Scolaris Content Display Scolaris Content Display

Cochrane Database of Systematic Reviews Review - Intervention

Non‐specialist health worker interventions for the care of mental, neurological and substance‐abuse disorders in low‐ and middle‐income countries

Authors' declarations of interest

Version published: 19 November 2013 Version history

https://doi.org/10.1002/14651858.CD009149.pub2

This is not the most recent version

view the current version 05 August 2021
Collapse all Expand all

Abstract

available in

Background

Many people with mental, neurological and substance‐use disorders (MNS) do not receive health care. Non‐specialist health workers (NSHWs) and other professionals with health roles (OPHRs) are a key strategy for closing the treatment gap.

Objectives

To assess the effect of NSHWs and OPHRs delivering MNS interventions in primary and community health care in low‐ and middle‐income countries.

Search methods

We searched the Cochrane Central Register of Controlled Trials (CENTRAL) (including the Cochrane Effective Practice and Organisation of Care (EPOC) Group Specialised Register) (searched 21 June 2012); MEDLINE, OvidSP; MEDLINE In Process & Other Non‐Indexed Citations, OvidSP; EMBASE, OvidSP (searched 15 June 2012); CINAHL, EBSCOhost; PsycINFO, OvidSP (searched 18 and 19 June 2012); World Health Organization (WHO) Global Health Library (searched 29 June 2012); LILACS; the International Clinical Trials Registry Platform (WHO); OpenGrey; the metaRegister of Controlled Trials (searched 8 and 9 August 2012); Science Citation Index and Social Sciences Citation Index (ISI Web of Knowledge) (searched 2 October 2012) and reference lists, without language or date restrictions. We contacted authors for additional studies.

Selection criteria

Randomised and non‐randomised controlled trials, controlled before‐and‐after studies and interrupted‐time‐series studies of NSHWs/OPHR‐delivered interventions in primary/community health care in low‐ and middle‐income countries, and intended to improve outcomes in people with MNS disorders and in their carers. We defined an NSHW as any professional health worker (e.g. doctors, nurses and social workers) or lay health worker without specialised training in MNS disorders. OPHRs included people outside the health sector (only teachers in this review).

Data collection and analysis

Review authors double screened, double data‐extracted and assessed risk of bias using standard formats. We grouped studies with similar interventions together. Where feasible, we combined data to obtain an overall estimate of effect.

Main results

The 38 included studies were from seven low‐ and 15 middle‐income countries. Twenty‐two studies used lay health workers, and most addressed depression or post‐traumatic stress disorder (PTSD). The review shows that the use of NSHWs, compared with usual healthcare services: 1. may increase the number of adults who recover from depression or anxiety, or both, two to six months after treatment (prevalence of depression: risk ratio (RR) 0.30, 95% confidence interval (CI) 0.14 to 0.64; low‐quality evidence); 2. may slightly reduce symptoms for mothers with perinatal depression (severity of depressive symptoms: standardised mean difference (SMD) ‐0.42, 95% CI ‐0.58 to ‐0.26; low‐quality evidence); 3. may slightly reduce the symptoms of adults with PTSD (severity of PTSD symptoms: SMD ‐0.36, 95% CI ‐0.67 to ‐0.05; low‐quality evidence); 4. probably slightly improves the symptoms of people with dementia (severity of behavioural symptoms: SMD ‐0.26, 95% CI ‐0.60 to 0.08; moderate‐quality evidence); 5. probably improves/slightly improves the mental well‐being, burden and distress of carers of people with dementia (carer burden: SMD ‐0.50, 95% CI ‐0.84 to ‐0.15; moderate‐quality evidence); 6. may decrease the amount of alcohol consumed by people with alcohol‐use disorders (drinks/drinking day in last 7 to 30 days: mean difference ‐1.68, 95% CI ‐2.79 to ‐0.57); low‐quality evidence).

It is uncertain whether lay health workers or teachers reduce PTSD symptoms among children. There were insufficient data to draw conclusions about the cost‐effectiveness of using NSHWs or teachers, or about their impact on people with other MNS conditions. In addition, very few studies measured adverse effects of NSHW‐led care ‐ such effects could impact on the appropriateness and quality of care.

Authors' conclusions

Overall, NSHWs and teachers have some promising benefits in improving people's outcomes for general and perinatal depression, PTSD and alcohol‐use disorders, and patient‐ and carer‐outcomes for dementia. However, this evidence is mostly low or very low quality, and for some issues no evidence is available. Therefore, we cannot make conclusions about which specific NSHW‐led interventions are more effective.

PICOs

Population
Intervention
Comparison
Outcome

The PICO model is widely used and taught in evidence-based health care as a strategy for formulating questions and search strategies and for characterizing clinical studies or meta-analyses. PICO stands for four different potential components of a clinical question: Patient, Population or Problem; Intervention; Comparison; Outcome.

See more on using PICO in the Cochrane Handbook.

Plain language summary

available in

The effect of non‐specialist health workers on people with mental, neurological and substance‐abuse disorders in developing countries

Background

In developing countries, most people with mental, neurological and substance‐abuse (MNS) disorders do not receive adequate care mainly because of a lack of mental health professionals. Non‐specialist health workers, but also other professionals with health roles, such as teachers, may therefore have an important role to play in delivering MNS health care.

Researchers in The Cochrane Collaboration carried out a review of the effects of using non‐specialist health workers or other professionals with health roles to help people with MNS disorders in developing countries. After searching for all relevant studies in scientific databases, they found 38 studies published before October 2012. Their findings are summarised below.

What is a non‐specialist health worker?

Any type of health worker (like a doctor, nurse or lay health worker) who is not a specialist in mental health or neurology but who may have had some training in these fields. We also looked at teachers, as they can be particularly important in the care of children and youths.

What the research says

The studies in this review were from 22 developing countries. In most studies, lay health workers delivered the mental health care, and addressed depression or anxiety (or both), or post‐traumatic stress disorder. The review shows that the use of non‐specialist health workers, compared with usual healthcare services:

· may increase the number of adults who recover from depression or anxiety (or both) two to six months after treatment;

· may slightly reduce symptoms formothers with depression;

· may slightly reduce the symptoms of adults with post‐traumatic stress disorder (non‐specialists and teachers were used in one study);

· probably slightly improves the symptoms of people with dementia;

· probably improves/slightly improves the mental well‐being, burden and distress of carers of people with dementia;

· may decrease the quantity of alcohol consumed by problem drinkers.

It is uncertain whether lay health workers or teachers reducepost‐traumatic stress disorder symptoms among children. There were too few studies to draw any conclusions about the cost‐effectiveness of using non‐specialist health workers or teachers, or about their impact on people with other MNS conditions such as epilepsy, schizophrenia, and alcohol and drug abuse problems. In addition, very few studies measured unintended consequences of non‐specialist health worker‐led care ‐ such effects could impact on the appropriateness and quality of care.

Quality of the evidence

Overall, non‐specialist health workers and teachers have some promising benefits in improving people's outcomes for general and perinatal depression, post‐traumatic stress disorder and alcohol‐use disorders, and patient and carer outcomes for dementia. However, this evidence is of low or very low quality in some areas, and for some issues no evidence is available. Therefore, we cannot make conclusions about which specific interventions using non‐specialist health workers to help people with MNS disorders are more effective.

Authors' conclusions

Implications for practice

Most results from the 38 studies suggest non‐specialist health workers (NSHW) delivering mental, neurological and substance‐use disorders (MNS) interventions have some impact on patients' outcomes, though the evidence is overall of low quality. Given the multitude of settings, disorders, interventions and health worker expertise covered in this review, there are still too few studies within each category to draw conclusions on specific intervention characteristics (such as type of health worker, duration of intervention, levels of training and supervision, etc.) that may impact on effectiveness.

The results show that in adults, lay health worker (LHW)‐led psychological interventions and collaborative care (a multidisciplinary team with NSHWs and specialists) may increase the number of adults who recover from depression or anxiety, or both, two to six months after treatment (low‐quality evidence). At seven to 12 months after treatment, it is uncertain whether the delivery of psychological treatment by LHWs alone is more effective than delivery by non‐specialists who are part of a multidisciplinary team (collaborative care). NSHWs may also slightly reduce symptoms for mothers with perinatal depression symptoms (low‐quality evidence).

Among the other disorders, NSHWs probably slightly improve the symptoms of people with dementia and the mental well‐being, burden and distress of carers of people with dementia (moderate‐quality evidence). They may also slightly reduce the symptoms of adults with post‐traumatic stress disorder (PTSD) and may decrease the amount of alcohol consumed by people with alcohol‐use disorders (low‐quality evidence).

It is uncertain whether LHWs or teachers reduce PTSD symptoms among children (very‐low‐quality evidence). There were insufficient data to draw conclusions about the cost‐effectiveness of using NSHWs or teachers, or about their impact on people with other MNS conditions such as epilepsy, schizophrenia, and alcohol and drug abuse problems. There is also insufficient evidence to determine which NSHW training or intervention strategies are likely to be most effective.

Implications for research

While this review has identified a large number of studies conducted in low‐ and middle‐income countries (LMICs), a number of important research questions remain. Research recommendations have been subdivided into those for trialists, systematic reviewers and other researchers.

Trialists

Trialists need to:

  • describe trial interventions better, for example in terms of training, supervision and incentives for NSHWs or other professionals with health roles (OPHRs). This will allow systematic reviewers to identify and compare characteristics that may help to explain the effects of NSHW interventions better;

  • conduct trials comparing interventions with different characteristics/types of NSHWs/OPHRs or modes of delivery, to be able to understand the effects of these variations. This is particularly applicable to collaborative care and other complex interventions where there may be several types of specialists and NSHWs, and several types of interventions on offer (such as stepped care);

  • compare NSHWs/OPHRs versus specialists to be able to assess the potential for task‐shifting;

  • include assessments of potential adverse effects or unintended consequences of NSHWs and OPHRs

  • design better quality trials, which includes more rigorous local validation of instruments and agreeing on standard instruments for specific outcomes and disorders to facilitate pooling and comparing data;

  • focus on clinical issues that have been poorly addressed to date, including epilepsy and other neurological disorders, severe mental disorders and substance abuse;

  • include economic data in their trials, as costs and cost‐effectiveness are important information for health planning.

Systematic reviewers

Further systematic reviews, drawing on a range of study designs (such as studies of effects, but also process evaluations, economic evaluations and qualitative work), are needed on:

  • factors affecting the sustainability of NSHW/OPHR interventions when scaled up;

  • the effectiveness of different approaches to ensure programme sustainability, including the use of different types of incentives and payment systems for NSHWs/OPHRs;

  • mechanisms for integrating LHW (subset of NSHW) programmes into the formal health system;

  • the equity impacts of these programmes.

Other researchers

Given the very broad range of NSHWs and OPHRs (with considerable variation in their characteristics (training, supervision, etc.), settings, interventions and delivery mechanisms in mental health care), there is a need to develop a comprehensive typology of NSHWs and OPHRs, as well as of the interventions they provide, which would help health planners and future researchers to have more standardised and comparable interventions and situations.

Summary of findings

Open in table viewer
Summary of findings for the main comparison. NSHW‐led psychological interventions compared with usual care in treating depression in adults in low‐ and middle‐income countries (RCTs)

What are the effects of NSHW‐led psychological interventions for treating depression in adults in low‐ and middle‐income countries?

