Effectiveness and implementation of Self-Help Plus (SH+) and Doing What Matters in Times of Stress (DWMS): protocol of a systematic review and meta-analysis

Abstract

Introduction

Global increases in armed conflict, forced displacement, pandemics and economic instability have contributed to rising levels of psychological distress worldwide, placing relevant segments of the population at increased risk of developing mental health conditions. This burden is particularly pronounced in humanitarian and low-resource settings where access to specialist mental health services is limited. Scalable, low-intensity, evidence-based psychological interventions are therefore urgently needed. In response, the WHO has developed transdiagnostic programmes, including Self-Help Plus (SH+) and Doing What Matters in Times of Stress (DWMS). Although these interventions are increasingly implemented across humanitarian and public health contexts, evidence for their effectiveness and implementation has not yet been systematically synthesised.

Methods and analysis

This preregistered systematic review and meta-analysis will be conducted in accordance with Cochrane Collaboration standards and reported following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines. We will include randomised controlled trials evaluating the effectiveness of SH+ or DWMS, alongside qualitative and mixed-methods studies examining their implementation among stressor-exposed individuals of any age. Outcomes will include symptoms of depression and anxiety, general distress and post-traumatic stress symptoms. Moreover, we will examine effects on well-being, psychosocial functioning, adverse events and implementation outcomes (eg, acceptability, feasibility, fidelity). We will search Cochrane CENTRAL, APA PsycNet, Web of Science Core Collection, Embase and Scopus for records published from 2016 onwards. Searches will be supplemented by hand-searching preprint repositories and citation tracking. Risk of bias will be assessed using the Revised Cochrane Risk of Bias Tool and a customised appraisal tool for studies on implementation. Quantitative data will be synthesised using random-effects multilevel meta-analyses, with meta-regression models applied to examine moderators. Bayesian meta-analyses will be conducted where appropriate as sensitivity analyses to assess the robustness of the findings. Certainty of evidence will be evaluated using the Grading of Recommendations, Assessment, Development and Evaluation (GRADE) approach.

Ethics and dissemination

Ethical approval is not required. Findings will be disseminated through an open-access peer-reviewed publication, a plain-language summary, and the Open Science Framework, where all materials will be made publicly available.

PROSPERO registration number

CRD420251168521.

STRENGTHS AND LIMITATIONS OF THIS STUDY.

• This systematic review will use rigorous, state-of-the-art methods for literature searching, study screening, and data extraction to synthesise evidence on the effectiveness and implementation of Self-Help Plus and Doing What Matters in Times of Stress.

• Planned multilevel approaches will enable the integration of diverse outcome measures and appropriately account for dependencies among effect estimates.

• The combination of qualitative and quantitative synthesis will allow for a comprehensive assessment of intervention effectiveness and implementation processes, including barriers and facilitators across diverse settings.

• All materials, including search strategies, datasets, and analytic code, will be made openly available via the Open Science Framework.

• Potential limitations include incomplete reporting in primary studies, heterogeneity in outcome measures, the inclusion of pilot and feasibility trials and a small number of trials for certain populations, which may reduce the precision and generalisability of pooled estimates.

Introduction

Over the past decade, societies worldwide have faced an accumulation of social, economic and environmental stressors—including armed conflicts, forced displacements, pandemics and financial crises.¹ Because exposure to chronic or acute stress is a major risk factor for mental health conditions, rising global stressor exposure places large numbers of people at heightened risk for developing mental health problems.^{2 3} The global burden of mental health conditions is substantial and increasing: more than one billion people are currently affected, accounting for nearly one-sixth of all years lived with disability.4,6 In addition to individual suffering, the economic consequences of mental health conditions are estimated to reach nearly US$1 trillion annually.⁵

