Skip to main content
PMC home page
JAMA Network logoLink to JAMA Network
. 2022 Mar 23;79(5):430–443. doi: 10.1001/jamapsychiatry.2022.0301

Association of Task-Shared Psychological Interventions With Depression Outcomes in Low- and Middle-Income Countries

A Systematic Review and Individual Patient Data Meta-analysis

Eirini Karyotaki 1,✉, Ricardo Araya 2, Ronald C Kessler 3, Ahmed Waqas 4, Arvin Bhana 5,6, Atif Rahman 7, Camila T Matsuzaka 8, Clara Miguel 1, Crick Lund 2,9, Emily C Garman 9, Etheldreda Nakimuli-Mpungu 10, Inge Petersen 5, John A Naslund 11, Marguerite Schneider 9, Siham Sikander 12,13, Mark J D Jordans 14, Melanie Abas 14, Pauline Slade 15, Stephen Walters 16, Traolach S Brugha 17, Toshi A Furukawa 18, Yagmur Amanvermez 1, Marcelo F Mello 19,20, Milton L Wainberg 21, Pim Cuijpers 1, Vikram Patel 11,22
PMCID: PMC8943620  PMID: 35319740

This systematic review and meta-analysis examines data for the outcomes and moderators of task-shared psychological interventions associated with depression in low- and middle-income countries.

Key Points

Question

What are the depression outcomes and moderators associated with task-shared psychological interventions, ie, those delivered by nonspecialist workers, in low- and middle-income countries (LMICs)?

Findings

This systematic review and individual patient data meta-analysis showed that task-shared psychological interventions were associated with significantly larger reduction in depression severity and enhanced response and remission rates compared with control conditions. These outcomes were associated with the presence of psychomotor symptoms, while no other significant associations were identified.

Meaning

The present findings underscore the need for scaling up interventions that use task sharing to reduce the burden of depression in LMICs.

Abstract

Importance

Task sharing, the training of nonspecialist workers with no formal experience in counseling, is a promising strategy for addressing the large gap in treatment for depression in low- and middle-income countries (LMICs).

Objective

To examine the outcomes and moderators of task-shared psychological interventions associated with depression severity, response, and remission.

Data Sources

Systematic literature searches in PubMed, Embase, PsycINFO, and Cochrane Library up to January 1, 2021.

Study Selection

Randomized clinical trials (RCTs) of task-shared psychological interventions compared with control conditions for adults with depressive symptoms in LMICs were included.

Data Extraction and Synthesis

Two researchers independently reviewed the titles, abstracts, and full text of articles from an existing generic meta-analytic database that includes all RCTs on psychotherapy for depression. A systematic review and individual patient data (IPD) meta-analysis was used to estimate the outcomes of task-shared psychological interventions across patient characteristics using mixed-effects models. Procedures for abstracting data and assessing data quality and validity followed the Preferred Reporting Items for Systematic Reviews and Meta-analyses reporting guideline.

Main Outcomes and Measures

Primary outcome was reduction in depression symptom severity measured by the 9-item Patient Health Questionnaire (PHQ-9). Response and remission rates were also estimated.

Results

Of 13 eligible trials, 11 (4145 participants) contributed IPD. Task-shared psychological interventions were associated with a greater decrease in depressive symptom severity than control conditions (Hedges g, 0.32; 95% CI, –0.26 to –0.38). Participants in the intervention groups had a higher chance of responding (odds ratio, 2.11; 95% CI, 1.60 to 2.80) and remitting (odds ratio, 1.87; 95% CI, 1.20 to 1.99). The presence of psychomotor symptoms was significantly associated with the outcomes of task-shared psychological interventions (β [SE], –1.21 [0.39]; P = .002). No other significant associations were identified. Heterogeneity among the trials with IPD was 74% (95% CI, 53%-86%).

Conclusions and Relevance

In this meta-analysis of IPD, task-shared psychological interventions were associated with a larger reduction in depressive symptom severity and a greater chance of response and remission than control conditions. These findings show potential for the use of task-sharing of psychological interventions across different groups of patients with depression. Further research would help identify which people are most likely to benefit and strengthen larger-scale implementation of this strategy to address the burden of depression in LMICs.

Introduction

Depression is a leading cause of the global burden of disease.1 Although psychological interventions effectively promote remission and are recommended as first-line treatment for depression by the World Health Organization, most affected persons in low- and middle-income countries (LMICs) do not have access to them.2,3 A major barrier to improving access to psychological interventions is the lack of skilled mental health practitioners.4,5 Task sharing to the front line, ie, delegating care tasks to community or primary care–based nonspecialist workers, has been advocated to address this barrier.6,7 Several studies have examined the effects of psychological interventions delivered by such workers.8 Recent trials in this field have demonstrated a range of effects in treating depression9,10,11,12 from moderate or large10,11,13 to no effect.12,14 Given the mixed evidence, there is still reluctance to scale up task sharing as a mental health care strategy.15

Moreover, critical outcomes for clinical decision making, such as intervention response and remission, are underreported by randomized clinical trials (RCTs). It also remains unclear whether patient-level factors may influence the responsiveness to task sharing. Notable examples of such factors include clinical and sociodemographic characteristics. Identifying patients who are more or less likely to benefit from these interventions could inform efforts to reach these individuals more efficiently and improve larger-scale implementation of task sharing.

The individual patient data meta-analytic approach, which uses raw data from RCTs, has been increasingly used to synthesize evidence across trials, improve the precision of overall estimates, and maximize the power to identify patient characteristics that moderate intervention outcomes.16 In the present study, we conducted a systematic review and individual patient data meta-analysis (IPD-MA) to examine the outcomes of task-shared psychological interventions (ie, reducing symptom severity, improving response and remission rates) compared with control conditions in adults with depression in LMICs. We also evaluated participant- and study-level characteristics as moderators of treatment outcomes.

Methods

This study was considered exempt from review by the Harvard Longwood Campus institutional review board (IRB). The study was registered with Open Science Framework (https://osf.io/h4kf3) and reported according to Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) reporting guidelines for IPD-MA.17

Eligibility Criteria

We included RCTs that were conducted in LMICs on (1) task-shared psychological interventions that were (2) compared with controls such as treatment as usual (3) for adults (≥18 years old) with depression as established by either a diagnostic interview or cutoff scores on self-report measures (eg, 9-item Patient Health Questionnaire [PHQ-9]18). Psychological interventions were included if they were delivered by nonspecialists (eg, lay counselors, health workers, peers) who were not mental health experts (ie, psychiatrists, psychologists, or psychiatric nurses).

We excluded studies about collaborative care, defined as coordinated multidisciplinary teams with assigned roles and tasks working together to draw individualized plans for patients according to World Health Organization definition.19 Further, self-help and telephone-administered interventions were not eligible for inclusion because they have a different format. We also excluded prevention trials because we focused on treatment. Trials focusing on comorbid depression with other mental health disorders (eg, alcohol misuse) were not excluded by the present study.

Identification of Studies

To identify eligible studies, we searched an existing generic meta-analytic database that includes all RCTs on psychotherapy for depression. This database has been developed based on comprehensive searches in PubMed, Embase, PsycINFO, and Cochrane Library from database inception to January 1, 2021. The full search string for PubMed is provided in the eMethods in the Supplement. In these searches, 2 reviewers (P.C. and E.K.) independently screened the titles, abstracts, and full text of retrieved articles. In case of disagreement, consensus was reached through discussion. A detailed description of this database can be found elsewhere (https://osf.io/825c6/). This generic meta-analytic database was searched by 2 independent reviewers (E.K. and Y.A.) using the eligibility criteria of the present study. Disagreements between the reviewers were resolved through discussion. In addition, we screened meta-analyses of psychological interventions in LMICs20,21,22,23,24 (“reference tracking”) and invited the primary authors of the identified RCTs to indicate any other relevant study they were aware of. Neither reference tracking nor primary author queries resulted in additional RCTs that were not previously identified through our searches.

