OR the entry criteria for an EIP service for first‐episode psychosis (ICD‐10) 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: unclear Special subgroup: first‐episode patients, children and adolescents, positive symptoms N: 61 Gender: 30 men, 29 women Age: mean 16.3 years CBT without antipsychotic arm: participants total: 18, participants male: 8, participants female: 10, age: mean 16.3 years (SD = 1.4), age range: 14 to 18 years, duration of untreated psychosis (months): 12, % taking antipsychotics: 33%, baseline PANSS total M = 72.9, SD = 9.7, N = 18 CBT + antipsychotic arm: participants total: 21, participants male: 10, participants female: 10, age: mean 16.2 years (SD = 1.3), age range: 14 to 18 years, duration of untreated psychosis (months): 8, % taking antipsychotics: 57.1%, baseline PANSS total M = 75.9, SD = 14.8, N = 21 Antipsychotics arm: participants total: 22, participants male: 12, participants female: 9, age: mean 16.4 (SD = 1.3), age range: 14 to 18 years, duration of untreated psychosis (months): 9, % taking antipsychotics: 72.7%, baseline PANSS total M = 74.8, SD = 12.2, N = 22
Interventions	1. Psychological intervention (CBT + optional family intervention) (CBT), 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, 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. After an initial session involving assessment, formulation sharing, and agreeing goals and problems to be worked on, family intervention involved aspects such as psycho educational work, provision of normalising information and recovery‐oriented information, problem solving, and relapse prevention planning. Therapy session records were completed by therapists throughout the delivery of psychological interventions, and therapists received supervision once a week from two MAPS group members (APM and SB). Audio‐recorded CBT sessions, taped with the patient’s consent, were rated regularly via rotational sampling of tapes with the Cognitive Therapy Scale–Revised (by APM and SB) to ensure fidelity to the protocol." Number of sessions: 26 CBT + 6 family (optional) + 4 CBT (booster); median of 14 actually received Sessions frequency and duration: 4 per month, 1 family Therapist: appropriately trained therapist Therapist experience: NA Supervision: yes Modality: individual Delivery: face‐to‐face Treatment manualised: yes Pharmacotherapy allowed: yes 2. CBT plus antipsychotics, N = 21 Description of treatment: "Participants allocated to receive antipsychotics plus psychological intervention were offered all treatments as described for the monotherapy groups." Number of sessions: 26 CBT + 6 family (optional) + 4 CBT (booster); median of 15 actually received Sessions frequency and duration: 4, 1 family Therapist: appropriately qualified psychological therapist + psychiatrist Therapist experience: NA Supervision: yes Modality: individual Delivery: face‐to‐face Treatment manualised: yes Pharmacotherapy allowed: yes 3. Antipsychotics, 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 (DM, MRB, NH, AJ, PMH, RW, RU, and FP) 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 Pharmacotherapy allowed: yes
Outcomes	Mental state – general: mean endpoint PANSS total (< 6 months, < 1 year) Mental state – general: mean endpoint overall symptoms scales (PANSS/BPRS) (< 6 months, < 1 year) Mental state – clinically important change in general mental state (< 6 months, < 1 year) Mental state – specific: mean endpoint PANSS positive (< 6 months, < 1 year) Mental state – specific: mean endpoint positive symptoms scales (< 6 months, < 1 year) Mental state – specific: mean endpoint PANSS negative (< 6 months, < 1 year) Mental state – specific: mean endpoint negative symptoms scales (< 6 months, < 1 year) Mental state – specific: mean endpoint HADS Depression (< 6 months, < 1 year) Mental state – specific: mean endpoint depression scales (< 6 months, < 1 year) Mental state – specific: mean endpoint HADS Anxiety (< 6 months, < 1 year) Mental state – specific: mean endpoint anxiety scales (< 6 months, < 1 year) Global state – relapse (< 6 months, < 1 year) Service use – admission to hospital (< 6 months) Leaving the study early – for any reason (< 6 months, < 1 year) Functioning – mean endpoint FESFS (< 6 months, < 1 year) Quality of life – average endpoint or change score on quality of life scale – EQ‐5D‐5L endpoint (< 6 months, < 1 year) Quality of life – mean endpoint on quality of life scales (< 6 months, < 1 year) Behaviour – occurrence of violent incidents (< 1 year) Behaviour – self‐injury (< 1 year) Behaviour – suicide attempt (< 1 year) Adverse events/effects – at least 1 adverse event/effect (< 1 year) Number of participants with serious adverse events (< 6 months, < 1 year) Adverse events/effects – specific adverse events (< 6 months, < 1 year) Scales for side effects – ANSSERS endpoint (< 6 months, < 1 year)
Notes	Authors were contacted and provided additional information.

ANNSERS: Antipsychotic Non‐Neurological Side Effects Rating Scale

BAI: Beck Anxiety Inventory

BDI: Beck Depression Inventory

BPRS: Brief Psychiatric Rating Scale

CBCM: cognitive behavioural case management

CBT: cognitive behavioural therapy

CGI: Clinical Global Impression Scale

DSM‐IV: Diagnostic and Statistical Manual of Mental Disorders, 4th Edition

DUP: duration of untreated psychosis

FESFS: First Episode Social Functioning Scale

HADS: Hospital Anxiety and Depression Scale

HAM‐A: Hamilton Anxiety Rating Scale

HAM‐D: Hamilton Depression Scale

ICD‐10: International Classification of Diseases 10th Revision

ITT: intention to treat

M: mean

NA: not available

PANSS: Positive and Negative Syndrome Scale

PSP: Personal and Social Performance Scale

SANS: Scale for the Assessment of Negative Symptoms

SOFAS: Social and Occupational Functioning Assessment Scale

SD: standard deviation

TAU: treatment as usual

WHOQOL‐BREF: World Health Organization Quality of Life abbreviated form

Characteristics of excluded studies [ordered by study ID]

Study	Reason for exclusion
Morrison 2012	Design: not randomised
Morrison 2016	Intervention: no CBT without antipsychotics arm
Morrison 2018 – FOCUS	Intervention: no CBT without antipsychotics arm

Differences between protocol and review

In the protocol we planned to conduct several sensitivity and subgroup analyses; however, this was not possible, since all analyses included a maximum of three studies only.

We added the outcome 'average endpoint or change on adverse effects scale' as a possible measure of adverse events/effects.

Contributions of authors

IB: 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

OC: 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; GRADE assessment; interpretation of data; writing of the review

NHS: search and selection of studies for inclusion in the review; interpretation of data; writing of the review

SL: conception of the review; design of the review; interpretation of data; writing of the review

Sources of support

Internal sources

• Freistaat Bayern, Germany

  The employer of Irene Bighelli and Stefan Leucht

• National Institute for Health and Care Research (NIHR), UK

  Provided funding for Cochrane Schizophrenia Group

External sources

• DAAD ASFE, Germany

  Doctoral scholarship of Nurul Husna Salahuddin

Declarations of interest

IB is the Deputy Co‐ordinating Editor of the Cochrane Schizophrenia Group. She was not involved in the editorial process of the current review (see Acknowledgements for details about editorial process).

OC: none

NHS: none

SL: 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, Gedeon Richter, Janssen, Karuna, Kynexis, Lundbeck, Medichem, Medscape, Mitsubishi, Otsuka, Novo Nordisk, Recordati, Rovi, Teva. He has no declaration of interest related to cognitive behavioural therapy. SL is an editor of the Cochrane Schizophrenia Group. He was not involved in the editorial process of the current review (see Acknowledgements for details about editorial process).

These authors should be considered joint first author

New