Patient or population: Adults with depression
Settings: Low‐ and middle‐income countries (Taiwan, Pakistan, Uganda)
Intervention: NSHWs conducting psychological interventions
Comparison: Usual care

Outcomes

Illustrative comparative risks* (95% CI)

Effect estimate
(95% CI)

No of Participants
(studies)

Quality of the evidence
(GRADE)

Comments

Assumed risk

Corresponding risk

Usual care

NSHWs

Prevalence of depression (adults), short term (0‐8 weeks)
measured using various depression rating scales1

300 per 1000

91 per 1000

RR 0.30
(0.14 to 0.64)

1082
(3 studies)

⊕⊕⊝⊝
low2,3

‐

*The basis for the assumed risk is the mean control group risk across studies for pooled results and the control group risk for single studies. The corresponding risk (and its 95% CI) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).
CI: confidence interval; DSM: Diagnostic and Statistical Manual of Mental Disorders; NSHW: non‐specialist health worker; RCT: randomised controlled trial; RR: risk ratio.

GRADE Working Group grades of evidence
High quality: Further research is very unlikely to change our confidence in the estimate of effect.
Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.
Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.
Very low quality: We are very uncertain about the estimate.

1Bolton 2003 C‐RCT Uganda: DSM‐IV criteria A, C and E; Rahman 2008 CRCT Pakistan: Hamilton Depression Rating scale; Chen 2000 RCT Taiwan: Taiwanese Beck Depression Inventory.

2 Serious study limitations: Two of the three studies were at risk of bias. Bolton 2003 C‐RCT Uganda was judged unclear for allocation concealment, and quasi‐randomisation of individuals within clusters (though randomisation was in clusters) could have introduced bias; Chen 2000 RCT Taiwan was unclear for sequence generation and allocation concealment, all outcomes were self reported, there was possible contamination and the dropout rate after randomisation was high, with no analysis of differences in dropouts versus non‐dropouts. These two studies contributed 62% of the weight in the pooled analysis. Downgraded by 1.

3Serious inconsistency: I2 was 81%. However, the inconsistency related to the magnitude of benefit favouring collaborative care rather than in the direction of effect. Downgraded by 1.

Open in table viewer
Summary of findings 2. Collaborative care model (NSHWs plus specialist) compared with usual care in treating common mental disorders in adults in low‐ and middle‐income countries (RCTs)

What are the effects of a collaborative care model (NSHW plus specialist supervision) for mental health care in adults with common mental disorders low‐ and middle‐income countries?

Patient or population: Adults (≥ 18 years) with CMDs (includes anxiety or depression, or both)
Settings: Middle‐income countries (Chile, India)
Intervention: Collaborative care model (NSHW plus specialist supervision)

Comparison: Enhanced usual care

Outcomes

Illustrative comparative risks* (95% CI)

Effect estimate
(95% CI)

No of Participants
(studies)

Quality of the evidence
(GRADE)

Comments

Assumed risk

Corresponding risk

Usual care

Collaborative care model

Prevalence of CMDs, short term (2‐6 months)
measured using various CMD/depression rating scales1

205 per 1000

140 per 1000

RR 0.63
(0.44 to 0.90)

2380
(3 studies)

⊕⊕⊝⊝
low2,3

In Patel 2010 C‐RCT India; collaborative care reduced the prevalence of CMDs at 6 months in a subgroup of people treated at public health facilities (RR 0.57, 95% CI 0.42 to 0.78; 1528 participants). This effect was not seen in people treated at private facilities (RR 1.12, 95% CI 0.68 to 1.84; 823 participants)

*The basis for the assumed risk is the mean control group risk across studies for pooled results and the control group risk for single studies. The corresponding risk (and its 95% CI) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).
CI: confidence interval; CIS: Clinical Interview Schedule; CMD: common mental disorder; EPDS: Edinburgh Postnatal Depression Scale; GP: general practitioner; HDRS: Hamilton Depression Rating Scale; ICD: International Classification of Diseases; NSHW: non‐specialist health worker; RCT: randomised controlled trial; RR: risk ratio.

GRADE Working Group grades of evidence
High quality: Further research is very unlikely to change our confidence in the estimate of effect.
Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.
Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.
Very low quality: We are very uncertain about the estimate.

1Araya 2003 RCT Chile: HDRS; Patel 2010 C‐RCT India: CIS‐R generated ICD‐10 diagnosis for CMD; Rojas 2007 RCT Chile: EPDS with a 6‐point reduction in score indicating recovery.

2Serious study limitations: In Araya 2003 RCT Chile, GPs provided both intervention and control treatments, so there was a high risk of contamination. Downgraded by 1.

3Serious inconsistency: I2 was 79% with Araya 2003 RCT Chile clearly an outlier, contributing to this unexplained inconsistency. However, the inconsistency related to the magnitude of benefit favouring collaborative care rather than in the direction of effect. Downgraded by 1.

Open in table viewer
Summary of findings 3. NSHWs compared with usual care for treating maternal depression (RCTs)

What are the effects of NSHW‐led interventions for treating maternal depression in low‐ and middle‐income countries?

Patient or population: Adult women with maternal depression
Settings: Low‐ and middle‐income countries (Chile, Jamaica, Pakistan, Taiwan)
Intervention: NSHW‐led interventions
Comparison: Usual care

Outcomes

Illustrative comparative risks* (95% CI)

Estimate effect
(95% CI)

No of Participants
(studies)

Quality of the evidence
(GRADE)

Comments

Assumed risk

Corresponding risk

Usual care

NSHWs

Severity of symptoms of perinatal depression, (short and medium term: 0‐12 months)
measured using various depression rating scales1

‐

The mean severity of symptoms of perinatal depression ‐ medium term with NSHW‐led interventions was
0.42 standard deviations lower
(0.58 to 0.26 lower)

SMD ‐0.42 (‐0.58 to ‐0.26)

1213
(4 studies)

⊕⊕⊝⊝
low2,3

Note that a small clinically appreciable benefit was set at SMD < 0.2, and a moderate benefit at SMD of 0.5 to 0.8 (Cohen 1988)

*The basis for the assumed risk is the mean control group risk across studies for pooled results and the control group risk for single studies. The corresponding risk (and its 95% CI) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).
BDI: Becks Depression Inventory; CES‐D: Center for Epidemiologic Studies Depression Scale; CI: confidence interval; EPDS: Edinburgh Postnatal Depression Scale; HDRS: Hamilton Depression Rating Scale; NSHW: non‐specialist health worker; RCT: randomised controlled trial; SMD: standardised mean difference.

GRADE Working Group grades of evidence
High quality: Further research is very unlikely to change our confidence in the estimate of effect.
Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.
Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.
Very low quality: We are very uncertain about the estimate.

1Baker‐H 2005 CRCT Jamaica CES‐D; Chen 2000 RCT Taiwan Taiwanese BDI; Rahman 2008 CRCT Pakistan: HDRS; Rojas 2007 RCT Chile: EPDS.

2 Serious study limitations: Baker‐H 2005 CRCT Jamaica; Chen 2000 RCT Taiwan has study limitations and together contributed 24% weight to the pooled estimates. Removal of these trials altered the results to favour NSHW‐led interventions strongly. Downgraded by 1.

3 Serious imprecision: The 95% CI of the SMD indicated appreciable and non‐appreciable benefit for NSHW‐led interventions. Downgarded by 1.

Open in table viewer
Summary of findings 4. NSHWs compared with specialists in treating depression in adults in low‐ and middle‐income countries (CBAs)

What are the effects of NSHWs compared with specialists in treating depression for mental health care in low‐ and middle‐income countries?

Patient or population: Adults with depression
Settings: Middle‐income countries (Hungary and Argentina)

Intervention: NSHWs providing pharmacological intervention
Comparison: Specialists providing pharmacological intervention

Outcomes

Illustrative comparative risks* (95% CI)

Relative effect
(95% CI)

No of Participants
(studies)

Quality of the evidence
(GRADE)

Comments

Assumed risk

Corresponding risk

Specialists

NSHWs

Severity of depression, short term (0‐56 days)
measured using HDRS
Follow‐up: 56 days

The mean score (SD) on the HDRS was 9.6 (2.1)

The mean severity of depression ‐ short term (2 months post intervention) in the NSHW group was
0.9 lower
(1.2 to 0.6 lower)

MD ‐0.90 (‐1.20 to ‐0.60)

768
(1 study)

⊕⊝⊝⊝
very low1,2

Note that a small clinically appreciable benefit was set at SMD < 0.2, and a moderate benefit at SMD of 0.5 to 0.8 (Cohen 1988)

*The basis for the assumed risk is the risk in the control group. The corresponding risk (and its 95% CI) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).
CBA: controlled before‐and‐after; CI: confidence interval; HDRS: Hamilton Depression Rating Scale; MD: mean difference; NSHW: non‐specialist health worker; RR: risk ratio; SD: standard difference; SMD: standardised mean difference.

GRADE Working Group grades of evidence
High quality: Further research is very unlikely to change our confidence in the estimate of effect.
Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.
Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.
Very low quality: We are very uncertain about the estimate.

1 Very serious study limitations: Lyketsos1999CBA Argentina was a CBA study so selection bias was likely. There was a risk of contamination and outcome assessments were done by same physicians doing the intervention. Downgraded by 2.

2 Serious imprecision: The MD on the HDRS was < 1 point and this is not clinically a meaningful difference on the HDRS; and the 95% CI of the MD indicated only non‐appreciable benefits with NSHW intervention versus specialist intervention. However, the data came from only one study, so estimate is imprecise. Downgraded by 1.

Open in table viewer
Summary of findings 5. NSHW‐led psychological interventions compared with usual care in treating adults with PTSD (NRCT)

What are the effects of NSHWs compared with usual mental health care in low‐ and middle‐income countries for data from an NRCT in adults with PTSD?

Patient or population: Adults with PTSD

Settings: Low‐ and middle‐income countries (Bosnia, Burundi, Uganda)
Intervention: NSHWs and OPHRs delivering psychological interventions (narrative exposure therapy, trauma counselling and workshops with psychoeducation)
Comparison: Usual care

Outcomes

Illustrative comparative risks* (95% CI)

Relative effect
(95% CI)

No of Participants
(studies)

Quality of the evidence
(GRADE)

Comments

Assumed risk

Corresponding risk

Usual care

NSHWs/OPHRs

Severity of PTSD symptoms in LHW/teacher‐led psychological interventions (trauma counselling, workshop with psychoeducation, mother intervention) in the short term (2 weeks to 6 months)
measured using various PTSD symptom scales1

The mean severity of PTSD with psychological interventions in the short term (within 6 months post‐intervention) was
0.36 standard deviations lower
(0.67 to 0.05 lower)

SMD ‐0.36 (‐0.67 to ‐0.05)

223
(3 studies)

⊕⊕⊝⊝
low2,3

*The basis for the assumed risk is the median control group risk or mean control group risk across studies for pooled estimates and the control group risk for single studies. The corresponding risk (and its 95% CI) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).
CI: confidence interval; LHW: lay health workers; NRCT: non‐randomised controlled trial; NSHW: non‐specialist health worker; OPHR: other professionals with health roles; PTSD: post‐traumatic stress disorder; SMD: standardised mean difference.

GRADE Working Group grades of evidence
High quality: Further research is very unlikely to change our confidence in the estimate of effect.
Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.
Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.
Very low quality: We are very uncertain about the estimate.

1Neuner 2008 NRCT Uganda: Post‐traumatic Stress Diagnostic Scale; Yeomans 2010 RCT Burundi: Harvard Trauma Questionnaire; Dybdahl 2001 RCT Bosnia: Impact of Events Scale.

2Serious study limitations: Neuner 2008 NRCT Uganda no allocation concealment, randomisation had no sequence generation. High dropout rate and different between groups, different baseline characteristics and likely contamination; Yeomans 2010 RCT Burundi: unvalidated Harvard Trauma Questionnaire in the local context (only validated in Burundi) so may affect reliability of outcomes. Dybdahl 2001 RCT Bosnia: incomplete outcome reporting, Impact of Events Scale not previously validated in this setting. Downgraded by 1.
3Serious imprecision: The 95% CI of the effect estimates demonstrated appreciable and non‐appreciable benefit with NSHW care. Downgraded by 1.

