Adjunctive Psychotherapy for Bipolar Disorder: A Systematic Review and Component Network Meta-analysis | Bipolar and Related Disorders | JAMA Psychiatry | JAMA Network
[Skip to Navigation]
Sign In
Figure 1.  PRISMA Flow Diagram
PRISMA Flow Diagram
Figure 2.  Geometry of Networks for Treatment-Level Comparisons
Geometry of Networks for Treatment-Level Comparisons

Network structure for the 4 outcomes examined in this article. Nodes denote treatments, and lines denote trials performing the corresponding treatment comparison. The size of a node is proportional to the number of studies that included the corresponding treatment. The thickness of the lines corresponds to the number of studies performing each comparison (also indicated by the number on each line). CBT indicates cognitive behavioral therapy; IPSRT, interpersonal and social rhythm therapy; and TAU, treatment as usual.

Table 1.  Pairwise and Network Meta-analysis for Treatment-Level Comparisons of Recurrence (Odds Ratios)a
Pairwise and Network Meta-analysis for Treatment-Level Comparisons of Recurrence (Odds Ratios)a
Table 2.  Pairwise and Network Meta-analysis for Treatment-Level Comparisons of Depression End-Point Scores (Standardized Mean Differences)a
Pairwise and Network Meta-analysis for Treatment-Level Comparisons of Depression End-Point Scores (Standardized Mean Differences)a
Table 3.  Data on Incremental Effectiveness of Treatment Components Regarding Outcome Variablesa
Data on Incremental Effectiveness of Treatment Components Regarding Outcome Variablesa
Supplement.

eTable 1. Online Search Strategies

eTable 2. Psychosocial Interventions for Bipolar Disorder: Criteria for Classification Nodes

eTable 3. Study Outcome Instruments

eTable 4. Details of Component Network Meta-Analysis

eTable 5A. Study Characteristics: Treatments and Effect Sizes

eTable 5B. Study Characteristics: Risk of Bias

eTable 6. Grouping of Experimental Treatments vs Control Treatments

eTable 7. Assessing Inconsistency for Primary Analysis of Recurrence

eTable 8. Network Meta-Analysis at Treatment-Level for Recurrence Rates: SUCRA Rankings of Treatments (Primary Analysis)

eTable 9. Sensitivity Analysis: NMA at Treatment Level in a Complete Case Analysis

eTable 10. SUCRA Rankings for Complete Case Analysis

eTable 11. Network Meta-Analysis at Treatment-Level for Depression: SUCRA Rankings of Treatments (Primary Analysis)

eTable 13. NMA at Treatment Level for Depression End Points: Imputing Missing Information on Number of Patients

eTable 14. SUCRA Rankings: NMA With Imputation of Missing Cases

eTable 15. Pairwise and Network Meta-Analysis for Treatment-Level Comparisons on Mania End Point Scores (Standardized Mean Differences)

eTable 16. Network Meta-Analysis at Treatment-Level for Mania: SUCRA Rankings of Treatments (Primary Analysis)

eTable 17. Assessing Inconsistency (Mania Outcomes)

eTable 18. NMA at Treatment Level for Mania End Points: Imputing Missing Information on Number of Patients

eTable 19. Sensitivity Analysis for Recurrence: Fitting the Component NMA Model in a Complete Case Analysis

eTable 20. Sensitivity Analysis for Recurrence: Fitting the Component NMA Model in a Frequentist Setting

eTable 21. Sensitivity Analysis for Depression: Fitting the Component NMA Model in a Complete Case Analysis

eTable 22. Sensitivity Analysis for Depression: Fitting the Component NMA Model in a Frequentist Setting

eTable 23. Sensitivity Analysis for Mania: Fitting the Component NMA Model in a Complete Case Analysis

eTable 24. Sensitivity Analysis for Mania: Fitting the Component NMA Model in a Frequentist Setting

eTable 25. Pairwise and Network Meta-Analyses for Acceptability (Study Attrition)

eTable 26. Network Meta-Analysis at Treatment-Level for Acceptability: SUCRA Rankings of Treatments (Primary Analysis)

eTable 27. Assessing Inconsistency for Comparisons on Acceptability

eTable 28. Sensitivity Analysis for Acceptability: Fitting Component NMA in a Frequentist Setting

eFigure 1. Locations and Publication Years of Included Trials

eFigure 2. Contour-Enhanced Funnel Plot Meta-Analysis Comparing Experimental vs Control Treatments: Assessment of Small Study Effects and Publication Bias (Primary Analysis)

eFigure 3. Network Meta-Analysis at Treatment-Level (Depression): Assessment of Small Study Effects and Publication Bias (Primary Analysis)

eFigure 4. Network Geometry for Depression Sensitivity/Imputation Analysis

eFigure 5. Network Meta-Analysis at Treatment-Level (Mania): Assessment of Small Study Effects and Publication Bias (Primary Analysis)

eFigure 6. Network Geometry for Mania: Imputation of Missing Subject Data

eFigure 7. Analyses of Small Study Effects and Publication Bias: Comparison of all Experimental Treatments vs All Control Treatments on Acceptability

eReferences.

