The Effectiveness of Community-Based Mental Health Interventions Delivered by Lay Health Workers in Sub-Saharan Africa: A Systematic Review and Meta-Analysis

Rationale

Prevalent psychiatric conditions, including depressive and anxiety disorders, constitute an escalating global health burden with particularly severe consequences for lower-income nations.^[1,2] Current estimates from the World Health Organization (WHO) indicate that more than one billion individuals experience mental health conditions, with depression ranking among the foremost contributors to disability-adjusted life years globally.^[3]Within sub-Saharan African regions, this burden is especially pronounced, exacerbated by socioeconomic deprivation, communicable disease pandemics including HIV/AIDS, armed conflicts, and fragile healthcare infrastructure.^[4,5]

A pronounced disparity between need and service provision persists, with more than 85% of individuals with psychiatric disorders in lower-income settings remaining untreated.^[6,7]his therapeutic void stems predominantly from critical shortages of specialized mental health personnel. Across sub-Saharan Africa, psychiatrist density averages below 0.1 per 100,000 individuals, contrasting sharply with rates exceeding 9 per 100,000 in affluent nations.^[8] Such workforce constraints render specialist-dependent care models impractical and economically untenable.

In response to this challenge, the WHO established the Mental Health Gap Action Program (mhGAP), promoting task redistribution whereby mental healthcare responsibilities are transferred from specialized clinicians to non-specialized personnel, including nursing staff, community health agents, and trained volunteers.^[9] This strategy has gained recognition as a potentially effective mechanism for broadening the availability of empirically supported psychological treatments in under-resourced contexts. Prior evidence syntheses have documented favorable outcomes for task-redistributed interventions targeting common psychiatric disorders across lower-income countries.^[10,11] Still, a focused synthesis of evidence from SSA is needed to inform regional policy and practice.

Objectives

The present systematic review with meta-analysis seeks to consolidate available evidence regarding the efficacy of community-implemented, non-specialist-delivered psychological treatments for prevalent mental disorders within sub-Saharan African populations. The principal aim was to quantify the aggregate impact of such interventions on depressive symptomatology relative to standard care or waitlist conditions. Secondary objectives included examining the heterogeneity of impact across different intervention types, populations, and settings.

Protocol and registration

This review adhered to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) 2020 reporting standards. Prior to study commencement, the protocol received prospective registration with PROSPERO under identifier CRD420251141329.

Eligibility criteria

Study selection

Electronic database searches yielded 4,708 unique citations. Following the removal of 1,584 duplicate records, 3,124 titles and abstracts underwent screening, resulting in the exclusion of 2,989 records. Full-text review was conducted for 135 articles, with 125 subsequently excluded based on: non-randomized design (n=45), facility-based delivery (n=28), specialist-delivered interventions (n=22), focus on severe psychiatric disorders (n=12), protocol publications without outcome data (n=8), inadequate statistical reporting (n=7), and absence of standard care comparator (n=3). Ten trials satisfied all eligibility criteria and were incorporated into the systematic review and quantitative synthesis [Figure 1 and Supplementary file S3]].

Close

Figure 1:

PRISMA Flow diagram. RCT: Randomized controlled trials

Export to PPT

Study characteristics

The ten eligible randomized trials were implemented during 2014-2022, spanning seven sub-Saharan African nations: Zimbabwe (n=1), South Africa (n=2), Uganda (n=2), Tanzania (n=2), Nigeria (n=1), Kenya (n=1), and Zambia (n=1). Collectively, these trials enrolled 4,761 participants, with individual study samples ranging from 34 to 1,140. Implementation contexts included primary healthcare facilities (n = 5), HIV treatment centers (n = 4), educational institutions (n = 1), and prenatal care clinics (n = 1). Geographic distribution comprised six urban studies, three with mixed urban-rural recruitment, and one exclusively rural trial. Target populations included general adult samples with depressive or common mental disorders (n=5), persons living with HIV (n=4), orphaned and vulnerable youth (n=1), school-enrolled adolescents (n=1), and pregnant women with HIV (n=1). Therapeutic modalities included problem-solving therapy (n=2), interpersonal psychotherapy (n=1), group support psychotherapy (n=2), cognitive-behavioral and trauma-focused cognitive behavioral therapy (CBT) approaches (n=2), stepped care frameworks (n=2), and positive psychology (n=1). All interventions were administered by community workers who received abbreviated training (ranging from 5 days to 4 weeks). Treatment intensity varied from 4 to 12 sessions (median: 8) delivered over 4 weeks to 12 months (median: 3 months). Comparison conditions included enhanced usual care (n = 4), standard care (n = 4), group health education (n = 2), and study skills training (n = 1). Primary endpoint assessments occurred between 4 weeks and 12 months post-baseline, with a median follow-up of 6 months [Table 1].^[13-22]

Risk of bias in studies

A detailed bias assessment is presented in Supplementary file S4. Eight trials (80%) demonstrated low risk across all evaluated domains. Two studies (20%) elicited concerns regarding substantial participant loss (Petersen et al.,^[13]2014) or limited sample size (Osborn et al.,^[14] 2021). No trials received overall high-risk ratings. All ten trials employed appropriate randomization methods, receiving low-risk ratings for allocation procedures. Analytic approaches (intention-to-treat or modified intention-to-treat) were judged appropriate across all studies. Eight trials demonstrated low attrition risk, while two raised concerns due to substantial dropout (Petersen et al.,^[13] 2014, 45% retention) or imbalanced attrition (Osborn et al.,^[14] 2021).Validated outcome instruments were used universally, yielding a low risk of measurement bias. All trials reported pre-registered outcomes, indicating a low risk of selective reporting.

Results of individual studies

Among the 10 included studies, six (60%) reported statistically significant benefits of the intervention compared to the control condition. Chibanda et al.^[15] (2016) in Zimbabwe evaluated the Friendship Bench intervention, a problem-solving therapy delivered by lay health workers in primary care clinics. Among 521 participants, the intervention group showed a significantly greater reduction in symptoms of common mental disorders on the Shona Symptom Questionnaire (SSQ-14) at 6 months compared to the enhanced usual care group, with a large effect size (SMD = -1.12, 95% CI -1.31 to -0.94, p<0.001). Petersen et al.^[13] (2014) in South Africa conducted a pilot RCT of group-based interpersonal psychotherapy for HIV-positive adults with depression. Among the 34 enrolled participants, the treatment group exhibited markedly lower Patient Health Questionnaire-9 (PHQ-9) scores at 3 months compared to standard care, reflecting a large effect magnitude (SMD = -1.35, 95% CI -2.09 to -0.60, p < 0.01). NakimuliMpungu et al.^[16] and colleagues (2015) investigated group support psychotherapy for depression among HIV-positive individuals in Uganda. Within a cohort of 618 participants, the intervention arm exhibited significant symptom improvement on the Self-Reporting Questionnaire-20 (SRQ-20) at 6-month assessment compared with group health education, demonstrating a moderate-to-large effect (SMD = -0.73, 95% CI -1.19 to -0.31, p < 0.01). Murray et al.^[17] and colleagues (2015) examined trauma-focused cognitive behavioral therapy for orphaned and vulnerable youth in Zambia. The study, involving 257 participants, found a very large effect on trauma symptoms post-treatment (SMD = -2.39, 95% CI -2.71 to -2.07, p<0.001) compared to treatment as usual. Nakimuli-Mpungu et al.^[20] and colleagues (2020) subsequently confirmed group support psychotherapy efficacy for HIV-associated depression in a larger Ugandan trial. Among 1,018 participants, the intervention produced a very large reduction in major depression odds at 6 months (SMD = -1.28, 95% CI -1.42 to -1.14, p<0.001) relative to group health education. Kaaya et al.^[22], 2022 evaluated a stepped care model for perinatal depression in HIV-positive Tanzanian women. While the primary outcome at 9 months was not significant, the secondary outcome at 6 weeks postpartum showed a significant and medium-to-large effect on depression symptoms (SMD = -0.61, 95% CI -0.76 to -0.46, p<0.001) compared to enhanced usual care [Table 2].