Open in table viewer
Summary of findings 6. NSHWs compared with usual care in improving dementia patients' and carers' outcomes in low‐ and middle‐income countries (RCTs)

What are the effects of NSHW‐led care in improving dementia patients' and carers' outcomes for mental health care in low‐ and middle‐income countries?

Patient or population: People with dementia and their carers
Settings: Middle‐income countries (India, Russia)
Intervention: NSHWs delivering brief intervention
Comparison: Usual care

Outcomes

Illustrative comparative risks* (95% CI)

Estimate effect
(95% CI)

No of Participants
(studies)

Quality of the evidence
(GRADE)

Comments

Assumed risk

Corresponding risk

Usual care

NSHWs

Severity of patient behavioural problems, short term (6 months)
measured using the behavioural symptom scale (NPI‐S)

The mean severity of patient behavioural problems with this brief carer intervention was
0.26 standard deviations lower
(0.60 lower to 0.08 higher)

SMD ‐0.26 (‐0.60 to 0.08)

134
(2 studies)

⊕⊕⊕⊝
moderate1,2

Note that a small clinically appreciable benefit was set at SMD < 0.2, and a moderate benefit at SMD of 0.5‐0.8 (Cohen 1988)

*The basis for the assumed risk is the mean control group risk across studies for pooled results and the control group risk for single studies. The corresponding risk (and its 95% CI) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).
CI: confidence interval; NPI‐S: Neuropsychiatric Inventory ‐ Severity; NSHW: non‐specialist health worker; RCT: randomised controlled trial; SMD: standardised mean difference.

GRADE Working Group grades of evidence
High quality: Further research is very unlikely to change our confidence in the estimate of effect.
Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.
Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.
Very low quality: We are very uncertain about the estimate.

1 No serious study limitations: Gavrilova 2009 RCT Russia was unclear whether allocation concealed. Dias 2008 RCT India was at low risk of bias and contributed > 60% of the weight to the pooled estimates. Removal of the former study did not alter the results. Not downgraded.
2 Serious imprecision: The 95% CI for the pooled estimates indicates appreciable benefit for NSHW care and non‐appreciable benefit for usual care. Downgraded by 1.

Open in table viewer
Summary of findings 7. NSHW‐led brief alcohol interventions compared with usual care for adults with alcohol‐use disorders (RCTs)

What are the effects of NSHWs in delivering brief alcohol interventions in RCTs for alcohol‐use disorders?

Patient or population: People with alcohol‐use disorders
Settings: Low‐ and middle‐income countries (Thailand, Kenya)
Intervention: NSHWs in delivering brief alcohol interventions
Comparison: Usual care

Outcomes

Illustrative comparative risks* (95% CI)

Relative effect
(95% CI)

No of Participants
(studies)

Quality of the evidence
(GRADE)

Comments

Assumed risk

Corresponding risk

Usual care

NSHWs

Amount of alcohol consumed, short term (3‐6 months)
measured using the number of drinks/drinking day (in past week to 30 days)

The mean amount of alcohol consumed in the intervention groups was 1.68 lower (2.79 lower to 0.57 lower)

MD ‐1.68 (‐2.79 to ‐0.57)

167
(2 studies)

⊕⊕⊝⊝
low1,2

*The basis for the assumed risk is the mean control group risk across studies for pooled data or the control group risk for individual studies. The corresponding risk (and its 95% CI) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).
CI: confidence interval; MD: mean difference; NSHW: non‐specialist health worker; RCT: randomised controlled trial; RR: risk ratio.

GRADE Working Group grades of evidence
High quality: Further research is very unlikely to change our confidence in the estimate of effect.
Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.
Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.
Very low quality: We are very uncertain about the estimate.

1 Serious study limitations: Noknoy 2010 RCT Thailand: high dropout rate with no information on whether they are different to completers, no validated tools in the setting, so unreliable primary outcomes. Papas 2011 RCT Kenya: unclear about whether the non‐blinding of outcome assessors would have impacted on study. Downgraded by 1.
2 Serious imprecision: The 95% CI of the MD in number of drinks indicates marginal benefit and no appreciable benefit with interventions. The sample size was also low. Downgraded by 1.

Open in table viewer
Summary of findings 8. NSHWs/OPHRs compared with usual care in conducting interventions for children with post‐traumatic stress disorder and depression (RCTs)

What are the effects of NSHWs/OPHRs conducting interventions for children with PTSD from RCTs in low‐ and middle‐income countries?

Patient or population: Children/adolescents with PTSD and related depressive/anxiety symptoms
Settings: Low‐ and middle‐income countries (Bosnia, Kosovo, Sri Lanka)
Intervention: NSHWs/OPHRs delivering psychological and psychosocial interventions

Comparison: Usual care

Outcomes

Illustrative comparative risks* (95% CI)

Estimate effect
(95% CI)

No of Participants
(studies)

Quality of the evidence
(GRADE)

Comments

Assumed risk

Corresponding risk

Usual care

NSHWs/OPHRs

Severity of PTSD symptoms in LHW/teacher‐led interventions, short term (1‐6 months)
measured using various PTSD severity of symptom scales1

The mean severity of PTSD symptoms in children in teacher‐led intervention groups was
1.2 standard deviations lower
(1.52 to 0.88 lower)

SMD ‐0.89 (‐1.49 to ‐0.30)

298
(3 studies)

⊕⊝⊝⊝
very low2,3

Note that a small clinically appreciable benefit was set at SMD < 0.2, a moderate benefit at SMD of 0.5‐0.8, and a large benefit > 0.8 (Cohen 1988)

*The basis for the assumed risk the mean control group risk across studies for pooled results and the control group risk for single studies. The corresponding risk (and its 95% CI) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).
CI: confidence interval; LHW: lay health workers; NSHW: non‐specialist health worker; OPHR: other professionals with health roles; PTSD: post‐traumatic stress disorder; RCT: randomised controlled trial; SMD: standardised mean difference; UCLA: University of California at Los Angeles.

GRADE Working Group grades of evidence
High quality: Further research is very unlikely to change our confidence in the estimate of effect.
Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.
Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.
Very low quality: We are very uncertain about the estimate.

1Berger2009 CRCT SriLanka: UCLA PTSD scale; Gordon 2008 RCT Kosovo: Harvard Trauma Questionnaire; Ertl 2011 RCT Uganda: Clinician‐administered PTSD scale (CAPS).
2 Very serious study limitations: Gordon 2008 RCT Kosovo no allocation concealment, also likely contamination, and no blinding of outcome assessments; Berger2009 CRCT SriLanka no allocation concealment, likely contamination and outcomes not adjusted for clustering. Two of the three trials are at risk of bias and contribute to > 60% weight to the pooled results. Downgraded by 2.
3 Serious inconsistency: I2 = 78%. The inconsistency is not related to the direction of effect. Downgraded by 1.

Background

Description of the condition

The global burden of mental, neurological and substance‐abuse (MNS) illnesses is high. The latest global burden of disease estimates have shown that mental, behavioural and neuropsychiatric disorders all feature in the top 30 causes of all years lived with disability, the highest contributors being major depression (ranked second), anxiety (ranked seventh) and substance‐use disorders (ranked twelfth) (Vos 2012). The contribution of major depressive disorders to worldwide disability‐adjusted life years (DALYs) has increased by 37% from 1990 to 2010 and is predicted to rise further (Murray 2012; Prince 2007). Furthermore, self inflicted injuries and alcohol‐related disorders are likely to increase in the ranking of disease burden due to the decline in communicable diseases and because of a predicted increase in war and violence. The disease burden due to Alzheimer's disease is also increasing, linked to the demographic transition towards an ageing population (Vos 2012).

These illnesses also come with substantial economic costs. One recent report on the global economic burden of non‐communicable diseases (NCDs) suggests that by the early 2030s, mental health conditions alone will account for the loss of an additional USD16.1 trillion with dramatic impact on productivity and quality of life (Bloom 2011). Data remain poor on the macro‐economic costs for low‐ and middle‐income country (LMIC) settings (Hu 2006). However, the economic and social costs for individuals and families are substantial. High direct costs are incurred in countries where health spending is met largely through private, as opposed to public, spending and where health insurance and employer‐met health payments are insubstantial (Patel 2007a). High indirect costs are also incurred due to informal care‐giving and lost work opportunities, as well as due to untreated disorders and their associated disability (Chisholm 2000a; WHO 2003a).

The gap between those who could benefit from MNS health interventions and those who receive such care is very large (WHO 2008; WHO 2010); in LMICs up to 90% of people needing care do not receive it (Demyttenaere 2004; Saxena 2007). This is despite the existence of a range of cost‐effective interventions in mental health care (Patel 2007b; WHO 2010). Major barriers to closing the treatment gap are the huge scarcity of skilled human resources, large inequities and inefficiencies in resource distribution and utilisation, and the significant stigma associated with psychiatric illness (Saxena 2007). Some papers have advocated for scaling up evidence‐based services and for the task‐shifting of mental health interventions to non‐specialists as key strategies for closing the treatment gap (Jacob 2007; Lancet 2007; Patel 2007b; Prince 2007; Saraceno 2007; Saxena 2007).

Description of the intervention

Non‐specialist health workers (NSHWs) are first‐level providers who have received general rather than specialist mental health training. Cadres included are professionals (doctors, nurses and other general paraprofessionals) and non‐professionals (such as lay providers). NSHWs do not include, for example, psychiatrists, neurologists, psychologists, psychiatric nurses or mental health social workers. Other professionals with health roles (OPHRs), such as teachers and community‐level workers, are a further human resource used in delivering mental health care and are also included in this review. These OPHRs have an important role, particularly in the promotion of mental health and the detection of mental disorders (Patel 2007b; Patel 2008b; WHO 2003b).

NSHWs and OPHRs have been used in various services, including those delivered by governmental, private and non‐governmental organisations (NGOs) in clinics, half‐way homes and communities. They have been involved in a variety of activities and roles, including detecting, diagnosing, treating and preventing common and severe mental disorders, epilepsy and mental retardation. Their roles differ according to their level of training. For example, lay health workers (LHW) have been involved in supporting carers, befriending, ensuring adherence and in detection of mental health problems (Chatterjee 2003; Dias 2008 RCT India; Rahman 2008 CRCT Pakistan). Nurses, social workers and lay workers may also take on follow‐up or educational/promotional roles (Araya 2003 RCT Chile; Chatterjee 2003; Patel 2008b). In addition, doctors with general mental health training have been involved in the identification, diagnosis, treatment and referral of complex cases (Murthy 1987; Patel 2008b; Saxena 2007).

How the intervention might work

In many LMICs, training and retaining sufficient numbers of specialists is not feasible in the near future. It is, therefore, important in these settings to consider options for expanding access to mental health services. The use of NSHWs, who are far more numerous and affordable than specialists, is one such option that is of high relevance to LMICs.

Training these NSHWs to deliver MNS interventions may be a way of expanding provision of mental health services as well as making these services more accessible to communities. It has been suggested that interventions that rely on NSHWs could deliver general health and mental health interventions that are at least as effective and acceptable as those delivered by specialist health workers (Chatterjee 2003; Lewin 2008; McKenzie 2004; Thornicroft 2004; WHO 2001; Wiley‐Exley 2007). In addition, NSHW interventions often have lower up‐front costs compared with reliance on professional specialist health workers. However, it is possible that these savings may be cancelled out by higher downstream resource use (Chisholm 2000a), and this review will, therefore, include data on the costs and cost‐effectiveness of NSHW interventions.

