Interventions for helping people recognise early signs of recurrence in bipolar disorder

Objectives

This is a protocol for a Cochrane Review (intervention). The objectives are as follows:

1. To evaluate the effectiveness of EWS plus TAU or EWS plus psychological therapy versus TAU alone or psychological treatment (without EWS) independently on time to recurrence of any bipolar episode and hospitalisation, and other clinically relevant outcome measures.

2. To evaluate the effectiveness of intermittent medication used on recognition of EWS without continued mood‐stabilising medication versus TAU involving continued mood‐stabilising medication on time to recurrence of any bipolar episodes.

Background

Description of the condition

Bipolar disorder is a common psychiatric illness. The lifetime prevalence of bipolar I disorder (mania and depression) is estimated at 1% of adults, and bipolar II disorder (hypomania and depression) affects approximately 0.4% of adults (NICE 2014). The disorder is characterised by periods of depression, mania/hypomania, or mixed episodes. Depression is characterised by core symptoms of depressed mood and loss of interest for at least two weeks associated with other symptoms such as changes in sleep pattern, appetite and concentration; low self‐esteem; hopelessness; self‐blame; and suicidal ideation. Hypomania is characterised by periods of elation or irritable mood lasting at least four days coupled with other symptoms of overactivity, reduced need for sleep, disinhibition, increased self‐esteem or grandiosity, and increased talkativeness. Mixed episode is characterised by concurrent presence of both depressive and manic symptoms.

Bipolar disorder is the 18th leading cause of lifelong disability measured by disability‐adjusted life years in adults worldwide (Whiteford 2013). In addition to reducing psychosocial functioning, it is also associated with loss of approximately 10 to 20 years of life expectancy (McIntyre 2020).

Bipolar disorder is a recurrent illness, and most people experience many episodes in their lifetime. Approximately 60% people experience a recurrence of bipolar disorder in the first two years, ‎and about 75% experience a recurrence over five years following the initial episode. However, bipolar disorder remains a very unpredictable illness and prediction of course and outcome is still very challenging. The nature of the initial episode, onset of illness, and family history of bipolar disorders are often used in clinical practice to predict the course, but with limited success (Gelenberg 1989; Gitlin 1995; Treuer 2010). A significant proportion of people continue to experience subsyndromal symptoms between episodes and they spend significantly more time in depressed phases in comparison to manic or hypomanic phases (Judd 2002; Judd 2003). Moreover, depression is associated with significant dysfunction. Hence, it is not surprising that up to 60% of people do not return to a premorbid level of functioning after the initial episodes (Bauer 2001; Keck 1998).

Description of the intervention

Pharmacological treatments are the mainstay of the treatment, and it includes treatment of acute affective episodes and as well as for prevention of future episodes (NICE 2014). Non‐compliance with the treatment is common like any other chronic medical or psychiatric disorders (Semahegn 2020). Moreover, many people experience recurrence of disorder despite being compliant with the medications (Goldberg 1995).

Addition of psychological treatments has been recommended by all professional guidelines including the National Institute of Health and Care Excellence (NICE) and British Association of Psychopharmacology (Goodwin 2016; NICE 2014). The 2014 NICE guidelines recommend that after every affective episode, clinicians should discuss with the patients and carers the long‐term management of the condition, including pharmacological and psychological therapies to prevent relapse and reduce symptoms (NICE 2014). The psychological therapies take into account potential triggers for relapse, early warning symptoms (EWS), and self‐management strategies. The recommended psychological therapies for prevention are family‐focussed therapy (FFT) and evidence‐based structured psychological interventions developed for bipolar disorder.

Most people with bipolar disorder experience EWS, two to four weeks before a full episode (Jackson 2003). These symptoms vary between individuals, are different for manic and depressive episodes, but are consistent within each individual from episodes of mania to episode of mania or episode of depression to depression. Gradually the nature and timing of EWS may evolve. EWS before manic episodes are qualitatively different from usual experience and relatively acute in onset so they are rarely confused with residual symptoms of bipolar disorder. The most common EWS are increased self‐esteem, increased brightness, loudness of senses, decreased need for sleep, and increased goal‐directed activities (Wong 1999). Decreased need for sleep and increased goal‐directed activities are also frequently identified and reported by family members. Self‐recognition of these symptoms depends on the insight of patients, which may require a brief psychological intervention to achieve. In contrast, EWS before depressive episodes are very gradual and are often the build up to a greater frequency and severity of residual symptoms that are commonly experienced between bipolar episodes (Sierra 2007). Compared to EWS for mania‐like episodes, EWS for depression may be more difficult to identify, especially by family members, although EWS for depression can sometimes be quite distinct (e.g. playing a particular music track before each episode of depression). EWS for mixed affective episodes may be distinct or similar to EWS for mania. Regular checking of personal EWS of mania and depression is associated with improved social and occupational function (Lobban 2011).