Four studies (40%) found no statistically significant difference between the intervention and control groups: Gureje et al.^[18] (2019) in Nigeria conducted a large cluster RCT of a stepped-care intervention for major depression in primary care. This trial, encompassing 1,035 participants, revealed no significant difference in depression remission at 12 months between stepped care and an enhanced usual care condition that also incorporated mhGAP-IG training (SMD = 0.02, 95% CI -0.10 to 0.15, p = 0.163). This null result suggests enhanced usual care may achieve comparable effectiveness to more intensive stepped-care models in this setting. Dow et al.^[19] and colleagues (2020) examined trauma-focused CBT for HIV-positive adolescents in Tanzania. Among the 93 enrolled participants, a small and non-significant effect on depressive symptoms was observed at 6 months (SMD = -0.28, 95% CI -0.69 to 0.14, p = 0.19). Osborn et al.^[14] and colleagues (2021) conducted a pilot trial in Kenya evaluating the Shamiri positive psychology intervention for school-based adolescents. Among 51 participants, the intervention yielded a small and non-significant effect on depression and anxiety at post-treatment relative to the study skills control (SMD = -0.35, 95% CI -0.91 to 0.21, p = 0.22). Petersen et al.^[21]2017 examined a collaborative care model for depression in hypertensive patients in South Africa. This large pragmatic cluster trial, involving 925 participants, detected no significant difference in depression response at 12 months between collaborative care and usual care arms (SMD = -0.20, 95% CI -0.33 to -0.07, p = 0.583).

Results of syntheses

A meta-analysis of all 10 studies (4,761 participants) revealed that community-based interventions delivered by lay health workers significantly reduced symptoms of depression compared to control conditions (pooled SMD = -0.82, 95% CI -1.24 to -0.40, p < 0.001). This represents a large effect according to Cohen's conventions (small: 0.2, medium: 0.5, large: 0.8).

Heterogeneity

Statistical heterogeneity was considerable (I² = 97.7%, Q = 392.39, df = 9, p<0.001), indicating substantial variability in effect sizes across studies. This heterogeneity is expected given the diversity of interventions (problem-solving therapy, interpersonal psychotherapy, cognitive behavioral therapy, stepped care, positive psychology), populations (general adults, HIV+, perinatal women, adolescents, orphans), settings (primary care, HIV clinics, schools, antenatal clinics), and outcome measures (PHQ-9, SSQ-14, SRQ-20, etc.). We conducted subgroup analyses to explore potential sources of this heterogeneity.

Subgroup analyses: We conducted subgroup analyses to explore sources of heterogeneity, examining the influence of population, intervention, and comparator types. The results are presented in Table 3. The analysis by comparator type revealed a notable difference: studies comparing the intervention to standard/minimal care showed a larger pooled effect size (SMD = -1.08) than those comparing to enhanced usual care (SMD = -0.47), although this difference was not statistically significant (p=0.12). No significant differences were found for population type or intervention type [Figure 2].

Close

Figure 2:

Forest plot of depression outcomes. CI: Confidence interval

Export to PPT

Reporting biases

Visual inspection of the funnel plot [Figure 3] suggested a relatively symmetrical distribution of studies, indicating a low risk of publication bias. This was supported by formal statistical testing. Egger's regression test for funnel plot asymmetry was not significant (intercept = -3.92, p = 0.51), and Begg's rank correlation test also showed no evidence of publication bias (Kendall's tau = -0.33, p = 0.22).