Data Extraction and Acquisition

We extracted a range of study-level data from the published reports of the trials, including type of psychological intervention, type of control, trial setting, target group, country where the study was conducted, World Bank classification of the country, and data related to the risk-of-bias assessment. We gathered and synthesized all available sociodemographic and clinical characteristics (see the list of moderators with respective definitions in eTable 1 in the Supplement). Individual patient-level variables were chosen based on their availability in the included studies.25 To gather these variables, we contacted the corresponding author of each eligible study to request access to the raw trial data. If there was no response after 1 month, the trial was excluded as unavailable. After checking each data set (no issues identified), we merged the data into the IPD-MA data set.

Quality Assessment

To assess risk of bias in the included studies, we used the revised risk-of-bias tool of the Cochrane Collaboration.26 This tool examines bias arising from (1) the randomization process, (2) deviations from intended interventions, (3) missing outcome data, (4) measurement of the outcome, and (5) selection of reported results. Because the present study is an IPD-MA, we did not evaluate criteria 3 and 5. Incomplete outcome data were addressed by the IPD-MA, and selective reporting was not relevant for our study because we had access to the full data sets. Risk of bias was evaluated based on the information provided in the published articles. If items were unclear, we consulted the authors. Thus, each aspect of the assessment tool was evaluated as low or high risk of bias. The risk of bias was determined by 2 reviewers independently (E.K. and C.M.).

Statistical Analysis

All analyses were conducted with Stata (version 16.0) and R (version 4.0.3) using the “meta” package.27 Our primary outcome was reduction in depressive symptom severity on PHQ-918 postintervention because PHQ-9 was the most commonly used scale across the trials (8/11). Other depression scales were converted into PHQ-9 using conversion algorithms.28,29 To test the effect of the conversion on outcomes, we performed a sensitivity analysis including only the studies that used the PHQ-9 scale. We also examined response rates (50% reduction of baseline depression symptoms) and remission (score less than cutoff that indicated mild depressive symptoms, eg, PHQ-9 < 5) postintervention. Response and remission rates were calculated based on the original depression scales used by the trials.

To examine whether there is a difference between the effects of the studies that provided IPD and those that did not, we performed a conventional meta-analysis using data from the published articles. Regarding the IPD-MA, all analyses were conducted according to the intention-to-treat (ITT) principle. We used multiple imputation to handle incomplete outcome data postintervention (missing-at-random assumption, 20 imputations). We conducted a sensitivity analysis using complete cases to test the robustness of our findings. To calculate the outcomes of task-shared psychological interventions, we merged the IPD from all available studies using the 1-stage IPD-MA with participants nested within trials while adjusting for baseline depression symptom severity.30,31 Under the random-effects model, we performed a mixed-effect linear or logistic regression (depending on whether the outcome was continuous or dichotomous) using the Stata functions xtmixed and meqrlogit, respectively. Symptom severity, response, and remission were the dependent variables; treatment group was the independent variable. The resulting outcome of the mixed effect linear and logistic regressions is a β coefficient, which shows how many SD the dependent variable changes per each SD change in the independent variable. The higher the β value is, the greater the effect. To test the robustness of the findings of the 1-stage IPD-MA, we replicated all outcomes using a 2-stage IPD-MA in which the outcomes per each trial are calculated separately and then pooled together using the random-effects model.16 We also calculated the Hedges g32 for continuous outcomes and number needed to treat (NNT)33 and odds ratio (OR) for binary outcomes to allow a better understanding of the current findings in comparison with previous literature. We converted the main β coefficient to Hedges g based on the procedures described by Lipsey and Wilson.34

We tested whether sociodemographic and clinical variables moderate intervention outcomes postintervention. To examine potential moderators, we added the interaction term between each moderator variable and depression severity, response, and remission rates into the mixed-effects linear or logistic regression model. Each potential moderator variable was added into separate bivariate models. To adjust for multiple testing, we performed the Bonferroni correction,35 and the new P value was .0026 (P = .05 divided by 19, maximum number of moderator analyses = .0026). To examine study-level variables, we ran a series of subgroup analyses, including type of psychological intervention, type of control condition, target group, type of outcome measure, depression diagnosis, income of country, and region.

We measured heterogeneity across the included studies using the I2 statistic with values of 0% indicating no observed heterogeneity and values of 25%, 50%, and 75% indicating low, moderate, and high heterogeneity, respectively. Using the noncentral χ2-based approach,36 we calculated 95% CIs around I2 to give the full magnitude of heterogeneity. We also calculated 95% prediction intervals (PIs) around the pooled effect sizes, showing the range within which the effect of a future study would fall.37 We examined possible publication bias by inspecting the funnel plot on primary outcome measures (also known as a test for small study effects38). If asymmetry due to publication bias was suspected, we tested whether the observed asymmetry was significant by performing an Egger test39 and adjusted the effect for possible publication bias using the Duval and Tweedie trim-and-fill procedure.40

To evaluate the certainty of our main results, we performed the GRADE methodology (eTable 6 in the Supplement).

Results

Study Selection

The systematic literature search resulted in 13 eligible RCTs9,10,11,12,13,14,41,42,43,44,45,46,47 of the 3238 full-text articles screened. We obtained IPD from most of the eligible trials (11/13) and were able to synthesize approximately 94% of all existing IPD (4145/4419 patients). Two data sets9,47 were not available because of data loss9 and no response47 (Figure 1).

Figure 1. PRISMA Individual Patient Data (IPD) Diagram of Study Selection Process.

Figure 1.

Open in a new tab

aDocuments that could not be retrieved from the university library, which were mainly abstracts published for conferences.

bDocuments that did not match the description of the other exclusion categories (eg, trial registrations, replies to letters to the editor).

Study Characteristics

Table 1 shows the characteristics of the included studies. Most of the included studies (10/11) recruited participants through clinical samples, while 1 trial12 recruited participants through the community. Six studies included participants based on elevated depressive symptoms on a self-report measure,10,11,12,14,41,42 and 5 used a diagnostic interview.13,43,44,45,46 Most of the included studies examined mainly the effects of cognitive behavioral therapy–based interventions10,12,14,41,42,43 against enhanced treatment as usual10,11,12,14,41,42,44 in 3 target groups, ie, adults with depression in general,10,42,43,44 women with perinatal depression,11,12,13,14 and people living with HIV and depression.41,45,46 (eTable 2 in the Supplement shows the interventions’ content.) The interventions were delivered by lay counselors,10,41,42,45,46 nonspecialist health workers,14,43,44 or peers.11,12,13 The studies were conducted in 4 low-income countries,41 1 lower-middle income country,10,11,42 and 2 upper-middle income countries.14,44,46

Table 1. Characteristics of Included Studies.