It is known that bipolar disorder is more responsive to treatment at an early stage and the responsiveness to pharmacological treatment is less in people with multiple episodes (Agren 2007). There is evidence that adding a psychological intervention involving the recognition and management of EWS for mania and depression to pharmacological treatment might improve the course of bipolar disorder with greater benefits in people earlier in the course of bipolar disorder (Colom 2003; Colom 2009; Morriss 2016). These interventions are effective after first episodes and recurrent episodes of bipolar disorders (de Barros Pellegrinelli 2013). However, even in people with refractory bipolar disorder there may still be benefits (Isasi 2010). Such interventions may be cost saving (Camacho 2017).

National implementation of EWS interventions for mania and depression might reduce bipolar episodes, admissions to hospital, and involuntary admissions to hospital compared to usual care (Joas 2020). The longer‐term effects of EWS interventions for bipolar disorder are unclear.

Psychological therapies that teach people to recognise and manage EWS for mania and depression can be provided in individual or group setting, with or without family members (Bauer 2006; Cardoso 2015; Husain 2017; Madigan 2012). They can be delivered digitally or face‐to‐face (Depp 2015; Lobban 2017). The psychological therapies are generally recommended as an add‐on to pharmacological therapies (NICE 2014), although there may be benefits in some people with bipolar disorder who are not taking pharmacotherapy (Todd 2014). The evidence to support the efficacy or effectiveness of the above interventions in the form of psychoeducation about EWS is mostly positive but is very variable (Bond 2015), and this is also true when they are combined with cognitive behavioural therapy (CBT) or FFT (Oud 2016).

For this review, EWS of recurrence refer to initial symptoms of impending relapse or recurrence of mania or depression in a person already experiencing bipolar disorder. Therefore, it is a secondary prevention approach to this long‐term condition rather than primary prevention. This is different from prodromal symptoms, which also refer to the symptoms that patients experience prior to the first episode of their illness. In other words, the term 'prodromal symptoms' is used before someone is diagnosed with an illness and refers to early symptoms that might lead to full‐blown syndrome or disorder.

How the intervention might work

Enabling people to recognise EWS for impending relapses is a useful strategy for management of chronic relapsing physical and psychiatric illnesses. In bipolar disorder, its possible advantages include better compliance with treatment, delayed relapse, early interventions that might abort an impending episode or might lead to use of effective treatment at an early stage ensuring effective control of the episodes, in terms of severity and duration and effective management of risky behaviours (Colom 2009; Husain 2017; Morriss 2016; Scott 2007). Moreover, many people with bipolar disorder also develop effective coping mechanisms to deal with EWS of mania or depression. For example, if a patient feels that they are sleeping less and having racing thoughts as part of EWS for mania then they might decide to intentionally decrease their social activities, increase resting or sleeping time, or take a low dose of an antipsychotic drug (Perry 1999).

Many people with bipolar disorder report important psychological and functional improvements with EWS. They report feeling more hope about the future and to be proactive in reducing the harm that might arise from mania and depression so they are ready to try out activities that they might not have done such as changing job, returning to work, or going on holiday (Lobban 2017; Morriss 2016; Todd 2014). As a result, they are more likely to report feeling healthier and functional outcomes are often improved (Morriss 2016; Perry 1999). However, recognising EWS and monitoring mood without plans to address EWS may increase anxiety, fear, and rumination about relapses, and lead to increased service use (Morriss 2007). Mood monitoring alone might prolong depression episodes (Faurholt‐Jepsen 2015), while recognition of EWS for depression followed by antidepressant use is likely to be ineffective and increase the unnecessary use of antidepressants (Perry 1999).

EWS interventions are usually a part of psychoeducation programmes (these also include providing education about illness, medications, enhancement of general coping strategies and lifestyle modifications). EWS is also included in other psychological interventions developed/adapted for bipolar disorders such as CBT, FFT, and interpersonal and social rhythm therapy (IPSRT).

Why it is important to do this review

Since the last review by Morris et al in 2007 (which had included the studies published until 25 October 2005; Morriss 2007), many studies have been performed to elucidate the effectiveness of EWS recognition in bipolar disorders, either as a part of psychoeducation or other formal psychological therapies such as CBT or FFT. Most pioneering work has been by Colom and colleagues in Barcelona (Spain), who have published the effect of 21 sessions of group psychoeducation up to five years (Colom 2003; Colom 2009). Their psychoeducation programme has been adapted and tried in many other countries, with inconsistent results (Candini 2013; Cardoso 2015; de Barros Pellegrinelli 2013; Husain 2017; Maczka 2010; Morriss 2016). Other psychoeducational programmes have also been developed and tried with variable results (Bauer 2006; Castle 2010). Moreover, apart from time to relapse, time to hospitalisation, or effect on symptoms, more‐recent studies have used other useful outcome measures such socio‐occupational functioning, quality of life, and cost‐effectiveness (de Barros Pellegrinelli 2013; Morriss 2016; Smith 2011; Todd 2014). Another development is mode of delivery of psychological interventions, many web‐based interventions such as 'Beating bipolar' (Smith 2011), 'Living with bipolar' (Todd 2014), and MoodSwings (Lauder 2015) have been developed.