Close

Figure 3:

Funnel plot: Assessment of publication bias

Export to PPT

Certainty of evidence

Using GRADE criteria, the certainty of evidence for the primary outcome of depression symptoms was rated as moderate This rating reflects a two-level downgrade due to inconsistency resulting from very high heterogeneity (I² = 97.7%), while no downgrades were applied to other domains. Risk of bias was considered not serious, as most studies were at low risk; indirectness was not serious, given the direct relevance of populations, interventions, and outcomes; imprecision was not serious due to the large sample size and narrow confidence intervals; and publication bias was undetected, as there were insufficient studies for formal assessment. The observed heterogeneity appears to stem from genuine diversity in interventions, populations, and settings, rather than methodological flaws. The direction of effect consistently favored the intervention, despite variability in its magnitude [Supplementary file S2].

Summary of main findings

This systematic review and meta-analysis provide robust evidence that community-based mental health interventions delivered by lay health workers are effective in reducing symptoms of common mental disorders in sub-Saharan Africa. The pooled effect size (SMD = -0.82, 95% CI -1.24 to -0.40, p<0.001) represents a clinically meaningful benefit that is comparable to effects observed for specialist-delivered psychotherapy in high-income countries. Of the 10 included studies involving 4,761 participants from seven countries, six studies (60%) demonstrated statistically significant benefits of the intervention, while four studies (40%) found no significant difference between intervention and control groups. The considerable heterogeneity observed in this meta-analysis (I² = 97.7%) reflects the diversity of interventions, populations, and contexts included in the review. This heterogeneity should be viewed as a strength rather than a limitation, as it demonstrates that lay health worker-delivered interventions can be effective across a range of real-world implementation scenarios, including different intervention types (problem-solving therapy, interpersonal psychotherapy, cognitive behavioral therapy, stepped care), diverse populations (people living with HIV, perinatal women, adolescents, orphans, adults with hypertension), and varied settings (primary care clinics, HIV clinics, schools, antenatal clinics, community centers).

Interpretation of results in context

The findings show that six out of ten studies demonstrated significant benefits, while four showed null results, which warrants careful interpretation. Notably, three of the four null studies (Gureje et al.^[18], 2019; Petersen et al.^[21] 2017; Kaaya et al.^[22] 2022 compared the intervention to enhanced usual care that included some mental health components. For example, in the Gureje et al.^[18], 2019 trial in Nigeria, both the stepped-care intervention group and the control group received enhanced usual care that involved training primary care providers in the WHO Mental Health Gap Action Programme Intervention Guide (mhGAP-IG). The null finding in this study suggests that enhanced usual care with provider training alone may be as effective as a more intensive stepped-care model, at least in the context of well-resourced primary care settings. Similarly, Petersen et al.^[21] 2017 compared collaborative care to usual care in a pragmatic effectiveness trial, and the smaller effect size may reflect the challenges of implementing complex interventions in real-world settings with limited resources and competing demands.

In contrast, the studies that showed large effects (Chibanda et al.^[15] 2016, Petersen et al.^[13] 2014, Nakimuli-Mpungu et al.^[16] 2015, Murray et al.^[17] 2015, Nakimuli-Mpungu et al.^[20] 2020) typically compared the intervention to minimal usual care or waitlist controls, representing a more stringent test of intervention efficacy. The very large effect sizes observed in Murray et al.^[17] (2015) (SMD = -2.39) and Nakimuli-Mpungu (2020) (SMD = -1.28) are particularly noteworthy, as they demonstrate that lay health workers can achieve outcomes comparable to or exceeding those typically observed with specialist-delivered care.

Our findings align closely with systematic reviews examining task-shifting for mental health in low- and middle-income countries (LMICs) more broadly. A landmark meta-analysis by Singla et al. .^[10] synthesized evidence from 27 RCTs across multiple LMICs, including studies from South Asia (India, Pakistan, Nepal), Latin America (Chile, Peru), and Africa. They found that psychological treatments delivered by non-specialists were effective for depression (SMD = -0.49, 95% CI -0.62 to -0.36) and anxiety (SMD = -0.59, 95% CI -0.76 to -0.42). Our pooled effect size of SMD = -0.62 for depression symptoms in sub-Saharan Africa is slightly larger than the global LMIC estimate, suggesting that task-shifting approaches may be particularly effective in the African context, possibly due to the strong community orientation and cultural acceptability of lay health workers in many African societies.