1.
Goodwin  GM, Haddad  PM, Ferrier  IN,  et al.  Evidence-based guidelines for treating bipolar disorder: Revised third edition recommendations from the British Association for Psychopharmacology.   J Psychopharmacol. 2016;30(6):495-553. doi:10.1177/0269881116636545PubMedGoogle ScholarCrossref
2.
Geddes  JR, Miklowitz  DJ.  Treatment of bipolar disorder.   Lancet. 2013;381(9878):1672-1682. doi:10.1016/S0140-6736(13)60857-0 PubMedGoogle ScholarCrossref
3.
Salcedo  S, Gold  AK, Sheikh  S,  et al.  Empirically supported psychosocial interventions for bipolar disorder: current state of the research.   J Affect Disord. 2016;201:203-214. doi:10.1016/j.jad.2016.05.018 PubMedGoogle ScholarCrossref
4.
Miklowitz  DJ, Otto  MW, Frank  E,  et al.  Psychosocial treatments for bipolar depression: a 1-year randomized trial from the Systematic Treatment Enhancement Program.   Arch Gen Psychiatry. 2007;64(4):419-426. doi:10.1001/archpsyc.64.4.419 PubMedGoogle ScholarCrossref
5.
Cipriani  A, Higgins  JP, Geddes  JR, Salanti  G.  Conceptual and technical challenges in network meta-analysis.   Ann Intern Med. 2013;159(2):130-137. doi:10.7326/0003-4819-159-2-201307160-00008 PubMedGoogle ScholarCrossref
6.
Efthimiou  O, Debray  TPA, van Valkenhoef  G,  et al; GetReal Methods Review Group.  GetReal in network meta-analysis: a review of the methodology.   Res Synth Methods. 2016;7(3):236-263. doi:10.1002/jrsm.1195 PubMedGoogle ScholarCrossref
7.
Welton  NJ, Caldwell  DM, Adamopoulos  E, Vedhara  K.  Mixed treatment comparison meta-analysis of complex interventions: psychological interventions in coronary heart disease.   Am J Epidemiol. 2009;169(9):1158-1165. doi:10.1093/aje/kwp014 PubMedGoogle ScholarCrossref
8.
Pompoli  A, Furukawa  TA, Efthimiou  O, Imai  H, Tajika  A, Salanti  G.  Dismantling cognitive-behaviour therapy for panic disorder: a systematic review and component network meta-analysis.   Psychol Med. 2018;48(12):1945-1953. doi:10.1017/S0033291717003919 PubMedGoogle ScholarCrossref
9.
Chatterton  ML, Stockings  E, Berk  M, Barendregt  JJ, Carter  R, Mihalopoulos  C.  Psychosocial therapies for the adjunctive treatment of bipolar disorder in adults: network meta-analysis.   Br J Psychiatry. 2017;210(5):333-341. doi:10.1192/bjp.bp.116.195321 PubMedGoogle ScholarCrossref
10.
Miklowitz  DJ, Cipriani  A, Goodwin  GM.  Network meta-analysis: drawing conclusions regarding trials of psychosocial interventions for bipolar disorder.   Br J Psychiatry. 2017;211(6):334-336. doi:10.1192/bjp.bp.117.202739 PubMedGoogle ScholarCrossref
11.
Moher  D, Liberati  A, Tetzlaff  J, Altman  DG; PRISMA Group.  Preferred reporting items for systematic reviews and meta-analyses: the PRISMA statement.   PLoS Med. 2009;6(7):e1000097. doi:10.1371/journal.pmed.1000097 PubMedGoogle Scholar
12.
Hutton  B, Salanti  G, Caldwell  DM,  et al.  The PRISMA extension statement for reporting of systematic reviews incorporating network meta-analyses of health care interventions: checklist and explanations.   Ann Intern Med. 2015;162(11):777-784. doi:10.7326/M14-2385 PubMedGoogle ScholarCrossref
13.
Cipriani  A, Miklowitz  D, McMahon  H,  et al. Comparative efficacy and acceptability of psychological interventions in the long term treatment of bipolar disorder: a network meta-analysis. PROSPERO 2015: International Prospective Register of Systematic Reviews. CRD42015016085. Accessed August 27, 2020. https://www.crd.york.ac.uk/prospero/display_record.php?RecordID=16085
14.
Higgins  JPT, Thomas  J. Cochrane Handbook for Systematic Reviews of Interventions, version 6.0. Cochrane Reviews. 2019. Accessed May 21, 2020. https://training.cochrane.org/handbook/current
15.
American Psychiatric Association.  Diagnostic and Statistical Manual of Mental Disorders. 3rd ed. American Psychiatric Association; 1980.
16.
American Psychiatric Association.  Diagnostic and Statistical Manual of Mental Disorders. 3rd ed, revised. American Psychiatric Association; 1987.
17.
American Psychiatric Association. Diagnostic and Statistical Manual of Mental Disorders. 4th ed. American Psychiatric Association; 1994.
18.
American Psychiatric Association. Diagnostic and Statistical Manual of Mental Disorders. 4th ed, text revision. American Psychiatric Association; 2000.
19.
American Psychiatric Association.  Diagnostic and Statistical Manual of Mental Disorders. 5th ed. American Psychiatric Association; 2013.
20.
World Health Organization.  The ICD-10 Classification of Mental and Behavioural Disorders: Clinical Descriptions and Diagnostic Guidelines. 1st ed. World Health Organization; 1992.
21.
Spitzer  RL, Endicott  J, Robins  E.  Research Diagnostic Criteria: rationale and reliability.   Arch Gen Psychiatry. 1978;35(6):773-782. doi:10.1001/archpsyc.1978.01770300115013 PubMedGoogle ScholarCrossref
22.
Miklowitz  DJ, Goodwin  GM, Bauer  MS, Geddes  JR.  Common and specific elements of psychosocial treatments for bipolar disorder: a survey of clinicians participating in randomized trials.   J Psychiatr Pract. 2008;14(2):77-85. doi:10.1097/01.pra.0000314314.94791.c9 PubMedGoogle ScholarCrossref
23.
Hedges  L, Olkin  I.  Statistical Methods for Meta-Analysis. New York: Academic Press; 1985.
24.
Balduzzi  S, Rücker  G, Schwarzer  G.  How to perform a meta-analysis with R: a practical tutorial.   Evid Based Ment Health. 2019;22(4):153-160. https://ebmh.bmj.com/content/22/4/153. doi:10.1136/ebmental-2019-300117PubMedGoogle ScholarCrossref
25.
Rücker  G, Krahn  U, König  J, Efthimiou  O, Schwarzer  G. netmeta: Network meta-analysis using frequentist methods. R package version 1.2-1. 2020. Accessed July 20, 2020. https://cran.r-project.org/web/packages/netmeta/index.html
26.
Salanti  G, Ades  AE, Ioannidis  JP.  Graphical methods and numerical summaries for presenting results from multiple-treatment meta-analysis: an overview and tutorial.   J Clin Epidemiol. 2011;64(2):163-171. doi:10.1016/j.jclinepi.2010.03.016 PubMedGoogle ScholarCrossref
27.
Altman  DG, Bland  JM.  Detecting skewness from summary information.   BMJ. 1996;313(7066):1200. doi:10.1136/bmj.313.7066.1200 PubMedGoogle ScholarCrossref
28.
Furukawa  TA, Barbui  C, Cipriani  A, Brambilla  P, Watanabe  N.  Imputing missing standard deviations in meta-analyses can provide accurate results.   J Clin Epidemiol. 2006;59(1):7-10. doi:10.1016/j.jclinepi.2005.06.006 PubMedGoogle ScholarCrossref
29.
Rhodes  KM, Turner  RM, Higgins  JP.  Predictive distributions were developed for the extent of heterogeneity in meta-analyses of continuous outcome data.   J Clin Epidemiol. 2015;68(1):52-60. doi:10.1016/j.jclinepi.2014.08.012 PubMedGoogle ScholarCrossref
30.
Turner  RM, Jackson  D, Wei  Y, Thompson  SG, Higgins  JPT.  Predictive distributions for between-study heterogeneity and simple methods for their application in Bayesian meta-analysis.   Stat Med. 2015;34(6):984-998. doi:10.1002/sim.6381 PubMedGoogle ScholarCrossref
31.
White  IR, Barrett  JK, Jackson  D, Higgins  JPT.  Consistency and inconsistency in network meta-analysis: model estimation using multivariate meta-regression.   Res Synth Methods. 2012;3(2):111-125. doi:10.1002/jrsm.1045 PubMedGoogle ScholarCrossref
32.
König  J, Krahn  U, Binder  H.  Visualizing the flow of evidence in network meta-analysis and characterizing mixed treatment comparisons.   Stat Med. 2013;32(30):5414-5429. doi:10.1002/sim.6001 PubMedGoogle ScholarCrossref
33.
Peters  JL, Sutton  AJ, Jones  DR, Abrams  KR, Rushton  L.  Comparison of two methods to detect publication bias in meta-analysis.   JAMA. 2006;295(6):676-680. doi:10.1001/jama.295.6.676 PubMedGoogle ScholarCrossref
34.
Lunn  DJ, Thomas  A, Best  N, Spiegelhalter  D.  WinBUGS—a Bayesian modelling framework: concepts, structure, and extensibility.   Stat Comput. 2000;10:325-337. doi:10.1023/A:1008929526011 Google ScholarCrossref
35.
Miklowitz  DJ, Axelson  DA, Birmaher  B,  et al.  Family-focused treatment for adolescents with bipolar disorder: results of a 2-year randomized trial.   Arch Gen Psychiatry. 2008;65(9):1053-1061. doi:10.1001/archpsyc.65.9.1053 PubMedGoogle ScholarCrossref
36.
Miklowitz  DJ, Schneck  CD, George  EL,  et al.  Pharmacotherapy and family-focused treatment for adolescents with bipolar I and II disorders: a 2-year randomized trial.   Am J Psychiatry. 2014;171(6):658-667. doi:10.1176/appi.ajp.2014.13081130 PubMedGoogle ScholarCrossref
37.
Rea  MM, Tompson  MC, Miklowitz  DJ, Goldstein  MJ, Hwang  S, Mintz  J.  Family-focused treatment versus individual treatment for bipolar disorder: results of a randomized clinical trial.   J Consult Clin Psychol. 2003;71(3):482-492. doi:10.1037/0022-006X.71.3.482 PubMedGoogle ScholarCrossref
38.
Cuijpers  P, Noma  H, Karyotaki  E, Vinkers  CH, Cipriani  A, Furukawa  TA.  A network meta-analysis of the effects of psychotherapies, pharmacotherapies and their combination in the treatment of adult depression.   World Psychiatry. 2020;19(1):92-107. doi:10.1002/wps.20701 PubMedGoogle ScholarCrossref
39.
Keller  MB, Lavori  PW, Friedman  B,  et al.  The longitudinal interval follow-up evaluation: a comprehensive method for assessing outcome in prospective longitudinal studies.   Arch Gen Psychiatry. 1987;44(6):540-548. doi:10.1001/archpsyc.1987.01800180050009 PubMedGoogle ScholarCrossref
40.
Post  RM, Denicoff  KD, Leverich  GS,  et al.  Morbidity in 258 bipolar outpatients followed for 1 year with daily prospective ratings on the NIMH life chart method.   J Clin Psychiatry. 2003;64(6):680-690. doi:10.4088/JCP.v64n0610 PubMedGoogle ScholarCrossref
41.
Pilhatsch  M, Glenn  T, Rasgon  N,  et al.  Regularity of self-reported daily dosage of mood stabilizers and antipsychotics in patients with bipolar disorder.   Int J Bipolar Disord. 2018;6(1):10-18. doi:10.1186/s40345-018-0118-8 PubMedGoogle ScholarCrossref
42.
Miklowitz  DJ, Price  J, Holmes  EA,  et al.  Facilitated integrated mood management for adults with bipolar disorder.   Bipolar Disord. 2012;14(2):185-197. doi:10.1111/j.1399-5618.2012.00998.x PubMedGoogle ScholarCrossref
43.
Fristad  MA, MacPherson  HA.  Evidence-based psychosocial treatments for child and adolescent bipolar spectrum disorders.   J Clin Child Adolesc Psychol. 2014;43(3):339-355. doi:10.1080/15374416.2013.822309 PubMedGoogle ScholarCrossref
44.
Yatham  LN, Kennedy  SH, Parikh  SV,  et al.  Canadian Network for Mood and Anxiety Treatments (CANMAT) and International Society for Bipolar Disorders (ISBD) 2018 guidelines for the management of patients with bipolar disorder.   Bipolar Disord. 2018;20(2):97-170. doi:10.1111/bdi.12609 PubMedGoogle ScholarCrossref
45.
Clark  DM.  Implementing NICE guidelines for the psychological treatment of depression and anxiety disorders: the IAPT experience.   Int Rev Psychiatry. 2011;23(4):318-327. doi:10.3109/09540261.2011.606803 PubMedGoogle ScholarCrossref
46.
Depressive and Bipolar Support Alliance. Preferences for the treatment of bipolar disorder survey. DBSA Survey Center. Published May 2017. Accessed May 21, 2020. https://www.dbsalliance.org/wp-content/uploads/2019/02/DBSA_Survey_Center_BPD_Treatment_Preferences_FINAL.pdf
Limit 200 characters
Limit 25 characters
Conflicts of Interest Disclosure

