[Skip to Content]
Sign In
Individual Sign In
Create an Account
Institutional Sign In
OpenAthens Shibboleth
[Skip to Content Landing]
Figure 1.
PRISMA Flowchart
PRISMA Flowchart

Flowchart of the inclusion of studies in the review. IPD indicates individual participant data.

Figure 2.
Study-Level Analysis Examining the Effect of the Presence Chronic Physical Conditions on the Effectiveness of Collaborative Care
Study-Level Analysis Examining the Effect of the Presence Chronic Physical Conditions on the Effectiveness of Collaborative Care

Study-level data and pooled effects across 36 comparisons. A, Studies that did not explicitly recruit patients with chronic physical conditions. B, Studies that explicitly recruited patients with chronic physical conditions; mixed-effects model used. Weights are from random-effects analysis. ML indicates maximum likelihood.

Figure 3.
Individual Participant–Level Analysis Examining the Effect of the Presence of Chronic Physical Conditions on the Effectiveness of Collaborative Care
Individual Participant–Level Analysis Examining the Effect of the Presence of Chronic Physical Conditions on the Effectiveness of Collaborative Care

Individual participant data and pooled effects across 30 comparisons. A, Studies in which participants did not have a chronic physical condition. B, Studies in which participants had a chronic physical condition. Mixed-effects model used. Weights are from random-effects analysis. REML indicates restricted maximum likelihood.

Figure 4.
Individual Participant–Level Analysis Examining the Effect of the Number of Chronic Physical Conditions on the Effectiveness of Collaborative Care
Individual Participant–Level Analysis Examining the Effect of the Number of Chronic Physical Conditions on the Effectiveness of Collaborative Care

Individual participant data and pooled effects across 30 comparisons. A, Main effect of the individual participant data analysis. B, Interaction effect of the study group and the number of physical chronic conditions. Mixed-effects model used. Weights are from random-effects analysis. REML indicates restricted maximum likelihood.