Studies from Asia have provided particularly strong evidence for task-shifting mental health interventions. The MANAS trial in India (Patel et al.^[5], which involved lay health counselors supervised by mental health specialists delivering collaborative stepped care for depression and anxiety in primary care, found significant benefits at 12 months (OR for recovery = 2.18, 95% CI 1.27 to 3.75). In South Asia, the Thinking Healthy Program in Pakistan trained community health workers to deliver cognitive-behavioral therapy for perinatal depression in rural settings.^[23] This cluster RCT with 818 mothers demonstrated large, sustained effects, with the intervention group showing significantly lower rates of major depression at both 6 months (23% vs 53%, OR 0.22, 95% CI 0.14-0.36) and 12 months (27% vs 59%, OR 0.23, 95% CI 0.15-0.36) compared to usual care.^[23] The adjusted mean difference in depression scores was -5.86 at 6 months and -6.65 at 12 months, representing effect sizes comparable to those observed in several of our included SSA studies. These findings suggest that the effectiveness of task-shared interventions is not unique to SSA but represents a generalizable phenomenon across diverse LMIC contexts.

Our findings are also comparable to meta-analyses of psychotherapy effectiveness in high-income countries, suggesting that the quality of care delivered by lay health workers in sub-Saharan Africa is not inferior to that provided by specialists in resource-rich settings. A comprehensive meta-analysis by Cuijpers et al.^[24] Examining psychotherapy for major depression in adults in high-income countries found an overall effect size of SMD = -0.70 (95% CI -0.75 to -0.64), which is only slightly larger than our finding of SMD = -0.62 in sub-Saharan Africa. This comparison is critical because it challenges the assumption that mental health care in low-resource settings is necessarily of lower quality or effectiveness than care in high-income countries.

Furthermore, a meta-analysis by Karyotaki et al.^[11] Specifically examined task-shared psychological interventions globally, including studies from both high-income and low-income countries. They found that task-shared interventions were associated with significant reductions in depression (SMD = -0.52) and anxiety (SMD = -0.50) symptoms, with no significant difference in effectiveness between task-shared and specialist-delivered interventions. This finding reinforces our conclusion that lay health workers can deliver mental health care that is as effective as care provided by specialists.

Studies from the United States and Europe have also demonstrated the effectiveness of collaborative care models that involve non-specialist providers. The IMPACT trial in the United States (Unützer et al.^[25], which involved care managers (often social workers or nurses without specialized mental health training) working alongside primary care physicians and psychiatric consultants, found significant improvements in depression outcomes (effect size, d = 0.50). A systematic review by Hoeft et al.^[26] Examining task-sharing approaches in rural areas of high-income countries found that community health workers and primary care providers could effectively deliver mental health interventions with appropriate training and supervision.

Mechanisms of effectiveness

Several mechanisms may explain the effectiveness of lay health worker-delivered interventions in sub-Saharan Africa. First, lay health workers are often members of the communities they serve, which may enhance cultural appropriateness, reduce stigma, and improve therapeutic alliance. Cultural concordance between providers and patients has been shown to improve engagement and outcomes in mental health care. Second, community-based delivery reduces barriers related to transportation, cost,and stigma associated with visiting specialist mental health facilities. Third, the use of evidence-based, manualized interventions with clear protocols and supervision structures ensures that lay health workers deliver care with fidelity. Fourth, integrating mental health care into existing health platforms (such as HIV clinics, antenatal clinics, and primary care) leverages existing infrastructure and relationships, making mental health care more accessible and acceptable. The role of supervision and ongoing support for lay health workers cannot be overstated. All included studies provided regular supervision by mental health specialists or trained supervisors, which likely contributed to the interventions' effectiveness. Supervision provides opportunities for case consultation, skill development, emotional support, and quality assurance.