Study Inclusion criteriaa Target group Setting Intervention (No. of participants) Control (No. of participants) Country Region Incomeb
Abas et al,41 2018 PHQ-9 ≥ 5 Adults with HIV HIV clinics PST (14) eTAU (18) Zimbabwe Sub-Saharan Africa Low
Chowdhary et al,42 2016 PHQ-9 > 14 Adults in general Primary care BA&PST (24) eTAU (31) India South Asia Lower-middle
Fuhr et al,11 2019 PHQ-9 > 9 Perinatal depression Antenatal clinics BA&PST (140) eTAU (140) India South Asia Lower-middle
Jordans et al,43 2019 Depression diagnosisc Adults in general Primary care BA (60) TAU (60) Nepal South Asia Low
Lund et al,14 2020 EPDS >12 Perinatal depression Antenatal clinics BA&PST (216) eTAU (209) South Africa Sub-Saharan Africa Upper-middle
Matsuzaka et al,44 2017 MDD, dysthymia (MINI) Adults in general Primary care IPT (43) eTAU (43) Brazil Latin America Upper-middle
Nakimuli-Mpungu et al,45 2020 Depression (MINI) Adults with HIV HIV clinics SUP (578) HIV-c (562) Uganda Sub-Saharan Africa Low
Patel et al,10 2017 PHQ-9 > 14 Adults in general Primary care BA&PST (245) eTAU (248) India South Asia Lower-middle
Petersen et al,46 2014 MDD (SCID)d Adults with HIV HIV clinics IPT (41) HIV-c (35) South Africa Sub-Saharan Africa Upper-middle
Rahman et al,13 2008 MDD (SCID)e Perinatal depression Primary care CBT (463) TAU (440) Pakistan South Asia Low
Sikander et al,12 2019 PHQ-9 > 9 Perinatal depression Villages BA&PST (283) eTAU (287) Pakistan South Asia Low
Open in a new tab

Abbreviations: BA, behavioral activation; CBT, cognitive behavioral therapy; EPDS, Edinburgh Postnatal Depression Scale; eTAU, enhanced treatment as usual; HIV-c, HIV counseling; IPT, interpersonal psychotherapy; MDD, major depressive disorder; MINI, Mini-International Neuropsychiatric Interview; PHQ-9, 9-item Patient Health Questionnaire; PST, problem-solving therapy; SCID, Structural Clinical Interview; SUP, supportive psychotherapy; TAU, treatment as usual.

a

This is based on the eligibility criteria of the studies and does not include all depressive measures assessed by these studies (eg, 3 studies used PHQ-9 to measure depressive symptoms but did not use it as an inclusion criterion).

b

Income level of the country at the time of the study publication was based on the World Bank classification.

c

Inclusion was determined by health worker diagnosis using the Mental Health Gap Action Program (‎mhGAP)‎ guidelines of the World Health Organization for assessment and clinical decision making.

d

The SCID was conducted by a clinical psychologist.

e

The SCID was conducted by a psychiatrist.

Participant Characteristics

Among the 4145 participants, the mean (SD) age was 33 (9.8) years, 2180 (52%) were male, 1750 completed primary education, 3546 (85.5%) were in a relationship, and 1669 (46.8%) were unemployed. Across the included studies, 11.5% of values (479/4145) were missing postintervention, indicating a small study dropout rate (13% in the intervention groups and 10% in the control groups). Mean (SD) score on PHQ-9 was 14.3 (6.5) at baseline and 5.3 (6.2) at the primary end point (mean [SD], 3.7 [1.8] months; range, 2-6 months). Overall, at the primary end point, 67% (2453/3661) of participants showed response and 61.6% (2254/3661) remission. Response rates were 75.4% (1361/1806) for the intervention and 59% (1092/1855) for the control condition whereas remission rates were 69% (1246/1806) for the intervention and 54.3% (1008/1855) for the control condition.

Risk of Bias

Overall, all included studies were at low risk of bias across most domains, except for bias in measurement of the outcome. All trials were at low risk of bias arising from the randomization process and deviation from the intended intervention. (Descriptions of training and supervision of nonspecialists appear in eTable 3 in the Supplement.) Missing data were handled by the present IPD-MA using multiple imputation, while the percentage of missing values was small across the studies (up to 20.7%) and acceptably balanced between the intervention and control conditions. Most of the studies used measures administered by a blind assessor, while 2 did not perform blinding (eTable 3 in the Supplement).

Results of Conventional Meta-analysis

The conventional meta-analysis of the 13 eligible trials showed that task-shared psychological interventions resulted in a significantly larger reduction in depressive symptom severity compared with control conditions postintervention (Hedges g, 0.48; 95% CI, 0.26-0.68; P < .001). Heterogeneity was high I2 = 86% (95% CI, 78%-91%). We found no evidence of a difference between studies providing IPD and those that did not (between subgroups P = .52).

Results of the IPD-MA

Table 2 presents the findings of the 1-stage IPD-MA on depressive symptom severity. Task-shared psychological interventions were significantly associated with greater reduction in depressive symptom severity compared with control conditions (β [SE], –2.11 [0.51]; g, 0.32; 95% CI, 0.26-0.38; P < .001). Complete case and sensitivity analyses including only the studies that originally used PHQ-9 showed similar outcomes. Of the individual participant-level factors, only the presence of psychomotor symptoms at baseline (n = 2628 participants experienced either agitation or retardation) was associated with intervention outcome (β [SE], –1.21 [0.39]; P = .002), suggesting that the outcomes of intervention are more pronounced when individuals experience psychomotor symptoms at baseline. This association was confirmed in both complete case analysis and sensitivity analysis including only the studies that originally used PHQ-9. No other significant associations were identified.

Table 2. Mixed-Effects Restricted Maximum Likelihood Model Outcomes on Depressive Symptom Severity, 1-Stage IPD-MAa.