The review is limited to LMICs where the need for NSHWs is greater than in high‐income settings. The prevalence of psychiatrists and psychiatric nurses is much lower in LMICs (the median number of psychiatrists is 172 times lower in low‐income countries (LICs) than high‐income countries (HICs) (Kakuma 2011; Mental Health Atlas 2011)) and the organisation and resourcing of mental health services is poorer. These differences in the organisation of mental health services between LMICs and HICs, with poorer countries having little or no mental health service structures in primary care or the community, means that the problem of providing mental health care is different in such settings. NSHWs may need to work with little or no support from specialist mental health services and fewer options for referral. Consequently, NSHWs interventions might be expected to function differently in many LMICs compared with HICs.

Why it is important to do this review

The continuing shortage of specialist human resources for health in LMICs has made the need to involve non‐specialists in MNS healthcare provision more urgent. Reliable evidence is needed on the effectiveness of NSHWs and OPHRs in scaling up mental health interventions, including for the detection, treatment and rehabilitation of MNS disorders. This systematic review will provide the evidence needed to inform policy development for the sustainable scaling up of mental health services in LMICs (Cohen 2003; Murthy 2008).

The intention of this review is to examine which non‐specialised cadres of healthcare providers can effectively deliver different aspects of treatment interventions.

Objectives

To assess the effectiveness of the delivery of mental, neurological and substance abuse (MNS) interventions by non‐specialist health workers (NSHWs) and other professionals with health roles (OPHRs) in LMICs. This includes the effects on patient and health delivery outcomes of NSHWs and OPHRs:

  • delivering acute MNS interventions;

  • delivering long‐term follow‐up and rehabilitation for people with MNS disorders;

  • detecting MNS disorders.

For each of these areas, we have also examined the impacts of delivery by NSHWs and OPHRs on the resource use and costs associated with MNS healthcare provision in LMICs.

Methods

Criteria for considering studies for this review

Types of studies

We included randomised controlled trials (RCT), non‐randomised controlled trials (NRCT), controlled before‐and‐after (CBA) studies and interrupted time series (ITS) studies. We only included CBAs with at least two control sites and two intervention sites. We included controlled and non‐controlled ITS that had at least three time points before the intervention and three time points after the intervention (as per the Cochrane Effective Practice and Organisation of Care (EPOC) review group criteria) (Ballini 2010). We only included studies conducted in LMICs, as defined by the World Bank.

We also included economic studies conducted as part of included effectiveness studies. We considered full economic evaluations (cost‐effectiveness analyses, cost‐utility analyses or cost‐benefit analyses), cost analyses or comparative resource utilisation studies. We extracted and reported only cost and resource usage outcomes from these studies.

Types of participants

We included children (aged below 18 years) or adults with any MNS seeking first‐level care/primary care or who were detected in the community in LMICs. Additionally we included carers of people with MNS disorders (i.e. any relative or friend of any age who defined themselves as a key supporter to a person with an MNS disorder) as some interventions may be directed at the carers rather than at patients themselves ‐ for example interventions to alleviate carer burden.

(See Table 1 for further definitions of participants, 'LMIC' and 'primary care'.)

Open in table viewer
Table 1. Definitions

Adult

Patients who were ≥ 18 years old. However, if some studies had an age range from, for example, 16 years upwards and the majority of participants are over 18 years, we included these study participants as adults.

Children and adolescents

Children (from birth to 18 years) were considered as a separate group of participants as they have 1. different patterns of psychopathology/mental disorders; 2. different help‐seeking behaviours that would, therefore, require different interventions, in different settings (e.g. schools) and a different approach to careworker interventions (such as teacher‐led interventions).

Mental, neurological and substance‐abuse (MNS) disorders

This review included MNS disorders as defined by any criteria within included papers. For the purpose of subgroup analysis, we subcategorised these disorders using the International Classification of Diseases (ICD)‐10 criteria for mental and behavioural disorders and epilepsy in adults (the related ICD‐10 code is listed in brackets). These categories are most likely to be used in LMIC mental health service delivery, and are based on Patel's classification (Patel 2003c), and the World Health Organization (WHO) MNS disorder categorisation (WHO 2008)

1. Common mental disorders

Mild to moderate mood (affective) disorders (F32‐38)

Neurotic, stress‐related and somatoform disorders  (F40‐49)

Behavioural syndromes associated with physiological disturbances and physical factors (F50‐59)

2. Severe mental disorders

Schizophrenia, schizotypal and delusional disorders (F20‐F29)

Bipolar affective disorder (F31)

Severe depressive episode with/without psychosis (F32.2, F32.3)

3. Neuropsychiatric disorders

Organic, including symptomatic, mental disorders (includes dementia) (F1‐9)

Mental retardation (F70‐79)

Epilepsy (G40)

4. Disorders caused by substance abuse

Mental and behavioural disorders due to psychoactive substance use (F10‐19)

5. Mental disorders specifically related to childhood/development

Conduct disorders

Developmental disorders

Eating disorders

Pervasive developmental disorders

The diagnosis could be made in clinical practice or in the context of the trial.

First level care, primary care and community

First level of contact with formal health services were community‐based interventions or  primary care interventions (or both), on their own or attached to hospital settings, provided they had no specialist input apart from supervision (modified from Wiley‐Exley 2007). This would include individuals with mental illness living in the community and programmes in outpatient clinics or primary care practices. This would not include programmes in hospitals unless the programmes in the hospitals were providing care to outpatients (i.e. generalists in outpatient departments).

Community: as mentioned above detection of mental disorders in all age groups were often done outside the health facility, for example through school, training and other community settings. Therefore, we considered interventions outside the health sector.

Low‐ and middle‐income country (LMIC)

Any country that has ever been an LMIC, as defined by the World Bank lists of LMICs

Non‐specialist health workers (NSHWs)

Health workers who were not specialised in MNS disorders or have not received in‐depth professional specialist training in this clinical area. These included doctors, nurses, auxiliary nurses, lay health workers, as well as allied health personnel such as social workers, occupational therapists. This category did not include professional specialist health workers such as psychiatrists, neurologists, psychiatric nurses or mental health social workers. For inclusion, NSHWs received some training in MNS disorders (in either the control or the intervention group), but this would not constitute a professional category. The authors made a judgement of what constitutes 'some training'. Examples of 'some training' may be an undergraduate module or a short course in mental health.

Other professionals with health roles (OPHRs)

People who were involved as community‐level workers but were not within the health sector, as many people, particularly adolescents and young adults, have low contact with health workers. This category included teachers/trainers/support workers from schools and colleges, and other volunteers or workers within community‐based networks or non‐governmental organisations. These OPHRs have an important role particularly in the promotion of mental health and detection of mental disorders (Patel 2007c; Patel 2008a; WHO 2003a)

We excluded studies that looked at informal care provided by family members or extended members only to members of his or her own family (i.e. who were unavailable to other members of the community) from this review. As previously highlighted in Lewin;s Cochrane review, "these interventions are qualitatively different from other LHW [lay health worker] interventions included in this review given that parents or spouses have an established close relationship with those receiving care which could affect the process and effects of the intervention" (Lewin 2010).

Clinical interventions

1. Detection (recognition and diagnosis) of illness, including screening

2. Acute interventions: drug treatment, non‐drug treatment/care (such as specific psychological therapies, or interventions with psychosocial components like counselling, psychoeducation, coping skills, etc.), referral

3. Follow‐up, rehabilitation

Service interventions

These include change in staffing, or change in mechanism of mental health service delivery (e.g. extension of mental health services through camps and such other outreach services, mobile vans, etc.).

Types of interventions

Clinical (medical and psychological) and service interventions delivered in primary care or the community by NSHWs or OPHRs, and intended to improve MNS disorders were included (see Table 1 for definitions of OPHR and NSHW and types of interventions). We did not include social interventions (such as income generation or general social support) if the trial did not also include a specific MNS intervention.

We included interventions delivered for any MNS disorder. Acute interventions delivered by NSHWs/OPHRs could include various forms of psychotherapy or pharmacological treatment. Long‐term interventions delivered by NSHWs/OPHRs could include roles in follow‐up or rehabilitation of people with chronic severe mental disorders, and roles in detecting and dealing with relapse/recurrence, compliance issues, side effects of treatment or psychosocial problems.

We considered the following comparisons:

  • provision of MNS care by NSHWs/OPHRs with some MNS care training compared with usual/no care;

  • provision of MNS care by NSHWs/OPHRs trained and supervised in MNS care (i.e. the highest level of training for NSHWs) compared with mental health specialists in primary care and the community;

  • provision of MNS care by NSHWs/OPHRs with some MNS care training compared with non‐trained NSHWs/OPHRs.

We included studies where a specialist teaches NSHW/OPHRs about psychiatric illness and its management. The only interventions of this type that we excluded were those where there were no patient outcomes (i.e. where they only assessed knowledge or attitude changes, such as pre‐post training interventions).

We included studies that considered the effect of detection, screening or case‐finding of MNS disorders by NSHWs or OPHRs on subsequent patient and health provider outcomes, compared with NSHWs/OPHRs not actively detecting cases, or where specialists did the detection.

The identification methods used by NSHWs could include 'naturalistic' detection (i.e. detection in the course of a routine clinical consultation), or detection using a validated screening/detection tool (e.g. in the context of a trial). We did not examine diagnostic accuracy between these NSHWs and specialists, as this was likely to be confounded by the screening/detection tools used. Therefore, it would be difficult to differentiate between the effect of the screening tool and the skills of the health worker (specialist or non‐specialist).

Types of outcome measures

We organised these outcomes into categories drawing on the Cochrane Consumers and Communication Review Group's outcome taxonomy (La Trobe 2008), and consultation with co‐reviewers and service users from the Movement for Global Mental Health discussion board. Where studies reported more than one measure for each relevant outcome, we abstracted the primary or main measure (as defined by the study authors). We separately documented the other measures used, as necessary.

We grouped outcomes into two sets of time points:

  • up to six months post intervention (to detect illness recovery/symptom reduction);

  • six to 12 months post intervention (which indicates medium‐ to long‐term avoidance of recurrence and chronicity).

For depression and other common mental disorders, we did not group results up to three months post intervention. This time point would normally elicit whether the length of a depressive episode would be shortened compared with spontaneous recovery (which occurs for 50% of people with depression at three months after treatment initiation and for 65% of people with depression at six months) (Spijker 2002). However, most of these studies had very variable lengths of interventions (zero to 18 months) and it was difficult to ascertain how long the depression had been present when treatment started (we could assume that people who have not recovered naturally within three months seek help). Pooled results up to three months post intervention would, therefore, not reflect whether the intervention shortened recovery from depression to less than or equal to a spontaneous recovery.

Primary outcomes

  1. Improvement of symptoms (e.g. level of anxiety, depression, psychosis).

  2. Psychosocial functioning and impairment (e.g. levels of self esteem, perception of coping, level of dependency, self care ability).

  3. Quality of life outcomes (including disability).

We changed the definitions of outcomes 2 and 3 during our analysis from those stated in the protocol, as many scales measured both impairment and functioning and were considered part of the same spectrum. Quality of life outcomes were deemed different from outcomes related to psychosocial functioning as the former encompass a summary of many other aspects of life in addition to psychosocial functioning.

For the detection component of the review, we aimed to consider the outcomes for the patient, the carer, the health provider, or a combination of these people, not the accuracy of diagnosis among NSHWs, compared with specialists, as this is likely to be confounded by the screening/detection tools used. Therefore, it would be difficult to differentiate between the effect of the screening tool and the skills of the health worker (specialist or non‐specialist). We did not base inclusion decisions on whether a reference or validated standard measure (either a screening instrument or psychiatric assessment) had been used in studies to differentiate between those correctly and incorrectly diagnosed by NSHWs, but this featured as part of the assessment of the quality of evidence (within study limitations).