Identify all potential conflicts of interest that might be relevant to your comment.

Conflicts of interest comprise financial interests, activities, and relationships within the past 3 years including but not limited to employment, affiliation, grants or funding, consultancies, honoraria or payment, speaker's bureaus, stock ownership or options, expert testimony, royalties, donation of medical equipment, or patents planned, pending, or issued.

Err on the side of full disclosure.

If you have no conflicts of interest, check "No potential conflicts of interest" in the box below. The information will be posted with your response.

Not all submitted comments are published. Please see our commenting policy for details.

Limit 140 characters
Limit 3600 characters or approximately 600 words
    Original Investigation
    October 14, 2020

    Adjunctive Psychotherapy for Bipolar Disorder: A Systematic Review and Component Network Meta-analysis

    Author Affiliations
    • 1Department of Psychiatry and Biobehavioral Sciences, David Geffen School of Medicine, University of California, Los Angeles
    • 2Department of Psychiatry, Warneford Hospital, University of Oxford, Oxford, United Kingdom
    • 3Institute of Social and Preventive Medicine, University of Bern, Bern, Switzerland
    • 4Department of Health Promotion and Human Behavior, Kyoto University Graduate School of Medicine/School of Public Health, Kyoto, Japan
    • 5Department of Clinical Epidemiology, Kyoto University Graduate School of Medicine/School of Public Health, Kyoto, Japan
    • 6Brain and Mind Centre, The University of Sydney, Sydney, Australia
    • 7Institute of Neuroscience, Newcastle University, Newcastle, United Kingdom
    • 8Oxford Health National Health Service Foundation Trust, Warneford Hospital, Oxford, United Kingdom
    JAMA Psychiatry. 2021;78(2):141-150. doi:10.1001/jamapsychiatry.2020.2993
    Key Points

    Question  Which psychosocial interventions are associated with an improved course and medium-term outcomes of bipolar disorder?

    Findings  In a systematic review and network meta-analysis of 39 randomized clinical trials of adjunctive psychotherapy, there was evidence that family, cognitive behavioral, and psychoeducational therapies were associated with reduced episode recurrence vs treatment as usual in individuals with bipolar disorder. Cognitive behavioral therapy was associated with greater stabilization of residual symptoms of depression compared with treatment as usual.

    Meaning  This study suggests that outpatients with bipolar disorder receiving pharmacotherapy should also be offered psychosocial treatments that emphasize illness management strategies and enhance coping skills; delivering these components in family or group format may be especially advantageous.

    Abstract

    Importance  Several psychotherapy protocols have been evaluated as adjuncts to pharmacotherapy for patients with bipolar disorder, but little is known about their comparative effectiveness.

    Objective  To use systematic review and network meta-analysis to compare the association of using manualized psychotherapies and therapy components with reducing recurrences and stabilizing symptoms in patients with bipolar disorder.

    Data Sources  Major bibliographic databases (MEDLINE, PsychInfo, and Cochrane Library of Systematic Reviews) and trial registries were searched from inception to June 1, 2019, for randomized clinical trials of psychotherapy for bipolar disorder.

    Study Selection  Of 3255 abstracts, 39 randomized clinical trials were identified that compared pharmacotherapy plus manualized psychotherapy (cognitive behavioral therapy, family or conjoint therapy, interpersonal therapy, or psychoeducational therapy) with pharmacotherapy plus a control intervention (eg, supportive therapy or treatment as usual) for patients with bipolar disorder.

    Data Extraction and Synthesis  Binary outcomes (recurrence and study retention) were compared across treatments using odds ratios (ORs). For depression or mania severity scores, data were pooled and compared across treatments using standardized mean differences (SMDs) (Hedges-adjusted g using weighted pooled SDs). In component network meta-analyses, the incremental effectiveness of 13 specific therapy components was examined.

    Main Outcomes and Measures  The primary outcome was illness recurrence. Secondary outcomes were depressive and manic symptoms at 12 months and acceptability of treatment (study retention).

    Results  A total of 39 randomized clinical trials with 3863 participants (2247 of 3693 [60.8%] with data on sex were female; mean [SD] age, 36.5 [8.2] years) were identified. Across 20 two-group trials that provided usable information, manualized treatments were associated with lower recurrence rates than control treatments (OR, 0.56; 95% CI, 0.43-0.74). Psychoeducation with guided practice of illness management skills in a family or group format was associated with reducing recurrences vs the same strategies in an individual format (OR, 0.12; 95% CI, 0.02-0.94). Cognitive behavioral therapy (SMD, −0.32; 95% CI, −0.64 to −0.01) and, with less certainty, family or conjoint therapy (SMD, −0.46; 95% CI, −1.01 to 0.08) and interpersonal therapy (SMD, –0.46; 95% CI, −1.07 to 0.15) were associated with stabilizing depressive symptoms compared with treatment as usual. Higher study retention was associated with family or conjoint therapy (OR, 0.46; 95% CI, 0.26-0.82) and brief psychoeducation (OR, 0.44; 95% CI, 0.23-0.85) compared with standard psychoeducation.