Full sample Complete case analysisb
Nobs (Ns) β coefficient (SE) P value Nobs (Ns) β coefficient (SE) P value
Main effects: depression severity
Baseline severity 4118 0.13 (0.02) <.001 3660 0.13 (0.02) <.001
Group (11) –2.11 (0.51) <.001 (11) –2.37 (0.53) <.001
Sensitivity analysis (PHQ-9 studies only)
Baseline severity 0.35 (0.05) <.001 1469 0.34 (0.04) <.001
Group –2.29 (0.65) <.001 (8) –2.54 (0.65) <.001
Moderators
Age
Baseline severity 4118 0.13 (0.02) <.001 3660 0.13 (0.02)
Group (11) –1.64 (0.84) .005 (11) –2.14 (0.83) .01
Age (continuous) 0.03 (0.01) .03 0.02 (0.01) .07
Age × group –0.01 (0.02) .50 –0.01 (0.02) .72
Sex
Baseline severity 4118 0.13 (0.02) <.001 3660 0.13 (0.02) <.001
Group (11) –2.06 (0.55) <.001 (11) –2.31 (0.56) <.001
Men 0.25 (0.37) .49 0.07 (0.35) .84
Sex × treatment group –0.13 (0.51) .80 –0.17 (0.49) .72
Educational levelc
Baseline severity 4118 0.13 (0.02) <.001 3660 0.13 (0.02) <.001
Group (11) –2.33 (0.66) <.001 (11) –2.54 (0.68) <.001
Primary –0.65 (0.36) .07 –0.75 (0.34) .03
Secondary –0.87 (0.40) .03 –0.90 (0.37) .01
Tertiary –1.54 (0.76) .04 –1.51 (0.70) .03
Other 0.47 (1.27) .71 1.02 (1.22) .41
Primary × group 0.74 (0.49) .13 0.83 (0.48) .08
Secondary × group –0.06 (0.54) .92 –0.20 (0.52) .70
Tertiary × group –0.27 (1.05) .79 –0.65 (1.03) .53
Other × group –1.16 (1.81) .52 –1.73 (1.75) .32
P value of educational level × group .43 .19
Relationship status
Baseline severity 4118 0.13 (0.02) <.001 3660 0.13 (0.02) <.001
Group (11) –2.42 (0.67) <.001 (11) –2.64 (0.67) <.001
In a relationship 0.067 (0.37) .86 0.02 (0.37) .96
Relationship × group 0.38 (0.54) .48 0.33 (0.52) .53
Employment statusd
Baseline severity 3537 0.12 (0.02) <.001 3194 0.12 (0.02) <.001
Group (10) –2.35 (0.65) <.001 (10) –2.56 (0.67) <.001
Employed 0.09 (0.42) .82 0.20 (0.39) .62
Student –0.76 (0.97) .44 –0.75 (0.93) .42
Other 0.65 (0.40) .10 0.77 (0.38) .04
Employed × group 0.39 (0.58) .50 0.32 (0.57) .57
Student × group 0.95 (1.47) .52 0.81 (1.38) .56
Other × group –0.57 (0.55) .30 –0.70 (0.53) .18
P value of employment status × group .28 .17
Baseline severity of depression
Baseline severity 4118 0.16 (0.03) <.001 3660 0.16 (0.02) <.001
Group (11) –1.35 (0.73) .06 (11) –1.52 (0.73) .04
Baseline severity × group –0.05 (0.04) .15 –0.06 (0.03) .10
Depression duration
Baseline severity 1645 0.29 (0.04) <.001 1405 0.31 (0.04) <.001
Group (4) –2.02 (0.86) .02 (4) –2.47 (0.90) .01
Duration in months 0.003 (0.003) .35 0.003 (0.003) .33
Duration × group 0.001 (0.01) .72 0.002 (0.005) .66
Loss of interest in daily activities
Baseline severity 4113 0.13 (0.02) <.001 3655 0.13 (0.02) <.001
Group (11) –2.16 (0.73) .003 (11) –2.40 (0.72) .001
Loss of interest (yes) 0 0.07 (0.41) .87 0.08 (0.39) .84
Loss of interest × group 0.06 (0.59) .92 0.03 (0.55) .92
Depressed mood
Baseline severity 4113 0.13 (0.02) <.001 3655 0.13 (0.02) <.001
Group (11) –1.78 (0.76) .02 (11) –1.93 (0.76) .01
Depressed mood (yes) 0.17 (0.44) .70 0.22 (0.43) .60
Depressed mood × group –0.35 (0.62) .56 –0.48 (0.61) .43
Sleep problems
Baseline severity 4111 0.13 (0.02) <.001 3653 0.13 (0.02) <.001
Group (11) –1.61 (0.63) .01 (11) –1.66 (0.64) .009
Sleep problems (yes) 0.64 (0.32) .05 0.79 (0.31) .01
Sleep problems × group –0.61 (0.45) .17 –0.86 (0.43) .05
Tiredness
Baseline severity 4026 0.11 (0.02) <.001 3652 0.11 (0.02) <.001
Group (11) –1.53 (0.62) .01 (11) –1.65 (0.62) .008
Tiredness (yes) 1.60 (0.32) <.001 1.75 (0.31) <.001
Tiredness × group –0.71 (0.44) .11 –0.83 (0.43) .05
Concentration problems
Baseline severity 4112 0.13 (0.02) <.001 3654 0.13 (0.02) <.001
Group (11) –1.87 (0.63) .003 (11) –2.13 (0.64) .001
Concentration (yes) 0.50 (0.34) .14 0.54 (0.32) .09
Concentration × group –0.29 (0.47) .54 –0.31 (0.46) .51
Appetite change
Baseline severity 4113 0.13 (0.02) <.001 3655 0.13 (0.02) <.001
Group (11) –2.31 (0.61) <.001 (11) –2.57 (0.62) <.001
Appetite change (yes) 0.19 (0.31) .54 0.19 (0.29) .53
Appetite change × group –0.26 (0.43) .61 0.25 (0.41) .54
Sense of worthlessness/guilt
Baseline severity 4112 0.13 (0.02) <.001 3654 0.13 (0.02) <.001
Group (11) –1.68 (0.60) .005 (11) –1.91 (0.62) .002
Sense of worthlessness/guilt (yes) 0.18 (0.31) .56 0.24 (0.29) .41
Sense of worthlessness/guilt × group –0.57 (0.42) .16 –0.64 (0.40) .11
Psychomotor symptoms
Baseline severity 4111 0.13 (0.02) <.001 3653 0.14 (0.02) <.001
Group (11) –1.36 (0.54) .001 (11) –1.49 (0.55) .007
Psychomotor symptoms (yes) 0.56 (0.28) .05 0.68 (0.26) .01
Psychomotor × group –1.21 (0.39) .002e –1.45 (0.37) <.001e
Suicidal ideation
Baseline severity 4111 0.12 (0.02) <.001 3653 0.11 (0.02) <.001
Group (11) –1.85 (0.53) <.001 (11) –2.12 (0.26) .001
Suicidal ideation (yes) 0.83 (0.28) .003 0.89 (0.26) .001
Suicidal ideation × group –0.63 (0.37) .09 –0.63 (0.36) .08
Domestic violence
Baseline severity 1560 0.04 (0.02) .06 1401 0.03 (0.02) .04
Group (2) –0.48 (0.29) .09 (2) –0.67 (0.24) .005
Domestic violence (yes) 0.79 (0.27) .004 0.90 (0.26) .001
Domestic violence × group –0.16 (0.47) .73 –0.07 (0.41) .86
Problematic alcohol drinking
Baseline severity 2509 0.08 (0.02) <.001 2278 0.07 (0.02) <.001
Group (8) –1.69 (0.55) .002 (8) –1.89 (0.55) .001
Problematic alcohol drinking (yes) 0.64 (0.40) .11 0.76 (0.37) .04
Alcohol × group –0.09 (0.58) .88 –0.25 (0.53) .64
Comorbid physical disorder
Baseline severity 1327 0.01 (0.01) .45 1259 0.01 (0.01) .27
Group (5) –1.64 (1.34) .22 (5) –1.45 (1.26) .25
Comorbid physical disorder (yes) 0.11 (0.92) .91 0.38 (0.79) .63
Comorbid physical disorder × group –1.11 (1.38) .42 –1.65 (1.19) .16
Open in a new tab

Abbreviations: Nobs, number of observations; Ns, number of studies; PHQ-9, 9-item Patient Health Questionnaire.

a

Parameters are standardized β weights of the composite of PHQ-9 scores; 2-tailed P values are presented.

b

This sensitivity analysis was conducted only with participants who completed a postintervention depression questionnaire.

c

Reference group was illiteracy.

d

Reference group was unemployment.

e

Significant association.

The 2-stage IPD-MA resulted in a g of 0.32 (95% CI, 0.18-0.46; P < .001) in favor of task-shared psychological interventions. The PIs ranged from g = −0.12 to 0.76. Heterogeneity was 74% (95% CI, 53%-86%), and there was no indication of publication bias. Similar outcomes were observed in complete case and sensitivity analyses. Subgroup analyses showed no evidence of a difference between target patient groups, studies that originally used PHQ-9 and those that did not, types of interventions, control conditions, income of country, and region. Results of the 2-stage IPD-MA are presented in Figure 2 and in eTable 4 and eFigure 1 in the Supplement.

Figure 2. Effects of Task-Shared Psychological Interventions Compared With Controls on Depression Symptom Severity at Postintervention Assessment.

Figure 2.

Open in a new tab

Table 3 presents the findings of the 1-stage IPD-MA on response and remission. Overall, the likelihood of response and remission was significantly higher in the intervention compared with control groups (response: β [SE], 0.75 [0.14]; OR, 2.11, 95% CI, 1.60-2.80; remission: β [SE], 0.63 [0.15]; OR, 1.87; 95% CI, 1.20-1.99; P < .001) with broad PIs (eFigures 2 through 5 in the Supplement). Complete case analyses resulted in comparable outcomes. Moderator analysis showed that the chance of remission and response after task-shared psychological interventions was significantly higher among individuals with psychomotor symptoms. Moreover, the 2-stage IPD-MA resulted in identical findings with the 1-stage IPD-MA for both response and remission. Similar results were observed in complete case and sensitivity analyses. No evidence of a difference was observed between the examined subgroups. Furthermore, we found no evidence of publication bias (eTable 5 and eFigures 2 through 5 in the Supplement).

Table 3. Mixed-Effects Maximum Likelihood Model Outcomes on Response and Remission, 1-Stage IPD-MAa .