Table.  
Comparison of Studies Providing Data for the IPD Analyses and Those Not Providing Data
Comparison of Studies Providing Data for the IPD Analyses and Those Not Providing Data
1.
Murray  CJL, Vos  T, Lozano  R,  et al.  Disability-adjusted life years (DALYs) for 291 diseases and injuries in 21 regions, 1990-2010: a systematic analysis for the Global Burden of Disease Study 2010.  Lancet. 2012;380(9859):2197-2223.PubMedGoogle ScholarCrossref
2.
Whiteford  HA, Degenhardt  L, Rehm  J,  et al.  Global burden of disease attributable to mental and substance use disorders: findings from the Global Burden of Disease Study 2010.  Lancet. 2013;382(9904):1575-1586.PubMedGoogle ScholarCrossref
3.
Moussavi  S, Chatterji  S, Verdes  E, Tandon  A, Patel  V, Ustun  B.  Depression, chronic diseases, and decrements in health: results from the World Health Surveys.  Lancet. 2007;370(9590):851-858.PubMedGoogle ScholarCrossref
4.
Freedland  KE, Carney  RM.  Depression as a risk factor for adverse outcomes in coronary heart disease.  BMC Med. 2013;11:131.PubMedGoogle ScholarCrossref
5.
Carney  RM, Blumenthal  JA, Catellier  D,  et al.  Depression as a risk factor for mortality after acute myocardial infarction.  Am J Cardiol. 2003;92(11):1277-1281.PubMedGoogle ScholarCrossref
6.
Panagioti  M, Stokes  J, Esmail  A,  et al.  Multimorbidity and patient safety incidents in primary care: a systematic review and meta-analysis.  PLoS One. 2015;10(8):e0135947.PubMedGoogle ScholarCrossref
7.
Dickens  C, Katon  W, Blakemore  A,  et al.  Does depression predict the use of urgent and unscheduled care by people with long term conditions? a systematic review with meta-analysis.  J Psychosom Res. 2012;73(5):334-342.PubMedGoogle ScholarCrossref
8.
Naylor  C, Parsonage  M, McDaid  D, Knapp  M, Fossey  M, Galea  A.  Long-term Conditions and Mental Health: The Cost of Co-morbidities. London, England: King’s Fund and Centre for Mental Health; 2012.
9.
Gunn  J, Diggens  J, Hegarty  K, Blashki  G.  A systematic review of complex system interventions designed to increase recovery from depression in primary care.  BMC Health Serv Res. 2006;6:88.PubMedGoogle ScholarCrossref
10.
Archer  J, Bower  P, Gilbody  S,  et al.  Collaborative care for depression and anxiety problems.  Cochrane Database Syst Rev. 2012;10:CD006525.PubMedGoogle Scholar
11.
Katon  W, Von Korff  M, Lin  E, Simon  G.  Rethinking practitioner roles in chronic illness: the specialist, primary care physician, and the practice nurse.  Gen Hosp Psychiatry. 2001;23(3):138-144.PubMedGoogle ScholarCrossref
12.
Coventry  PA, Hudson  JL, Kontopantelis  E,  et al.  Characteristics of effective collaborative care for treatment of depression: a systematic review and meta-regression of 74 randomised controlled trials.  PLoS One. 2014;9(9):e108114.PubMedGoogle ScholarCrossref
13.
Community Preventive Services Task Force.  Recommendation from the Community Preventive Services Task Force for use of collaborative care for the management of depressive disorders.  Am J Prev Med. 2012;42(5):521-524.PubMedGoogle ScholarCrossref
14.
Thota  AB, Sipe  TA, Byard  GJ,  et al; Community Preventive Services Task Force.  Collaborative care to improve the management of depressive disorders: a community guide systematic review and meta-analysis.  Am J Prev Med. 2012;42(5):525-538.PubMedGoogle ScholarCrossref
15.
NICE.  Depression in Adults With a Chronic Physical Health Problem: Treatment and Management. Leicester, England: British Psychological Society; 2010.
16.
Bower  P, Gilbody  S, Richards  D, Fletcher  J, Sutton  A.  Collaborative care for depression in primary care: making sense of a complex intervention: systematic review and meta-regression.  Br J Psychiatry. 2006;189:484-493.PubMedGoogle ScholarCrossref
17.
Bartels  SJ, Coakley  EH, Zubritsky  C,  et al; PRISM-E Investigators.  Improving access to geriatric mental health services: a randomized trial comparing treatment engagement with integrated versus enhanced referral care for depression, anxiety, and at-risk alcohol use.  Am J Psychiatry. 2004;161(8):1455-1462.PubMedGoogle ScholarCrossref
18.
Blanchard  MR, Waterreus  A, Mann  AH.  The effect of primary care nurse intervention upon older people screened as depressed.  Int J Geriatr Psychiatry. 1995;10(4):289-298.Google ScholarCrossref
19.
Bruce  ML, Ten Have  TR, Reynolds  CF  III,  et al.  Reducing suicidal ideation and depressive symptoms in depressed older primary care patients: a randomized controlled trial.  JAMA. 2004;291(9):1081-1091.PubMedGoogle ScholarCrossref
20.
Ciechanowski  P, Wagner  E, Schmaling  K,  et al.  Community-integrated home-based depression treatment in older adults: a randomized controlled trial.  JAMA. 2004;291(13):1569-1577.PubMedGoogle ScholarCrossref
21.
Ell  K, Unützer  J, Aranda  M, Gibbs  NE, Lee  PJ, Xie  B.  Maaging depression in home health care: a randomized clinical trial.  Home Health Care Serv Q. 2007;26(3):81-104.PubMedGoogle ScholarCrossref
22.
McCusker  J, Cole  M, Yaffe  M,  et al.  Project Direct: pilot study of a collaborative intervention for depressed seniors.  Can J Commun Ment Health. 2008;27(2):201-218.Google ScholarCrossref
23.
Unützer  J, Katon  W, Callahan  CM,  et al; IMPACT Investigators.  Improving mood-promoting access to collaborative treatment: collaborative care management of late-life depression in the primary care setting: a randomized controlled trial.  JAMA. 2002;288(22):2836-2845.PubMedGoogle ScholarCrossref
24.
Cole  MG, McCusker  J, Elie  M, Dendukuri  N, Latimer  E, Belzile  E.  Systematic detection and multidisciplinary care of depression in older medical inpatients: a randomized trial.  CMAJ. 2006;174(1):38-44.PubMedGoogle ScholarCrossref
25.
Chew-Graham  CA, Lovell  K, Roberts  C,  et al.  A randomised controlled trial to test the feasibility of a collaborative care model for the management of depression in older people.  Br J Gen Pract. 2007;57(538):364-370.PubMedGoogle Scholar
26.
Stewart  GB, Altman  DG, Askie  LM, Duley  L, Simmonds  MC, Stewart  LA.  Statistical analysis of individual participant data meta-analyses: a comparison of methods and recommendations for practice.  PLoS One. 2012;7(10):e46042.PubMedGoogle ScholarCrossref
27.
Clarke  MJ, Stewart  LA.  Obtaining data from randomised controlled trials: how much do we need for reliable and informative meta-analyses?  BMJ. 1994;309(6960):1007-1010.PubMedGoogle ScholarCrossref
28.
Riley  RD, Lambert  PC, Abo-Zaid  G.  Meta-analysis of individual participant data: rationale, conduct, and reporting.  BMJ. 2010;340:c221.PubMedGoogle ScholarCrossref
29.
Harrison  M, Reeves  D, Harkness  E,  et al.  A secondary analysis of the moderating effects of depression and multimorbidity on the effectiveness of a chronic disease self-management programme.  Patient Educ Couns. 2012;87(1):67-73.PubMedGoogle ScholarCrossref
30.
Fried  TR, O’Leary  J, Towle  V, Goldstein  MK, Trentelange  M, Martin  DK.  The effects of comorbidity on the benefits and harms of treatment for chronic disease: a systematic review.  PLoS One. 2014;9(11):e112593.PubMedGoogle ScholarCrossref
31.
Stewart  LA, Clarke  M, Rovers  M,  et al; PRISMA-IPD Development Group.  Preferred Reporting Items for Systematic Review and Meta-analyses of Individual Participant Data: the PRISMA-IPD statement.  JAMA. 2015;313(16):1657-1665.PubMedGoogle ScholarCrossref
32.
Tierney  JF, Vale  C, Riley  R,  et al.  Individual participant data (IPD) meta-analyses of randomised controlled trials: guidance on their use.  PLoS Med. 2015;12(7):e1001855.PubMedGoogle ScholarCrossref
33.
Beck  AT, Steer  RA.  Internal consistencies of the original and revised Beck Depression Inventory.  J Clin Psychol. 1984;40(6):1365-1367.PubMedGoogle ScholarCrossref
34.
Radloff  LS.  The CES-D scale: a self-report depression scale for research in the general population.  Appl Psychol Meas. 1977;1(3):385-401.Google ScholarCrossref
35.
Hamilton  M.  Rating depressive patients.  J Clin Psychiatry. 1980;41(12, pt 2):21-24.PubMedGoogle Scholar
36.
Kroenke  K, Spitzer  RL, Williams  JB.  The PHQ-9: validity of a brief depression severity measure.  J Gen Intern Med. 2001;16(9):606-613.PubMedGoogle ScholarCrossref
37.
Charlson  ME, Pompei  P, Ales  KL, MacKenzie  CR.  A new method of classifying prognostic comorbidity in longitudinal studies: development and validation.  J Chronic Dis. 1987;40(5):373-383.PubMedGoogle ScholarCrossref
38.
Gadbury  GL, Coffey  CS, Allison  DB.  Modern statistical methods for handling missing repeated measurements in obesity trial data: beyond LOCF.  Obes Rev. 2003;4(3):175-184.PubMedGoogle ScholarCrossref
39.
Donders  AR, van der Heijden  GJ, Stijnen  T, Moons  KG.  Review: a gentle introduction to imputation of missing values.  J Clin Epidemiol. 2006;59(10):1087-1091.PubMedGoogle ScholarCrossref
40.
Rubin  DB.  Multiple Imputation for Nonresponse in Surveys. New York, NY: Wiley & Sons; 1987.
41.
Schafer  J.  Analysis of Incomplete Multivariate Data. Boca Raton, FL: Chapman & Hall; 1997.
42.
Stata Press.  Multiple-Imputation Reference Manual. College Station, TX: Stata Press; 2014.
43.
Stewart  LA, Parmar  MK.  Meta-analysis of the literature or of individual patient data: is there a difference?  Lancet. 1993;341(8842):418-422.PubMedGoogle ScholarCrossref
44.
Vickers  AJ, Cronin  AM, Maschino  AC,  et al; Acupuncture Trialists’ Collaboration.  Acupuncture for chronic pain: individual patient data meta-analysis.  Arch Intern Med. 2012;172(19):1444-1453.PubMedGoogle ScholarCrossref
45.
Simmonds  MC, Higgins  JP.  Covariate heterogeneity in meta-analysis: criteria for deciding between meta-regression and individual patient data.  Stat Med. 2007;26(15):2982-2999.PubMedGoogle ScholarCrossref
46.
Whitehead  A.  Meta-analysis of Controlled Clinical Trials. New York, NY: John Wiley & Sons; 2002.