These challenges are particularly pronounced in low- and middle-income countries (LMICs). Individuals living in humanitarian settings are frequently exposed to multiple and prolonged stressors, including poverty, political instability, limited access to healthcare and heightened exposure to trauma, loss and displacement.⁷ Consequently, mental health needs are substantially elevated in these contexts. For example, a systematic review reported that approximately 30% of refugees meet criteria for depression and around 40% for post-traumatic stress disorder, prevalence rates that markedly exceed those observed in general populations in high-income countries (HICs).⁶

Despite this substantial burden, access to evidence-based psychological interventions remains severely limited in LMICs. It is estimated that up to 85% of individuals with mental health conditions in these settings receive no formal treatment.⁸ In addition, in conflict-affected contexts, the density of trained mental health professionals often falls below one per 100 000 inhabitants, compared with approximately 55 per 100 000 in HICs.⁹ This pronounced treatment gap underscores the urgent need for scalable, low-intensity psychological interventions that can be delivered feasibly by non-specialist providers.

To address these systemic inequities in mental healthcare access, the WHO and its research partners have developed a set of transdiagnostic, low-intensity psychological interventions aimed at reducing distress and strengthening coping capacity in stressor-exposed populations. These interventions—including Self-Help Plus (SH+),¹⁰ Doing What Matters in Times of Stress (DWMS),¹¹ Problem Management Plus (PM+),¹² Step-by-Step (SbS)¹³ and Early Adolescent Skills for Emotions (EASE)¹⁴—share several core characteristics: (1) brief, structured delivery formats; (2) grounding in evidence-based psychotherapeutic principles; (3) low-threshold accessibility, with delivery possible as self-help or by non-specialist providers and (4) cultural adaptability across a wide range of settings.¹⁵ These scalable approaches form a key component of global mental health strategies and represent a pragmatic means of reducing the treatment gap in settings where specialist resources are limited.¹⁶ Recent evidence syntheses have evaluated the effectiveness of PM+,17,20 SbS¹⁷ and EASE,^{21 22} while such a synthesis is still missing for SH+ and DWMS.

SH+ is a five-session, group-based guided self-help intervention for adults experiencing psychological distress, with the primary aim of reducing depressive symptoms.¹⁰ Each session lasts approximately 90 to 120 min and is delivered through pre-recorded audio materials accompanied by an illustrated handbook for participants. Sessions are facilitated by trained non-specialist providers, enabling large-scale implementation with minimal infrastructural requirements. SH+ is grounded in Acceptance and Commitment Therapy, emphasising mindfulness, acceptance of internal experiences and engagement in values-based actions to enhance psychological flexibility.²³

DWMS is an illustrated self-help guide that constitutes the core textual component of SH+.¹⁰ Research on DWMS as a stand-alone intervention emerged during the COVID-19 pandemic, when restrictions on social contact rendered group-based interventions such as SH+ infeasible.²⁴ In response, DWMS was adapted as a scalable self-help intervention that could be delivered without direct interpersonal contact. As a result, DWMS has been evaluated not only in low-resource and humanitarian settings but also in HICs, including studies conducted among healthcare professionals during the COVID-19 pandemic.^{25 26 27}

To date, evidence on the effects of SH+ has only been synthesised in a single individual participant data (IPD) meta-analysis,²⁸ which included three trials conducted among refugees and asylum seekers. This IPD meta-analysis found that SH+ was not superior to enhanced care as usual (CAU) at post-intervention assessment but demonstrated greater effectiveness at 5–6-month follow-up in improving depressive symptoms, self-identified problems and well-being. Moreover, effects tended to be more favourable among participants with a higher baseline burden, including those with greater initial distress and unemployed individuals. However, a comprehensive synthesis of the effects of SH+ and DWMS across other populations, as well as a systematic summary of their implementation—a key challenge especially in low-resource settings^{29 30}—is currently lacking.