Secondary outcomes
1. For studies evaluating the detection of mental disorders and the delivery of acute and chronic mental health interventions  

Patient/carer‐oriented outcomes and societal outcomes

  • Patient or carer satisfaction and involvement in decision‐making processes.

  • Patient health behaviour outcomes: such as rates of patient adherence or treatment/follow‐up compliance, utilisation of primary level services.

  • Adverse clinical outcomes: such as adverse effects rates, suicide/deliberate self harm rates, relapse or recurrence, hospital admission/readmission rates.

  • Patient social outcomes: return to work, offending rates, perception of social inclusion.

  • Carer outcomes: such as mental health outcomes, quality of life and functioning.

Health provider and service delivery related outcomes

  • Measures of changes in management (such as referral rates, prescribing patterns and appropriateness).

  • Measures of health worker behaviour (such as improvement in knowledge/skills, attitude/acceptability, retention rates, absenteeism).

  • Measures of service delivery change (such as number of supervision sessions, effect on other health services provided).

2. For studies of costs and resource use

We considered:

  • direct and indirect costs to the patient and health services (including opportunity costs);

  • resource use (such as the patient's lost productivity, and health service personnel's time allocated/number of consultations).

The economic outcome measures considered were informed by the training material of, and discussion with, the Campbell & Cochrane Economics Methods Group (CCEMG 2010). We included only measures related to resource use and costs in this review. We recognise that costs and resource use are intertwined but divided the outcomes in this way to make it clear which outcomes we intended to assess.

Search methods for identification of studies

Electronic searches

We searched the following electronic databases for primary studies:

  • the Cochrane Central Register of Controlled Trials (CENTRAL) 2012, Issue 6 (including the Cochrane EPOC Group Specialised Register (searched 21 June 2012);

  • MEDLINE, 1946 to June week 1 2012, OvidSP (searched 15 June 2012);

  • MEDLINE In‐Process & Other Non‐Indexed Citations 14 June 2012, OvidSP (searched 15 June 2012);

  • EMBASE, 1980 to 2012 week 23, OvidSP (searched 15 June 2012);

  • CINAHL (Cumulative Index to Nursing and Allied Health Literature), 1980 to 19 June 2012, EBSCOhost (searched 19 June 12);

  • PsycINFO, 1806 to June week 2 2012, OvidSP (searched 18 June 2012);

  • Latin American and Caribbean Health Sciences database (LILACS), Virtual Health Library (VHL) (searched 9 August 2012);

  • WHO Global Health Library (World Health Organization Library Information System (WHOLIS), AIM (AFRRO), IMEMR (EMRO), IMSEAR (SEARO, WPRIM, WPRO) (searched 29 June 2012);

  • Science Citation Index and Social Sciences Citation Index, ISI Web of Knowledge (searched 2 October 2012).

The EPOC Trials Search Co‐ordinator (TSC) (Marit Johansen), in consultation with the authors, developed the search strategies.

Search strategies were comprised of keywords and controlled vocabulary terms (selected index terms and free‐text terms relating to NSHWs and mental health).

We applied no language limits. We searched all databases from database start date to date of search. 

We used a combination of two methodology search filters to limit retrieval to appropriate study designs: a modified version of the Cochrane Highly Sensitive Search Strategy (sensitivity‐ and precision‐maximising version ‐ 2008 revision) to identify RCTs (cf. Cochrane Handbook for Systematic Reviews of Interventions Section 6.4d); and an EPOC methodology filter to identify NRCT designs.

Searching other resources

Grey Literature

Trial Registries

We also searched:

  • the reference lists of existing reviews (De Vet 2008);

  • other grey literature (unpublished material), through contacting experts;

  • conducted cited reference searches for all included studies in ISI Web of Knowledge.

We did not search for economic analyses. We retrieved potentially eligible economic analyses when screening records generated from the various searches reported above, but only selected those performed alongside identified effects studies. We contacted the authors of all included effects studies for information on any published or unpublished economic studies related to their trials. We also scanned the reference lists of eligible trials and economic analyses (where these were reported separately to the eligible trials), and other related reviews and papers, for further eligible studies.

See Appendix 1 for all search strategies used.

Data collection and analysis

Selection of studies

Four review authors (NvG, GR, MSM, JP) and a Chinese researcher for the Chinese included study double‐screened all records obtained from the searches. We retrieved full‐text copies of all articles identified as potentially relevant by at least one review author. Two review authors checked each full paper for inclusion criteria. We resolved disagreements on inclusion by discussion. If no agreement was reached, we asked a third review author to make an independent assessment (SL). Where appropriate, we contacted the study authors for further information.

Data extraction and management

Five review authors (NvG, GR, MSM, JP, PT) and the Chinese and Spanish researchers independently extracted descriptive and outcome data for each paper using an adapted version of the EPOC data collection checklist.Two review authors together or by one and cross‐checked by another (except the Chinese paper, which relied on one researcher's data extraction only) extracted data. Review authors obtained any missing data by contacting trial authors. Review authors entered the final agreed descriptive extracted data into the relevant tables of characteristics in Review Manager 5 (RevMan 2012). One review author (NvG) entered the checked outcome data into Review Manager 5 for meta‐analysis and this was checked by PT (RevMan 2012). 

We extracted the following information for all included studies, in the form that this was reported in the original text:

  • details of the intervention: the type and length of each of the clinical, psychosocial and service interventions; a full description of cadre(s) of NSHW/OPHRs consulting with the patient, including details of their training and supervision/support; and the length, frequency and type of intervention delivered by each NSHW/OPHR; description of the specialist providing care (type, experience, training in using reference standard);

  • participants: a full description of the participants (sex, age, socioeconomic status, ethnicity), including details of the MNS condition being treated;

  • setting: country; type of health service (e.g. government funded, NGO, etc.), organisation of the primary care and specialist services; specialist outreach or generalist;

  • results: organised into patient, provider and process outcomes (see above).

Assessment of risk of bias in included studies

Five review authors (NvG, GR, MSM, JP, PT) and the Chinese researcher working in pairs independently assessed each study for risk of bias. NvG and PT independently checked assessments for all studies. We followed the Cochrane EPOC group format (Ballini 2010) (which follows the Cochrane Collaboration approach (Higgins 2009)) to assess risk of bias for each of the study designs (RCT, CBA, NRCT, ITS). For two of the EPOC risk of bias criteria, we did the following:

  • divided detection bias into two categories, assessing whether subjective (requiring a judgement, such as clinical improvement) and objective outcomes (such as number of hospitalised days, etc.) were assessed blindly;

  • assessed attrition bias for two types of outcome: efficacy outcomes and safety outcomes (e.g. adverse events and unintended consequences).

For economic studies, we adapted the Consensus on Health Economic Criteria (CHEC) criteria list (see Appendix 2) to include an extra question on the sources of data used, and we excluded some questions that were already covered as part of the main risk of bias assessment described above.

We incorporated risk of bias assessments by generating 'Risk of bias' summary graphs and figures using Review Manager 5 (RevMan 2012).

Measures of treatment effect

Measures of intervention effect regarding clinical (medical and psychological) and service interventions

For dichotomous outcomes, we used risk ratios (RR). For continuous outcomes, we used the mean difference (MD), standardised mean difference (SMD) or mean change difference (MCD). We expressed all effect estimates with their 95% confidence intervals (CI). For SMDs, we used the Cochrane Handbook for Systematic Reviews of Interventions to interpret their clinical relevance: 0.2 represented a small effect, 0.5 a moderate effect, and 0.8 a large effect (Cohen 1988). We attempted to establish minimally important differences per outcome (as suggested in Guyatt 2013) but this was not possible due to the wide variety of instruments used.

Measures of effect of detection of MNS disorders interventions

We aimed to report the effects of detection of MNS disorders by NSHWs or OPHRs by assessing patient outcomes, looking at the proportion of patients who recovered or improved over a specific length of time as described in the included studies. We aimed to measure health worker outcomes by examining changes in prescribing rates, referral rates and treatment initiation rates.

Unit of analysis issues

Where possible, we re‐analysed studies that randomised or allocated clusters (patients, health professionals, healthcare settings or geographical areas) but did not account for clustering in their analysis (Ukoumunne 1999). We adjusted the results for clustering by multiplying the standard errors of the estimates by the square root of the design effect where the design effect is calculated as DEff = 1 + (M ‐ 1) ICC, where M is the mean cluster size and ICC is the intracluster correlation coefficient. All of the included studies reported the ICCs that we needed.

We combined the adjusted measures of effects of cluster‐randomised trials with the results of non‐cluster trials, if it was possible to adjust adequately the results of the cluster trials. There were too few studies per meta‐analysis to perform sensitivity analyses comparing the effects estimates with and without the inclusion of the cluster trials.

We contacted authors when we needed additional information for the analysis.

Dealing with missing data

For missing or unclear information, we contacted the study investigators for clarification or additional information. We were able to access all required authors for the purpose of statistical information. Some remaining missing information on the qualitative description of the interventions that we did not get despite several attempts at following up with study authors, is highlighted in the Characteristics of included studies tables. To reduce the risk of overly positive answers, we use open‐ended questions (as recommended in the Cochrane Handbook for Systematic Reviews of Interventions, Higgins 2009).

Where possible, we extracted data to allow an intention‐to‐treat (ITT) analysis in which all randomised participants were analysed in the groups to which they were originally assigned. If ITT data were not present, where possible, we did a full ITT analysis where we considered four scenarios in which the people reassigned to the control and intervention groups either had the condition or not. For studies that reported continuous data but did not report standard deviations, we either calculated these from other available data such as standard errors, or imputed these using the methods suggested in Higgins 2009. We did not make any assumptions about loss to follow‐up for continuous data and we analysed results for those who completed the trial.

Assessment of heterogeneity

We first made a qualitative assessment of the extent to which the studies assessing a particular comparison where similar to one another. This included an assessment of the settings, the interventions, the participants and outcomes to determine whether meta‐analysis was appropriate. We obtained an initial visual overview of statistical heterogeneity through scrutinising the forest plots, looking at the overlap between CIs around the estimate for each included study. To quantify the inconsistency across studies, and thus the impact of heterogeneity on the meta‐analysis, we used the I2 statistic, and defined an I2 greater than 50% as indicative of substantial heterogeneity. We then considered these assessments when interpreting the results of a pooled analysis: the importance of an observed I2 was interpreted in light of 1. the magnitude and direction of effects and, 2. the strength of evidence for heterogeneity (e.g. a CI for the I2, or the P value from the Chi2 test).

Assessment of reporting biases

To reduce possible publication bias, we employed strategies to search for and include relevant unpublished studies. These strategies included searching the grey literature and prospective trial registration databases to overcome time‐lag bias.

We used funnel plots for the outcomes with more than four studies to visualise whether there was asymmetry. None of them showed asymmetry. We performed no statistical testing for funnel plot asymmetry as none of the pooled outcomes included more than 10 studies.

Data synthesis

We grouped the studies for comparison by type of disorders (common mental disorders, severe mental disorders, neurological and substance‐abuse disorders); by mix of healthcare providers (NSHW‐led, collaborative, NSHWs and OPHRs); and by types of community intervention (pharmacological, non‐pharmacological and mixed approach). We did this as these categories fit with current models of service delivery in LMICs.

The number of comparisons was larger than anticipated at the protocol stage and we have outlined each comparison in the results section below. For each comparison (groups of disorders), we created tables of summary statistics according to study designs (RCTs, NRCTs and CBAs). These tables included study design, baseline and follow‐up summary statistics, effect estimates and their statistical significance. We used forest plots to display the data graphically.