47.
Sutton  AJ, Kendrick  D, Coupland  CA.  Meta-analysis of individual- and aggregate-level data.  Stat Med. 2008;27(5):651-669.PubMedGoogle ScholarCrossref
48.
StataCorp.  Stata Statistical Software: Release 12. College Station, TX: StataCorp LP; 2011.
49.
Kontopantelis  E, Reeves  D.  A short guide and a forest plot command (ipdforest) for one-stage meta-analysis.  Stata J. 2013;13(3):574-587.Google Scholar
50.
Higgins  JP, Thompson  SG.  Quantifying heterogeneity in a meta-analysis.  Stat Med. 2002;21(11):1539-1558.PubMedGoogle ScholarCrossref
51.
Ahmed  I, Sutton  AJ, Riley  RD.  Assessment of publication bias, selection bias, and unavailable data in meta-analyses using individual participant data: a database survey.  BMJ. 2012;344:d7762.PubMedGoogle ScholarCrossref
52.
Higgins  JPT, Altman  DG, Gøtzsche  PC,  et al; Cochrane Bias Methods Group; Cochrane Statistical Methods Group.  The Cochrane Collaboration’s tool for assessing risk of bias in randomised trials.  BMJ. 2011;343:d5928.PubMedGoogle ScholarCrossref
53.
Schulz  KF, Chalmers  I, Hayes  RJ, Altman  DG.  Empirical evidence of bias: dimensions of methodological quality associated with estimates of treatment effects in controlled trials.  JAMA. 1995;273(5):408-412.PubMedGoogle ScholarCrossref
54.
Pildal  J, Hróbjartsson  A, Jørgensen  KJ, Hilden  J, Altman  DG, Gøtzsche  PC.  Impact of allocation concealment on conclusions drawn from meta-analyses of randomized trials.  Int J Epidemiol. 2007;36(4):847-857.PubMedGoogle ScholarCrossref
55.
McMahon  L, Foran  KM, Forrest  SD,  et al.  Graduate mental health worker case management of depression in UK primary care: a pilot study.  Br J Gen Pract. 2007;57(544):880-885.PubMedGoogle ScholarCrossref
56.
Katon  W, Von Korff  M, Lin  E,  et al.  Collaborative management to achieve treatment guidelines. Impact on depression in primary care.  JAMA. 1995;273(13):1026-1031.PubMedGoogle ScholarCrossref
57.
Katon  W, Robinson  P, Von Korff  M,  et al.  A multifaceted intervention to improve treatment of depression in primary care.  Arch Gen Psychiatry. 1996;53(10):924-932.PubMedGoogle ScholarCrossref
58.
Katon  W, Von Korff  M, Lin  E,  et al.  Stepped collaborative care for primary care patients with persistent symptoms of depression: a randomized trial.  Arch Gen Psychiatry. 1999;56(12):1109-1115.PubMedGoogle ScholarCrossref
59.
Katon  W, Rutter  C, Ludman  EJ,  et al.  A randomized trial of relapse prevention of depression in primary care.  Arch Gen Psychiatry. 2001;58(3):241-247.PubMedGoogle ScholarCrossref
60.
Simon  GE, VonKorff  M, Rutter  C, Wagner  E.  Randomised trial of monitoring, feedback, and management of care by telephone to improve treatment of depression in primary care.  BMJ. 2000;320(7234):550-554.PubMedGoogle ScholarCrossref
61.
Simon  GE, Ralston  JD, Savarino  J, Pabiniak  C, Wentzel  C, Operskalski  BH.  Randomized trial of depression follow-up care by online messaging.  J Gen Intern Med. 2011;26(7):698-704.PubMedGoogle ScholarCrossref
62.
Simon  GE, Ludman  EJ, Tutty  S, Operskalski  B, Von Korff  M.  Telephone psychotherapy and telephone care management for primary care patients starting antidepressant treatment: a randomized controlled trial.  JAMA. 2004;292(8):935-942.PubMedGoogle ScholarCrossref
63.
Melville  JL, Reed  SD, Russo  J,  et al.  Improving care for depression in obstetrics and gynecology: a randomized controlled trial.  Obstet Gynecol. 2014;123(6):1237-1246.PubMedGoogle ScholarCrossref
64.
Buszewicz  M, Griffin  M, McMahon  EM, Beecham  J, King  M.  Evaluation of a system of structured, pro-active care for chronic depression in primary care: a randomised controlled trial.  BMC Psychiatry. 2010;10(1):61.PubMedGoogle ScholarCrossref
65.
Gensichen  J, von Korff  M, Peitz  M,  et al; PRoMPT (PRimary care Monitoring for depressive Patients Trial).  Case management for depression by health care assistants in small primary care practices: a cluster randomized trial.  Ann Intern Med. 2009;151(6):369-378.PubMedGoogle ScholarCrossref
66.
Menchetti  M, Sighinolfi  C, Di Michele  V,  et al.  Effectiveness of collaborative care for depression in Italy. A randomized controlled trial.  Gen Hosp Psychiatry. 2013;35(6):579-586.PubMedGoogle ScholarCrossref
67.
Huijbregts  KM, de Jong  FJ, van Marwijk  HW,  et al.  A target-driven collaborative care model for Major Depressive Disorder is effective in primary care in the Netherlands. A randomized clinical trial from the depression initiative.  J Affect Disord. 2013;146(3):328-337.PubMedGoogle ScholarCrossref
68.
Vlasveld  MC, van der Feltz-Cornelis  CM, Adèr  HJ,  et al.  Collaborative care for major depressive disorder in an occupational healthcare setting.  Br J Psychiatry. 2012;200(6):510-511.PubMedGoogle ScholarCrossref
69.
Smit  A, Kluiter  H, Conradi  HJ,  et al.  Short-term effects of enhanced treatment for depression in primary care: results from a randomized controlled trial.  Psychol Med. 2006;36(1):15-26.PubMedGoogle ScholarCrossref
70.
Patel  V, Weiss  HA, Chowdhary  N,  et al.  Effectiveness of an intervention led by lay health counsellors for depressive and anxiety disorders in primary care in Goa, India (MANAS): a cluster randomised controlled trial.  Lancet. 2010;376(9758):2086-2095.PubMedGoogle ScholarCrossref
71.
Wells  KB, Sherbourne  C, Schoenbaum  M,  et al.  Impact of disseminating quality improvement programs for depression in managed primary care: a randomized controlled trial [published correction appears in JAMA. 2000;283(24):3204].  JAMA. 2000;283(2):212-220.PubMedGoogle ScholarCrossref
72.
Adler  DA, Bungay  KM, Wilson  IB,  et al.  The impact of a pharmacist intervention on 6-month outcomes in depressed primary care patients.  Gen Hosp Psychiatry. 2004;26(3):199-209.PubMedGoogle ScholarCrossref
73.
Richards  DA, Lovell  K, Gilbody  S,  et al.  Collaborative care for depression in UK primary care: a randomized controlled trial: Corrigendum.  Psychol Med. 2009;39(4):701.Google ScholarCrossref
74.
Richards  DA, Hill  JJ, Gask  L,  et al.  Clinical effectiveness of collaborative care for depression in UK primary care (CADET): cluster randomised controlled trial.  BMJ. 2013;347:f4913.PubMedGoogle ScholarCrossref
75.
Katon  WJ, Von Korff  M, Lin  EH,  et al.  The Pathways Study: a randomized trial of collaborative care in patients with diabetes and depression.  Arch Gen Psychiatry. 2004;61(10):1042-1049.PubMedGoogle ScholarCrossref
76.
Katon  WJ, Lin  EH, Von Korff  M,  et al.  Collaborative care for patients with depression and chronic illnesses.  N Engl J Med. 2010;363(27):2611-2620.PubMedGoogle ScholarCrossref
77.
Davidson  KW, Bigger  JT, Burg  MM,  et al.  Centralized, stepped, patient preference-based treatment for patients with post-acute coronary syndrome depression: CODIACS vanguard randomized controlled trial.  JAMA Intern Med. 2013;173(11):997-1004.PubMedGoogle ScholarCrossref
78.
Ell  K, Xie  B, Quon  B, Quinn  DI, Dwight-Johnson  M, Lee  P-J.  Randomized controlled trial of collaborative care management of depression among low-income patients with cancer.  J Clin Oncol. 2008;26(27):4488-4496.PubMedGoogle ScholarCrossref
79.
Ell  K, Katon  W, Xie  B,  et al.  Collaborative care management of major depression among low-income, predominantly Hispanic subjects with diabetes: a randomized controlled trial.  Diabetes Care. 2010;33(4):706-713.PubMedGoogle ScholarCrossref
80.
Huffman  JC, Mastromauro  CA, Sowden  GL, Wittmann  C, Rodman  R, Januzzi  JL.  A collaborative care depression management program for cardiac inpatients: depression characteristics and in-hospital outcomes.  Psychosomatics. 2011;52(1):26-33.PubMedGoogle ScholarCrossref
81.
Huffman  JC, Mastromauro  CA, Beach  SR,  et al.  Collaborative care for depression and anxiety disorders in patients with recent cardiac events: the Management of Sadness and Anxiety in Cardiology (MOSAIC) randomized clinical trial.  JAMA Intern Med. 2014;174(6):927-935.PubMedGoogle ScholarCrossref
82.
Rollman  BL, Belnap  BH, LeMenager  MS,  et al.  Telephone-delivered collaborative care for treating post-CABG depression: a randomized controlled trial.  JAMA. 2009;302(19):2095-2103.PubMedGoogle ScholarCrossref
83.
Tierney  JF, Pignon  J-P, Gueffyier  F,  et al; Cochrane IPD Meta-analysis Methods Group.  How individual participant data meta-analyses have influenced trial design, conduct, and analysis.  J Clin Epidemiol. 2015;68(11):1325-1335.PubMedGoogle ScholarCrossref
84.
Piette  JD, Kerr  EA.  The impact of comorbid chronic conditions on diabetes care.  Diabetes Care. 2006;29(3):725-731.PubMedGoogle ScholarCrossref
85.
Simmonds  MC, Higgins  JPT, Stewart  LA, Tierney  JF, Clarke  MJ, Thompson  SG.  Meta-analysis of individual patient data from randomized trials: a review of methods used in practice.  Clin Trials. 2005;2(3):209-217.PubMedGoogle ScholarCrossref
86.
Altman  DG, Deeks  JJ.  Meta-analysis, Simpson’s paradox, and the number needed to treat.  BMC Med Res Methodol. 2002;2:3.PubMedGoogle ScholarCrossref
87.
Kievit  RA, Frankenhuis  WE, Waldorp  LJ, Borsboom  D.  Simpson’s paradox in psychological science: a practical guide.  Front Psychol. 2013;4:513.PubMedGoogle ScholarCrossref
88.
Vale  CL, Rydzewska  LH, Rovers  MM, Emberson  JR, Gueyffier  F, Stewart  LA; Cochrane IPD Meta-analysis Methods Group.  Uptake of systematic reviews and meta-analyses based on individual participant data in clinical practice guidelines: descriptive study.  BMJ. 2015;350:h1088.PubMedGoogle ScholarCrossref
89.
McGregor  M, Lin  EH, Katon  WJ.  TEAMcare: an integrated multicondition collaborative care program for chronic illnesses and depression.  J Ambul Care Manage. 2011;34(2):152-162.PubMedGoogle ScholarCrossref
Original Investigation
Meta-analysis
September 2016