Objectives

The present project aims to address this evidence gap by providing a high-quality synthesis of the effectiveness of SH+ and DWMS, alongside a comprehensive summary of their implementation across diverse settings. In doing so, the project aligns with the objective of the WHO to establish a portfolio of low-threshold, transdiagnostic psychosocial interventions to address the substantial global need for accessible mental healthcare.^{12 31} By strengthening the evidence base on both effectiveness and implementation, this project supports the broader goal of enabling these interventions to achieve sustained public health impact through large-scale dissemination and implementation.³²

The project will address the following research questions:

1. What are the effects of SH+ and DWMS on psychological symptoms (eg, depressive symptoms, anxiety symptoms, general distress and post-traumatic stress symptoms) and on indicators of positive mental health (eg, quality of life, well-being, self-efficacy and psychological flexibility)?

2. What evidence is available regarding adverse events or unintended negative effects associated with participation in SH+ or DWMS?

3. Which study-level and participant-level characteristics moderate intervention effects?

4. What is known about implementation outcomes for SH+ and DWMS (eg, acceptability, feasibility, fidelity³³)?

5. Which factors have been reported to hinder or facilitate delivery, scalability and sustainment of SH+ and DWMS across settings?

Methods and analysis

Preregistration and reporting

The methodological framework of this systematic review is based on standards established by the Cochrane Collaboration.³⁴ The review protocol was developed in accordance with the Preferred Reporting Items for Systematic Review and Meta-Analysis Protocols (PRISMA-P) guidelines (PRISMA-P³⁵; see online supplemental material 1) and the PRISMA extension for reporting literature searches (PRISMA-S³⁶; see online supplemental material 2). The completed review will be reported in line with the PRISMA statement,³⁷ as well as the PRISMA recommendations for reporting harms in systematic reviews.³⁸ The review protocol was prospectively registered with PROSPERO (ID: CRD420251168521) on 2 December 2025. In line with principles of research transparency and reproducibility, any major protocol amendments or methodological updates will be documented and made publicly available via the Open Science Framework (OSF, https://osf.io/r5w3p/).

Eligibility criteria

Based on preliminary literature searches and previous reviews of comparable WHO interventions,^{17 21} the review will include stressor-exposed populations of any age. Eligible stressors will encompass both major (macro-level) stressors—such as humanitarian crises, natural disasters, armed conflicts or pandemics—and minor (micro-level) stressors or daily hassles (eg, time pressure or interpersonal conflicts).³⁹ Eligibility will not be restricted by gender, ethnicity, nationality or mental health status, including whether participants meet diagnostic criteria for a mental health condition or score above or below established screening thresholds. Similarly, inclusion will not depend on physical health status and will encompass individuals with or without diagnosed medical conditions, as well as those receiving any form of pharmacotherapy. Studies explicitly describing populations as not exposed to stressors will be excluded.

Intervention

Studies evaluating SH+¹⁰ and/or DWMS¹¹ will be eligible for inclusion. Multicomponent interventions that include SH+ and/or DWMS as an initial component will also be eligible for inclusion, provided that outcomes are reported following delivery of SH+ or DWMS and prior to the introduction of subsequent intervention components. Eligibility will not be restricted by intervention setting (eg, group-based or individual formats), mode of delivery (eg, face-to-face, telephone-based, online or mobile-based), implementation site (eg, home visits, refugee camps, healthcare facilities, community, educational or clinical settings), type of facilitator (eg, self-guided, coach-guided or peer-guided) or intervention characteristics (eg, number or duration of sessions). Studies evaluating mental health interventions other than SH+ or DWMS will be excluded, as will stepped-care or multistage interventions in which SH+ or DWMS are implemented only as secondary or subsequent components (eg, interventions starting with progressive muscle relaxation followed by SH+ or DWMS).

Comparator

Eligible comparators will be defined in accordance with Cochrane taxonomy³⁴ and must be drawn from the same target population as the intervention group. The following comparator types will be included: No intervention controls, including minimal-contact or assessment-only conditions; waitlist controls; CAU and enhanced CAU (ECAU); attention or placebo controls, designed to control for non-specific intervention effects, provided they are not intended as head-to-head comparisons of specific psychosocial interventions.