Response Remission
Full sample Complete case analysisb Full sample Complete case analysisb
Nobs (Ns) β coefficient (SE) P value Nobs (Ns) β (SE) P value Nobs (Ns) β (SE) P value Nobs (Ns) β (SE) P value
Main effects 4118 3661 4118 3661
Group (11) 0.75 (0.14) <.001 (11) 0.89 (0.16) <.001 (11) 0.63 (0.15) <.001 (11) 0.79 (0.17) <.001
Moderators
Age
Group 4118 –0.02 (.001) .01 3661 –0.01 (0.01) .01 4118 0.31 (0.34) .37 3661 0.53 (0.36) .14
Age (11) 0.44 (0.32) .17 (11) 0.71 (0.35) .04 (11) –0.02 (0.01) .003 (11) –0.02 (0.01) .001
Age × group 0.01 (0.01) .28 0.01 (0.01) .55 0.01 (0.01) .30 0.01 (0.01) .42
Sex
Group 4118 0 .66 (0.17) <.001 3661 0.79 (0.19) <.001 4118 0.58 (0.18) .001 3661 0.74 (0.20) <.001
Men (11) –0.18 (0.18) .31 (11) –0.10 (0.18) .56 (11) –0.13 (0.22) .54 (11) 0.05 (0.23) .81
Sex × group 0.19 (0.24) .42 0.24 (0.25) .34 0.11 (0.25) .66 0.15 (0.28) .60
Educational levelc
Group 4118 0.84 (0.23) <.001 3661 0.97 (0.25) <.001 4118 0.65 (0.24) .006 3661 0.77 (0.27) .004
Primary (11) 0.17 (0.15) .26 (11) 0.24 (0.15) .13 (11) 0.16 (0.17) .33 (11) 0.22 (0.17) .20
Secondary 0.16 (0.16) .30 0.21 (0.16) .21 0.21 (0.17) .20 0.24 (0.17) .16
Tertiary 0.30 (0.28) .29 0.31 (0.30) .30 0.25 (0.29) .40 0.18 (0.31) .55
Other –0.07 (0.50) .88 –0.13 (0.51) .78 0.10 (0.49) .83 0.10 (0.51) .85
Primary × group –0.28 (0.22) .20 –0.33 (0.23) .15 –0.23 (0.23) .32 –0.25 (0.24) .29
Secondary × group –0.001 (0.23) .99 0.07 (0.24) .77 0.17 (0.23) .45 0.31 (0.25) .20
Tertiary × group 0.37 (0 .47) .42 0.63 (0.49) .20 0.50 (0.43) .25 0.80 (0.47) .09
Other × group 0.21 (0.74) .77 0.28 (0.75) .71 –0.62 (0.71) .38 –0.62 (0.74) .40
P value of educational level × group .48 .24
Relationship status
Group 4118 0.97 (0.27) <.001 3661 1.12 (0.29) <.001 4118 0.73 (0.29) .01 3661 0.87 (0.31) .006
In a relationship (11) 0.01 (0.18) .05 (11) 0.04 (0.18) .20 (11) –0.01 (0.22) .94 (11) –0.01 (0.23) .95
Relationship × group –0.27 (0.27) .27 –0.27 (–0.95) .34 –0.12 (0.30) .68 –0.09 (0.31) .79
Employment statusd
Group 3537 0.90 (0.18) <.001 3195 1.03 (0.20) <.001 3537 0.82 (0.20) <.001 3195 0.95 (0.21) <.001
Employed (10) 0.08 (0.18) .64 (10) 0.08 (0.18) .68 (10) –0.14 (0.19) .46 (10) –0.24 (0.20) .24
Student 0.38 (0.37) .30 0.44 (0.38) .26 0.46 (0.38) .22 0.55 (0.38) .15
Other –0.13 (0.20) .51 –0.22 (0.21) .30 –0.22 (0.23) .34 –0.41 (0.25) .11
Employed × group –0.19 (0.26) .47 –0.18 (0.27) .51 –0.15 (0.26) .56 –0.09 (0.28) .74
Student × group –.058 (0.54) .28 –0.66 (0.57) .25 –1.06 (0.54) .05 –1.17 (0.57) .04
Other × group 0.07 (0.27) .78 0.17 (0.29) .55 –0.03 (0.29) .92 –0.09 (0.32) .76
P value of employment status × group .85 .50 .28 .08
Baseline severity of depression
Group 4118 0.57 (0.29) .05 3661 0.65 (0.32) .04 4118 0.62 (0.31) .05 3661 0.79 (0.35) .02
Baseline severity (11) 0.03 (0.01) <.001 (11) 0.03 (0.01) .01 (11) –0.07 (0.01) <.001 (11) –0.08 (0.01) <.001
Baseline severity × group 0.01 (0.02) .55 0.01 (0.02) .42 0.004 (0.02) .83 0.005 (0.20) .81
Depression duration
Group 1645 0.79 (0.20) <.001 1405 0.98 (0.20) <.001 1645 0.58 (0.24) .01 1405 0.69 (0.25) .006
Duration in months (4) –0.001 (0.001) .64 (4) –0.001 (0.001) .56 (4) –0.001 (0.001) .43 (4) –0.001 (0.001) .48
Duration × group –0.002 (0.002) .35 –0.002 (0.002) .26 0.001 (0.002) .66 0.001 (0.002) .67
Loss of interest
Group 4113 0.72 (0.30) .02 3656 0.80 (0.31) .01 4113 0.55 (0.30) .06 3656 0.70 (0.33) .03
Loss of interest (yes) (11) 0.02 (0.20) .92 (11) 0.001 (0.20) .99 (11) –0.35 (0.21) .10 (11) –0.42 (0.22) .06
Loss of interest × group 0.03 (0.29) .92 0.10 (0.29) .73 0.08 (0.28) .78 0.10 (0.31) .76
Depressed mood
Group 4113 0.52 (0.30) .08 3656 0.64 (0.31) .04 4113 0.70 (0.30) .02 3656 0.90 (0.33) .006
Depressed mood (yes) (11) –0.09 (0.20) .66 (11) –0.14 (0.20) .50 (11) –0.29 (0.20) .14 (11) –0.36 (0.21) .09
Depressed mood × group 0.25 (0.29) .40 0.28 (0.30) .34 –0.078 (0.29) .79 –0.12 (0.30) .70
Sleep problems
Group 4111 0.57 (0.21) .008 3654 0.64 (0.23) .005 4111 0.69 (0.22) .002 3654 0.81 (0.24) .001
Sleep problems (yes) (11) –0.01 (0.13) .94 (11) –0.07 (0.13) .60 (11) –0.31 (0.14) .03 (11) –0.40 (0.14) .004
Sleep problems × group 0.22 (0.19) .26 0.31 (0.19) .12 –0.07 (0.20) .74 –0.01 (0.20) .96
Tiredness
Group 4026 0.59 (0.22) .01 3653 0.67 (0.24) .005 4026 0.50 (0.22) .03 3653 0.61 (0.24) .01
Tiredness (yes) (11) –0.56 (0.14) <.001 (11) –0.65 (0.15) <.001 (11) –0.76 (0.14) <.001 (11) –0.86 (0.15) <.001
Tiredness × group 0.20 (0.21) .35 0.26 (0.21) .22 0.17 (0.20) .39 0.22 (0.21) .29
Concentration problems
Group 4112 0.88 (0.24) <.001 3655 1.11 (0.25) <.001 4112 0.59 (0.25) .02 3655 0.72 (0.26) .007
Concentration (yes) (11) 0.18 (0.15) .24 (11) 0.20 (0.15) .20 (11) –0.38 (0.16) .02 (11) –0.46 (0.16) .005
Concentration × group –0.17(0.24) .47 –0.27 (0.24) .26 0.05 (0.23) .81 0.10 (0.24) .68
Appetite change
Group 4113 0.99 (0.21) <.001 3656 1.15 (0.23) <.001 4113 0.80 (0.22) <.001 3656 1.00 (0.23) <.001
Appetite change (yes) (11) 0.13 (0.13) .29 (11) 0.13 (0.13) .31 (11) –0.05 (0.13) .73 (11) –0.03 (0.14) .84
Appetite change × group –0.31 (0.19) .11 –0.33 (0.20) .10 –0.21 (0.19) .26 –0.26 (0.20) .19
Sense of worthlessness/guilt
Group 4112 0.67 (0.20) .001 3655 0.80 (0.22) <.001 4112 0.47 (0.21) .03 3655 0.57 (0.23) .01
Worthlessness/guilt (yes) (11) 0.17 (0.13) .18 (11) 0.17 (0.13) .18 (11) –0.26 (0.13) .05 (11) –0.34 (0.14) .01
Worthlessness/guilt × group 0.11 (0.19) .57 0.14 (0.19) .46 0.22 (0.19) .25 0.30 (0.20) .12
Psychomotor symptoms
Group 4111 0.41 (0.16) .01 3654 0.46 (0.18) .01 4111 0.31 (0.17) .07 3654 0.36 (0.18) .05
Psychomotor symptoms (yes) (11) –0.20 (0.11) .09 (11) –0.28 (0.12) .01 (11) –0.39 (0.12) .002 (11) –0.51 (0.12) <.001
Psychomotor symptoms × group 0.56 (0.16) .001e 0.72 (0.17) <.001e 0.55 (0.17) .002e 0.74 (0.17) <.001e
Suicidal ideation
Group 4111 0.69 (0.16) <.001 3654 0.84 (0.18) <.001 4111 0.62 (0.16) <.001 3654 0.77 (0.19) <.001
Suicidal ideation (yes) (11) –0.04 (0.11) .70 (11) –0.07 (0.12) .57 (11) –0.32 (0.12) .008 (11) –0.40 (0.13) .002
Suicidal ideation × group 0.12 (0.17) .45 0.1 (0.17) .49 0.07 (0.17) .66 0.10 (0.17) .56
Domestic violence
Group 1560 0.47 (0.32) .14 1401 0.90 (0.51) .08 1560 0.11 (0.19) .56 1401 0.20 (0.20) .31
Domestic violence (yes) (2) 0.002 (0.29) .99 (2) 0.05 (0.30) .86 (2) –0.23 (0.31) .46 (2) –0.31 (0.36) .39
Domestic violence × group –0.12 (0.48) .81 –0.44 (0.48) .36 –0.004 (0.47) .99 0.01 (0.53) .98
Problematic alcohol drinking
Group 2509 0.75 (0.18) <.001 2279 0.92 (0.21) <.001 2509 0.64 (0.17) <.001 2279 0.80 (0.19) <.001
Problematic alcohol drinking (yes) (8) –0.35 (0.22) .11 (8) –0.45 (0.23) .04 (8) –0.38 (0.26) .14 (8) –0.62 (0.27) .02
Alcohol × group .014 (0.37) .97 –0.03 (0.38) .94 –0.17 (0.36) .63 0.02 (0.41) .95
Open in a new tab