A few studies have shown positive effects of early recognition of EWS especially in preventing manic or depression (or both) episodes, but there are many studies that found no significant advantage of the addition of psychological therapies with recognition of early warning signs to treatment as usual (TAU) (Castle 2010; Colom 2003; Colom 2009; de Barros Pellegrinelli 2013; Morriss 2016; Perry 1999; Smith 2011).

We propose to update the previous systematic review and meta‐analysis of studies that have assessed the efficacy of recognition of EWS in bipolar disorder.

Objectives

1. To evaluate the effectiveness of EWS plus TAU or EWS plus psychological therapy versus TAU alone or psychological treatment (without EWS) independently on time to recurrence of any bipolar episode and hospitalisation, and other clinically relevant outcome measures.

2. To evaluate the effectiveness of intermittent medication used on recognition of EWS without continued mood‐stabilising medication versus TAU involving continued mood‐stabilising medication on time to recurrence of any bipolar episodes.

Methods

Criteria for considering studies for this review

Types of studies

Inclusion criteria

We will include randomised controlled trials regardless of language, country of origin, or publication status.

Exclusion criteria

We will exclude studies comparing two active interventions that both contain EWS interventions and studies providing only one session of intervention.

Types of participants

Inclusion criteria

Adults with a diagnosis of bipolar disorder or associated diagnoses (including bipolar type I and type II) based on standardised psychiatric criteria such as Research Diagnostic Criteria, Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM‐5; APA 2013), and the International Classification of Diseases 10th Revision (ICD‐10; WHO 1992). Adults will be included at any stage of the disorder and in any treatment setting. Studies including patients based on self‐reported diagnosis will also be included, but these will be removed in sensitivity analyses.

Exclusion criteria

The review will exclude studies:

1. where the primary diagnosis is another disorder (e.g. panic, eating disorders);

2. where participants are selected based on a non‐psychiatric medical condition (e.g. cancer, human immune deficiency syndrome) or the presence of another mental disorder (e.g. substance‐use disorder);

3. with 'at risk' populations who have not yet received a definite diagnosis of bipolar disorder.

Types of interventions

Experimental interventions

1. Interventions designed to systematically train people with a diagnosis of bipolar disorder to recognise early warning signs of recurrence of episodes. Each person in the study must have received at least one hour of training specifically in recognising and managing EWS. EWS interventions require detailed history taking, with or without additional techniques such as diary keeping and card sorting techniques, and a plan of action based on the EWS.

2. EWS intervention training provided online or virtually.

3. EWS interventions either as a major focus or as one component of an intervention package.

4. Individually based or group based.

5. Interventions involving family members or carers that include the participant.

6. Interventions in community, primary care, or secondary care settings.

During abstract review, in the absence of an explicit description of early warning signs training, we will consider studies where the intervention is described as cognitive, family, psychosocial, education, self‐monitoring, interpersonal, relapse prevention, self‐help, problem‐solving, counselling, or targeted/intermittent medication as potentially including EWS interventions.

Comparator interventions

Comparators are broad and will include TAU, other psychological and pharmacological interventions. TAU will mostly involve pharmacological intervention with or without informal psychoeducation about bipolar disorder. However, we acknowledge that non‐pharmacological informal intervention will vary depending on the provisions of local health services. There may be studies where both intervention and comparator include EWS, if these meet the inclusion criteria they will be dealt with as a sub‐analysis as it is unlikely that we will be a ble to compare the effect of the EWS against the other. Patients who fulfil the inclusion criteria are equally eligible to be randomised to any of the interventions we plan to compare.

Types of outcome measures

Primary outcomes

1. Time to first recurrence of any manic, depressive, and all bipolar episodes.

Secondary outcomes

1. Percentage of people who were hospitalised as a direct consequence of worsening of their illness.

2. Functioning: social functioning/interpersonal functioning, occupational functioning MRC Social Performance Scale (Hurry 1983), Social Adjustment Scale (Weissman 1978), Social and Occupational Functioning Assessment Scale (SOFAS; Goldman 1992), the Global Assessment of Functioning (GAF; American Psychiatric Association 2013), Work and Social Adjustment Scale (Mundt 2002).