Studies will be excluded if comparators are drawn from a different population than the intervention group, if they involve head-to-head comparisons with other structured psychosocial interventions (eg, comparisons of SH+ or DWMS with PM+), or if no control condition is included in effectiveness studies.

For studies examining implementation outcomes or barriers and facilitators to implementation, a comparator will not be required.

Outcomes

Effectiveness outcomes

Studies will be eligible if they report outcomes related to mental distress. These include depressive symptoms, anxiety symptoms, general distress, post-traumatic stress symptoms and personally identified or self-reported problems. Outcomes reflecting positive mental health—such as well-being, life satisfaction, (health-related) quality of life, psychological flexibility and resilience-related constructs (eg, self-efficacy)—will also be eligible. Measures of functioning, including daily or overall level of functioning, will also be considered. Additional outcomes of interest will include physical health and somatic symptoms, economic outcomes (eg, cost-effectiveness, economic efficiency or healthcare costs) and health service or care utilisation.

Harms

In line with Cochrane guidance on the assessment of adverse effects,⁴⁰ all reported adverse events (eg, hospitalisation, suicidal ideation, suicide attempts or death) will be extracted irrespective of their presumed relationship to SH+ or DWMS.

Implementation outcomes

Studies reporting implementation outcomes will be included and categorised according to the implementation outcomes framework proposed by Proctor et al.³³ Outcomes encompass acceptability, feasibility, fidelity, implementation costs, sustainability, appropriateness, penetration and adoption (see table 1 for a brief definition). Both qualitative evidence on barriers and facilitators to implementation⁴¹ will also be eligible.

Table 1. Implementation outcomes.

Implementation outcome	Definition (Proctor et al33)	Typical measures/indicators
Acceptability	The perception among stakeholders that a given treatment, service, practice or innovation is agreeable, palatable and satisfactory	Satisfaction surveys, perceived usefulness, qualitative interviews
Feasibility	The extent to which an intervention can be successfully used or carried out within a given setting	Completion rates, feasibility questionnaires, resource requirements
Fidelity	The degree to which an intervention was implemented as it was prescribed in the original protocol or manual	Adherence checklists, observations, protocol compliance
Implementation cost	The incremental cost of implementing an intervention, including expenses related to adoption and delivery	Personnel time, training costs, materials, infrastructure expenses
Sustainability	The extent to which an intervention is maintained or institutionalised within a setting over time	Continued use post-implementation, long-term routine data
Appropriateness	The perceived fit, relevance or compatibility of the intervention for a given setting, provider or client	Fit or relevance scales, contextual appropriateness ratings
Penetration	The integration of a practice within a service setting; often expressed as a proportion	Proportion of eligible providers or clients receiving the intervention
Adoption	The intention, initial decision or action to try or employ an intervention	Uptake rates, time to first use, proportion of providers adopting

Definitions reported based on Ceccarelli et al.30

Study design

For the synthesis of intervention effects, we will include randomised controlled trials (RCTs), including cluster RCTs (cRCTs), irrespective of follow-up duration or risk of bias, provided they evaluate SH+¹⁰ and/or DWMS¹¹ and report at least one eligible effectiveness outcome. In accordance with Cochrane recommendations,³⁴ quasi-randomised trials (eg, allocation based on date of birth, alternation or administrative sequence) will be excluded from the effectiveness synthesis due to the high risk of selection bias.⁴²

For the synthesis of implementation outcomes, we will include a broader range of study designs. In addition to RCTs and cRCTs, quasi-randomised trials, non-randomised quantitative studies and mixed-methods studies will be eligible, provided that they examine the implementation or delivery of SH+¹⁰ and/or DWMS¹¹ and report implementation outcomes and/or barriers and facilitators to implementation.^{33 41}

Pilot and feasibility studies will be eligible for inclusion in either synthesis if they report quantitative data for at least one relevant effectiveness outcome and/or findings related to implementation. This decision reflects the small number of eligible primary studies identified in preliminary searches and the need to maximise the available evidence base. As effect estimates derived from pilot and feasibility studies may be at increased risk of bias, the impact of including or excluding these studies will be examined in sensitivity analyses.