Abbreviations: Nobs, number of observations; Ns, number of studies.

a

Parameters are standardized β weights of the composite of 9-item Patient Health Questionnaire scores; 2-tailed P values are presented.

b

This sensitivity analysis was conducted only with participants who completed a postintervention depression questionnaire.

c

Reference group was illiteracy.

d

Reference group was unemployment.

e

Significant association.

The GRADE assessment of main outcomes (Grading of Recommendations, Assessment, Development and Evaluations) showed moderate strength of the resulting evidence (eTable 6 in the Supplement).

Discussion

In this study, we analyzed individual patient data from 11 RCTs to study the depression outcomes of task-shared psychological interventions for adults with depression in LMICs and to identify moderators of these outcomes. Task-shared psychological interventions were associated with a larger reduction in depressive symptom severity and a greater chance of response and remission than control measures (moderate strength of evidence). We also found that the presence of psychomotor symptoms was associated with more pronounced effects of task-shared psychological interventions. None of the other participant- or study-level factors were associated with the intervention outcomes.

The present findings are in line with previous reviews on interventions delivered by nonspecialists for common mental disorders in LMICs.7,8,23,24 However, our novel methodological approach provides more robust estimates of the diverse outcomes of task-shared psychological interventions associated with depression, including response, remission, NNTs, and participant- and study-level moderators, which to our knowledge have not been reported earlier. We found that 7 individuals need to be treated to expect 1 individual with a 50% reduction in baseline depressive symptoms, while the NNT for remission was 8. Although these NNTs are relatively large, their magnitude should be interpreted considering that the delivery model of these interventions is through the lowest-cost human resource in the community, and control participants often received enhanced treatment as usual. Such NNTs are still promising because task-shared psychological interventions may have a significant effect when scaled up and delivered to large populations. Notably, the NNTs found by the present IPD-MA are comparable with those of 2 of the most common antidepressant medications, based on previous research mainly conducted in high-income countries, ie, paroxetine (NNT = 5.6 based on standardized mean difference [SMD] = –0.32) and fluoxetine (NNT = 7.7 based on SMD = –0.23), when compared with pill placebo.48

To our knowledge, the association of psychomotor symptoms with intervention outcomes has not been identified by previous literature on task sharing for depression. However, previous research has suggested that presence of psychomotor retardation is associated with functional impairment, depression severity, and treatment prognosis.25,49 The higher response in patients with psychomotor symptoms may be partly associated with the type of intervention. Most of the included studies evaluated a cognitive behavioral therapy intervention that involved behavioral activation, a skill that may be particularly relevant to patients with psychomotor symptoms. Nevertheless, future studies are needed to replicate this finding to draw robust conclusions on the association of psychomotor symptoms with the response to task-shared psychological interventions.

Limitations

The present findings should be interpreted considering several limitations. First, the included studies were conducted across 7 LMICs, suggesting that our findings cannot be generalized to all LMICs. Second, although we could test the association of a wide range of participant characteristics with the intervention outcomes, our analysis was limited to variables examined by the included studies. Thus, we could not investigate the role of some clinically important variables associated with depression prognosis50 (eg, number of previous episodes, existence of other psychiatric conditions such as anxiety, substance use disorders, neurocognitive impairments, etc). Third, some of the examined moderators (eg, domestic violence) were available only in a small number of trials, limiting our conclusions for the respective associations. Nevertheless, the number of participants was large in all moderator analyses (>1300), suggesting that the statistical power was adequate. Fourth, similar to previous meta-analyses on studies in LMICs,21 we found moderate to large heterogeneity and broad prediction intervals across most of our analyses, which might be associated with various reasons, including the differences between the examined settings (ie, primary care, antenatal clinics, HIV clinics, and community), comorbidities, type of care worker and the quality of their training, and contextual determinants. However, we did not confirm such differences in subgroup analyses (eg, target group). Thus, the present findings should be interpreted cautiously because of the unexplained heterogeneity.

Fifth, most of the examined interventions involved cognitive behavioral therapy techniques. Still, in some of the included studies, these techniques had to be simplified and adapted for use in settings where participants and care workers have limited general or health literacy or training. Nevertheless, this is a commonly done practice in these and other settings,51 as adaptation to local contexts is an essential step in the design of intervention studies. Sixth, we observed high response and remission rates among participants in the control groups. Such rates are possibly associated with the active control groups used by most of the included trials (ie, enhanced usual care and HIV counseling). It is therefore possible that participants in the control groups received more substantial care than they would typically receive in these low-resource settings. This hypothesis needs further investigation in future research. Further, although we excluded collaborative care studies, some collaborative care strategies may have been implicit in both groups of the trials we included, for example, because of trial procedures requiring certain types of participants to be reviewed by a physician (eg, in case of suicidal risk). These strategies would have been equally applicable in both groups. Further, in this analysis, we focused only on depression, but patients in these settings may concurrently experience other common mental health problems such as anxiety and posttraumatic stress. Future research should examine the effects of task-shared psychological interventions in patients with common mental disorders in LMICs.

Conclusions

Despite these limitations, our results showed that task-shared psychological interventions were associated with promising depression outcomes and may be particularly well-suited to patients with psychomotor symptoms. Moreover, these outcomes were not associated with several other patient- and study-level factors that were assessed in the examined trials, suggesting the generalizability of the findings to diverse populations.