3. Continuous measures of depressive symptoms on: Beck Depression Inventory (BDI; Beck 1961); Hamilton Depression Rating Scale (HDRS; Hamilton 1960); Montgomery‐Asberg Depression Rating Scale (MADRS; Montgomery 1979); Hospital Anxiety and Depression Scale (HADS; Zigmond 1983).

4. Continuous measures of manic symptoms on: Bech Rafaelson Mania Scale (Bech 1978); Young Mania Rating Scale (YMRS; Young 1978); Longitudinal Interval Follow‐up Evaluation (LIFE; Keller 1987); Internal States Scale (Bauer 1991); Altman Mania Scale (Altman 1997); Quick Inventory of Depressive Symptomatology (QIDS; Rush 2003); Patient Health Questionnaire‐9 (PHQ‐9; Kroenke 2001).

5. Satisfaction with treatment approach (as a measure of acceptance of the intervention).

6. Perceived control over the illness.

7. Recovery‐Bipolar Recovery Questionnaire (Jones 2013).

8. Quality of life, using validated measures such as EuroQol Five Dimension Scale (EQ‐5D; Rabin 2001), Short Form Six Dimension (SF‐6D; Brazier 2002) (which is derived from the 36‐item Short Form 36 (SF‐36); Ware 1992) and ICEpop CAPability measure for Adults (ICECAP‐A; Al‐Janabi 2012).

Measures of symptoms and functioning where available at three, six, 12, 18, 24, 60 months or greater after baseline.

Search methods for identification of studies

Electronic searches

An Information Specialist will conduct searches on the following bibliographic databases, using relevant subject headings (controlled vocabularies) and search syntax, appropriate to each resource:

1. Cochrane Central Register of Controlled Trials (CENTRAL; current issue) in the Cochrane Library;

2. Ovid MEDLINE (from 1946 onwards) (Appendix 1);

3. Ovid Embase (from 1974 onwards);

4. Ovid PsycINFO (all available years).

There will be no restrictions on date, language, or publication status applied to the searches.

We will search the World Health Organization International Clinical Trials Registry Platform (ICTRP; trialsearch.who.int/Default.aspx) and (clinicaltrials.gov) to identify unpublished or ongoing trials.

Searching other resources

We will search grey literature using the following sources (primarily for dissertations and theses):

1. Open Grey (opengrey.eu);

2. ProQuest Dissertations & Theses Global (search.proquest.com/pqdtglobal/dissertations/);

3. DART‐Europe E‐theses Portal (dart-europe.eu/basic-search.php);

4. Networked Digital Library of Theses and Dissertations (NDLTD) (search.ndltd.org/);

5. EThOS – the British Libraries e‐theses online service;

6. Open Access Theses and Dissertations (OATD) (oatd.org).

Other

To help identify further published, unpublished or ongoing research we will scan the reference lists of included studies and any relevant systematic reviews.

We may contact study authors to request any missing data or information, as required.

Data collection and analysis

Selection of studies

Two review authors (from SG, LRA, and OO) will independently read titles and abstracts with reference to the criteria specified under the selection criteria and decide whether to retrieve the full‐text report of the study.

During the abstract assessment stage, in the absence of an explicit description of EWS training, we will consider studies where the intervention was described as cognitive, family, psychosocial, education, self‐monitoring, interpersonal, relapse prevention, self‐help, problem‐solving, counselling, or targeted/intermittent medication as potentially including EWS interventions.

Definition of intervention: the essential elements of EWS monitoring appear to be:

1. education to increase awareness of EWS and identification of these EWS;

2. people with bipolar disorder learning to self‐monitor for EWS (e.g. mood monitoring); or carers or health professionals (or both) learning to monitor on behalf of the person with bipolar disorder;

3. early action to prevent the development of recurrence (e.g. early help‐seeking or self‐coping methods).

We will retrieve the full‐text reports of all studies that meet or potentially meet, the inclusion criteria and, if necessary, contact the study authors for further clarification to assess eligibility after title/abstract screening or by reviewing protocol or pilot studies (if available). Two review authors (LRA and OO) will independently inspect full‐text studies, and decide whether to include or exclude, noting reasons for exclusion based on a predefined list of categories for exclusion in the characteristics of excluded studies table. We will resolve any disagreements through discussion, where necessary with help from a third review author (SG). Where this is not possible, we will attempt to contact the study authors for clarification and request missing data and add these studies to the studies awaiting classification table.

Where studies have multiple publications, we will collate the reports of the same study so that each study, rather than each report, is the unit of interest for the review, and such studies have a single identifier with multiple references.

We will record the process in sufficient detail to complete a PRISMA study flow diagram.

Data extraction and management

We will design a formal data extraction form and pilot it on three studies and amend it accordingly. We will extract the following data from each included study.