The following study types will be excluded: non-RCTs that do not report implementation outcomes or barriers and facilitators to implementation; studies without a control group and without an implementation focus, including single-arm trials or other non-comparative designs (eg, interrupted time-series studies); studies reporting exclusively on intervention development⁴³; and theoretical or discussion papers, letters to the editor without empirical data, editorials and commentaries.

Systematic reviews and meta-analyses will not be included in the synthesis; however, they will be screened to identify additional potentially eligible primary studies.

Information sources and search strategy

The search strategy focuses exclusively on primary studies. Following preregistration, a comprehensive systematic literature search was conducted on 2 December 2025 across five electronic databases: the Cochrane Central Register of Controlled Trials (CENTRAL), APA PsycNet (including PsycINFO, PsycArticles and PsycExtra), Web of Science Core Collection, Embase and Scopus. As the first article on SH+ was published in 2016,¹⁰ this year was defined as the start date for the literature search. Search updates will be conducted periodically, with a final update planned no later than 2 months prior to completion of the review. Detailed search strategies for all databases, including applied filters, are provided in online supplemental material 3.

The search strategy comprises four conceptual clusters. The first cluster includes terms related to SH+ and DWMS (eg, ‘Self-Help Plus’, ‘Doing What Matters in Times of Stress’). The second cluster consists of terms related to psychosocial interventions (eg, intervention OR self-help), while the third cluster captures mental health-related terms (eg, mental health OR distress). To improve specificity, a fourth cluster includes exclusion terms for abbreviations that overlap with ‘SH+’ but are unrelated to the interventions of interest (eg, superhydrophob OR superheterodyn). Within each cluster, terms are combined using the Boolean operator OR, and clusters 1 to 3 are combined using AND.

Where appropriate, proximity operators (eg, NEAR or database-specific equivalents) are applied to link conceptually related terms that would otherwise yield overly broad results (eg, (WHO) NEAR/3 (self NEAR/1 help)).

To capture the most recent evidence, additional searches will be conducted in the preprint databases medRxiv, PsyArXiv and the meta-registry OSF Preprints. To maximise sensitivity, reference lists of relevant systematic reviews and eligible primary studies will be screened. Forward citation tracking of included trials and the original introductory publications on SH+¹⁰ and DWMS¹¹ will be performed via Web of Science.

Study selection

After deduplication using the automated Deduplicator tool,⁴⁴ study selection will be conducted in Rayyan.⁴⁵ Prior to formal screening, all reviewers conducted a calibration exercise on a sample of records to ensure consistent interpretation of the eligibility criteria. After removal of duplicates, two reviewers—comprising at least one psychologist with a minimum of a master’s degree and at most one student team member—will independently screen titles and abstracts to assess eligibility. Records that clearly do not meet the eligibility criteria will be excluded at this stage.

Subsequently, full-text screening will be performed independently and in duplicate by two reviewers. Following PRISMA recommendations,³⁷ reasons for exclusion at the full-text level will be documented. Disagreements at any stage of the selection process will be resolved through discussion; if consensus cannot be reached, a third reviewer will be consulted.

For the main project, no automated or machine-learning tools will be used during the screening process (see Exploratory use of Elicit). Inter-rater reliability will be assessed for both the title/abstract and full-text screening phases using Cohen’s kappa. The study selection process will be reported in accordance with PRISMA guidelines using a PRISMA flow diagram.³⁷

Data extraction

Data extraction will be performed using a standardised electronic data extraction form. Data will be extracted by one reviewer and independently verified by a second reviewer for accuracy and completeness. Any discrepancies will be resolved through discussion; if agreement cannot be reached, a third reviewer will adjudicate.