Association Between Chronic Physical Conditions and the Effectiveness of Collaborative Care for Depression: An Individual Participant Data Meta-analysis

Author Affiliations
  • 1National Institute of Health Research School for Primary Care Research, Centre for Primary Care, Institute of Population Health, University of Manchester, Manchester, England
  • 2School of Nursing, Midwifery and Social Work, University of Manchester, Manchester, England
  • 3Mental Health and Addiction Research Group, Department of Health Sciences, Hull York Medical School, University of York, York, England
  • 4Institute of Health Research, University of Exeter Medical School, Exeter, England
  • 5National Institute of Health Research Collaboration for Leadership in Applied Health Research and Care for the South West Peninsula, University of Exeter, Exeter, England
  • 6Group Health Research Institute, Seattle, Washington
  • 7Ethnicity and Poverty, School of Social Work, University of Southern California, Los Angeles
  • 8Harvard Medical School, General Hospital/Blake 11, Boston, Massachusetts
  • 9Department of Psychiatry and Behavioral Sciences, Faculty of Social and Behavioral Sciences, Tilburg University, the Netherlands
  • 10Department of Psychiatry, Tufts Medical Center, Boston, Massachusetts
  • 11Department of Psychiatry, Weill Cornell Medical College, White Plains, New York
  • 12Institute of Epidemiology and Health, Faculty of Population and Health Sciences, University College London, London, England
  • 13Department of Psychiatry, St. Mary's Hospital Center, McGill University, Montreal, Quebec, Canada
  • 14Center for Behavioral Cardiovascular Health, Department of Medicine, Columbia University, New York, New York
  • 15Interdisciplinary Center Psychopathology and Emotion Regulation, University Medical Center Groningen, Groningen, the Netherlands
  • 16Institute of General Practice, Friedrich-Schiller-University, School of Medicine, University Hospital, Jena, Germany
  • 17Netherlands Institute of Mental Health and Addiction, Trimbos Institute, Utrecht, the Netherlands
  • 18Department of Medical and Surgical Sciences, University of Bologna, Bologna, Italy
  • 19Public Health Foundation of India, Gurgaon, India
  • 20Psychiatry, Biomedical Informatics, and Clinical and Translational Science, Department of Biomedical Informatics, University of Pittsburgh, Pittsburgh, Pennsylvania
  • 21Department of Psychology, University of Colorado, Denver
  • 22Mental Health and Addiction Research Group, Department of Health Sciences, University of York, York, England
  • 23Centre for Reviews and Dissemination, University of York, York, England
JAMA Psychiatry. 2016;73(9):978-989. doi:10.1001/jamapsychiatry.2016.1794
Abstract

Importance  Collaborative care is an intensive care model involving several health care professionals working together, typically a physician, a case manager, and a mental health professional. Meta-analyses of aggregate data have shown that collaborative care is particularly effective in people with depression and comorbid chronic physical conditions. However, only participant-level analyses can rigorously test whether the treatment effect is influenced by participant characteristics, such as chronic physical conditions.

Objective  To assess whether the effectiveness of collaborative care for depression is moderated by the presence, type, and number of chronic physical conditions.

Data Sources  Data were obtained from MEDLINE, EMBASE, PubMed, PsycINFO, CINAHL Complete, and Cochrane Central Register of Controlled Trials, and references from relevant systematic reviews. The search and collection of eligible studies was ongoing until May 22, 2015.