1. Study design.

2. Number of participants.

3. Participant characteristics (e.g. age, gender, axis I or II comorbidity).

4. EWS intervention only or as a part of another psychological intervention.

5. Target of intervention: patient recognition of EWS, with or without carers.

6. Type of EWS intervention.

7. Type of control group.

8. Information on delivery of intervention: individual versus group format, type of provider, mode of delivery, place of delivery, duration, and frequency.

9. Outcomes (noting where primary and secondary outcomes are not measured, or measured but not reported).

10. Source of funding.

We will tabulate information regarding methods, participants, comparison groups, interventions, and outcomes. We will record data to assess potential sources of clinical heterogeneity including study design, participant characteristics, and aspects of intervention content and delivery. We will rate all included studies according to methodological quality criteria (see following section).

One review author (OO) will extract data on study characteristics, and a second review author (LRA) will check a random 20% of the entered data. Two review authors (OO and LRA) will independently extract outcome data, discussing and resolving discrepancies between them, or will consult a third review author if necessary.

Determining whether described interventions include training in recognising and managing EWS

The essential elements of early warning signs (EWS) monitoring are defined as:

1. education to increase awareness of EWS and identification of these EWS;

2. person with bipolar disorder learning to self‐monitor for EWS (e.g. mood monitoring), or carers or health professionals (or both) learning to monitor on behalf of the person with bipolar disorder;

3. early action to prevent development of recurrence (e.g. early help‐seeking or self‐coping methods).

Where the information is ambiguous in the original report, we will attempt to contact the corresponding study author, or the first or most senior author when the corresponding author is not designated. We will send a standard letter by post or e‐mail requesting the authors to clarify whether the intervention included early warning signs training, and if yes, whether the intervention was systematically employed with all participants. We will send a further letter if the author does not respond within one month to the initial letter or email. If there is no response to the second mailing, we will exclude the study from the analysis.

All studies will be categorised by:

1. target of intervention:

   1. interventions targeted at participant's own recognition of EWS;

   2. interventions targeted at recognition of EWS by carers or health professionals;

   3. interventions targeted at both participant and carer;

2. individual or group treatment;

3. focus of intervention: EWS was primary focus of intervention or one component of a multifaceted intervention;

4. intervention as an addition or replacement for TAU:

   1. intervention targeting EWS plus TAU versus TAU not including psychological treatment;

   2. EWS intervention plus TAU versus another psychological treatment without EWS intervention plus TAU;

   3. EWS intervention with intermittent medication use in presence of EWS without continued mood‐stabilising medication versus TAU using continued mood‐stabilising medication;

5. EWS intervention involved only help‐seeking from others, or EWS intervention involved other self‐directed coping;

6. RCT involved participants with euthymic bipolar disorder or participants with bipolar disorder who were in a depressive episode;

7. RCT included or excluded participants with bipolar disorder with other psychiatric comorbidities.

Assessment of risk of bias in included studies

Two review authors (OO, ALR) will independently assess risk of bias using Cochrane's RoB 1 tool for assessing risk of bias (Higgins 2011). We will consider the following domains.

1. Sequence generation: was the allocation sequence adequately generated?

2. Allocation concealment: was allocation adequately concealed?

3. Blinding of outcome assessors for each main outcome or class of outcomes: was knowledge of the allocated treatment adequately prevented from outcome assessors during the study?

4. Incomplete outcome data for each main outcome or class of outcomes: were incomplete outcome data adequately addressed?

5. Selective outcome reporting: are reports of the study free of any suggestion of selective outcome reporting?

6. Other sources of bias: was the study apparently free of other problems that could put it at high risk of bias? Additional items to be included are therapist qualifications, treatment fidelity, and researcher allegiance/conflict of interest.

We will make a judgement on the risk of bias for each domain within and across studies, based on the following three categories.

1. Low risk of bias.

2. Unclear risk of bias.

3. High risk of bias.

We will resolve any disagreement by discussion or by consulting two review authors (SG, RM).

For cluster‐RCTs, we will assess the risk of bias by considering recruitment bias, baseline imbalance, loss of cluster, incorrect analysis, and comparability with individual RCTs.

We will record the level of risk of bias in both the review text and risk of bias summary figures.

Measures of treatment effect

The treatment outcomes relevant to this study are measured as time to recurrence of episodes, measures of symptoms, and outcomes presented as dichotomous (hospitalisation/no hospitalisation) or continuous (means and standard deviations on rating scales) data.

Where trial authors have analysed the data using a Cox proportional hazards model (or similar regression model for time to event data), we will summarise time‐to‐recurrence data as hazard ratios (HRs) and their 95% confidence intervals (CI). If HRs and their standard errors (SE) are not reported in the publication, where possible we will estimate these statistics using standard methods (Tierney 2007).