The data extraction form is based on the former Effective Practice and Organisation of Care (EPOC) Review Group data collection template⁴⁶ and has been adapted from tools used in previous, related review projects conducted by the research team.^{17 21 47}

Information will be systematically coded across the following domains: general study information (eg, citation details, year of publication, country, funding, open science practices); sample characteristics (eg, mental health status, age, gender distribution, exposure to stressors); methodological characteristics (eg, study design, inclusion and exclusion criteria, randomisation procedures); intervention characteristics (eg, SH+ or DWMS, delivery mode, setting, number and duration of sessions, type of provider); comparator characteristics (eg, type, intensity and duration of comparator); outcome data (eg, outcome domains, measurement instruments, assessment time points, data required for effect size calculation, adverse events); implementation outcomes (eg, acceptability, feasibility, fidelity), barriers and facilitators to implementation (eg, factors related to the outer setting, inner setting and individual level); and information required for quality appraisal. Where reported information is incomplete or unclear, corresponding authors will be contacted for clarification, with a reminder email being sent after 4 and 8 weeks.

A preliminary overview of the data items to be extracted is provided in online supplemental material 4 and will be made publicly available via the OSF (https://osf.io/r5w3p/).

Quality appraisal

Risk of bias

Risk of bias for effectiveness outcomes will be assessed independently by two reviewers, with disagreements resolved through discussion or consultation with a third reviewer. The assessment will be conducted using the Revised Cochrane Risk of Bias Tool for Randomised Trials (RoB 2),⁴⁸ applying an adapted version developed within the MetaPsy project (version 1.0–0, February 2025; https://www.metapsy.org), which is specifically tailored to the evaluation of psychological interventions.

For individual RCTs, the following bias domains will be evaluated: (1) bias arising from the randomisation process; (2) bias due to deviations from intended interventions; (3) bias due to missing outcome data; (4) bias in measurement of the outcome and (5) bias in selection of the reported result. For cRCTs, an additional domain assessing bias arising from the identification or recruitment of participants into clusters will be evaluated. Each domain will be judged as ‘low risk of bias’, ‘some concerns’ or ‘high risk of bias’, leading to an overall risk-of-bias judgement for each outcome.

The evaluation of risk of bias for implementation outcomes as well as barriers and facilitators to implementation will be conducted separately from the RoB 2 assessment. For this purpose, a reporting quality tool will be developed, based on a previously published review of implementation barriers and facilitators⁴⁹ and adapted to the current review to additionally capture implementation outcomes.³³

Publication bias

The potential influence of publication bias will be examined for the synthesis of intervention effectiveness using visual inspection of contour-enhanced funnel plots,⁵⁰ regression-based tests appropriate for multilevel meta-analyses⁵¹ and worst-case meta-analyses to assess the robustness of pooled effect estimates under conservative assumptions.⁵²

Data synthesis

All included studies will be summarised narratively and in tabular form, providing an overview of key study characteristics (eg, intervention type, number of participants per arm, study duration, setting). Primary studies will be grouped according to relevant characteristics, such as intervention type (SH+ vs DWMS), target population and follow-up duration, to facilitate both qualitative and quantitative synthesis.

For implementation outcomes and barriers and facilitators to implementation, a narrative synthesis will be conducted. For studies also reporting intervention effects, both qualitative and quantitative syntheses will be undertaken.

Statistical analyses will be performed using R (V.4.5.2 or later),⁵³ primarily employing the packages metafor,⁵⁴ clubSandwich⁵⁵ and PublicationBias.⁵² Bayesian analyses, planned as sensitivity analyses, will be conducted using JASP,⁵⁶ with Bayesian model averaging based on the RoBMA R package.⁵⁷ Additional software packages will be used as required by specific modelling approaches.