Study Selection  This was an update to a previous meta-analysis. Two independent reviewers were involved in the study selection process. Randomized clinical trials that compared the effectiveness of collaborative care with usual care in adults with depression and reported measured changes in depression severity symptoms at 4 to 6 months after randomization were included in the analysis. Key search terms included depression, dysthymia, anxiety, panic, phobia, obsession, compulsion, posttraumatic, care management, case management, collaborative care, enhanced care, and managed care.

Data Extraction and Synthesis  Individual participant data on baseline demographics and chronic physical conditions as well as baseline and follow-up depression severity symptoms were requested from authors of the eligible studies. One-step meta-analysis of individual participant data using appropriate mixed-effects models was performed.

Main Outcomes and Measures  Continuous outcomes of depression severity symptoms measured using self-reported or observer-rated measures.

Results  Data sets from 31 randomized clinical trials including 36 independent comparisons (N = 10 962 participants) were analyzed. Individual participant data analyses found no significant interaction effects, indicating that the presence (interaction coefficient, 0.02 [95% CI, −0.10 to 0.13]), numbers (interaction coefficient, 0.01 [95% CI, −0.01 to 0.02]), and types of chronic physical conditions do not influence the treatment effect.

Conclusions and Relevance  There is evidence that collaborative care is effective for people with depression alone and also for people with depression and chronic physical conditions. Existing guidance that recommends limiting collaborative care to people with depression and physical comorbidities is not supported by this individual participant data meta-analysis.

Introduction

Depression is the leading global cause of disease burden accounting for most disability-adjusted life years.1,2 The combination of depression with a chronic physical condition (hereinafter termed physical condition), such as cardiovascular diseases, respiratory diseases, and diabetes, is associated with the greatest decrements in quality of life, years lost owing to disability,3 patient safety failures, and unscheduled care.4-8

A promising intervention for depression in primary care is collaborative care,9 which involves the use of a case manager (usually nonmedical but also medical; eg, clinical social workers) working with primary care professionals, often supervised by a mental health specialist and supported by care management systems. A Cochrane review10 showed that collaborative care is more effective than usual care for depression.

Finding feasible and effective ways of integrating care for patients with depression and comorbid physical conditions remains a critical goal for health systems worldwide. There has been significant interest in the ability of collaborative care to improve care for people with depression and physical conditions.11,12 In the United States, the Community Preventive Services Task Force13,14 recommends collaborative care for the treatment of major depression in adults but concedes that there are evidence gaps about the effectiveness of this approach in people with comorbid physical conditions. In the United Kingdom, the English organization responsible for clinical guidelines (National Institute for Health and Care Excellence)15 recommends that collaborative care should be considered only for people with depression and comorbid physical conditions based on results from aggregate-data meta-analyses of 2 sets of trials: collaborative care for patients with depression and collaborative care for patients with depression and physical conditions. Although these recommendations were based on the best available evidence at the time, conducting 2 separate meta-analyses means that any differences in effectiveness may be confounded by differences between the trials (eg, location, quality, and intervention format) and may not be due to physical conditions.16 In addition, some collaborative care trials that recruited people with depression would not necessarily have excluded those with physical conditions, especially trials conducted in older populations in whom such conditions are highly prevalent.17-25

To reach international consensus about the most effective ways to manage depression it is critical that guidelines reflect the most robust analysis of the most current data. There is increasing recognition that individual participant data (IPD) meta-analysis is a better basis for modeling treatment effects than are aggregate data meta-analyses.26 Individual participant data meta-analysis is more precise because it involves the application of standardized analyses across multiple data sets, overcomes sample size and reporting issues, and allows more sophisticated modeling of moderator effects.26-28 In this context, IPD allows for more accurate coding of comorbidity based on actual patient health.

In the present study, we used IPD meta-analysis to test whether physical conditions moderate the effectiveness of collaborative care for depression outcomes. First, we examined whether studies that recruited individuals with physical conditions as part of inclusion criteria demonstrated greater effect of collaborative care on depression outcomes (ie, a study-level moderator analysis). Subsequently, we examined whether the effectiveness of collaborative care on depression outcomes was moderated by the presence, number, and type of physical conditions reported by individual participants within trials (ie, participant-level moderator analyses). The results will provide a rigorous basis for recommendations about the types of people most likely to benefit from collaborative care and contribute to the wider debate about how multimorbidity influences treatment effectiveness.29,30

Box Section Ref ID

Key Points

  • Question Is collaborative care more effective for patients with depression and chronic physical conditions compared with patients with depression alone?

  • Findings This meta-analysis of individual participant data from 10 960 patients found no significant differences in the treatment effects of collaborative care between patients with depression alone and patients with depression and chronic physical conditions.

  • Meaning Collaborative care is an equally effective way to deliver depression care for patient with and without comorbid chronic physical conditions; existing guidance that recommends limiting access to collaborative care for patients with depression and chronic physical conditions should be updated.

Methods

This IPD meta-analysis was conducted and reported according to published methodological guidelines.31,32 The PRISMA-D was completed (eTable 1 in the Supplement).

Information Sources

The published Cochrane review10 of collaborative care was used to identify eligible randomized clinical trials (RCTs). The Cochrane review searches were updated in March 2014 (MEDLINE, EMBASE, PsycINFO, Cochrane Central Register of Controlled Trials, and CINAHL Complete) but search and collection of eligible studies was ongoing until May 22, 2015. Key search terms included depression, dysthymia, anxiety, panic, phobia, obsession, compulsion, posttraumatic, care management, case management, collaborative care, enhanced care, and managed care. The reference lists of reports of all included studies were screened for additional studies. We also asked authors of studies reviewed to identify additional published studies and trials in progress.

Eligibility Criteria

We used the same inclusion criteria reported by the Cochrane review,10 except we excluded studies conducted in adolescents and studies that did not report a depression outcome (eMethods in the Supplement). The following factors were evaluated in study selection:

  1. Adults (age ≥18 years) with depression or mixed anxiety and depressive disorder,

  2. Collaborative care interventions (multiprofessional approach to patient care, structured management plan, scheduled patient follow-ups, and enhanced interprofessional communication),

  3. Comparison (usual or enhanced usual care),

  4. Outcome (continuous depression scores), and

  5. Research design (RCTs or cluster RCTs).

Measuring Depression and Physical Conditions

All studies provided continuous depression scores measured using validated scales, including the Beck Depression Inventory,33 the Center for Epidemiologic Studies Depression Scale,34 the Hamilton Rating Scale for Depression,35 and the Patient Health Questionnaire–9.36 We focused on short-term depression outcomes reported between 4 and 6 months after randomization. Depression scores were standardized within each study, using the baseline SD and the follow-up mean score.

At study level, 9 of the 36 comparisons recruited participants with a physical condition. Thirty of the 36 comparisons reported data on the presence and number of physical conditions at individual participant level. Of these, most comparisons (21 [58%]) used validated comorbidity indices, such as the Charlson Comorbidity Index,37 and 6 (17%) used empirical lists of physical conditions. Seventeen comparisons (47%) identified the type of physical conditions among participants. Based on these data, we were able to create 5 categories of physical conditions: cancer (10 comparisons), cardiac disease (16 comparisons), diabetes (17 comparisons), hypertension (11 comparisons), and respiratory disease (11 comparisons).

Data Extraction and Preparation

We contacted the study authors to obtain primary data sets for the following data: treatment group, age, sex, baseline and follow-up depression scores, and number and types of physical conditions (eTable 2 in the Supplement). Received data were cleaned, converted into the same reporting format, and aggregated into a single data set. Initial separate analyses on depression outcomes were conducted for each study to ensure that our analyses were consistent with those reported in the original study. We extracted data from the published reports of all the eligible studies using a standardized Excel (Microsoft Inc) data extraction form. Data on populations, interventions, chronic conditions (used as moderator in the analyses), risk of bias, and outcome effect sizes were also extracted. Studies with data available to us were compared with studies with unavailable data in terms of outcome effect sizes and moderator analyses.

Missing Data

Multiple imputation techniques were used to obtain more complete data sets and to better protect against bias due to data missing at random mechanisms. Missing values for age and depression scores at follow-up were imputed with a multivariate imputation algorithm (mi impute mvn in Stata, version 14; StataCorp LP) using Markov chain Monte Carlo.38,39 This process produces several data sets, each of which is analyzed separately using the prespecified model; the results are then combined using Rubin’s rules,40 which account for uncertainty in imputed values. A total of 1000 new data sets with the observed and imputed scores for age and depression at follow-up were generated based on values obtained from study identification number, treatment group, baseline depression score, and sex. The range of imputed values was limited to the range of observed values of the variables. Time series and autocorrelation plots of the worst linear function were performed to monitor the convergence of the generated imputation algorithms.41,42 We examined whether baseline variables (study, treatment group, age, sex, and baseline depression scores) predicted missing data to confirm that the assumptions underlying imputations were met. Sensitivity analyses were performed using only cases with available data; no significant differences were detected in any of the reported results.