Addressing heterogeneity

The high heterogeneity observed in this meta-analysis (I² = 97.7%) warrants further discussion. While high heterogeneity can be a limitation in meta-analyses, in the context of implementation science and global mental health, it can also be interpreted as evidence of the intervention's robustness across diverse real-world settings. The consistent direction of the effect in favor of the intervention, despite the variability in its magnitude, suggests that these interventions are effective across a wide range of populations, intervention types, and contexts. However, this heterogeneity also underscores the need for future research to identify the specific intervention components and implementation strategies that are most effective in different settings. The included studies varied across multiple dimensions: intervention type (PST, IPT, CBT, stepped care), population (general adults, HIV-positive individuals, perinatal women, adolescents, orphans), setting (primary care, HIV clinics, schools, community centers), comparison condition (usual care, enhanced usual care, waitlist), and follow-up duration (immediate post-treatment to 12 months). We did not conduct formal subgroup analyses due to the small number of studies in each subgroup. Still, visual inspection of the forest plot suggests that studies with more intensive interventions (e.g., Murray et al.^[17] 2015 with TF-CBT for trauma, Nakimuli-Mpungu et al.^[20] 2020 with group support psychotherapy) tended to show larger effects than studies with less intensive interventions or those comparing to enhanced usual care. Future research with larger numbers of studies should conduct meta-regression analyses to identify moderators of intervention effectiveness, such as intervention intensity, provider training duration, supervision frequency, and comparison condition type.

Implications for policy and practice

These findings have important implications for mental health policy and practice in sub-Saharan Africa. First, they provide strong evidence to support the scale-up of task-shifting approaches to address the mental health treatment gap. With fewer than one psychiatrist per 500,000 people in many African countries, relying solely on specialist providers is not a viable strategy for expanding access to mental health care. Task-shifting to lay health workers, when implemented with appropriate training, supervision, and quality assurance mechanisms, can significantly expand the reach of evidence-based mental health interventions.

The findings demonstrate that mental health interventions can be successfully integrated into existing health platforms, including HIV care, maternal and child health services, and primary care. This integration approach is consistent with the WHO's mhGAP and can leverage existing infrastructure, relationships, and funding streams. Integration also helps to reduce stigma by normalizing mental health care as part of routine health services.

The cost-effectiveness of lay health worker-delivered interventions makes them particularly attractive for resource-limited settings. While only a subset of the included studies reported economic data, those that did (e.g., NakimuliMpungu et al.^[20], 2020) found that the interventions were highly cost-effective, with costs per disability-adjusted life year (DALY) averted ranging from US$13 to US$50, which is well below the WHO threshold for cost-effectiveness in low-income countries.

The evidence supports the need for investment in training, supervision, and support systems for lay health workers. Effective task-shifting requires more than simply delegating tasks to less-trained providers; it necessitates a systematic approach that encompasses competency-based training, ongoing supervision, quality assurance mechanisms, and supportive policies and regulations.

Limitations

This review has several significant limitations that should be considered when interpreting the findings. First, the high heterogeneity (I² = 97.7%) limits our ability to identify specific intervention components or implementation strategies associated with greater effectiveness. While we view this heterogeneity as evidence of effectiveness across diverse contexts, it also means that we cannot make definitive recommendations about which specific interventions or approaches are most effective. Future research should conduct component analyses and implementation research to identify the active ingredients of effective interventions.