Where the outcomes assessed and the settings and interventions were very diverse (as agreed by at least two review authors), we did not consider it appropriate to combine the results quantitatively. For these results, we have presented a descriptive summary of data.

For all data syntheses, we used the generic inverse‐variance model of analysis as this allows the analysis of continuous and dichotomous data and allows clustered and non‐clustered data to be combined. We based the choice of whether to use a fixed‐effect or random‐effects model on the extent to which studies were similar, or homogeneous, based on their PICOS characteristics (population, intervention, comparators, outcomes and settings). No studies were homogeneous enough to apply the fixed‐effect model.

We reported the results separately for RCTs and for NRCTs. No ITS studies were included in the review. We used effect estimates adjusted for confounding (baseline differences in control and intervention groups) where possible, and used the methods described in Reeves 2009 to guide data synthesis.

Economic data

We conducted all the elements of the economics component of this review according to current guidance on the use of economics methods in the preparation and maintenance of Cochrane reviews (Shemilt 2009). We classified the included economic evaluations based on an established system (Drummond 2005). We summarised the characteristics and results of included economic evaluations using additional tables, supplemented by a narrative summary that compared and evaluated methods used and principal results between studies.

We displayed resource use and cost data in a table, along with unit cost data (where available). A unit cost was defined as the cost of each specific resource input calculated by multiplying the measured number of units (quantities) of an item of resource use (e.g. the number of hours of time provided by a senior teacher) by an applicable unit cost (e.g. the salary cost of one hour of senior teacher time). We reported the currency and price year applicable to measures of costs and unit costs in each original study. Measures of costs are highly likely to vary across and within study settings, and over time. This is the product of variations in the underlying quantities of resource use and variations in the underlying unit costs.

Because the data on resource use and costs were very heterogeneous, meta‐analysis was not appropriate and we presented the findings narratively. We discussed the limitations of this approach below.

Subgroup analysis and investigation of heterogeneity

Within each comparison, the following subgroups were considered: by category of health worker (professionals: e.g. doctors, nurses), OPHRs and non‐professionals (LHWs); by types of community intervention (e.g. collaborative versus psychological interventions in comparison 3); and by setting (government versus non‐government). We were not able to perform subgroup analyses to check if the intervention effect varied with different population characteristics as the number of included studies for each comparison was not sufficient. Where applicable, we have described subgroup differences narratively under Main results.

For random‐effects meta‐analyses, we used the formal Chi2 test and I2 statistic for subgroup differences in RevMan 2012 to detect statistically significant subgroup differences.

Sensitivity analysis

It was not possible to compare intervention effects according to risk of bias using meta‐regression due to insufficient data. We conducted sensitivity analyses based on attempting to reduce clinical heterogeneity.

Summarising and interpreting results

We used the GRADE approach to assess the quality of evidence related to each of the key outcomes (Schünemann 2009). We used the GRADE profiler (GRADE 2007), to import data from Review Manager 5 (RevMan 2012) and create 'Summary of findings' tables.

For assessments of the overall quality of evidence for each outcome that included pooled data from RCTs only, we downgraded the evidence from 'high quality' by one level for serious (or by two for very serious) study limitations (risk of bias), indirectness of evidence, serious inconsistency, imprecision of effect estimates or potential publication bias. Data from observational studies started at low quality. None were upgraded to moderate or high quality as no pooled estimates revealed a large magnitude of effect, negligible concerns about confounders or a strong dose‐response gradient.

We used these assessments, along with the evidence for absolute benefit or harm of the interventions and the sum of available data on all critical and important outcomes from each study included for each comparison, to draw conclusions about the effectiveness of NSHWs in mental healthcare provision in LMICs. 'Summary of findings' tables consisted of critically important clinical and functional outcomes identified in the selected trials.

When judging the importance of SMDs, we acknowledged that 0.2 represents a slight effect, 0.5 a moderate effect, and 0.8 a significant effect; and chose a threshold of 0.5 to indicate a minimum clinically important difference (Guyatt 2008; Higgins 2011).

Results

Description of studies

Results of the search

We included 38 studies in this review. Including the four consecutive searches performed in January 2011, May 2011, June 2012 and August 2012, we screened 11,825 titles and abstracts (excluding duplicates), of which we sourced 739 full texts to check inclusion criteria and we sourced 90 relevant references to screen their bibliographies (Figure 1).

Figure 1
Study flow diagram.

Study flow diagram.

Included studies

Study design

Of the 38 included studies, 17 were RCTs, 10 were cluster RCTs, nine were CBA studies and two were NRCTs. Analysis was by ITT in  eight  studies (Bolton 2007 RCT Uganda; Ertl 2011 RCT Uganda; Hirani 2010 CRCT Pakistan; Jenkins 2012 C‐RCT Kenya; Jordans 2010 C‐RCT Nepal; Tiwari 2010 RCT China; Tol 2008 C‐RCT Indonesia; Tol 2012 C‐RCT SriLanka), and was unclear in one (Neuner 2008 NRCT Uganda). It was not possible to do an ITT for the remaining studies (see Dealing with missing data).

Setting

Fifteen included studies were conducted in seven LICs: Burundi (one study), Kenya (two studies), Nepal (one study), Pakistan (three studies), Rwanda (two studies), Sri Lanka (two studies) and Uganda (four studies). Twenty‐three studies were from 15 middle‐income countries: Argentina (one study), Bosnia (one study), Chile (three studies), China (three studies), Hungary (one study), India (two studies), Indonesia (two studies), Jamaica (one study), Kosovo (one study), Malaysia (one study), Palestinian Territories (two studies), Russia (one study), Thailand (two studies), Turkey (one study) and Vietnam (one study). These LIC and middle‐income country assignments are based on the World Bank's classification of countries by gross national income per capita in 2010.

In this section, as well as following sections (participants, interventions, etc.), the numbers when added up may exceed 38 due to double counting. There were 16 studies from rural, 23 from urban and five from refugee camp settings. Most interventions were delivered in community groups/centres (11 studies). Others were delivered at home (nine studies), in primary healthcare (PHC) centres (eight studies), in schools (seven studies) and in other health clinics (three centres).

Participants

Twenty‐seven studies included adults. Of the studies including children, 10 included children up to the age of 12 years, and eight focused on adolescents (aged 12 to 17 years). Most studies covered common mental disorders (18 included depression, anxiety, maternal depression) and PTSD (12 studies). See 'Effects of interventions' for details of these by analysis groups.

Interventions

NSHWs and OPHRs: various cadres were used: LHWs (22 studies), doctors (nine studies), nurses (six studies), teachers (six studies) and social workers (three studies). The educational level of the LHWs was poorly documented, but of the 15 studies that did specify this, eight selected LHWs with a minimum of secondary school education, three used illiterate LHWs and three included LHWs who had primary school education and who were or were not literate. Remuneration was generally poorly described. The training and supervision of these providers are described in detail under 'Effects of interventions'.

Interventions: many studies combined different types of interventions. The eight interventions providing pharmacotherapy also provided follow‐up to check adherence, the effect of medication and side effects (provided by a LHW (four studies), a nurse/clinical officer (one study), a social worker (one study) or a doctor (two studies). Twenty‐five studies had some form of psychosocial intervention (which included psycho‐education, various support and general counselling/coping skills interventions and stimulation programmes for children). Sixteen studies used specific psychological interventions on their own or as part of a collaborative care model (e.g. cognitive behavioural therapy (CBT), interpersonal therapy (IPT), motivational interviewing). One study evaluated economic skills building as a second arm to the trial, which were expected to have an effect on mental health outcomes. No studies examined detection by NSHWs or OPHRs and none reported health worker outcomes. More details on these are provided under 'Effects of interventions'.

Economic studies

Three economic studies were conducted alongside included RCTs (Araya 2003 RCT Chile; Jordans 2011 (which is linked to Tol 2008 C‐RCT Indonesia; Tol 2012 C‐RCT SriLanka) and Zambori 2002 CBA Hungary). One further study noted that the financial burden and severity of schizophrenia decreased marginally for both intervention and control groups, but did not reach statistical significance; however, it did not measure costs (Paranthaman2010CBAMalaysi). In addition, one study mentioned they had collected cost data but results were not yet available before the end of the search period (Patel 2010 C‐RCT India). This was subsequently published (Buttorff 2012). We aim to include these data in a future update.

Excluded studies

We excluded 701 studies, of which 289 were of interest to this area of study but did not fulfill all inclusion criteria. These 289 studies, together with their reasons for exclusion, are documented in Characteristics of excluded studies.

Thirteen studies that included economic data on MNS conditions, but were not linked to studies included in this review, are reviewed in Appendix 3.

Risk of bias in included studies

The most often identified biases across studies were allocation concealment, random sequence generation, reliability of primary outcomes and blinding of outcome assessment (Figure 2; Figure 3).

Figure 2
Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies.

Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies.

Figure 3
Risk of bias summary: review authors' judgements about each risk of bias item for each included study.

Risk of bias summary: review authors' judgements about each risk of bias item for each included study.

Allocation

Only 13 of the 38 included studies met the ‘low risk of bias’ criteria for allocation concealment. Of the remaining studies, 13 explicitly did not conceal allocation (of which 10 were not RCTs). For 12 studies, the risk of bias was unclear due to poor reporting.

Eleven studies did not utilise randomised sequence generation. One RCT was also at high risk of bias with regard to allocation sequence generation because they had a combination of random and non‐random sequence generation (Sutcliffe2009RCT Thailand).

Several studies did not have similar subjective or objective outcome measurements (such as numbers of days in hospital) at baseline between the two arms (subjective outcomes: seven unclear and 10 not similar; objective outcomes: five unclear and five not similar) or did not have similar baseline characteristics (seven not similar and three unclear). The studies in which two or three of the baseline characteristics were not similar included the following CBA studies (Loughry 2006 CBA Palestin; Lyketsos1999CBA Argentina; Paranthaman2010CBAMalaysi; Thabet 2005 CBA Palestine; Zambori 2002 CBA Hungary), and RCTs (Li 1989 RCT China; Sutcliffe2009RCT Thailand).

Blinding

We divided the blinding domain into blinding of participants and personnel, and blinding of outcome assessment. All studies reported blinding of outcome assessment, one study did not blind participants/personnel (Neuner 2008 NRCT Uganda), and for four studies it was unclear if participants/personnel were blinded (Dybdahl 2001 RCT Bosnia; Ertl 2011 RCT Uganda; Loughry 2006 CBA Palestin; Lyketsos1999CBA Argentina).

Incomplete outcome data

We considered incomplete outcome data separately for efficacy and for adverse outcomes. For most studies, outcome data were complete. However, for six studies, this was unclear and seven had incomplete outcome data. Twenty‐two studies did not clearly report whether they had data on adverse outcomes, and an additional four studies stated explicitly that they had not collected adverse outcome data (or we obtained this information from the authors). This made analysis of adverse outcomes difficult for most comparisons.

Selective reporting

For 26 of the 38 studies, there appeared to be no selective reporting, based on the outcomes listed in the methods section of these papers, and from contacting authors where there was doubt. In only one study was it clear that there had been selective reporting (Dias 2008 RCT India). In 11 studies, this was not clear (see Characteristics of included studies tables).

Other potential sources of bias

Risk of contamination was quite common among both RCTs and CBA studies. We assessed six studies as unclear because insufficient information was available regarding whether contamination across groups was likely and conclusive information on this from the authors could not be obtained (Chen 2000 RCT Taiwan; Dias 2008 RCT India; Dybdahl 2001 RCT Bosnia; Gavrilova 2009 RCT Russia; Hirani 2010 CRCT Pakistan; Li 1989 RCT China). We assessed an additional six studies as being at high risk of contamination (Araya 2003 RCT Chile; Berger2009 CRCT SriLanka; Bolton 2007 RCT Uganda; Loughry 2006 CBA Palestin; Neuner 2008 NRCT Uganda; Sutcliffe2009RCT Thailand).