Statistical Analysis

One-step meta-analysis was undertaken because it is less susceptible to bias, is most efficient in terms of power, and allows for sophisticated modeling of covariates (age, sex, and baseline depression scores in the present study).43,44 A 1-step IPD meta-analysis constructs a model for the hypothesized treatment-covariate associations across all IPD sets while statistically accounting for clustering at the level of each included data set.28,45

Appropriate mixed-effects models (fixed study-specific intercepts, random treatment effect, and fixed study-specific effects for baseline depression) were used to meta-analyze the participant-level data and estimate the between-study and within-study variances and the effects of covariates.46 Clustered RCTs were statistically accounted for in the model by adhering to the methodologic recommendations of Sutton et al.47 The Stata, version 14, command mixed was used through the ipdforest command to summarize the evidence by study and obtain forest plots.48,49 A 1-stage meta-analysis variant of the I2 statistic was used to assess heterogeneity.50 We examined the impact of covariates separately, building a model based on both statistical and theoretical criteria. If studies included multiple treatment groups and a single control group, the treatment groups were treated as separate comparisons in the analyses, whereas, the control groups were halved at random to avoid double counting in the analyses. In accordance with published guidelines,51 funnel plots were constructed to assess the potential for publication bias.

Several prespecified primary analyses were performed. One analysis examined a study-level moderator (binary variable; participants with physical conditions as part of the study’s inclusion criteria: yes, 1; no, 0). The other analyses examined moderators at patient level including the presence (binary variable; present, 1; absent, 0), number (continuous variable), and types of physical conditions (binary variables for each condition; present, 1; absent, 0).

We conducted 2 secondary analyses to examine the robustness of the results and assessed whether the results remained the same after controlling for the risk of bias scores of the studies (based on allocation concealment item). Allocation concealment was selected as an indicator of risk of bias because it is the most sensitive item to changes in the treatment effect, especially when based on self-reported outcomes.52-54 We also explored whether the main effects were influenced by the measure used to assess physical conditions (use of validated comorbidity severity indices).

Results

Figure 1 presents the flowchart of the study selection process. A total of 76 RCTs (22 284 patients) including 86 independent comparisons were eligible for inclusion in the IPD meta-analysis. We found no evidence of asymmetry in the funnel plot for these studies (Egger regression test intercept [SE], −0.54 [0.42]; P = .21) (eFigure 1 in the Supplement). We collected data from 32 trials (11 531 [51.7%] of the total number of participants) that included 37 comparisons. One RCT55 (a pilot study based on 49 participants) was excluded from the analyses because it did not include data on age and sex, leaving 36 comparisons. A total of 569 individual cases (4.9%) were excluded from the analyses because of missing baseline values on depression or age, leaving 10 962 unique cases (of which 1819 [16.6%] were imputed using multiple imputations).

Baseline Characteristics and Comparisons Between Available and Unavailable Data

Eighteen studies were conducted in the United States, 11 in Europe, 1 in Canada, and 1 in India. Most participants were women (7749 [70.7%]); mean (SD) age was 51 (15) years (range, 17-97 years). Most participants (8099 [78.1%]) had at least 1 physical condition with a mean of 2.5 (2.3) conditions. We identified no important issues while checking the IPD. Details on the characteristics of the studies are presented in eTable 3 and eTable 4 in the Supplement.

Available and unavailable studies were compared in terms of population, intervention, and risk of bias characteristics, as well as outcome data. We selected these specific characteristics based on the results of 2 previous reviews12,16 that applied meta-regression analyses to identify moderators of the main effect of collaborative care interventions. As reported in the Table, the only statistically significant difference identified was for the intervention content; a larger proportion of trials that incorporated psychological interventions made data available.

Effects of Collaborative Care on 3 Depressive Symptoms at 4- to 6-Month Follow-up

Collaborative care was associated with a small but significant effect on depression outcomes compared with usual care (standardized mean difference [SMD], −0.22 [95% CI, −0.25 to −0.18]; I2 = 0.8%; 0.3%-3.5%) (eFigure 2 in the Supplement) equal to a drop of approximately 2 points on the Patient Health Questionnaire–9 above the change in the controls. This effect size was smaller but not significantly different from the effect size reported in a Cochrane review10 (−0.28 [95% CI, −0.31 to −0.25]; P = .227) (Table).

Association Between Systematic Identification of Patients with Chronic Physical Conditions and the Effectiveness of Collaborative Care on Depression Scores

A statistically significant interaction was found between systematic identification of participants with physical conditions in the study and treatment effect (interaction coefficient, −0.12 [95% CI, −0.23 to −0.02]) (Figure 2).12,19,21,24,56-82 Randomized clinical trials that explicitly recruited people with physical conditions were associated with significantly larger treatment effects for depression (SMD, −0.29 [95% CI, −0.37 to −0.21]) compared with RCTs that did not explicitly recruit people with physical conditions (SMD, −0.19 [95% CI, −0.23 to −0.15]). The moderating effect of inclusion of physical conditions was even larger in trials with adequate concealment of allocation (interaction coefficient, −0.14 [95% CI, −0.26 to −0.02]).

Association Between Presence of Chronic Physical Conditions and the Effectiveness of Collaborative Care on Depression Scores

When we compared the effects of collaborative care in participants with and without physical conditions, the interaction term with the treatment effect was nonsignificant (interaction coefficient, 0.02 [95% CI, −0.10 to 0.13]) (Figure 3).12,19,21,24,56,57,59,60,63,65-72,74-82 We could not demonstrate any statistically significant moderating effect of the presence of physical conditions on depression outcomes at follow-up (effect in those with physical conditions: SMD, −0.21 [95% CI, −0.27 to −0.15], in those without physical conditions: SMD, −0.23 [95% CI, −0.32 to −0.12]) (Figure 3).

This result was not sensitive to allocation concealment ratings (adequate: interaction coefficient, −0.06 [95% CI, −0.04 to 0.02]) or to the measure used to assess physical conditions (validated: 0.05 [95% CI, −0.08 to 0.10]).

Association Between Number of Chronic Physical Conditions and the Effectiveness of Collaborative Care on Depression Scores

The interaction term between number of physical conditions and treatment effect was nonsignificant (interaction coefficient, 0.00 [95% CI, −0.01 to 0.02]) (Figure 4).12,19,21,24,56,57,59,60,63,65-72,74-82 This finding suggests that the improvement of depression scores at follow-up does not differ according to the number of physical conditions. The interaction effect was not significantly affected by the allocation concealment rating (interaction coefficient, −0.00 [95% CI, −0.03 to 0.03]) or by the measure used to assess physical conditions (interaction coefficient, −0.01 [95% CI, −0.04 to 0.02]).

Association Between Types of Chronic Physical Conditions and the Effectiveness of Collaborative Care on Depression Scores

We found no evidence that the effects of collaborative care on depression outcomes are moderated by the types of physical conditions among participants. None of the interaction effects between treatment effect and types of physical conditions were significant: cancer (interaction coefficient, −0.11 [95% CI, −0.02 to 0.22]), cardiac disease (interaction coefficient, −0.02 [95% CI, −0.14 to 0.09]), diabetes (interaction coefficient, −0.02 [95% CI, −0.08 to 0.09]), hypertension (interaction coefficient, −0.09 [95% CI, −0.21 to 0.03]), and respiratory disease (interaction coefficient, −0.08 [95% CI, −0.21 to 0.07]). These findings suggest that the benefits derived by collaborative care do not differ significantly across subgroups of people with certain types of physical conditions. None of the interaction effects were affected by allocation concealment or by the measure used to assess physical conditions.

Discussion

Analyzing data from 36 comparisons of collaborative care and 10 962 participants, this IPD meta-analysis showed that collaborative care is associated with significant short-term improvements in depression outcomes across all people with or without comorbid physical conditions. At study level, trials that recruited only participants with comorbid physical conditions were associated with larger treatment effects compared with trials that did not include those with comorbid conditions, confirming previous findings.12,15 However, when a more accurate analysis at the individual participant level was undertaken, the presence, number, and type of physical conditions did not moderate the main effect of collaborative care on depression outcomes. Overall, the findings of this IPD meta-analysis do not support existing recommendations based on meta-analyses of aggregate data that collaborative care should be considered only for patients with comorbid depression and physical conditions. Our findings highlight the importance of undertaking IPD analyses in developing rigorous recommendations, especially for subgroups of complex patients.

Strengths and Limitations

To our knowledge, this study is the most methodologically rigorous test of the influence of physical conditions on the effects of collaborative care on depression outcomes.26,83 However, there are several limitations. Individual participant data meta-analysis remains vulnerable to important sources of bias, including publication, study or reviewer selection, and data availability bias.51 No funnel plot asymmetry was detected, suggesting that publication bias is not likely to be present in the overall data set. Study selection bias was minimized by including studies through multiple sources (ie, an existing standard Cochrane review,10 top-up database searches, and author requests) using strict prespecification of trial eligibility criteria. These efforts facilitated access to data from approximately half of the participants included in published RCTs of collaborative care for depression, which is below the recommended recruitment target (80% of data requested).51 We observed some differences between available and unavailable studies, but these differences rarely reached statistical significance. For example, the overall effect size was smaller than that found in the previous Cochrane review.10 This difference is likely to be explained by the fact that less than half of all collaborative care trials were included in this IPD analysis, and these trials were generally larger than those that were not included. It is important to continue to develop effective methods and agreements about data sharing to ensure that future analyses have better access to data.

Another limitation of the study is that the evaluation of physical conditions was not pursued in an entirely consistent manner across trials. Most of the trials assessed the number of physical conditions using validated comorbidity indexes that contain extensive lists of physical conditions, but some trials used less-comprehensive lists of physical conditions that were empirically devised.37 Although we separately examined the influence of 5 common physical conditions, participants could have more than 1 of these conditions. The experience and interactions of multiple concordant or discordant conditions84 is another possible factor to recognize why people might respond differently to depression treatment.