We will calculate binary outcomes using risk ratios (RR) and 95% CI, as odds ratios are often difficult to interpret. In addition, we will calculate the number needed to treat for an additional beneficial outcome (NNTB) for the primary outcome from the risk difference (www.nntonline.net/visualrx/) (Cates 2002; Deeks 2001).

We will calculate continuous outcomes as standardised mean differences (SMD) with 95% CI as we expect a variety of outcome measures to be used across included studies.

Unit of analysis issues

Cluster‐randomised controlled trials

Should we identify any cluster‐RCTs for inclusion, we will include them provided we can undertake proper adjustment for intracluster correlation, as described in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2021).

Cross‐over trials

Due to the risk of carry‐over effects in cross‐over trials, we will only include data from the first phase before cross‐over of treatments between arms.

Studies with multiple treatment groups

Where studies have additional arms that are not psychological therapies, we will only include the data relating to the therapy and one control arm in the review. However, we will list all treatment arms in the 'Characteristics of included studies' table, even if they are not used in the review.

If a study has more than two intervention arms that meet the inclusion criteria, for example two psychological therapies and a control arm, we will combine these intervention arms into one group. For dichotomous outcomes, we will sum sample sizes and number of events across intervention groups. For continuous outcomes, we will use standard formulae to combine means and standard deviations (Cochrane Handbook for Systematic Reviews of Interventions; Higgins 2021).

For subgroup analysis three (EWS intervention delivered to participant only or participant and carer or health professional), it will not be possible to combine intervention groups. Therefore, we will divide the shared control groups approximately evenly between the two comparisons.

Dealing with missing data

When possible, we will calculate missing standard deviations from reported SEs, P values, or CIs using the methods described in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2021). We will approach trial authors to obtain other types of missing data. Where reported, we will include intention‐to‐treat (ITT) data in the analyses where authors have used appropriate methods for handling missing data (e.g. multiple imputation, statistical models allowing for missing data). Where ITT analyses are not reported, we will contact authors to request these results. Where trial data include a combination of observed case and imputed missing data, we will conduct sensitivity analyses including observed‐case data only.

Assessment of heterogeneity

1. Clinical heterogeneity

We will inspect study characteristics and participant characteristics to identify trials that clearly deviate from what would be expected. Such occurrences will be taken into account in the interpretation of the review results.

2. Methodological heterogeneity

We will inspect the methodology of all included studies to identify trials that clearly deviate from what would be expected. Such occurrences will be taken into account in the interpretation of the review results. In this review, we would expect heterogeneity due to mode of delivery (online versus face‐to‐face, individual versus group interventions, or EWS interventions alone versus EWS interventions with other psychological interventions), intensity and duration of EWS interventions and early or late interventions (in terms of duration of illness). These variables would be the important effect modifiers.

3. Statistical heterogeneity

3.1 Visual inspection

We will visually inspect data from individual studies and any graphs derived from these data to explore the possibility of statistical heterogeneity.

3.2 Using the I² statistic

We will investigate 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 heterogeneity rather than to chance (Higgins 2021). The importance of the observed value of an I² statistic value depends on:

1. magnitude and direction of effects; and

2. strength of evidence for heterogeneity (e.g. P value from Chi² test, or a CI for the I² statistic).

For example, we will interpret an I² statistic estimate greater than or equal to 50% accompanied by a statistically significant Chi² test statistic as evidence of substantial levels of heterogeneity (Higgins 2021).

We will further investigate heterogeneity through subgroup analyses and sensitivity analyses.

4. Transitivity assumption (consistency between direct and indirect evidence) in standard network meta‐analysis

The standard network meta‐analysis model described below (see data synthesis section) makes a transitivity assumption between 'direct evidence' and 'indirect evidence' (see HIggins 2021, chapter 11). For example:

effect of B versus C = (effect of A versus C) ‐ (effect of A versus B)

Therefore trials with indirect evidence comparing treatments B versus C (through the common comparator A) can inform the B versus C comparison, along with trials that directly compare treatment B versus C (direct evidence). We will assess inconsistency between direct and indirect sources of evidence in several steps. First, we will assess distribution of potential effect modifiers across treatment comparisons based on an examination of participant, intervention and methodological characteristics. Second, we will conducted a global test of inconsistency using the design‐by‐treatment interaction model (Higgins 2012).Third, where there is evidence of potential inconsistency, we will investigate this using side splitting in netmeta in R (Rücker 2020).For each network, we assumed a common between‐trial heterogeneity variance of the relative treatment effects for everytreatment comparison.

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 7 of the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2021). We are aware that funnel plots may be useful in investigating reporting biases but are of limited power to detect small‐study effects. We plan to use funnel plots, except for outcomes where there are fewer than 10 studies included for each outcome, or where all studies are of similar sizes.