    Conclusions and Relevance  This study suggests that outpatients with bipolar disorder may benefit from skills-based psychosocial interventions combined with pharmacotherapy. Conclusions are tempered by heterogeneity in populations, treatment duration, and follow-up.

    Introduction

    There is increasing recognition that pharmacotherapy alone cannot prevent recurrences of bipolar disorder or fully alleviate postepisode symptoms or functional impairment.1 Psychotherapy, when provided at all, is viewed as an adjunctive treatment.2 Evidence from randomized clinical trials (RCTs) indicates that combining pharmacotherapy with manualized psychotherapies—including cognitive behavioral therapy (CBT), family-focused therapy, interpersonal and social rhythm therapy (IPSRT), and group psychoeducation—is more effective than pharmacotherapy alone in stabilizing symptoms and reducing recurrences among outpatients with bipolar disorder.1,3,4 The comparative effectiveness of these approaches, however, has received scant attention.

    Unfortunately, even carefully delivered psychosocial interventions are not effective for many patients with bipolar disorder, suggesting the importance of examining which therapy components (strategies, techniques, or formats) are essential for clinical effectiveness. The existing trial literature on bipolar disorder is heterogeneous, with few direct comparisons of modalities and a lack of clarity as to which treatments are effective in acute stabilization and which are effective in recurrence prevention. Network meta-analysis (NMA) aims to synthesize evidence across clinical trials so that all treatment options can be compared with each other, increasing the precision of effect estimates between interventions.5,6 Component NMA, an extension of standard NMA, allows for the decomposition of complex interventions and estimates the effectiveness of their constituent components.7,8

    There has been only 1 NMA of psychotherapies for bipolar disorder, to our knowledge; this NMA concluded that only caregiver-focused interventions (without patient participation) were associated with a significant reduction in the risk of recurrences in patients.9 One editorial described a number of limitations of this analysis, notably that the primary conclusion was based on only 2 trials and that other key effectiveness trials (often featuring different interventions) were excluded.10

    We performed a systematic review of the RCT literature and used NMA and component NMA to examine (1) whether any psychosocial interventions are associated with reduced episode recurrence and stabilizing symptoms in patients with bipolar disorder and (2) the outcomes of different therapy components (eg, provision of illness information or guided practice of illness management strategies or coping skills) or formats (ie, individual, family, or group). The primary outcome was illness recurrence, with secondary analyses of posttreatment depressive and manic symptom severity and acceptability (study attrition for any reason).

    Methods
    Search Methods for Identification of Studies

    We followed the specifications of the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA)11 statement and its extension for NMAs.12 We focused on RCTs comparing an experimental psychotherapy plus pharmacotherapy with another form of psychotherapy plus pharmacotherapy or treatment as usual (TAU, defined as pharmacotherapy with routine monitoring visits) for adults or adolescents with bipolar disorder. The literature was searched from inception to June 1, 2019 (PROSPERO registration number CRD42015016085).13 We searched MEDLINE, PsycInfo, Cochrane Library of Systematic Reviews, ClinicalTrials.gov, EU Clinical Trials Register, ISRCTN Registry, World Health Organization International Clinical Trials Registry, and the Australian New Zealand Clinical Trial Registry (search terms in eTable 1 in the Supplement). Trials were also located through searching reference lists of published and unpublished articles, conference proceedings, systematic reviews, and a prior NMA.9 No language restrictions were applied.

    Study Eligibility Criteria

    The included RCTs focused on the alleviation of mood symptoms and/or the prevention of recurrences. We included only studies in which participants received medications per standard clinical practice, as operationalized by the original investigators. We excluded quasi-randomized trials and studies that examined psychotherapy as 1 element of multicomponent systematic care (eTable 1 in the Supplement). Two independent raters (including R.M.) selected studies and extracted the outcome data, as well as information on potential effect modifiers: age, sex, bipolar subtype, blinding of outcome assessors, and year of publication. The Cochrane tool was used to classify risk of bias.14 When discrepancies occurred, a third rater (A.C.) was consulted, or the original study authors were contacted.

    Trial Participants

    Participants were outpatients or inpatients aged 12 years or older and of both sexes, with a primary diagnosis of bipolar disorder I or II or unspecified bipolar disorder according to DSM-III or DSM-III-R, DSM-IV or DSM-IV-TR, DSM-5, International Statistical Classification of Diseases and Related Health Problems, Tenth Revision, or Research Diagnostic Criteria.15-21 Participants could be in any clinical state or have any comorbid medical or psychiatric disorder at randomization.

    Types of Interventions

    Psychosocial interventions could be implemented with individuals, families, or groups and had to include in-person contact between the patient and a trained therapist (digital [ie, texting via smartphones], internet, or telephone formats were excluded). Two raters (D.J.M. and J.S.) classified each active intervention group as belonging to 1 of the following nodes (see operational definitions in eTable 2 in the Supplement): traditional CBT with cognitive restructuring, behavioral activation, and problem-solving (including dialectical behavior therapy components such as mindfulness and distress tolerance); standard-length psychoeducation (≥6 group or individual sessions); IPSRT; family or conjoint therapy (including family-focused therapy, multifamily groups, or caregiver-only groups); or functional remediation. The raters classified the control groups as the following: brief psychoeducation (≤3 sessions), supportive therapy, or TAU. Based on descriptions of each intervention and a prior review,22 2 raters (D.J.M. and J.S.) defined 18 therapy components (eg, cognitive restructuring and group format) and classified each trial group according to the presence or absence of each component.

    Types of Outcomes

    The primary outcome was the proportion of participants who experienced an episode recurrence of any type (depressed, manic, or mixed) during the first 12 months after randomization (or by the trial end point if follow-up was of shorter duration). When these data were unavailable, we imputed recurrence proportions at 12 months from survival curves or used the nearest available reported data. Secondary outcomes were depressive or manic symptoms at a common end point (12 months or the nearest available time point) and acceptability (study retention). Binary outcomes were compared using odds ratios (ORs). When studies used different rating scales to assess symptom severity (eTable 3 in the Supplement), data were pooled using standardized mean difference (SMD) scores (Hedges-adjusted g scores using weighted pooled SDs).23

    Data Synthesis

    Using the statistical package meta in R, version 4.0.2 (R Foundation for Statistical Computing),24 we performed standard pairwise meta-analyses using a random-effects model for (1) the omnibus comparison of all experimental interventions (CBT, IPSRT, family or conjoint therapy, and standard psychoeducation) with all control interventions (brief psychoeducation, supportive therapy, and TAU) and (2) direct comparisons of any 2 interventions occurring in at least 2 studies. We then performed random-effect NMA to synthesize evidence from the entire network by integrating direct and indirect estimates for each comparison into a single summary effect, using the netmeta command in R.25 League-tables with summary relative effect sizes (SMDs or ORs) for each possible pair of interventions were supplemented by an intervention effectiveness hierarchy using surface under the cumulative ranking curves.26

    The primary analyses of recurrences included only participants who completed the study. For sensitivity analyses, we assumed that patients who withdrew had a recurring outcome. For the secondary analyses of continuous symptom end points, we excluded studies that did not report the number of patients who completed the study. In sensitivity analyses, we imputed the number of patients who completed the study by multiplying the number randomized in the relevant study by the mean retention rate across the remaining studies. When P values, t values, 95% CIs, or SEs were reported, we calculated or imputed SDs.27,28 We assessed statistical heterogeneity in the entire network by comparing the magnitude of the heterogeneity variance parameter (τ2) from the NMA models with its empirical distribution.6,29,30 We evaluated the presence of inconsistency globally using the design-by-treatment test31 and locally using the back-calculation method comparing direct and indirect estimates.32 We assessed small study effects, including publication bias, by examining asymmetry in the funnel plots of all interventions vs TAU.33

    Component NMAs

    We performed component NMAs in which the effect of each composite therapy was expressed as the sum of the effects of its constituent components.7,8 The models estimate component-specific incremental odds ratios (iORs) for binary outcomes and incremental SMDs (iSMDs) for continuous outcomes. All component NMA models were conducted in a bayesian setting with analyses performed using OpenBUGS34 and uncertainty expressed by 95% credible intervals (CrIs). In sensitivity analyses, we repeated all component NMAs in a frequentist setting using the discomb command in netmeta (eTable 4 in the Supplement).