Also, most studies had relatively short follow-up periods (median, 6 months), which limited conclusions about the long-term sustainability of intervention effects. Mental health conditions are often chronic and recurrent, and it is essential to know whether the benefits of lay health worker-delivered interventions are maintained over time. Only three studies (Chibanda et al.^[15] 2016; Gureje et al.^[18], 2019; Petersen et al.^[21], 2017) had follow-up periods of 12 months or longer, and these studies yielded mixed results. Future research should include extended follow-up periods to assess the durability of intervention effects and the need for maintenance or booster sessions.

We were unable to systematically assess cost-effectiveness across all studies because only a subset reported economic data. While the available evidence suggests that lay health worker-delivered interventions are cost-effective, more rigorous economic evaluations are needed to inform resource allocation decisions and to compare the cost-effectiveness of different intervention approaches.

Most studies were conducted in urban or peri-urban settings, limiting generalizability to rural areas where the mental health treatment gap is often most significant and where access to supervision and support for lay health workers may be more challenging. Only one study (Dow et al.^[19] 2020) was conducted in a predominantly rural setting. Future research should prioritize rural settings and examine adaptations needed to implement lay health worker-delivered interventions in remote areas with limited infrastructure.

We were unable to systematically assess implementation outcomes (e.g., fidelity, acceptability, feasibility, sustainability) across studies because these outcomes were not consistently reported. Implementation research is crucial for understanding how to successfully scale up evidence-based interventions in real-world settings, and future studies should routinely assess and report implementation outcomes alongside clinical outcomes.

The meta-analysis combines different psychological treatment modalities (e.g., PST, IPT, CBT) into a single pooled effect, which assumes a degree of equivalence in their outcomes. While our subgroup analysis by intervention type provides some insight into the impact of different modalities, the small number of studies in each subgroup limits our ability to draw firm conclusions. Future research should include head-to-head trials to directly compare the effectiveness of different psychological interventions delivered by lay health workers.

Strengths

Despite these limitations, this review has several important strengths. First, we conducted a comprehensive and systematic search of multiple databases, including African-specific databases (African Index Medicus), and followed PRISMA 2020 guidelines rigorously. Second, we included only randomized controlled trials, which provide the strongest evidence for causal inference. Third, we assessed the risk of bias using the Cochrane RoB 2 tool and evaluated the certainty of evidence using the GRADE approach, providing transparency regarding the quality of the evidence. Fourth, we included diverse populations, interventions, and settings, enhancing the generalizability of our findings. Fifth, we extracted data directly from published articles and, when available, from the original PDFs, ensuring the accuracy of the data used in the meta-analysis.

Future research directions

Based on the findings and limitations of this review, several priorities for future research are recommended to strengthen the evidence base and inform policy and practice. Long-term follow-up studies lasting at least 24 months are needed to assess the sustainability of intervention effects and the potential need for maintenance or booster sessions. Implementation research should explore facilitators and barriers to scaling up lay health worker-delivered interventions, including training models, supervision strategies, quality assurance mechanisms, and integration approaches. Rigorous cost-effectiveness analyses are crucial for comparing different intervention types and delivery models, thereby informing resource allocation decisions. Research in rural and remote settings should be prioritized to understand the adaptations required for successful implementation in areas with limited infrastructure and supervision. Mechanism studies are needed to identify how these interventions achieve their effects, focusing on factors such as cultural concordance, therapeutic alliance, intervention fidelity, and supervision quality. Comparative effectiveness research should include head-to-head trials of different types of interventions (e.g., problem-solving therapy, interpersonal psychotherapy, cognitive-behavioral therapy) and delivery formats (e.g., individual vs. group, clinic-based vs. home-based) to determine the most effective and efficient approaches. Further research should also examine how to tailor evidence-based interventions while preserving their core components in a culturally sensitive manner. In addition, studies should investigate the impact of delivering mental health interventions on lay health workers themselves, including risks of burnout, secondary trauma, job satisfaction, and retention. Scale-up studies using pragmatic trials and implementation science methods are needed to evaluate effectiveness in routine care settings. Finally, future research should explore models for integrating mental health interventions with other health services such as HIV care, maternal and child health, and non-communicable disease management.