Considering the limited availability of mental health professionals in all countries of the world, and particularly so in LMICs,7,8 our study shows that it is possible and beneficial to use nonspecialist workers in the delivery of psychological interventions for most patients with depression. Scaling up this delivery model is probably a unique, low-cost, and widely accessible approach to reducing the burden of depression in LMICs.

Supplement.

eMethods. Search strings for PubMed

eTable 1. List of moderators with definitions

eTable 2. Description of interventions

eTable 3. Risk of bias assessment

eTable 4. Depression severity – Two stage IPDMA

eFigure 1. Depression severity – Two stage IPDMA, complete case sample

eTable 5. Response and remission – Two-stage IPDMA

eFigure 2. Response – Two stage-IPDMA, full imputed sample

eFigure 3. Response – Two stage-IPDMA, complete case sample

eFigure 4. Remission – Two stage-IPDMA, full imputed sample

eFigure 5. Remission – Two stage-IPDMA, complete case sample

eTable 6. GRADE assessment of main outcomes

eReferences

Click here for additional data file. (625.1KB, pdf)

References

  • 1.Herrman H, Kieling C, McGorry P, Horton R, Sargent J, Patel V. Reducing the global burden of depression: a Lancet-World Psychiatric Association Commission. Lancet. 2019;393(10189):e42-e43. doi: 10.1016/S0140-6736(18)32408-5 [DOI] [PubMed] [Google Scholar]
  • 2.Rathod S, Pinninti N, Irfan M, et al. Mental health service provision in low- and middle-income countries. Health Serv Insights. 2017;10:1178632917694350. doi: 10.1177/1178632917694350 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 3.Jacob KS. Mental health services in low-income and middle-income countries. Lancet Psychiatry. 2017;4(2):87-89. doi: 10.1016/S2215-0366(16)30423-0 [DOI] [PubMed] [Google Scholar]
  • 4.Patel V, Chowdhary N, Rahman A, Verdeli H. Improving access to psychological treatments: lessons from developing countries. Behav Res Ther. 2011;49(9):523-528. doi: 10.1016/j.brat.2011.06.012 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 5.Patel V. Mental health: in the spotlight but a long way to go. Int Health. 2019;11(5):324-326. doi: 10.1093/inthealth/ihz060 [DOI] [PubMed] [Google Scholar]
  • 6.Raviola G, Naslund JA, Smith SL, Patel V. Innovative models in mental health delivery systems: task sharing care with non-specialist providers to close the mental health treatment gap. Curr Psychiatry Rep. 2019;21(6):44. doi: 10.1007/s11920-019-1028-x [DOI] [PubMed] [Google Scholar]
  • 7.Papola D, Purgato M, Gastaldon C, et al. Psychological and social interventions for the prevention of mental disorders in people living in low-and middle-income countries affected by humanitarian crises. Cochrane Database Syst Rev. 2020;9(9):CD012417. doi: 10.1002/14651858.CD012417.pub2 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 8.van Ginneken N, Chin WY, Lim YC, et al. Primary-level worker interventions for the care of people living with mental disorders and distress in low-and middle-income countries. Cochrane Database Syst Rev. 2021;8(8):CD009149. doi: 10.1002/14651858.CD009149.pub3 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 9.Bolton P, Bass J, Neugebauer R, et al. Group interpersonal psychotherapy for depression in rural Uganda: a randomized controlled trial. JAMA. 2003;289(23):3117-3124. doi: 10.1001/jama.289.23.3117 [DOI] [PubMed] [Google Scholar]
  • 10.Patel V, Weobong B, Weiss HA, et al. The Healthy Activity Program (HAP), a lay counsellor-delivered brief psychological treatment for severe depression, in primary care in India: a randomised controlled trial. Lancet. 2017;389(10065):176-185. doi: 10.1016/S0140-6736(16)31589-6 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 11.Fuhr DC, Weobong B, Lazarus A, et al. Delivering the Thinking Healthy Programme for perinatal depression through peers: an individually randomised controlled trial in India. Lancet Psychiatry. 2019;6(2):115-127. doi: 10.1016/S2215-0366(18)30466-8 [DOI] [PubMed] [Google Scholar]
  • 12.Sikander S, Ahmad I, Atif N, et al. Delivering the Thinking Healthy Programme for perinatal depression through volunteer peers: a cluster randomised controlled trial in Pakistan. Lancet Psychiatry. 2019;6(2):128-139. doi: 10.1016/S2215-0366(18)30467-X [DOI] [PubMed] [Google Scholar]
  • 13.Rahman A, Malik A, Sikander S, Roberts C, Creed F. Cognitive behaviour therapy-based intervention by community health workers for mothers with depression and their infants in rural Pakistan: a cluster-randomised controlled trial. Lancet. 2008;372(9642):902-909. doi: 10.1016/S0140-6736(08)61400-2 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 14.Lund C, Schneider M, Garman EC, et al. Task-sharing of psychological treatment for antenatal depression in Khayelitsha, South Africa: effects on antenatal and postnatal outcomes in an individual randomised controlled trial. Behav Res Ther. 2020;130:103466. doi: 10.1016/j.brat.2019.103466 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 15.Fulton BD, Scheffler RM, Sparkes SP, Auh EY, Vujicic M, Soucat A. Health workforce skill mix and task shifting in low income countries: a review of recent evidence. Hum Resour Health. 2011;9:1. doi: 10.1186/1478-4491-9-1 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 16.Riley RD, Lambert PC, Abo-Zaid G. Meta-analysis of individual participant data: rationale, conduct, and reporting. BMJ. 2010;340:c221. doi: 10.1136/bmj.c221 [DOI] [PubMed] [Google Scholar]
  • 17.Stewart LA, Clarke M, Rovers M, et al. ; PRISMA-IPD Development Group . Preferred Reporting Items for a Systematic Review and Meta-Analysis of individual participant data: the PRISMA-IPD statement. JAMA. 2015;313(16):1657-1665. doi: 10.1001/jama.2015.3656 [DOI] [PubMed] [Google Scholar]
  • 18.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. doi: 10.1046/j.1525-1497.2001.016009606.x [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 19.World Health Organization . Depression and other common mental disorders: global health estimates. Published 2017. https://apps.who.int/iris/bitstream/handle/10665/254610/WHO-MSD-MER-2017.2-eng.pdf
  • 20.Chibanda D, Cowan FM, Healy JL, Abas M, Lund C. Psychological interventions for common mental disorders for people living with HIV in low- and middle-income countries: systematic review. Trop Med Int Health. 2015;20(7):830-839. doi: 10.1111/tmi.12500 [DOI] [PubMed] [Google Scholar]
  • 21.Cuijpers P, Karyotaki E, Reijnders M, Purgato M, Barbui C. Psychotherapies for depression in low- and middle-income countries: a meta-analysis. World Psychiatry. 2018;17(1):90-101. doi: 10.1002/wps.20493 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 22.Kohrt BA, Asher L, Bhardwaj A, et al. The role of communities in mental health care in low- and middle-income countries: a meta-review of components and competencies. Int J Environ Res Public Health. 2018;15(6):E1279. doi: 10.3390/ijerph15061279 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 23.Singla DR, Kohrt BA, Murray LK, Anand A, Chorpita BF, Patel V. Psychological treatments for the world: lessons from low-and middle-income countries. Annu Rev Clin Psychol. 2017;13:149-181. doi: 10.1146/annurev-clinpsy-032816-045217 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 24.van Ginneken N, Tharyan P, Lewin S, et al. Non-specialist health worker interventions for the care of mental, neurological and substance-abuse disorders in low-and middle-income countries. Cochrane Database Syst Rev. 2013;(11):CD009149. doi: 10.1002/14651858.CD009149.pub2 [DOI] [PubMed] [Google Scholar]