    Results
    Trial Characteristics

    An initial search retrieved 3255 abstracts, with further examination of 103 published articles (Figure 1). Of these, 25 articles met full inclusion criteria. A further search using narrower search terms generated 1074 abstracts, from which 38 articles were examined and 14 met inclusion criteria. Thus, 39 RCTs were included, 36 enrolling adults and 3 enrolling adolescents. Of the 39 trials, 37 compared 2 intervention groups (eTable 5A in the Supplement). Of the 3863 trial participants (mean [SD] age, 36.5 [8.2] years), sex was reported for 3693 (2247 female participants [60.8%] and 1446 male participants [39.2%]). Most articles did not report racial or ethnic sample compositions. The geographical distribution and the number of published articles per 5-year interval are displayed in eFigure 1 in the Supplement. Risk of bias was rated as low in 17 studies, moderate in 19, and high in 3 (eTable 5B in the Supplement).

    Transitivity is assumed when it is equally likely that any patient in a network of treatment comparisons could have been given any of the treatments in the network. In the studies, blinding of outcome assessors, publication year, and proportion of patients with bipolar disorder I or II were balanced across comparisons. However, 2 of 4 comparisons of family or conjoint therapy with standard or brief psychoeducation were conducted for adolescents35,36 and 1 such comparison was conducted for young adults.37 Otherwise, the assumption of transitivity appeared to be valid.

    Treatment-Level Comparisons on Prevention of Recurrences

    Across 20 two-group trials that provided usable information, experimental interventions were associated with a lower probability of recurrence than control interventions (OR, 0.56; 95% CI, 0.43-0.74) (eTable 6 in the Supplement). Statistical heterogeneity for this pairwise meta-analysis was τ2 = 0.16. Findings were nearly identical when we assumed recurrences for participants who withdrew. There was weak evidence of small study effects or publication biases (eFigure 2 in the Supplement).

    A total of 24 trials contained usable information for comparing associations of 2 therapy modalities with recurrence rates (Figure 2A; eTable 5A in the Supplement). There was no evidence of design-by-treatment inconsistency or local inconsistency (eTable 7 in the Supplement). Statistical heterogeneity (τ2) was 0.35, similar to the values of τ2 in Cochrane RCT reviews.30 When examined by a standard NMA, family or conjoint therapy (OR, 0.30; 95% CI, 0.17-0.53), CBT (OR, 0.52; 95% CI, 0.34-0.79), standard psychoeducation (OR, 0.52; 95% CI, 0.32-0.84), and brief psychoeducation (OR, 0.34; 95% CI, 0.16-0.74) were associated with a more favorable outcome compared with TAU (Table 1). The highest surface under the cumulative ranking curve ranking was obtained for family or conjoint therapy (95%) (eTable 8 in the Supplement). A sensitivity analysis in which missing data were imputed yielded similar results (eTables 9 and 10 in the Supplement).

    Treatment-Level Comparisons of Depressive or Manic Symptoms

    Twenty-one trials provided information on 12-month depression symptoms (Figure 2B). The common τ2 was 0.10, close to the empirical median.29 In treatment-level comparisons of depression end points, evidence suggested that CBT (SMD, –0.32; 95% CI, –0.64 to –0.01) and, with less certainty, family or conjoint therapy (SMD, –0.46; 95% CI, –1.01 to 0.08) and IPSRT (SMD, –0.46; 95% CI, –1.07 to 0.15) were associated with significantly improved outcomes compared with TAU (Table 2; eTable 11 in the Supplement). There was modest evidence for small study effects and publication bias favoring experimental interventions (eFigure 3 in the Supplement) and weak evidence for local inconsistency (eTable 12 in the Supplement). When imputing the number of patents who completed the study, sensitivity analyses favored the effects of CBT, IPSRT, and family or conjoint therapy vs TAU (eFigure 4, eTables 13 and 14 in the Supplement).

    Nineteen trials provided data for pairwise comparisons of treatment effects on 12-month mania symptoms (Figure 2C). When examined by a standard NMA, evidence suggested that CBT (SMD, –0.32; 95% CI, –0.65 to 0.01), psychoeducation (SMD, –0.31; 95% CI, –0.70 to 0.08), and family or conjoint therapy (SMD, –0.35; 95% CI, –0.86 to 0.17) were associated with significantly improved outcomes compared with TAU, although with substantial uncertainty (eTables 15 and 16 in the Supplement). There was minimal evidence of inconsistency and no evidence of small study effects (eFigure 5 and eTable 17 in the Supplement). The findings did not change when we recalculated the NMA with imputation of missing data (eFigure 6 and eTable 18 in the Supplement).

    Component Analysis

    Of 18 therapy components, 13 occurred in more than 2 intervention groups (Table 3). Interrater reliability on the presence or absence of these components in each group exceeded 80%. The specific components that were associated with lower recurrence rates were delivery of treatment in a family format (iOR, 0.16; 95% CrI, 0.02-1.22) and encouraging patients to monitor prodromal symptoms (iOR, 0.22; 95% CrI, 0.04-1.35). The estimated heterogeneity was τ2 = 0.19 (95% CrI, 0.00-1.42). Using the component NMA model, we estimated that psychoeducation with guided skill practice and self-monitoring delivered in a family or group format is more effective in reducing recurrences than the same 2 treatment components delivered in an individual format (OR, 0.12; 95% CrI, 0.02-0.94).

    Cognitive restructuring (iSMD, –1.26; 95% CrI, –2.10 to –0.35), regulating daily rhythms (iSMD, –0.78; 95% CrI, –1.28 to –0.24), and, with less certainty, communication training (iSMD, –0.84; 95% CrI, –1.81 to 0.23) were the most potent components for reducing severity of depression (Table 3). The combination of these 3 components was estimated to be more effective than TAU (iSMD, –2.89; 95% CrI, –4.70 to –0.91). The least potent components (albeit with greater uncertainty) were behavioral activation (iSMD, 0.92; 95% CrI, 0.11 to 1.71) and individual therapy format (iSMD, 1.01; 95% CrI, –0.12 to 2.07). Analogously, cognitive restructuring (iSMD, –1.00; 95% CrI, –2.15 to 0.16) and regulating daily rhythms (iSMD, –0.42; 95% CrI, –1.08 to 0.28) were associated with greater stabilization of manic symptoms, whereas behavioral activation was associated with lesser stabilization (iSMD, 0.98; 95% CrI, –0.10 to 2.03) (Table 3). Fitting the component models in complete case analyses or in a frequentist setting yielded similar results for recurrence, depression, and mania (eTables 19-24 in the Supplement).

    Treatment Acceptability

    A total of 36 trials provided data on acceptability (retention rate) (Figure 2D). There were no overall differences between experimental and control interventions with regard to acceptability (OR, 1.02; 95% CI, 0.75-1.39) and no evidence of small study effects or publication biases (eFigure 7 in the Supplement). In the NMA model comparing specific interventions (heterogeneity τ2 = 0.07), there was evidence that family or conjoint therapy (OR, 0.46; 95% CI, 0.26-0.82) and brief psychoeducation (OR, 0.44; 95% CI, 0.23-0.85) were associated with higher retention rates than standard-length (ie, ≥6 sessions) courses of psychoeducation (eTables 25 and 26 in the Supplement). There was less-certain evidence that family or conjoint therapy was associated with higher retention rates than CBT (OR, 0.64; 95% CI, 0.36-1.11) and TAU (OR, 0.61; 95% CI, 0.36-1.04), and no evidence of inconsistency (eTable 27 in the Supplement). In the component NMA, family format appeared to be the only component associated with a lower rate of attrition (iOR, 0.39; 95% CrI, 0.15-1.11; Table 3; eTable 28 in the Supplement).