The use of multiple mixed-effects regression analyses of IPD and controlling for covariates significantly reduces the possibility of bias present in aggregate data meta-analyses (eg, ecologic fallacy or Simpson paradox).26,83,85-87 However, we recommend interpreting these findings cautiously since some of these biases might still operate.

Finally, there were important between-study variations, including intervention content (collaborative care is generally heterogeneous)10 and depression measures. However, these study-level variations are unlikely to influence the participant-level analyses that showed that chronic physical conditions do not moderate the effectiveness of collaborative care. A post hoc sensitivity analysis indicated that the results were similar regardless of using self-reported or observer-rated measures for depression.

Comparison With Previous Systematic Reviews

Previous systematic reviews12,16 have examined moderators of the effect of collaborative care on depression outcomes but were based on aggregate data and used meta-regressions. Improved depressive outcomes were predicted by the inclusion of psychological interventions, and the use of antidepressant medication was predicted by recruiting people with physical conditions.12 Based on these findings, it was proposed12 that patients with physical conditions may derive greater benefits from collaborative care compared with patients without physical conditions. As noted above, these analyses are a less-robust basis for decision making because they are insensitive to variation in physical conditions at the level of individual patients. In this IPD meta-analysis, we found no support for this hypothesis.

Implications for Clinicians, Policymakers, and Researchers

This study suggests that depressed patients with and without comorbid physical conditions gain important improvements in depression outcomes from collaborative care. As such, our findings do not support the recommendation by the National Institute for Health and Care Excellence that collaborative care should be considered only for people with depression and comorbid physical conditions with functional impairment.15 Limiting collaborative care to people with depression and comorbid physical conditions does not appear to be a reasonable policy recommendation with important implications for patient benefit. Individual participant data meta-analyses are underused in the development of clinical guidelines, and the next iteration of guidelines for depression could be improved by using the most-reliable evidence available.88

Given that the effectiveness of collaborative care is confirmed, future research should focus on understanding how to optimize the delivery and outcomes of collaborative care. For instance, based on recent published evidence,89 the systematic measurement and management of physical health outcomes along with mental health outcomes has the potential to boost the effectiveness of collaborative care in people with physical comorbidities. However, the overall treatment benefits associated with collaborative care are modest.10 It is therefore legitimate to suggest that future treatment policy guidelines should also be grounded on the comparative cost-effectiveness of collaborative care to other types of interventions.

Conclusions

To our knowledge, this IPD meta-analysis represents the most rigorous and precise analysis to date on the extent to which physical conditions influence the effectiveness of collaborative care on depression outcomes. People with depression derive significant benefits from collaborative care regardless of the presence, number, or type of comorbid physical conditions. The core challenge now is to understand how to deliver these interventions at scale in routine settings and to better operationalize the treatment outcomes to maximize patient benefits.

Back to top
Article Information

Correction: This article was corrected on August 24, 2016, to fix the byline and Author Affiliations.

Corresponding Author: Peter A. Coventry, PhD, Mental Health and Addiction Research Group, Department of Health Sciences, University of York, York YO10 5DD, England (peter.coventry@york.ac.uk).

Accepted for Publication: June 14, 2016.

Published Online: August 17, 2016. doi:10.1001/jamapsychiatry.2016.1794.

Author Contributions: Dr Coventry had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Concept and design: Bower, Kontopantelis, Lovell, Gilbody, Waheed, Dickens, Archer, Coventry.

Acquisition, analysis, or interpretation of data: Panagioti, Bower, Kontopantelis, Lovell, Dickens, Simon, Huffman, Richards, van der Feltz-Cornelis, Adler, Bruce, Buszewicz, Cole, Davidson, de Jonge, Gensichen, Menchetti, Patel, Rollman, Shaffer, Coventry.

Drafting of the manuscript: Panagioti, Bower, Kontpantelis, Coventry.

Critical revision of the manuscript for important intellectual content: Bower, Kontopantelis, Gilbody, Lovell, Waheed, Dickens, Archer, Simon, Ell, Huffman, Richards, van der Feltz-Cornelis, Adler, Buszewicz, Cole, Davidson, de Jonge, Gensichen, Menchetti, Patel, Rollman, Shaffer, Zijlstra-Vlasveld, Coventry.

Statistical analysis: Panagioti, Kontopantelis.

Obtaining funding: Bower, Kontopantelis, Lovell, Gilbody, Waheed, Dickens, Archer, Coventry.

Administrative, technical, or material support: Panagioti, Bower, Lovell, Dickens, Huffman, van der Feltz-Cornelis, de Jonge, Huijbregts, Menchetti, Rollman, Shaffer, Zijlstra-Vlasveld.

Study supervision: Bower, Kontopantelis, Coventry.

Conflict of Interest Disclosures: Dr Bower is a paid consultant to the British Association of Counselling and Psychotherapy. The authors have been or are involved in the conduct of trials of collaborative care in the United Kingdom funded by the Medical Research Council and the National Institute for Health Research. No other disclosures were reported.

Funding/Support: This study was funded by grant 212 from the UK National Institute of Health Research (NIHR) School for Primary Care Research. The research team members were independent from the funding agency.

Role of the Funder/Sponsor: The NIHR 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 publication are those of the authors and not necessarily those of the National Health Service, the National Institute for Health Research, or the Department of Health.

Additional Contributions: We acknowledge the important contribution of Wayne Katon, MD, who died in March 2015 before approving this manuscript; he provided us access to 6 datasets of collaborative care trials. Harm van Marwijk, MD (National Institute of Health Research School for Primary Care Research, Centre for Primary Care, Institute of Population Health, University of Manchester), and Penny Bee, PhD (School of Nursing, Midwifery and Social Work, University of Manchester) provided comments on drafts of this manuscript. There was no financial compensation.