Where quantitative synthesis is not feasible (ie, when fewer than five effect estimates are available for a given outcome), results will be synthesised following the Cochrane Collaboration’s Synthesis Without Meta-analysis reporting guidelines.⁵⁸

Effect measures and time points

Where data permit, standardised mean differences (SMDs; Hedges’ g) with 95% CIs will be calculated for continuous outcomes, using post-intervention assessments (defined as within 2 weeks of intervention completion) and follow-up assessments (eg, 3 or 6 months post-intervention). Continuous outcomes will be prioritised. When only dichotomous data are available (eg, prevalence of mental health conditions), risk ratios with 95% CIs will be used as effect estimates. Where available, we will use intention-to-treat data for a conservative approach.

For cRCTs, unit-of-analysis issues will be addressed in accordance with Cochrane guidance.³⁴ Where trials appropriately account for clustering in their analyses, effect estimates will be extracted as reported. Based on previous review projects in the field,^{17 21 47} we expect to find many studies not considering clustering effects for effect estimates. In these cases, SEs will be adjusted using an effective sample size approach. Intracluster correlation coefficients (ICCs) reported in the primary studies will be applied where available; otherwise, external ICCs derived from comparable studies will be used to account for the design effect.⁵⁹

Planned analyses

The primary analyses will examine the effects of SH+ and DWMS across two outcome domains: mental distress and positive mental health, each comprising multiple outcome types (eg, depressive symptoms, anxiety symptoms or well-being and quality of life). Pairwise multilevel meta-analyses will be conducted to compare intervention and control groups at post-intervention assessments and, where available, at follow-up assessments, allowing evaluation of both short-term and longer-term between-group differences.

Primary analyses will adopt a frequentist approach using random-effects models, with effect estimates nested within primary studies and outcome types (eg, depressive symptoms, anxiety symptoms). Random effects will be specified at the study level and the outcome-within-study level, assuming normally distributed random effects with a mean of zero and unknown variances. Because the exact within-study covariance structure is typically unknown, cluster-robust variance estimators will be applied to obtain valid standard errors that are robust to misspecification of the within-study covariance matrix.⁶⁰ This modelling strategy accommodates correlations among outcomes within studies as well as similarities across studies reporting on the same outcome.⁶¹

Assessment of heterogeneity

Between-study and between-outcome heterogeneity will be assessed using visual inspection of forest plots, Cochran’s Q tests⁶² and the I² statistic.³⁴ To quantify the uncertainty around pooled effect estimates and to facilitate interpretation of heterogeneity, 95% prediction intervals (PIs) will be reported where at least five effect estimates are available for a given analysis.^{34 63}

Moderator and subgroup analyses

Where substantial heterogeneity is identified, moderator and subgroup analyses will be conducted to explore whether study-level or participant-level characteristics are associated with differential effects of SH+ or DWMS. Prespecified moderators will include intervention type (SH+ vs DWMS), setting (eg, LMICs vs HICs), type of control group, baseline symptom severity and participant characteristics (ie, mean age and gender distribution, operationalised as the percentage of women).

Moderator analyses will be conducted using mixed-effects meta-regression models, in which moderators will be entered as fixed effects while between-study heterogeneity will be modelled as a random effect. Subgroup analyses will be performed where moderators are categorical. The presence of a moderator effect will be evaluated using the QM test statistic.⁵⁴ All moderator analyses will be conducted at the level of the overall models for mental distress and positive mental health outcomes and will require five effect estimates per continuous moderator or category to ensure interpretability.

Results from moderator and subgroup analyses will be interpreted cautiously, recognising their observational and exploratory nature, the potential for residual confounding and the increased risk of type I error due to multiple testing.⁶⁴ Any moderators examined post hoc will be clearly labelled as exploratory.

Sensitivity analyses

Sensitivity analyses will be conducted to assess the robustness of findings across alternative analytic decisions. These analyses will compare intention-to-treat and complete-case analyses, the inclusion and exclusion of unpublished studies, pilot or feasibility trials and fully powered effectiveness trials, and high-intensity versus low-intensity control conditions, such as ECAU versus waitlist controls. The influence of study quality will also be examined by comparing findings across risk-of-bias categories, including low risk, some concerns and high risk of bias.