    Discussion

    In this NMA of 39 RCTs of patients with bipolar disorder, we confirm previous findings that pharmacotherapy in combination with manualized psychotherapy is associated with a more effective reduction in recurrences (OR, 0.56; 95% CI, 0.43-0.74) than pharmacotherapy with TAU. In addition, we demonstrate that family or conjoint therapy, CBT, and standard psychoeducation, with their focus on active skill training (eg, monitoring of prodromal symptoms), were each associated with a lower probability of recurrence than TAU. Family or conjoint therapy and brief psychoeducation were associated with lower attrition rates than standard psychoeducation. There was little evidence of inconsistency or small study effects in the networks, and heterogeneity was within expected ranges for all outcomes.29,30 Sensitivity analyses did not alter the findings. Our findings are similar to the NMA results of Cuijpers et al,38 who concluded that combining psychotherapy with pharmacotherapy is the best option for stabilizing episodes and preventing recurrences of major depressive disorder.

    Cognitive behavioral therapy, IPSRT, and family or conjoint therapy appeared to have comparable outcomes for depression stabilization, although there was greater precision for the effect size estimates for CBT, which was evaluated in a larger number of trials. Few trials recruited patients in an acute mood episode, suggesting that our findings concerning symptom end points pertain mainly to stabilization of interepisode symptoms. We share the conclusion of Chatterton et al9 that psychoeducation and CBT appear to be effective in stabilizing residual manic symptoms, and we add that regulating daily rhythms is more useful than behavioral activation in such circumstances. We stress that there is no evidence that cognitive restructuring is beneficial in acute mania. We do not share the conclusion reached by Chatterton et al9 that caregiver-focused psychoeducation (without patients present) is the most effective approach to recurrence prevention or that no intervention is effective in stabilizing depressive symptoms. In our analysis, family therapy, CBT, and group psychoeducation—all modalities that include patients as active participants—were associated with significantly improved outcomes compared with TAU with regard to recurrence prevention and depression stabilization.

    What do our findings suggest about treating outpatients with bipolar disorder? When the goals center on prevention of recurrences, patients should be engaged in family or group psychoeducation with guided skills training and active tasks to enhance coping skills (eg, monitoring and managing prodromal symptoms) rather than being passive recipients of didactic education. When the immediate goal is recovery from moderately severe depressive or manic symptoms, cognitive restructuring, regulating daily rhythms, and communication training may be associated with stabilization. It is unclear whether CBT techniques work best in an individual format; in this NMA, family and group formats were more closely associated with depression improvement than individual formats.

    Limitations

    The analyses were limited by small sample sizes and sparsely connected networks. Many of our conclusions are based on indirect rather than direct comparisons (eg, IPSRT vs TAU). We cannot draw conclusions about the comparative effectiveness of psychotherapy for patients with severe illness vs those with moderate or mild illness. Recruitment of acutely ill patients with bipolar disorder, particularly those with mania, into psychotherapy trials is neither feasible nor ethically justifiable unless therapy is initiated after stabilization of symptoms with pharmacotherapy. Randomized clinical trials that are adequately powered to examine interactions between treatments and levels of illness severity require considerable commitments of time and expense.

    In the 39 RCTs, the durations of therapy (3-12 months) and the follow-up intervals (6-60 months) were variable. Thus, we were unable to evaluate whether the associations of experimental interventions or their constituent components with outcomes were enduring or whether treatments would need to be revisited with booster sessions over time. Also, there was considerable variability in choice of assessment instruments. Consensus regarding a common assessment battery for drug or psychotherapy trials of bipolar disorder would enable cross-study findings to be compared more reliably. We recommend inclusion of clinician-rated assessments of weekly symptoms and mood polarity,39,40 as well as patient-rated electronic diaries of mood and medication use.41,42

    Most of the RCTs used study retention as a proxy for treatment acceptability. Other potentially informative definitions of acceptability, such as frequency of session attendance, medication adherence, or patients’ ratings of helpfulness, were rarely reported. We recommend that future trials include these more nuanced measures to offer insight into this important aspect of clinical management.

    Although our results suggest the effectiveness of family interventions, several of the relevant trials concerned adolescent or young-adult populations. Younger patients are more likely than older patients to have family supports,43 and adult patients without family members are less likely to recover from depressive episodes and more likely to be hospitalized than those with family members.4 For patients without family supports, group psychoeducation or CBT and IPSRT are recommended.

    None of the RCTs examined the comparative contributions of psychotherapy and pharmacotherapy to outcomes. Systematic examinations of whether pharmacotherapy regimens can be simplified (without loss of effectiveness) when combined with specific psychosocial protocols would be of considerable value. Last, although we were rigorous in our literature search, we cannot exclude the possibility that we failed to identify relevant published trials.

    Conclusions

    Despite these limitations, there is enough evidence from this NMA and other systematic reviews1-3,9 to conclude that health care systems should offer combinations of evidence-based pharmacotherapy and psychotherapy to outpatients with bipolar disorder. This recommendation is in line with the guidelines of the Canadian Network for Mood and Anxiety Treatments44 and the UK Improving Access to Psychological Therapies program.45 Implementing this recommendation will require a considerable reallocation of mental health resources. In a US survey of 1627 adults with bipolar disorder, 1448 (89%) were receiving medications for bipolar disorder, but only 820 (50.4%) were also receiving psychotherapy.46

    Psychotherapy in these RCTs was delivered by well-trained clinicians who received supervision throughout the trials. As with suboptimal pharmacotherapy, if the quality of therapy is substandard, benefits will not hold up in clinical practice. The widespread availability of evidence-based psychotherapies for bipolar disorder in community care will depend on the development of well-scaled methods for disseminating clinician training and monitoring treatment fidelity. Systematic studies of telehealth and internet-based psychotherapy may enhance progress toward these objectives. Finally, there is a need to evaluate the most effective combinations of therapy components for patients with different illness presentations treated across public and private settings. All of these strategies are required to translate the benefits of adjunctive psychotherapies into effective personalized treatments for individuals with bipolar disorder.

    Back to top
    Article Information

    Accepted for Publication: July 29, 2020.

    Corresponding Author: David J. Miklowitz, PhD, Department of Psychiatry and Biobehavioral Sciences, UCLA Semel Institute, David Geffen School of Medicine at UCLA, 760 Westwood Plaza, Room A8-256, Los Angeles, CA 90024 (dmiklowitz@mednet.ucla.edu).

    Published Online: October 14, 2020. doi:10.1001/jamapsychiatry.2020.2993

    Author Contributions: Drs Miklowitz and Efthimiou had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

    Concept and design: Miklowitz, Efthimiou, Furukawa, Scott, Geddes, Cipriani.

    Acquisition, analysis, or interpretation of data: All authors.

    Drafting of the manuscript: Miklowitz, Efthimiou, Furukawa, Scott, Geddes.

    Critical revision of the manuscript for important intellectual content: All authors.

    Statistical analysis: Miklowitz, Efthimiou, Scott, Geddes, Cipriani.

    Administrative, technical, or material support: Miklowitz, McLaren, Cipriani.

    Supervision: Miklowitz, Geddes, Cipriani.

    Conflict of Interest Disclosures: Dr Miklowitz reported receiving research support from the National Institute of Mental Health, the Danny Alberts Foundation, the Attias Family Foundation, the Carl and Roberta Deutsch Foundation, the Kayne Family Foundation, AIM for Mental Health, and the Max Gray Fund; receiving book royalties from Guilford Press and John Wiley and Sons; and serving as principal investigator on 4 of the included trials in this meta-analysis. Dr Furukawa reported receiving personal fees from Mitsubishi-Tanabe, MSD, and Shionogi; and receiving a grant from Mitsubishi-Tanabe, outside the submitted work; having a patent (2018-177688) pending; and being Diplomate of the Academy of Cognitive Therapy. Dr Scott reported being a distinguished founding fellow of the International Academy of CBT; receiving grant funding from the UK Medical Research Council and the UK National Institutes for Health Research for Patient Benefit programme; and serving as principal investigator on 2 of the included trials in this analysis. Dr Cipriani reported receiving research and consultancy fees from INCiPiT (Italian Network for Paediatric Trials), CARIPLO Foundation, and Angelini Pharma. No other disclosures were reported.