Data synthesis

We will undertake meta‐analyses only where this is meaningful, that is, if the treatments, participants, and the underlying clinical question are sufficiently similar for pooling to make sense. We will describe skewed data reported as medians and interquartile ranges narratively.

Pairwise meta‐analyses

Meta‐analyses will focus on the three objectives of this review.

1. Intervention targeted at recognition of EWS plus TAU versus TAU.

2. Intervention targeted at recognition of EWS plus psychological therapy versus TAU.

3. Intervention targeted at recognition of EWS coupled with intermittent medication use without continued mood‐stabilising medication versus TAU using continued mood‐stabilising medication.

We will use random‐effects models for all analyses, given the anticipated heterogeneity. We will perform analyses using Review Manager Web (Review Manager Web 2020). We will analyse time‐to‐recurrence data by entering log HR and their SEs into generic inverse variance meta‐analyses and report findings as HRs and 95% CIs. We will synthesise binary outcomes using RRs and 95% CIs and continuous outcomes as SMDs and 95% CIs.

Component network meta‐analyses

EWS are complex interventions that often include multiple interacting intervention components. Therefore we will use component network meta‐analyses to assess whether certain intervention components or combinations of components are associated with greater effectiveness. Using the R package netmeta, we propose to conduct the following analyses:

• standard network meta‐analysis model that will compare the following interventions (where available) including EWS intervention, EWS plus psychological intervention, EWS plus intermittent medication use, TAU

• additive model: each intervention component will be included in the analyses separately

• two‐way interaction model: where sufficient data are available, we will include pairs of intervention components to explore interactions between them

Model selection will be based on tests of goodness of fit (comparing the goodness of fit between standard and additive network meta‐analysis models and between standard and interaction models). For further details see Rücker 2020.

Subgroup analysis and investigation of heterogeneity

We plan to perform the following subgroup analyses.

1. EWS as primary focus of the intervention or as part of another intervention

2. EWS intervention delivered individually versus group intervention.

3.Online versus face to face interventions.

4. Early or late stage of bipolar disorder (five years or less versus more than five years' duration of illness).

Sensitivity analysis

We plan to perform the following sensitivity analyses.

1. Including only studies at low (or low and unclear) risk of bias.

2. Studies with imputed data versus studies with observed case data.

Summary of findings and assessment of the certainty of the evidence

We will use the CINeMA (Confidence in Network Meta‐Analysis) approach to interpret the certainty of evidence and to develop summary of findings tables (Nikolakopoulou 2020). These tables provide outcome‐specific information concerning the overall certainty of evidence from each included study 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. We will create one table for each of the comparisons under our Objectives should data be available.

We aim to select the following main outcomes (judged to be the most clinically meaningful) for inclusion in the summary of findings table.

1. Time to first recurrence of any manic, depressive, and all bipolar episode.

2. Percentage of people who were hospitalised as a direct consequence of worsening of their illness.

3. Functioning.

4. Quality of life.

What's new

Date	Event	Description
18 June 2025	New citation required and major changes	Changes to author team ‐ new authors added. Important changes to methodology ‐ added methods to conduct component network meta‐analyses.