For a number of studies, it was not clear whether the primary outcome measures were reliable: in 11 studies, these measures were not validated in the study context; and we assessed an additional six studies as 'unclear' because insufficient information was available on the validity of the measures.

Other sources of bias that were detected included:

  • the control and intervention arms potentially delivering interventions that were too similar, as mentioned by the authors (Sutcliffe2009RCT Thailand);

  • high likelihood of confounding: for example, due to incentives being provided to patients (Brown 2009 CBA Rwanda), or a teetotal religious festival occurring between baseline and follow‐up that may have had a greater impact on alcohol consumption than the motivational interviewing intervention in Noknoy 2010 RCT Thailand.

Economic studies ‐ risk of bias assessment with the adapted CHEC list criteria

All studies had significant risks of bias (Table 2), although we considered no study at high risk of bias on more than seven of the 23 adapted CHEC list criteria. The risk of biases identified were potentially important for the interpretation of costing, such as not discounting costs (Araya 2003 RCT Chile; Jordans 2011), not including the appropriate costs or outcomes and not valuing some outcomes appropriately.

Open in table viewer
Table 2. Risk of bias economic studies ‐ CHEC list criteria

Study

Risk of bias issues

Araya 2003 RCT Chile

‐ time horizon < 1 year

‐ a societal perspective would have been more appropriate

‐ not all relevant costs reported

‐ not all relevant outcomes included (only ambulatory, not hospital)

‐ no discounting

Jordans 2010 C‐RCT Nepal

‐ no discounting

‐ no sensitivity analysis

‐ not all important variables listed

‐ no discussion of ethical/distributional issues

Zambori 2002 CBA Hungary

‐ the competing alternatives were not described

‐ time horizon at 1 year was not appropriate (needs to be longer)

‐ not all relevant outcomes assessed (e.g. effect of treatment on severity, number of healthcare visits to psychiatrist)

‐ outcomes not measured appropriately (self reporting meant low response; standard prices used may not reflect actual prices)

‐ outcomes not valued (only the short‐term outcome)

‐ no sensitivity analysis

‐ conclusions do not all follow from results

Effects of interventions

See: Summary of findings for the main comparison NSHW‐led psychological interventions compared with usual care in treating depression in adults in low‐ and middle‐income countries (RCTs); Summary of findings 2 Collaborative care model (NSHWs plus specialist) compared with usual care in treating common mental disorders in adults in low‐ and middle‐income countries (RCTs); Summary of findings 3 NSHWs compared with usual care for treating maternal depression (RCTs); Summary of findings 4 NSHWs compared with specialists in treating depression in adults in low‐ and middle‐income countries (CBAs); Summary of findings 5 NSHW‐led psychological interventions compared with usual care in treating adults with PTSD (NRCT); Summary of findings 6 NSHWs compared with usual care in improving dementia patients' and carers' outcomes in low‐ and middle‐income countries (RCTs); Summary of findings 7 NSHW‐led brief alcohol interventions compared with usual care for adults with alcohol‐use disorders (RCTs); Summary of findings 8 NSHWs/OPHRs compared with usual care in conducting interventions for children with post‐traumatic stress disorder and depression (RCTs)

This review covered a wide range of NSHWs delivering a wide range of healthcare interventions for a variety of MNS disorders. However, no MNS detection studies were found that reported patient outcomes. We grouped studies by MNS disorders as different interventions and roles of NSHWs will in particular differ between severe and common mental disorders. These broad groups have, in turn, been subdivided into types of interventions that made clinical sense to group together (e.g. studies on depression have been divided into those involving collaborative care, where NSHWs are only one aspect of a complex intervention, and those involving psychological interventions provided by just one type of NSHW). We have further grouped studies by study design, and according to their comparator group (usual care or specialist care).

We performed meta‐analyses for eight groupings covering common mental disorders, PTSD, dementia and alcohol abuse. All analyses include the primary outcomes specified for this review, and some secondary outcomes. Below are the meta‐analysis groupings that we have reported:

  1. NSHW‐led psychological interventions versus usual care in treating common mental disorders in adults (RCTs);

  2. collaborative care model (NSHWs plus specialist) versus usual care in treating common mental disorders (RCTs and cluster RCTs);

  3. NSHWs versus usual care in treating maternal depression (RCTs);

  4. NSHWs versus specialist care in treating common mental disorders (CBA studies);

  5. NSHWs versus usual care in delivering PTSD interventions to adults (RCTs);

  6. NSHWs versus usual care in improving dementia patients' and carers' outcomes (RCTs);

  7. NSHW‐led brief alcohol interventions versus usual care in delivering interventions to adults with alcohol‐use disorders (RCTs);

  8. NSHWs/OPHRs versus usual care in delivering interventions for children with PTSD and depression (RCTs).

We could not pool the remaining studies, as they were individual studies of different disorders (severe mental disorders, epilepsy, drug abuse and child mental disorders other than PTSD and depression). We reported the results of these studies narratively in the text and in Table 3.

Open in table viewer
Table 3. Outcomes of studies not assigned to meta‐analyses

Study, and outcomes measured and tools

Intervention data [no. of participants]

Control data

Measure of effect (95% CI)

P value

Authors' conclusions

Brown 2009 CBA Rwanda(depression in youth)

Mentoring programme by LHW

Usual care

‐

‐

‐

Severity of depression at 2 years (mean) measured using CID‐S

Mean

[no. of participants]

23.27

[347]

Mean

[no. of participants]

23.28

[345]

‐

0.99

Reduction in intervention group but not in control group (at baseline higher score in intervention group). However, the score indicates continuing levels of depression in both groups

Levels of marginalisation at 2 years (mean) measured using a non‐validated marginalisation scale

3.35

3.13

‐

‐

Improved scores in intervention group, which are no different to control group

Levels of grief at 2 years (mean) measured using a non‐validated 7 point grief scale

3.42

3.38

‐

‐

Baseline lower levels of grief in the control group. No change at the end of the intervention though grief increased in control group and remained stable in the intervention group

Li 1989 RCT China(epilepsy ‐ adults and children)

Village doctors

Psychiatrists

‐

‐

‐

Effective epilepsy control with phenobarbital after 3 months

No. seizures/month

[no. of participants]

12

[20]

No. seizures/month

[no. of participants]

11

[20]

‐

‐

‐

Total number of adverse events after 3 months

No. events

[no. of participants]

19

[20]

No. events

[no. of participants]

39

[20]

‐

‐

‐

Paranthaman2010CBAMalaysi(people with schizophrenia and their carers)

Medical assistants/nurses

Usual care

MD (95% CI)

Pvalue

Authors' conclusions

Carer burden (activities in daily living) (mean) at 6 months. Measured using the Family Burden Interview schedule

Mean (SD)

[no. of participants]

9.41 (3.99)

[54]

Mean (SD)

[no. of participants]

8.93 (4.47)

0.48 (‐1.11 to 2.07)

0.55

Mostly there are similar scores between control and intervention groups.

Carer assistance in daily living severity ‐ ADL at 6 months
measured using the Family Burden Interview Schedule

‐

‐

0.83 (‐0.94 to 2.60)

‐

‐

Re‐admission rates

No. (events)

[no. of participants]

3

[54]

No. (events)

[no. of participants]

5

[55]

‐

0.47

‐

Defaulting from follow‐up

No. (events)

[no. of participants]

6

[54]

No. (events)

[no. of participants]

14

[55]

‐

0.03

important improvement in follow‐up rate for intervention group

Shin 2009 RCT Vietnam(children with intellectual disabilities)

Teacher‐led portage programme (OPHRs)

Usual care

MD (95% CI)

P value

Authors' conclusions

Functional impairment (motor skills) at 6 months (similar at 12 months) measured using the Vineland Adaptive Behaviour Scales

Mean (SD)

[no. of participants]

47.6 (16.8)

[16]

Mean (SD)

[no. of participants]

49 (15.4)

[14]

‐1.40 (‐12.93 to 10.13)

0.81

No significant difference for any mental outcomes but some improvement for motor and personal care outcomes if looked at time x effect interaction)

Functional impairment (social skills) at 6 months (similar at 12 months) measured using the Vineland Adaptive Behaviour Scales

47.1 (15.5)

[16]

46.3 (18.3)

[14]

0.80 (‐11.51 to 13.11)

0.93

‐

Behavioural changes at 6 months (similar at 12 months) measured using the Vineland Adaptive Behaviour Scales

55.6 (10.5)

[16]

55.7 (10)

[14]

‐0.10 (‐7.44 to 7.24)

0.98

‐

Sutcliffe2009RCT Thailand(people with drug abuse disorder)

Peer educator‐led psychoeducation (LHWs)

Usual care (life skills training)

RR/MD (95% CI)

P value

Authors' conclusions

Methamphetamine use at 6 months (similar results at 3, 9 and 12 months)

No.

[no. of participants]

272

[442]

No.

[no. of participants]

267

[440]

RR 1.01 (0.91 to 1.13)

0.79

Randomised peer education, social network intervention and control (social skills training) are both associated with reductions in methamphetamine use and increases in condom use over 12 months among a sample of young Thai people

Recovery of depressive symptoms at 12 months (index patient) measured using CES‐D score

Mean (SD)

[no. of participants]

15.7 (9.7)

[209]

Mean (SD)

[no. of participants]

17.9 (9.3)

[206]

MD ‐2.20 (‐4.03 to ‐0.37)

‐

The effect was strongly observed amount intervention index participants compared with both control and network participants

Recovery of depressive symptoms at 12 months (index and network patient combined) measured using CES‐D score

[no. of participants]

[495]

[no. of participants]

[488]

MD ‐1.05 [‐3.20 to 1.11]

‐

Contrary to expectation, mea and in CES‐D score change did not substantially differ between intervention network participants and control network participants. Thus, there is no evidence that the differential intervention effect on depression diffuses to network members

Prevalence of depression at 12 months (index patient) measured using CES‐D score

Events (No.)

[no. of participants]

57

[209]

Events (No.)

[no. of participants]

70

[206]

RR 0.80 (0.60 to 1.07)

‐

‐

Prevalence of depression at 12 months (index and network patient combined) measured using CES‐D score

[no. of participants]

[495]

[no. of participants]

[488]

RR 0.88 (0.73 to 1.06)

‐

‐

Hirani 2010 CRCT Pakistan(adults with depression, economic skills building intervention arm)

NSHW‐led economics skill building

n = 9

Usual care

n = 8

SMD (95% CI)

‐

Comment: these are presented as SMDs (calculated in RevMan, to compare with other SMDs in comparison 1.6 and 1.7)

Severity of depressive symptoms measured using Becks Depression Inventory II

Mean (SD)

20.1 (11.3)

Mean (SD)

27.63 (9.1)

SMD ‐0.69 (‐1.73 to 0.35)

‐

This study documents improved self efficacy and employment for women enrolled in economic skill‐building compared with general counselling and to control.

Functional impairment measured using the General Self‐Efficacy scale

28.7 (6.2)

21.63 (3.8)

SMD ‐1.29 (‐2.41 to ‐0.16)

‐

‐

CES‐D: Center for Epidemiological Studies Depression scale; CID‐S: Composite International Diagnostic‐Screener; CI: confidence interval; LHW: lay health workers; MD: mean difference; No.: number; OPHR: other professionals with health roles; RR: risk ratio; SD: standard deviation; SMD: standardised mean difference.