    Funding/Support: This study was supported in part by grant BRC-1215-20005 from the National Institute for Health Research (NIHR) Oxford Health Biomedical Research Centre (Drs Geddes and Cipriani). Dr Cipriani is also supported by the NIHR Oxford Cognitive Health Clinical Research Facility, by the NIHR Oxford and Thames Valley Applied Research Collaboration, and by an NIHR Research Professorship (grant RP-2017-08-ST2-006). Dr Miklowitz is supported by grant R34-MH117200 from the National Institute of Mental Health. Dr Efthimiou was supported by project grant No. 180083 from the Swiss National Science Foundation.

    Role of the Funder/Sponsor: The funding sources had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

    Disclaimer: The views expressed in this article are those of the authors and not necessarily those of the UK National Health Service, the National Institute for Health Research, the UK Department of Health, or the National Institute of Mental Health.

    Additional Contributions: Sarah Stockton, BA, and Hannah McMahon, BSc, MSc, Department of Psychiatry, University of Oxford, provided administrative support. Guy Goodwin, MD, Department of Psychiatry, University of Oxford; Georgia Salanti, MD, Institute of Social and Preventive Medicine, University of Bern; and Anna Chaimani, PhD, Center of Research in Epidemiology and Statistics, University of Paris, provided consultation on study procedures and statistical analyses. They were not compensated for their contributions.

    Additional Information: Data from this network meta-analysis are available by application to the first author Dr Miklowitz (dmiklowitz@mednet.ucla.edu) after evaluation of the request by the members of the research team according to an established procedure.