References
1.
Murray  CJL, Vos  T, Lozano  R,  et al.  Disability-adjusted life years (DALYs) for 291 diseases and injuries in 21 regions, 1990-2010: a systematic analysis for the Global Burden of Disease Study 2010.  Lancet. 2012;380(9859):2197-2223.PubMedGoogle ScholarCrossref
2.
Whiteford  HA, Degenhardt  L, Rehm  J,  et al.  Global burden of disease attributable to mental and substance use disorders: findings from the Global Burden of Disease Study 2010.  Lancet. 2013;382(9904):1575-1586.PubMedGoogle ScholarCrossref
3.
Moussavi  S, Chatterji  S, Verdes  E, Tandon  A, Patel  V, Ustun  B.  Depression, chronic diseases, and decrements in health: results from the World Health Surveys.  Lancet. 2007;370(9590):851-858.PubMedGoogle ScholarCrossref
4.
Freedland  KE, Carney  RM.  Depression as a risk factor for adverse outcomes in coronary heart disease.  BMC Med. 2013;11:131.PubMedGoogle ScholarCrossref
5.
Carney  RM, Blumenthal  JA, Catellier  D,  et al.  Depression as a risk factor for mortality after acute myocardial infarction.  Am J Cardiol. 2003;92(11):1277-1281.PubMedGoogle ScholarCrossref
6.
Panagioti  M, Stokes  J, Esmail  A,  et al.  Multimorbidity and patient safety incidents in primary care: a systematic review and meta-analysis.  PLoS One. 2015;10(8):e0135947.PubMedGoogle ScholarCrossref
7.
Dickens  C, Katon  W, Blakemore  A,  et al.  Does depression predict the use of urgent and unscheduled care by people with long term conditions? a systematic review with meta-analysis.  J Psychosom Res. 2012;73(5):334-342.PubMedGoogle ScholarCrossref
8.
Naylor  C, Parsonage  M, McDaid  D, Knapp  M, Fossey  M, Galea  A.  Long-term Conditions and Mental Health: The Cost of Co-morbidities. London, England: King’s Fund and Centre for Mental Health; 2012.
9.
Gunn  J, Diggens  J, Hegarty  K, Blashki  G.  A systematic review of complex system interventions designed to increase recovery from depression in primary care.  BMC Health Serv Res. 2006;6:88.PubMedGoogle ScholarCrossref
10.
Archer  J, Bower  P, Gilbody  S,  et al.  Collaborative care for depression and anxiety problems.  Cochrane Database Syst Rev. 2012;10:CD006525.PubMedGoogle Scholar
11.
Katon  W, Von Korff  M, Lin  E, Simon  G.  Rethinking practitioner roles in chronic illness: the specialist, primary care physician, and the practice nurse.  Gen Hosp Psychiatry. 2001;23(3):138-144.PubMedGoogle ScholarCrossref
12.
Coventry  PA, Hudson  JL, Kontopantelis  E,  et al.  Characteristics of effective collaborative care for treatment of depression: a systematic review and meta-regression of 74 randomised controlled trials.  PLoS One. 2014;9(9):e108114.PubMedGoogle ScholarCrossref
13.
Community Preventive Services Task Force.  Recommendation from the Community Preventive Services Task Force for use of collaborative care for the management of depressive disorders.  Am J Prev Med. 2012;42(5):521-524.PubMedGoogle ScholarCrossref
14.
Thota  AB, Sipe  TA, Byard  GJ,  et al; Community Preventive Services Task Force.  Collaborative care to improve the management of depressive disorders: a community guide systematic review and meta-analysis.  Am J Prev Med. 2012;42(5):525-538.PubMedGoogle ScholarCrossref
15.
NICE.  Depression in Adults With a Chronic Physical Health Problem: Treatment and Management. Leicester, England: British Psychological Society; 2010.
16.
Bower  P, Gilbody  S, Richards  D, Fletcher  J, Sutton  A.  Collaborative care for depression in primary care: making sense of a complex intervention: systematic review and meta-regression.  Br J Psychiatry. 2006;189:484-493.PubMedGoogle ScholarCrossref
17.
Bartels  SJ, Coakley  EH, Zubritsky  C,  et al; PRISM-E Investigators.  Improving access to geriatric mental health services: a randomized trial comparing treatment engagement with integrated versus enhanced referral care for depression, anxiety, and at-risk alcohol use.  Am J Psychiatry. 2004;161(8):1455-1462.PubMedGoogle ScholarCrossref
18.
Blanchard  MR, Waterreus  A, Mann  AH.  The effect of primary care nurse intervention upon older people screened as depressed.  Int J Geriatr Psychiatry. 1995;10(4):289-298.Google ScholarCrossref
19.
Bruce  ML, Ten Have  TR, Reynolds  CF  III,  et al.  Reducing suicidal ideation and depressive symptoms in depressed older primary care patients: a randomized controlled trial.  JAMA. 2004;291(9):1081-1091.PubMedGoogle ScholarCrossref
20.
Ciechanowski  P, Wagner  E, Schmaling  K,  et al.  Community-integrated home-based depression treatment in older adults: a randomized controlled trial.  JAMA. 2004;291(13):1569-1577.PubMedGoogle ScholarCrossref
21.
Ell  K, Unützer  J, Aranda  M, Gibbs  NE, Lee  PJ, Xie  B.  Maaging depression in home health care: a randomized clinical trial.  Home Health Care Serv Q. 2007;26(3):81-104.PubMedGoogle ScholarCrossref
22.
McCusker  J, Cole  M, Yaffe  M,  et al.  Project Direct: pilot study of a collaborative intervention for depressed seniors.  Can J Commun Ment Health. 2008;27(2):201-218.Google ScholarCrossref
23.
Unützer  J, Katon  W, Callahan  CM,  et al; IMPACT Investigators.  Improving mood-promoting access to collaborative treatment: collaborative care management of late-life depression in the primary care setting: a randomized controlled trial.  JAMA. 2002;288(22):2836-2845.PubMedGoogle ScholarCrossref
24.
Cole  MG, McCusker  J, Elie  M, Dendukuri  N, Latimer  E, Belzile  E.  Systematic detection and multidisciplinary care of depression in older medical inpatients: a randomized trial.  CMAJ. 2006;174(1):38-44.PubMedGoogle ScholarCrossref
25.
Chew-Graham  CA, Lovell  K, Roberts  C,  et al.  A randomised controlled trial to test the feasibility of a collaborative care model for the management of depression in older people.  Br J Gen Pract. 2007;57(538):364-370.PubMedGoogle Scholar
26.
Stewart  GB, Altman  DG, Askie  LM, Duley  L, Simmonds  MC, Stewart  LA.  Statistical analysis of individual participant data meta-analyses: a comparison of methods and recommendations for practice.  PLoS One. 2012;7(10):e46042.PubMedGoogle ScholarCrossref
27.
Clarke  MJ, Stewart  LA.  Obtaining data from randomised controlled trials: how much do we need for reliable and informative meta-analyses?  BMJ. 1994;309(6960):1007-1010.PubMedGoogle ScholarCrossref
28.
Riley  RD, Lambert  PC, Abo-Zaid  G.  Meta-analysis of individual participant data: rationale, conduct, and reporting.  BMJ. 2010;340:c221.PubMedGoogle ScholarCrossref
29.
Harrison  M, Reeves  D, Harkness  E,  et al.  A secondary analysis of the moderating effects of depression and multimorbidity on the effectiveness of a chronic disease self-management programme.  Patient Educ Couns. 2012;87(1):67-73.PubMedGoogle ScholarCrossref
30.
Fried  TR, O’Leary  J, Towle  V, Goldstein  MK, Trentelange  M, Martin  DK.  The effects of comorbidity on the benefits and harms of treatment for chronic disease: a systematic review.  PLoS One. 2014;9(11):e112593.PubMedGoogle ScholarCrossref
31.
Stewart  LA, Clarke  M, Rovers  M,  et al; PRISMA-IPD Development Group.  Preferred Reporting Items for Systematic Review and Meta-analyses of Individual Participant Data: the PRISMA-IPD statement.  JAMA. 2015;313(16):1657-1665.PubMedGoogle ScholarCrossref
32.
Tierney  JF, Vale  C, Riley  R,  et al.  Individual participant data (IPD) meta-analyses of randomised controlled trials: guidance on their use.  PLoS Med. 2015;12(7):e1001855.PubMedGoogle ScholarCrossref
33.
Beck  AT, Steer  RA.  Internal consistencies of the original and revised Beck Depression Inventory.  J Clin Psychol. 1984;40(6):1365-1367.PubMedGoogle ScholarCrossref
34.
Radloff  LS.  The CES-D scale: a self-report depression scale for research in the general population.  Appl Psychol Meas. 1977;1(3):385-401.Google ScholarCrossref
35.
Hamilton  M.  Rating depressive patients.  J Clin Psychiatry. 1980;41(12, pt 2):21-24.PubMedGoogle Scholar
36.
Kroenke  K, Spitzer  RL, Williams  JB.  The PHQ-9: validity of a brief depression severity measure.  J Gen Intern Med. 2001;16(9):606-613.PubMedGoogle ScholarCrossref
37.
Charlson  ME, Pompei  P, Ales  KL, MacKenzie  CR.  A new method of classifying prognostic comorbidity in longitudinal studies: development and validation.  J Chronic Dis. 1987;40(5):373-383.PubMedGoogle ScholarCrossref
38.
Gadbury  GL, Coffey  CS, Allison  DB.  Modern statistical methods for handling missing repeated measurements in obesity trial data: beyond LOCF.  Obes Rev. 2003;4(3):175-184.PubMedGoogle ScholarCrossref
39.
Donders  AR, van der Heijden  GJ, Stijnen  T, Moons  KG.  Review: a gentle introduction to imputation of missing values.  J Clin Epidemiol. 2006;59(10):1087-1091.PubMedGoogle ScholarCrossref
40.
Rubin  DB.  Multiple Imputation for Nonresponse in Surveys. New York, NY: Wiley & Sons; 1987.
41.
Schafer  J.  Analysis of Incomplete Multivariate Data. Boca Raton, FL: Chapman & Hall; 1997.
42.
Stata Press.  Multiple-Imputation Reference Manual. College Station, TX: Stata Press; 2014.
43.
Stewart  LA, Parmar  MK.  Meta-analysis of the literature or of individual patient data: is there a difference?  Lancet. 1993;341(8842):418-422.PubMedGoogle ScholarCrossref
44.
Vickers  AJ, Cronin  AM, Maschino  AC,  et al; Acupuncture Trialists’ Collaboration.  Acupuncture for chronic pain: individual patient data meta-analysis.  Arch Intern Med. 2012;172(19):1444-1453.PubMedGoogle ScholarCrossref
45.
Simmonds  MC, Higgins  JP.  Covariate heterogeneity in meta-analysis: criteria for deciding between meta-regression and individual patient data.  Stat Med. 2007;26(15):2982-2999.PubMedGoogle ScholarCrossref
46.
Whitehead  A.  Meta-analysis of Controlled Clinical Trials. New York, NY: John Wiley & Sons; 2002.
47.
Sutton  AJ, Kendrick  D, Coupland  CA.  Meta-analysis of individual- and aggregate-level data.  Stat Med. 2008;27(5):651-669.PubMedGoogle ScholarCrossref
48.
StataCorp.  Stata Statistical Software: Release 12. College Station, TX: StataCorp LP; 2011.
49.
Kontopantelis  E, Reeves  D.  A short guide and a forest plot command (ipdforest) for one-stage meta-analysis.  Stata J. 2013;13(3):574-587.Google Scholar
50.
Higgins  JP, Thompson  SG.  Quantifying heterogeneity in a meta-analysis.  Stat Med. 2002;21(11):1539-1558.PubMedGoogle ScholarCrossref
51.
Ahmed  I, Sutton  AJ, Riley  RD.  Assessment of publication bias, selection bias, and unavailable data in meta-analyses using individual participant data: a database survey.  BMJ. 2012;344:d7762.PubMedGoogle ScholarCrossref
52.
Higgins  JPT, Altman  DG, Gøtzsche  PC,  et al; Cochrane Bias Methods Group; Cochrane Statistical Methods Group.  The Cochrane Collaboration’s tool for assessing risk of bias in randomised trials.  BMJ. 2011;343:d5928.PubMedGoogle ScholarCrossref
53.
Schulz  KF, Chalmers  I, Hayes  RJ, Altman  DG.  Empirical evidence of bias: dimensions of methodological quality associated with estimates of treatment effects in controlled trials.  JAMA. 1995;273(5):408-412.PubMedGoogle ScholarCrossref
54.
Pildal  J, Hróbjartsson  A, Jørgensen  KJ, Hilden  J, Altman  DG, Gøtzsche  PC.  Impact of allocation concealment on conclusions drawn from meta-analyses of randomized trials.  Int J Epidemiol. 2007;36(4):847-857.PubMedGoogle ScholarCrossref
55.
McMahon  L, Foran  KM, Forrest  SD,  et al.  Graduate mental health worker case management of depression in UK primary care: a pilot study.  Br J Gen Pract. 2007;57(544):880-885.PubMedGoogle ScholarCrossref
56.
Katon  W, Von Korff  M, Lin  E,  et al.  Collaborative management to achieve treatment guidelines. Impact on depression in primary care.  JAMA. 1995;273(13):1026-1031.PubMedGoogle ScholarCrossref
57.
Katon  W, Robinson  P, Von Korff  M,  et al.  A multifaceted intervention to improve treatment of depression in primary care.  Arch Gen Psychiatry. 1996;53(10):924-932.PubMedGoogle ScholarCrossref
58.
Katon  W, Von Korff  M, Lin  E,  et al.  Stepped collaborative care for primary care patients with persistent symptoms of depression: a randomized trial.  Arch Gen Psychiatry. 1999;56(12):1109-1115.PubMedGoogle ScholarCrossref
59.
Katon  W, Rutter  C, Ludman  EJ,  et al.  A randomized trial of relapse prevention of depression in primary care.  Arch Gen Psychiatry. 2001;58(3):241-247.PubMedGoogle ScholarCrossref
60.
Simon  GE, VonKorff  M, Rutter  C, Wagner  E.  Randomised trial of monitoring, feedback, and management of care by telephone to improve treatment of depression in primary care.  BMJ. 2000;320(7234):550-554.PubMedGoogle ScholarCrossref
61.
Simon  GE, Ralston  JD, Savarino  J, Pabiniak  C, Wentzel  C, Operskalski  BH.  Randomized trial of depression follow-up care by online messaging.  J Gen Intern Med. 2011;26(7):698-704.PubMedGoogle ScholarCrossref
62.
Simon  GE, Ludman  EJ, Tutty  S, Operskalski  B, Von Korff  M.  Telephone psychotherapy and telephone care management for primary care patients starting antidepressant treatment: a randomized controlled trial.  JAMA. 2004;292(8):935-942.PubMedGoogle ScholarCrossref
63.
Melville  JL, Reed  SD, Russo  J,  et al.  Improving care for depression in obstetrics and gynecology: a randomized controlled trial.  Obstet Gynecol. 2014;123(6):1237-1246.PubMedGoogle ScholarCrossref
64.
Buszewicz  M, Griffin  M, McMahon  EM, Beecham  J, King  M.  Evaluation of a system of structured, pro-active care for chronic depression in primary care: a randomised controlled trial.  BMC Psychiatry. 2010;10(1):61.PubMedGoogle ScholarCrossref
65.
Gensichen  J, von Korff  M, Peitz  M,  et al; PRoMPT (PRimary care Monitoring for depressive Patients Trial).  Case management for depression by health care assistants in small primary care practices: a cluster randomized trial.  Ann Intern Med. 2009;151(6):369-378.PubMedGoogle ScholarCrossref
66.
Menchetti  M, Sighinolfi  C, Di Michele  V,  et al.  Effectiveness of collaborative care for depression in Italy. A randomized controlled trial.  Gen Hosp Psychiatry. 2013;35(6):579-586.PubMedGoogle ScholarCrossref
67.
Huijbregts  KM, de Jong  FJ, van Marwijk  HW,  et al.  A target-driven collaborative care model for Major Depressive Disorder is effective in primary care in the Netherlands. A randomized clinical trial from the depression initiative.  J Affect Disord. 2013;146(3):328-337.PubMedGoogle ScholarCrossref
68.
Vlasveld  MC, van der Feltz-Cornelis  CM, Adèr  HJ,  et al.  Collaborative care for major depressive disorder in an occupational healthcare setting.  Br J Psychiatry. 2012;200(6):510-511.PubMedGoogle ScholarCrossref
69.
Smit  A, Kluiter  H, Conradi  HJ,  et al.  Short-term effects of enhanced treatment for depression in primary care: results from a randomized controlled trial.  Psychol Med. 2006;36(1):15-26.PubMedGoogle ScholarCrossref
70.
Patel  V, Weiss  HA, Chowdhary  N,  et al.  Effectiveness of an intervention led by lay health counsellors for depressive and anxiety disorders in primary care in Goa, India (MANAS): a cluster randomised controlled trial.  Lancet. 2010;376(9758):2086-2095.PubMedGoogle ScholarCrossref
71.
Wells  KB, Sherbourne  C, Schoenbaum  M,  et al.  Impact of disseminating quality improvement programs for depression in managed primary care: a randomized controlled trial [published correction appears in JAMA. 2000;283(24):3204].  JAMA. 2000;283(2):212-220.PubMedGoogle ScholarCrossref
72.
Adler  DA, Bungay  KM, Wilson  IB,  et al.  The impact of a pharmacist intervention on 6-month outcomes in depressed primary care patients.  Gen Hosp Psychiatry. 2004;26(3):199-209.PubMedGoogle ScholarCrossref
73.
Richards  DA, Lovell  K, Gilbody  S,  et al.  Collaborative care for depression in UK primary care: a randomized controlled trial: Corrigendum.  Psychol Med. 2009;39(4):701.Google ScholarCrossref
74.
Richards  DA, Hill  JJ, Gask  L,  et al.  Clinical effectiveness of collaborative care for depression in UK primary care (CADET): cluster randomised controlled trial.  BMJ. 2013;347:f4913.PubMedGoogle ScholarCrossref
75.
Katon  WJ, Von Korff  M, Lin  EH,  et al.  The Pathways Study: a randomized trial of collaborative care in patients with diabetes and depression.  Arch Gen Psychiatry. 2004;61(10):1042-1049.PubMedGoogle ScholarCrossref
76.
Katon  WJ, Lin  EH, Von Korff  M,  et al.  Collaborative care for patients with depression and chronic illnesses.  N Engl J Med. 2010;363(27):2611-2620.PubMedGoogle ScholarCrossref
77.
Davidson  KW, Bigger  JT, Burg  MM,  et al.  Centralized, stepped, patient preference-based treatment for patients with post-acute coronary syndrome depression: CODIACS vanguard randomized controlled trial.  JAMA Intern Med. 2013;173(11):997-1004.PubMedGoogle ScholarCrossref
78.
Ell  K, Xie  B, Quon  B, Quinn  DI, Dwight-Johnson  M, Lee  P-J.  Randomized controlled trial of collaborative care management of depression among low-income patients with cancer.  J Clin Oncol. 2008;26(27):4488-4496.PubMedGoogle ScholarCrossref
79.
Ell  K, Katon  W, Xie  B,  et al.  Collaborative care management of major depression among low-income, predominantly Hispanic subjects with diabetes: a randomized controlled trial.  Diabetes Care. 2010;33(4):706-713.PubMedGoogle ScholarCrossref
80.
Huffman  JC, Mastromauro  CA, Sowden  GL, Wittmann  C, Rodman  R, Januzzi  JL.  A collaborative care depression management program for cardiac inpatients: depression characteristics and in-hospital outcomes.  Psychosomatics. 2011;52(1):26-33.PubMedGoogle ScholarCrossref
81.
Huffman  JC, Mastromauro  CA, Beach  SR,  et al.  Collaborative care for depression and anxiety disorders in patients with recent cardiac events: the Management of Sadness and Anxiety in Cardiology (MOSAIC) randomized clinical trial.  JAMA Intern Med. 2014;174(6):927-935.PubMedGoogle ScholarCrossref
82.
Rollman  BL, Belnap  BH, LeMenager  MS,  et al.  Telephone-delivered collaborative care for treating post-CABG depression: a randomized controlled trial.  JAMA. 2009;302(19):2095-2103.PubMedGoogle ScholarCrossref
83.
Tierney  JF, Pignon  J-P, Gueffyier  F,  et al; Cochrane IPD Meta-analysis Methods Group.  How individual participant data meta-analyses have influenced trial design, conduct, and analysis.  J Clin Epidemiol. 2015;68(11):1325-1335.PubMedGoogle ScholarCrossref
84.
Piette  JD, Kerr  EA.  The impact of comorbid chronic conditions on diabetes care.  Diabetes Care. 2006;29(3):725-731.PubMedGoogle ScholarCrossref
85.
Simmonds  MC, Higgins  JPT, Stewart  LA, Tierney  JF, Clarke  MJ, Thompson  SG.  Meta-analysis of individual patient data from randomized trials: a review of methods used in practice.  Clin Trials. 2005;2(3):209-217.PubMedGoogle ScholarCrossref
86.
Altman  DG, Deeks  JJ.  Meta-analysis, Simpson’s paradox, and the number needed to treat.  BMC Med Res Methodol. 2002;2:3.PubMedGoogle ScholarCrossref
87.
Kievit  RA, Frankenhuis  WE, Waldorp  LJ, Borsboom  D.  Simpson’s paradox in psychological science: a practical guide.  Front Psychol. 2013;4:513.PubMedGoogle ScholarCrossref
88.
Vale  CL, Rydzewska  LH, Rovers  MM, Emberson  JR, Gueyffier  F, Stewart  LA; Cochrane IPD Meta-analysis Methods Group.  Uptake of systematic reviews and meta-analyses based on individual participant data in clinical practice guidelines: descriptive study.  BMJ. 2015;350:h1088.PubMedGoogle ScholarCrossref
89.
McGregor  M, Lin  EH, Katon  WJ.  TEAMcare: an integrated multicondition collaborative care program for chronic illnesses and depression.  J Ambul Care Manage. 2011;34(2):152-162.PubMedGoogle ScholarCrossref
×