  • 25.Kessler RC, van Loo HM, Wardenaar KJ, et al. Using patient self-reports to study heterogeneity of treatment effects in major depressive disorder. Epidemiol Psychiatr Sci. 2017;26(1):22-36. doi: 10.1017/S2045796016000020 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 26.Higgins JP, 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. doi: 10.1136/bmj.d5928 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 27.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. doi: 10.1136/ebmental-2019-300117 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 28.Wahl I, Löwe B, Bjorner JB, et al. Standardization of depression measurement: a common metric was developed for 11 self-report depression measures. J Clin Epidemiol. 2014;67(1):73-86. doi: 10.1016/j.jclinepi.2013.04.019 [DOI] [PubMed] [Google Scholar]
  • 29.Furukawa TA, Reijnders M, Kishimoto S, et al. Translating the BDI and BDI-II into the HAMD and vice versa with equipercentile linking. Epidemiol Psychiatr Sci. 2020;29:e24. doi: 10.1017/S2045796019000088 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 30.Debray TP, Moons KG, Abo-Zaid GMA, Koffijberg H, Riley RD. Individual participant data meta-analysis for a binary outcome: one-stage or two-stage? PLoS One. 2013;8(4):e60650. doi: 10.1371/journal.pone.0060650 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 31.Stewart LA, Parmar MK. Meta-analysis of the literature or of individual patient data: is there a difference? Lancet. 1993;341(8842):418-422. doi: 10.1016/0140-6736(93)93004-K [DOI] [PubMed] [Google Scholar]
  • 32.Hedges LV, Olkin I. Statistical Methods for Meta-analysis. Academic Press; 2014. [Google Scholar]
  • 33.Kraemer HC, Kupfer DJ. Size of treatment effects and their importance to clinical research and practice. Biol Psychiatry. 2006;59(11):990-996. doi: 10.1016/j.biopsych.2005.09.014 [DOI] [PubMed] [Google Scholar]
  • 34.Lipsey MW, Wilson DB. Practical Meta-analysis. SAGE Publications, Inc; 2001. [Google Scholar]
  • 35.Bonferroni C. Teoria statistica delle classi e calcolo delle probabilita. Pubblicazioni del R Istituto Superiore di Scienze Economiche e Commerciali di Firenze. 1936;8:3-62. [Google Scholar]
  • 36.Heterogi: Stata module to quantify heterogeneity in a meta-analysis [computer program]. Revised January 25, 2006. Boston College Department of Economics; 2005. [Google Scholar]
  • 37.Borenstein M, Higgins JP, Hedges LV, Rothstein HR. Basics of meta-analysis: I2 is not an absolute measure of heterogeneity. Res Synth Methods. 2017;8(1):5-18. doi: 10.1002/jrsm.1230 [DOI] [PubMed] [Google Scholar]
  • 38.Sterne JA, Sutton AJ, Ioannidis JP, et al. Recommendations for examining and interpreting funnel plot asymmetry in meta-analyses of randomised controlled trials. BMJ. 2011;343:d4002. doi: 10.1136/bmj.d4002 [DOI] [PubMed] [Google Scholar]
  • 39.Egger M, Davey Smith G, Schneider M, Minder C. Bias in meta-analysis detected by a simple, graphical test. BMJ. 1997;315(7109):629-634. doi: 10.1136/bmj.315.7109.629 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 40.Duval S, Tweedie R. Trim and fill: a simple funnel-plot-based method of testing and adjusting for publication bias in meta-analysis. Biometrics. 2000;56(2):455-463. doi: 10.1111/j.0006-341X.2000.00455.x [DOI] [PubMed] [Google Scholar]
  • 41.Abas M, Nyamayaro P, Bere T, et al. Feasibility and acceptability of a task-shifted intervention to enhance adherence to HIV medication and improve depression in people living with HIV in Zimbabwe, a low income country in sub-Saharan Africa. AIDS Behav. 2018;22(1):86-101. doi: 10.1007/s10461-016-1659-4 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 42.Chowdhary N, Anand A, Dimidjian S, et al. The Healthy Activity Program lay counsellor delivered treatment for severe depression in India: systematic development and randomised evaluation. Br J Psychiatry. 2016;208(4):381-388. doi: 10.1192/bjp.bp.114.161075 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 43.Jordans MJD, Luitel NP, Garman E, et al. Effectiveness of psychological treatments for depression and alcohol use disorder delivered by community-based counsellors: two pragmatic randomised controlled trials within primary healthcare in Nepal. Br J Psychiatry. 2019;215(2):485-493. doi: 10.1192/bjp.2018.300 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 44.Matsuzaka CT, Wainberg M, Norcini Pala A, et al. Task shifting interpersonal counseling for depression: a pragmatic randomized controlled trial in primary care. BMC Psychiatry. 2017;17(1):225. doi: 10.1186/s12888-017-1379-y [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 45.Nakimuli-Mpungu E, Musisi S, Wamala K, et al. Effectiveness and cost-effectiveness of group support psychotherapy delivered by trained lay health workers for depression treatment among people with HIV in Uganda: a cluster-randomised trial. Lancet Glob Health. 2020;8(3):e387-e398. doi: 10.1016/S2214-109X(19)30548-0 [DOI] [PubMed] [Google Scholar]
  • 46.Petersen I, Hanass Hancock J, Bhana A, Govender K. A group-based counselling intervention for depression comorbid with HIV/AIDS using a task shifting approach in South Africa: a randomized controlled pilot study. J Affect Disord. 2014;158:78-84. doi: 10.1016/j.jad.2014.02.013 [DOI] [PubMed] [Google Scholar]
  • 47.Yuan J, Yin Y, Tang X, et al. Culturally adapted and lay-delivered cognitive behaviour therapy for older adults with depressive symptoms in rural China: a pilot trial. Behav Cogn Psychother. Published online October 29, 2020. doi: 10.1017/S1352465820000818 [DOI] [PubMed] [Google Scholar]
  • 48.Cipriani A, Furukawa TA, Salanti G, et al. Comparative efficacy and acceptability of 21 antidepressant drugs for the acute treatment of adults with major depressive disorder: a systematic review and network meta-analysis. Lancet. 2018;391(10128):1357-1366. doi: 10.1016/S0140-6736(17)32802-7 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 49.Bennabi D, Vandel P, Papaxanthis C, Pozzo T, Haffen E. Psychomotor retardation in depression: a systematic review of diagnostic, pathophysiologic, and therapeutic implications. Biomed Res Int. Published online October 30, 2013. doi: 10.1155/2013/158746 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 50.Bockting CL, Hollon SD, Jarrett RB, Kuyken W, Dobson K. A lifetime approach to major depressive disorder: the contributions of psychological interventions in preventing relapse and recurrence. Clin Psychol Rev. 2015;41:16-26. doi: 10.1016/j.cpr.2015.02.003 [DOI] [PubMed] [Google Scholar]
  • 51.Dias A, Azariah F, Anderson SJ, et al. Effect of a lay counselor intervention on prevention of major depression in older adults living in low-and middle-income countries: a randomized clinical trial. JAMA Psychiatry. 2019;76(1):13-20. doi: 10.1001/jamapsychiatry.2018.3048 [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplement.

eMethods. Search strings for PubMed

eTable 1. List of moderators with definitions

eTable 2. Description of interventions

eTable 3. Risk of bias assessment

eTable 4. Depression severity – Two stage IPDMA

eFigure 1. Depression severity – Two stage IPDMA, complete case sample

eTable 5. Response and remission – Two-stage IPDMA

eFigure 2. Response – Two stage-IPDMA, full imputed sample

eFigure 3. Response – Two stage-IPDMA, complete case sample

eFigure 4. Remission – Two stage-IPDMA, full imputed sample

eFigure 5. Remission – Two stage-IPDMA, complete case sample

eTable 6. GRADE assessment of main outcomes

eReferences

Click here for additional data file. (625.1KB, pdf)

Articles from JAMA Psychiatry are provided here courtesy of American Medical Association

RESOURCES