    References
    1.
    Goodwin  GM, Haddad  PM, Ferrier  IN,  et al.  Evidence-based guidelines for treating bipolar disorder: Revised third edition recommendations from the British Association for Psychopharmacology.   J Psychopharmacol. 2016;30(6):495-553. doi:10.1177/0269881116636545PubMedGoogle ScholarCrossref
    2.
    Geddes  JR, Miklowitz  DJ.  Treatment of bipolar disorder.   Lancet. 2013;381(9878):1672-1682. doi:10.1016/S0140-6736(13)60857-0 PubMedGoogle ScholarCrossref
    3.
    Salcedo  S, Gold  AK, Sheikh  S,  et al.  Empirically supported psychosocial interventions for bipolar disorder: current state of the research.   J Affect Disord. 2016;201:203-214. doi:10.1016/j.jad.2016.05.018 PubMedGoogle ScholarCrossref
    4.
    Miklowitz  DJ, Otto  MW, Frank  E,  et al.  Psychosocial treatments for bipolar depression: a 1-year randomized trial from the Systematic Treatment Enhancement Program.   Arch Gen Psychiatry. 2007;64(4):419-426. doi:10.1001/archpsyc.64.4.419 PubMedGoogle ScholarCrossref
    5.
    Cipriani  A, Higgins  JP, Geddes  JR, Salanti  G.  Conceptual and technical challenges in network meta-analysis.   Ann Intern Med. 2013;159(2):130-137. doi:10.7326/0003-4819-159-2-201307160-00008 PubMedGoogle ScholarCrossref
    6.
    Efthimiou  O, Debray  TPA, van Valkenhoef  G,  et al; GetReal Methods Review Group.  GetReal in network meta-analysis: a review of the methodology.   Res Synth Methods. 2016;7(3):236-263. doi:10.1002/jrsm.1195 PubMedGoogle ScholarCrossref
    7.
    Welton  NJ, Caldwell  DM, Adamopoulos  E, Vedhara  K.  Mixed treatment comparison meta-analysis of complex interventions: psychological interventions in coronary heart disease.   Am J Epidemiol. 2009;169(9):1158-1165. doi:10.1093/aje/kwp014 PubMedGoogle ScholarCrossref
    8.
    Pompoli  A, Furukawa  TA, Efthimiou  O, Imai  H, Tajika  A, Salanti  G.  Dismantling cognitive-behaviour therapy for panic disorder: a systematic review and component network meta-analysis.   Psychol Med. 2018;48(12):1945-1953. doi:10.1017/S0033291717003919 PubMedGoogle ScholarCrossref
    9.
    Chatterton  ML, Stockings  E, Berk  M, Barendregt  JJ, Carter  R, Mihalopoulos  C.  Psychosocial therapies for the adjunctive treatment of bipolar disorder in adults: network meta-analysis.   Br J Psychiatry. 2017;210(5):333-341. doi:10.1192/bjp.bp.116.195321 PubMedGoogle ScholarCrossref
    10.
    Miklowitz  DJ, Cipriani  A, Goodwin  GM.  Network meta-analysis: drawing conclusions regarding trials of psychosocial interventions for bipolar disorder.   Br J Psychiatry. 2017;211(6):334-336. doi:10.1192/bjp.bp.117.202739 PubMedGoogle ScholarCrossref
    11.
    Moher  D, Liberati  A, Tetzlaff  J, Altman  DG; PRISMA Group.  Preferred reporting items for systematic reviews and meta-analyses: the PRISMA statement.   PLoS Med. 2009;6(7):e1000097. doi:10.1371/journal.pmed.1000097 PubMedGoogle Scholar
    12.
    Hutton  B, Salanti  G, Caldwell  DM,  et al.  The PRISMA extension statement for reporting of systematic reviews incorporating network meta-analyses of health care interventions: checklist and explanations.   Ann Intern Med. 2015;162(11):777-784. doi:10.7326/M14-2385 PubMedGoogle ScholarCrossref
    13.
    Cipriani  A, Miklowitz  D, McMahon  H,  et al. Comparative efficacy and acceptability of psychological interventions in the long term treatment of bipolar disorder: a network meta-analysis. PROSPERO 2015: International Prospective Register of Systematic Reviews. CRD42015016085. Accessed August 27, 2020. https://www.crd.york.ac.uk/prospero/display_record.php?RecordID=16085
    14.
    Higgins  JPT, Thomas  J. Cochrane Handbook for Systematic Reviews of Interventions, version 6.0. Cochrane Reviews. 2019. Accessed May 21, 2020. https://training.cochrane.org/handbook/current
    15.
    American Psychiatric Association.  Diagnostic and Statistical Manual of Mental Disorders. 3rd ed. American Psychiatric Association; 1980.
    16.
    American Psychiatric Association.  Diagnostic and Statistical Manual of Mental Disorders. 3rd ed, revised. American Psychiatric Association; 1987.
    17.
    American Psychiatric Association. Diagnostic and Statistical Manual of Mental Disorders. 4th ed. American Psychiatric Association; 1994.
    18.
    American Psychiatric Association. Diagnostic and Statistical Manual of Mental Disorders. 4th ed, text revision. American Psychiatric Association; 2000.
    19.
    American Psychiatric Association.  Diagnostic and Statistical Manual of Mental Disorders. 5th ed. American Psychiatric Association; 2013.
    20.
    World Health Organization.  The ICD-10 Classification of Mental and Behavioural Disorders: Clinical Descriptions and Diagnostic Guidelines. 1st ed. World Health Organization; 1992.
    21.
    Spitzer  RL, Endicott  J, Robins  E.  Research Diagnostic Criteria: rationale and reliability.   Arch Gen Psychiatry. 1978;35(6):773-782. doi:10.1001/archpsyc.1978.01770300115013 PubMedGoogle ScholarCrossref
    22.
    Miklowitz  DJ, Goodwin  GM, Bauer  MS, Geddes  JR.  Common and specific elements of psychosocial treatments for bipolar disorder: a survey of clinicians participating in randomized trials.   J Psychiatr Pract. 2008;14(2):77-85. doi:10.1097/01.pra.0000314314.94791.c9 PubMedGoogle ScholarCrossref
    23.
    Hedges  L, Olkin  I.  Statistical Methods for Meta-Analysis. New York: Academic Press; 1985.
    24.
    Balduzzi  S, Rücker  G, Schwarzer  G.  How to perform a meta-analysis with R: a practical tutorial.   Evid Based Ment Health. 2019;22(4):153-160. https://ebmh.bmj.com/content/22/4/153. doi:10.1136/ebmental-2019-300117PubMedGoogle ScholarCrossref
    25.
    Rücker  G, Krahn  U, König  J, Efthimiou  O, Schwarzer  G. netmeta: Network meta-analysis using frequentist methods. R package version 1.2-1. 2020. Accessed July 20, 2020. https://cran.r-project.org/web/packages/netmeta/index.html
    26.
    Salanti  G, Ades  AE, Ioannidis  JP.  Graphical methods and numerical summaries for presenting results from multiple-treatment meta-analysis: an overview and tutorial.   J Clin Epidemiol. 2011;64(2):163-171. doi:10.1016/j.jclinepi.2010.03.016 PubMedGoogle ScholarCrossref
    27.
    Altman  DG, Bland  JM.  Detecting skewness from summary information.   BMJ. 1996;313(7066):1200. doi:10.1136/bmj.313.7066.1200 PubMedGoogle ScholarCrossref
    28.
    Furukawa  TA, Barbui  C, Cipriani  A, Brambilla  P, Watanabe  N.  Imputing missing standard deviations in meta-analyses can provide accurate results.   J Clin Epidemiol. 2006;59(1):7-10. doi:10.1016/j.jclinepi.2005.06.006 PubMedGoogle ScholarCrossref
    29.
    Rhodes  KM, Turner  RM, Higgins  JP.  Predictive distributions were developed for the extent of heterogeneity in meta-analyses of continuous outcome data.   J Clin Epidemiol. 2015;68(1):52-60. doi:10.1016/j.jclinepi.2014.08.012 PubMedGoogle ScholarCrossref
    30.
    Turner  RM, Jackson  D, Wei  Y, Thompson  SG, Higgins  JPT.  Predictive distributions for between-study heterogeneity and simple methods for their application in Bayesian meta-analysis.   Stat Med. 2015;34(6):984-998. doi:10.1002/sim.6381 PubMedGoogle ScholarCrossref
    31.
    White  IR, Barrett  JK, Jackson  D, Higgins  JPT.  Consistency and inconsistency in network meta-analysis: model estimation using multivariate meta-regression.   Res Synth Methods. 2012;3(2):111-125. doi:10.1002/jrsm.1045 PubMedGoogle ScholarCrossref
    32.
    König  J, Krahn  U, Binder  H.  Visualizing the flow of evidence in network meta-analysis and characterizing mixed treatment comparisons.   Stat Med. 2013;32(30):5414-5429. doi:10.1002/sim.6001 PubMedGoogle ScholarCrossref
    33.
    Peters  JL, Sutton  AJ, Jones  DR, Abrams  KR, Rushton  L.  Comparison of two methods to detect publication bias in meta-analysis.   JAMA. 2006;295(6):676-680. doi:10.1001/jama.295.6.676 PubMedGoogle ScholarCrossref
    34.
    Lunn  DJ, Thomas  A, Best  N, Spiegelhalter  D.  WinBUGS—a Bayesian modelling framework: concepts, structure, and extensibility.   Stat Comput. 2000;10:325-337. doi:10.1023/A:1008929526011 Google ScholarCrossref
    35.
    Miklowitz  DJ, Axelson  DA, Birmaher  B,  et al.  Family-focused treatment for adolescents with bipolar disorder: results of a 2-year randomized trial.   Arch Gen Psychiatry. 2008;65(9):1053-1061. doi:10.1001/archpsyc.65.9.1053 PubMedGoogle ScholarCrossref
    36.
    Miklowitz  DJ, Schneck  CD, George  EL,  et al.  Pharmacotherapy and family-focused treatment for adolescents with bipolar I and II disorders: a 2-year randomized trial.   Am J Psychiatry. 2014;171(6):658-667. doi:10.1176/appi.ajp.2014.13081130 PubMedGoogle ScholarCrossref
    37.
    Rea  MM, Tompson  MC, Miklowitz  DJ, Goldstein  MJ, Hwang  S, Mintz  J.  Family-focused treatment versus individual treatment for bipolar disorder: results of a randomized clinical trial.   J Consult Clin Psychol. 2003;71(3):482-492. doi:10.1037/0022-006X.71.3.482 PubMedGoogle ScholarCrossref
    38.
    Cuijpers  P, Noma  H, Karyotaki  E, Vinkers  CH, Cipriani  A, Furukawa  TA.  A network meta-analysis of the effects of psychotherapies, pharmacotherapies and their combination in the treatment of adult depression.   World Psychiatry. 2020;19(1):92-107. doi:10.1002/wps.20701 PubMedGoogle ScholarCrossref
    39.
    Keller  MB, Lavori  PW, Friedman  B,  et al.  The longitudinal interval follow-up evaluation: a comprehensive method for assessing outcome in prospective longitudinal studies.   Arch Gen Psychiatry. 1987;44(6):540-548. doi:10.1001/archpsyc.1987.01800180050009 PubMedGoogle ScholarCrossref
    40.
    Post  RM, Denicoff  KD, Leverich  GS,  et al.  Morbidity in 258 bipolar outpatients followed for 1 year with daily prospective ratings on the NIMH life chart method.   J Clin Psychiatry. 2003;64(6):680-690. doi:10.4088/JCP.v64n0610 PubMedGoogle ScholarCrossref
    41.
    Pilhatsch  M, Glenn  T, Rasgon  N,  et al.  Regularity of self-reported daily dosage of mood stabilizers and antipsychotics in patients with bipolar disorder.   Int J Bipolar Disord. 2018;6(1):10-18. doi:10.1186/s40345-018-0118-8 PubMedGoogle ScholarCrossref
    42.
    Miklowitz  DJ, Price  J, Holmes  EA,  et al.  Facilitated integrated mood management for adults with bipolar disorder.   Bipolar Disord. 2012;14(2):185-197. doi:10.1111/j.1399-5618.2012.00998.x PubMedGoogle ScholarCrossref
    43.
    Fristad  MA, MacPherson  HA.  Evidence-based psychosocial treatments for child and adolescent bipolar spectrum disorders.   J Clin Child Adolesc Psychol. 2014;43(3):339-355. doi:10.1080/15374416.2013.822309 PubMedGoogle ScholarCrossref
    44.
    Yatham  LN, Kennedy  SH, Parikh  SV,  et al.  Canadian Network for Mood and Anxiety Treatments (CANMAT) and International Society for Bipolar Disorders (ISBD) 2018 guidelines for the management of patients with bipolar disorder.   Bipolar Disord. 2018;20(2):97-170. doi:10.1111/bdi.12609 PubMedGoogle ScholarCrossref
    45.
    Clark  DM.  Implementing NICE guidelines for the psychological treatment of depression and anxiety disorders: the IAPT experience.   Int Rev Psychiatry. 2011;23(4):318-327. doi:10.3109/09540261.2011.606803 PubMedGoogle ScholarCrossref
    46.
    Depressive and Bipolar Support Alliance. Preferences for the treatment of bipolar disorder survey. DBSA Survey Center. Published May 2017. Accessed May 21, 2020. https://www.dbsalliance.org/wp-content/uploads/2019/02/DBSA_Survey_Center_BPD_Treatment_Preferences_FINAL.pdf
    ×