To complement the primary frequentist analyses, additional Bayesian random-effects meta-analyses will be conducted focusing on the broader outcome domains (ie, mental distress and positive mental health outcomes). Bayesian methods are particularly advantageous in contexts with a limited number of studies and small effect sizes, as they allow the incorporation of prior information to stabilise estimates when data are sparse.^{65 66} This approach is expected to provide deeper insight into the evidence for small but potentially meaningful effects of SH+ and DWMS, consistent with findings reported for comparable interventions in previous reviews.^{17 21}

Given the absence of strong prior evidence, weakly informative priors will be specified for all Bayesian models.^{67 68} Specifically, a normal prior distribution N(0, 1) will be applied to the overall intervention effect (μ), and a half-Cauchy prior HC(0, 0.5)—with a location parameter of zero and a scale parameter of 0.5—will be used for the between-study heterogeneity parameter (τ). SMDs at post-intervention and follow-up, adjusted for clustering effects, will be used as effect estimates and reported alongside 95% credible intervals. These intervals represent the range within which the true parameter value lies with 95% probability, conditional on the observed data and the specified prior distributions. In addition, 95% PIs will be reported to reflect the expected range of effects for a future study with 95% probability.⁶⁹

Model convergence and adequacy will be assessed using the potential scale reduction factor (R̂) and effective sample size (ESS). R̂ values below 1.01 and ESS values greater than 400 will be considered indicative of satisfactory convergence and sufficient sampling efficiency.⁷⁰

To evaluate the strength of evidence for the effects of SH+ and DWMS, Bayes factors (BF₁₀) will be calculated, with values greater than 1 indicating increasing support for the intervention relative to the control condition. In line with conventional interpretative guidelines, BF₁₀ values between 1 and 3 will be interpreted as anecdotal evidence, values between 3 and 10 as moderate evidence and values exceeding 10 as strong evidence in favour of SH+ or DWMS.⁷¹

Certainty of evidence

The overall certainty of the evidence will be assessed by two reviewers—including at least one psychologist with a master’s degree or higher—using the Grading of Recommendations, Assessment, Development and Evaluation (GRADE) Working Group framework.⁷² For the seven outcomes most frequently reported in the primary studies, the certainty of evidence will be rated as high, moderate, low or very low in accordance with the GRADE framework, considering the domains of risk of bias, inconsistency, indirectness, imprecision and publication bias.

Exploratory use of Elicit

In addition to the primary, manually conducted systematic search, screening, data extraction and synthesis procedures, this review will be carried out in parallel using the artificial intelligence (AI)-assisted tool Elicit⁷³ as an exploratory supplementary approach. Elicit is an AI research assistant designed to support literature search, screening and data extraction workflows in evidence synthesis processes by leveraging semantic search over academic papers and generating structured outputs to aid screening and extraction. The purpose of this component is to compare studies identified through conventional database searching with those retrieved or suggested via an AI-supported method, with the aim of examining potential differences in sensitivity, coverage of relevant literature and efficiency. All outputs generated via Elicit will be documented separately and systematically compared with the results of the primary review workflow. The use of Elicit will not influence the final study selection, data synthesis or conclusions of the review and is intended solely for exploratory and methodological purposes. Specifically, it will be used to evaluate the potential added value of AI-assisted tools for systematic reviews of psychosocial interventions, a methodological application that has not yet been systematically examined.

Patient and public involvement

Patients and members of the public will not be involved in the development, conduct, reporting or dissemination of this systematic review.

Ethics and dissemination

As this study is a systematic review based exclusively on published aggregate data and does not involve IPD, ethical approval was not required. The findings will be disseminated through publication in a peer-reviewed journal, with a preference for open-access publication. In addition, a plain language summary will be prepared to enhance accessibility for non-academic audiences.

To promote transparency and reproducibility, all materials used in the review—including the protocol, search strategies and data extraction materials—will be made publicly available via the OSF (https://osf.io/r5w3p/).