History

Protocol first published: Issue 11, 2022

Appendices

Appendix 1. MEDLINE search

Ovid MEDLINE(R) ALL <1946 to date of search>

1 "bipolar and related disorders"/ or bipolar disorder/ 42609
2 Mania/ 163
3 (Bipolar I or Bipolar II).tw,kf. 4487
4 bipolar.ti. or (bipolar adj1 (patients or inpatients or outpatients)).ab,kf. 28959
5 ((bipolar and (disorder* or depress* or mania* or manic* or hypomani* or rapid cycling or schizoaffective)) or BD‐NOS or BDNOS).tw,kf. 40872
6 cyclothymi*.hw,tw,kf. 1418
7 or/1‐6 64324
8 Recurrence/ 191534
9 Secondary Prevention/ 21782
10 ((few* or frequen* or limit* or prevent* or reduc* or risk* or stop* or symptom* or time or number or data or rate* or status or symptom*) adj5 (recur* or reoccur* or re‐occur* or relaps* or recrudesce* or new episode* or another episode or re‐emerg* or resurg* or re‐surg* or reappear* or re‐appear* or flare‐up)).tw,kf. 248677
11 relapsing.tw,kf. 31323
12 relaps*.ab. /freq=3 34664
13 ((future or repeat* or subsequent*) adj2 episod*).tw,kf. 5691
14 prodrom* symptom*.tw,kf. 1611
15 ((remain* or stay*) adj (free or well or without)).tw,kf. 14276
16 ((sustain* or remain* or stay*) adj2 (recovery or remission or remitted or remittance)).tw,kf. 7334
17 Patient Compliance/ 59627
18 ((drug* or medicat* or pharma* or psychopharma* or psychotropic* or treatment or patient) adj2 (complian* or comply)).tw,kf. 22173
19 ((subsyndrom* or sub‐syndrom* or prodrom*) and monitor*).tw,kf. 497
20 (monitor* adj3 (signs or symptoms or triggers)).tw,kf. 5152
21 (early adj1 (detect* or recogni* or interven* or symptom* or warning*)).tw,kf,hw. 161562
22 patient education as topic/ 87553
23 (patient* adj2 (educat* or train*)).tw,kf,hw. 123385
24 ((carer? or partner* or spouse or husband* or wife or wives or family or families) adj2 (educat* or intervention* or support* or therap* or train*)).tw,kf,hw. 47672
25 (psychoeducat* or psycho‐educat*).tw,kf. 7442
26 (cognitive adj2 (intervention* or therap* or psychotherap* or train*)).tw,kf,hw. 44643
27 or/8‐26 886792
28 (7 and 27) 6108
29 controlled clinical trial.pt. 94499
30 randomized controlled trial.pt. 548945
31 clinical trials as topic/ 197934
32 (randomi#e* or randomi#ation or randomi#ing).tw,kf. 714544
33 (RCT or "at random" or (random* adj3 (administ* or allocat* or assign* or class* or cluster or crossover or cross‐over or control* or determine* or divide* or division or distribut* or expose* or fashion or number* or place* or pragmatic or quasi or recruit* or split or subsitut* or treat*))).tw,kf. 631672
34 placebo.tw,kf. 229034
35 trial.ti. 250389
36 ((single or double or triple or treble) adj2 (blind* or mask* or dummy)).tw,kf. 185132
37 double‐blind method/ or random allocation/ or single‐blind method/ 293818
38 ((treatment‐as‐usual or (treatment* adj2 usual) or (standard adj2 care) or (standard adj2 treatment) or (standard adj2 control?) or (routine adj2 care) or (usual adj2 care) or (usual adj2 medication*) or TAU or CAU or waitlist* or wait* list* or WLC) adj5 (group* or control* or arm? or compar*)).tw,kf. 45180
39 (control* adj3 group*).ab. 591549
40 or/29‐39 1885431
41 exp animals/ not humans.sh. 4908071
42 (40 not 41) 1637024
43 (28 and 42) 1455

Contributions of authors

All authors made substantial contributions to the conception or design of the work and contributed to drafting the protocol.

SG led the drafting process for this updated protocol, which builds on the earlier protocol and review (Morriss 2004; Morriss 2007).

RS developed the search strategy.

All authors approved the final amended version to be published and agreed to be accountable for all aspects of the work

Sources of support

Internal sources

• Tees, Esk and Wear Valleys NHS Foundation Trust, UK

  Mental Health Trust

• University of Nottingham, UK

  Author institution support

External sources

• National Institute for Health Research (NIHR), UK

  Richard Morriss is an NIHR Senior Investigator, Co‐theme lead Nottingham NIHR Biomedical Research Centre Mental Health and Technology Theme, Adult Mental Health Theme lead NIHR MindTech MedTech and in‐Vitro Centre, NIHR Applied Research Collaboration Mental Health and Well‐being Theme Lead.

• Wellcome Trust, UK

  Wellcome Trust Doctoral Fellowship to LAW

• NIHR, UK

  NIHR Academic Clinical Fellowship to LAW

Declarations of interest

Sumeet Gupta: no relevant interests; Consultant Psychiatrist, Tees, Esk and Wear Valleys NHS Foundation Trust, Harrogate, UK.

Laurence Astill Wright: no relevant interests, Clinical Research Fellow at University of Nottingham, Nottingham, UK.

Ogba Onwuchekwa: no relevant interests; Senior Registrar working at Minster Ward, Foss Park Hospital, York, UK.

Lakshmi Ramana Alla: no relevant interests; Consultant Forensic Psychiatrist at Stockton Hall Hospital, Priory Group, York, UK.

Richard K Morriss: Magstim Inc. (Grant/Contract), Electromedical Products Inc, Texas (Grant / Contract), Novartis (Independent Contractor – Data And Safety Monitoring); published randomized controlled trials, clinical guidelines, and meta‐analyses, Universities of Manchester, Liverpool, and Nottingham (funding sources: North West Regional Health Authority, Medical Research Council, National Institute for Health Research).

Rachel Steele: none known.

Nitin Gupta: no relevant interests; Consultant Psychiatrist, Gupta Mind Healing and Counselling Centre, Chandigarh, India.

Nicholas Meader: no relevant interests; Deputy Coordinating Editor for Cochrane Common Mental Disorders (closed in March 2023) and was not involved in the editorial decision‐making for this protocol.

Ryan PW Kenny: none known.

Amended to reflect a change in scope (see 'What's new')