Comparison 1. Non‐specialist health workers‐led psychological interventions versus usual care in treating common mental disorders in adults (RCTs)

Setting: we identified seven studies from four countries: China (two studies) (Chen 2000 RCT Taiwan; Tiwari 2010 RCT China), Jamaica (one study) (Baker‐H 2005 CRCT Jamaica), Pakistan (three studies) (Ali 2003 RCT Pakistan; Hirani 2010 CRCT Pakistan; Rahman 2008 CRCT Pakistan), and Uganda (one study) (Bolton 2003 C‐RCT Uganda). Interventions were delivered in urban settings (Ali 2003 RCT Pakistan; Baker‐H 2005 CRCT Jamaica; Chen 2000 RCT Taiwan; Hirani 2010 CRCT Pakistan), rural settings (Bolton 2003 C‐RCT Uganda; Rahman 2008 CRCT Pakistan), and both (Tiwari 2010 RCT China).

Participants: participants were mostly from deprived backgrounds, though those in Ali (2003) were lower middle class and those in Chen (2000) were split equally between high‐, middle‐ and low‐income groups. Six studies included only women with depression (Ali 2003 RCT Pakistan; Hirani 2010 CRCT Pakistan; Tiwari 2010 RCT China), or perinatal depression (Baker‐H 2005 CRCT Jamaica; Rahman 2008 CRCT Pakistan). Studies including women tended to exclude adult women over the age of 50 years.

Intervention:NSHWs: there were four LHW‐led interventions (Ali 2003 RCT Pakistan; Baker‐H 2005 CRCT Jamaica; Bolton 2003 C‐RCT Uganda; Hirani 2010 CRCT Pakistan). The LHWs in these studies all had primary or no education, and some had high school or further education (Bolton 2003 C‐RCT Uganda; Rahman 2008 CRCT Pakistan). The group also includes one nurse‐led (Chen 2000 RCT Taiwan), and one social worker‐led (Tiwari 2010 RCT China), intervention. Most of the NSHWs were women, though Bolton had sex‐specific health workers for sex‐specific groups. In two studies, the NSHWs were employed by the government (Baker‐H 2005 CRCT Jamaica; Rahman 2008 CRCT Pakistan), and the others were salaried or volunteers within NGOs.

Training duration and intensity very varied from three days (Hirani 2010 CRCT Pakistan), to four weeks (Baker‐H 2005 CRCT Jamaica). Though information was often incomplete, most studies that reported the content of the training had a mixture of didactic and practical training.

Supervision was highly varied in terms of organisation and intensity from ad‐hoc checking (Ali 2003 RCT Pakistan; Tiwari 2010 RCT China), to structured meetings every two weeks (Baker‐H 2005 CRCT Jamaica). All training and supervision was done by the principal investigators or specialists (psychiatrists and psychologists), or both.

Description of interventions: LHWs provided psychological interventions: CBT‐like problem solving (Ali 2003 RCT Pakistan; Rahman 2008 CRCT Pakistan), and group interpersonal therapy (G‐IPT) (Bolton 2003 C‐RCT Uganda). LHWs also provided general counselling and economic skills building in one study (Hirani 2010 CRCT Pakistan). In two trials, non‐medical professionals delivered psychosocial counselling and problem solving (Chen 2000 RCT Taiwan; Tiwari 2010 RCT China). Interventions were delivered in community centres or groups (Baker‐H 2005 CRCT Jamaica; Bolton 2003 C‐RCT Uganda; Hirani 2010 CRCT Pakistan; Tiwari 2010 RCT China), in healthcare settings (Chen 2000 RCT Taiwan), and in homes (Ali 2003 RCT Pakistan; Rahman 2008 CRCT Pakistan).

Interventions varied in duration (30 to 120 minutes), in frequency (weekly to monthly, often with increasing intervals between sessions, e.g. Rahman 2008 CRCT Pakistan), and in total time (one month (Chen 2000 RCT Taiwan) to one year (Baker‐H 2005 CRCT Jamaica)). Three interventions included manuals for training and for conducting the intervention (Baker‐H 2005 CRCT Jamaica; Bolton 2003 C‐RCT Uganda; Rahman 2008 CRCT Pakistan).

Comparison groups included usual care without the addition of a NSHW (Ali 2003 RCT Pakistan; Bolton 2003 C‐RCT Uganda; Hirani 2010 CRCT Pakistan), or usual care where the NSHW was already present but was not trained to deliver the intervention (Baker‐H 2005 CRCT Jamaica; Chen 2000 RCT Taiwan; Rahman 2008 CRCT Pakistan; Tiwari 2010 RCT China).

Results
1. Prevalence of depression

LHW‐led psychological interventions may reduce depression prevalence within six months (RR 0.30, 95% CI 0.14 to 0.64, 3 studies, 1082 participants) but this evidence was of low quality due to heterogeneity (I2 = 81%; P value = 0.005) and selection bias (summary of findings Table for the main comparison) (Bolton 2003 C‐RCT Uganda; Chen 2000 RCT Taiwan; Rahman 2008 CRCT Pakistan). ITT analyses (looking at the four possible scenarios where re‐assigned participants are either assigned with improved outcomes or not) showed that these results varied from RR 0.20 (95% CI 0.09 to 0.45) to RR 1.11 (95% CI 0.56 to 2.21) indicating uncertainty of this result. Chen (2000) and Bolton (2003) varied widely through these four scenarios from favouring NSHW to favouring usual care, probably because of their relatively small sample size and large dropout rate. Rahman (2008) was least susceptible to change in figures, indicating possibly more reliable results (Figure 4).

Figure 4
Forest plot of comparison: 1 NSHW‐led psychological interventions versus usual care in treating CMDs in adults (RCTs), outcome: 1.1 Prevalence of depression (adults) (completers).

Forest plot of comparison: 1 NSHW‐led psychological interventions versus usual care in treating CMDs in adults (RCTs), outcome: 1.1 Prevalence of depression (adults) (completers).

2. Severity of common mental disorder symptoms (including anxiety and depression)

Seven studies reported severity of common mental disorder symptoms (including anxiety and depression). LHW‐led psychological interventions (Ali 2003 RCT Pakistan; Bolton 2003 C‐RCT Uganda; Hirani 2010 CRCT Pakistan; Rahman 2008 CRCT Pakistan), were pooled with nurse and social worker‐led interventions (Chen 2000 RCT Taiwan; Tiwari 2010 RCT China). It is uncertain whether these interventions lead to appreciable clinical benefit in common mental disorder symptom severity at six months post‐intervention, because despite an apparent clinical appreciable benefit (SMD ‐0.75, 95% CI ‐1.29 to ‐0.21, 1470 participants), the evidence was of very low quality due to high heterogeneity (I2 = 94%; P value < 0.00001) and selection bias. (Note that a small clinically appreciable benefit was set at SMD < 0.2, and a moderate benefit at SMD of 0.5 to 0.8) (Cohen 1988) (Table 4). One study, Bolton 2003, was an outlier (possibly because their LHWs performed single‐sex group interventions). When this study was excluded the heterogeneity reduced and suggested LHWs may have a clinically appreciable benefit (SMD ‐0.42, 95% CI ‐0.53 to ‐0.30, low‐quality evidence).

Open in table viewer
Table 4. SoF 1: NSHW‐led psychological interventions compared with usual care in treating common mental disorders in adults in low‐ and middle‐income countries (RCTs)

What are the effects of NSHW‐led psychological interventions for treating common mental disorders in adults in low‐ and middle‐ income countries? (additional outcomes to comparison 1)

Patient or population: Adults with common mental disorders (depression or anxiety, or both)
Settings: Low‐ and middle‐income countries (China, Jamaica, Pakistan, Taiwan, Uganda)
Intervention: NSHWs conducting psychological interventions
Comparison: Usual care

Outcomes

Illustrative comparative risks* (95% CI)

Effect estimate
(95% CI)

No of Participants
(studies)

Quality of the evidence
(GRADE)

Comments

Assumed risk

Corresponding risk

Usual care

NSHWs

Severity of CMD symptoms ‐ all interventions

short term (0‐6 months)
measured using various depression rating scales1

‐

The mean severity of CMD symptoms ‐ NSHW interventions short term was
0.75 standard deviations lower
(1.29 to 0.21 lower)

SMD ‐0.75 (‐1.29 to ‐0.21)

1470
(6 studies)

⊕⊝⊝⊝
very low2,3,4

‐

Severity of CMD symptoms ‐ all interventions

medium term (12 months)
measured using various CMD rating scales5

‐

The mean severity of CMD symptoms ‐ NSHW interventions medium term was
0.47 standard deviations lower
(0.60 to 0.34 lower)

SMD ‐0.47 (‐0.60 to ‐0.34)

923

(2 studies)

⊕⊕⊕⊝
moderate6,7

‐

Functional impairment/disability in adults with CMD ‐ NSHW interventions

short term (2‐6 months)
measured using various functional impairment scales8

‐

The mean functional impairment of adults with CMD ‐ NSHW interventions short term was
0.33 standard deviations lower

(0.80 lower to 0.13 higher)

SMD ‐0.33 (‐0.80 to 0.13)

1243
(4 studies)

⊕⊝⊝⊝
very low9,10,11

‐

Functional impairment/disability in depression/CMD (adults) ‐ NSHW interventions

medium term (2‐6 months)
measured using the Global Assessment of Functioning scale

‐

The mean functional impairment of adults with CMD ‐ NSHW interventions medium term was
0.56 standard deviations lower
(0.70 to 0.42 lower)

SMD ‐0.56 (‐0.70 to ‐0.42)

798
(1 study)

⊕⊕⊕⊝
moderate12

‐

*The basis for the assumed risk is the mean control group risk across studies for pooled results and the control group risk for single studies. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).
CMD: common mental disorders; CI: confidence interval; NSHW: non‐specialist health worker; SMD: standardised mean difference.

GRADE Working Group grades of evidence
High quality: Further research is very unlikely to change our confidence in the estimate of effect.
Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.
Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.
Very low quality: We are very uncertain about the estimate.

1Chen 2000 RCT Taiwan, Hirani 2010 CRCT Pakistan and Tiwari 2010 RCT China used the Beck's Depression Inventory; Rahman 2008 CRCT Pakistan used the Hamilton Depression Rating Scale; Ali 2003 RCT Pakistan used the AKUADS; Bolton 2003 C‐RCT Uganda used the Hopkins Symptom Checklist (HSCL).

2 Serious study limitations: Two of the six trials in this analysis were judged at high risk of bias and one was unclear about possible risk of bias. Chen 2000 RCT Taiwan had unclear sequence generation and allocation concealment, all were self reported outcomes, there was possible contamination and there was a high dropout rate after randomisation, with no analysis of dropout versus non‐dropout differences; Hirani 2010 CRCT Pakistan was unclear regarding allocation concealment, there was no blinding of outcome assessment (self reported outcomes), it was unclear if baseline measures and characteristics were similar in both groups; and the report provided no information on dropouts. Bolton 2003 C‐RCT Uganda was not clear about allocation concealment and quasi randomisation of individuals within clusters (though randomisation of clusters) may have introduced bias. The three trials contributed 45% of the weight in the pooled analysis. Downgraded by 1.

3Serious inconsistency: I2 statistic = 94%. However, the inconsistency related to the magnitude of benefit favouring NSHW interventions rather than in the direction of effect. Downgraded by 1.
4 Serious imprecision: The 95% CI for the pooled estimates indicates appreciable benefit and non‐appreciable benefit for collaborative care (appreciable SMD = ≥ 0.5; non‐appreciable benefit ≤ 0.2). Downgraded by 1.

5Baker‐H 2005 CRCT Jamaica used the CED‐S; Rahman 2008 CRCT Pakistan used the Hamilton Depression Rating Scale.

6 No serious study limitations: The CES‐D used in Baker‐H 2005 CRCT Jamaica is not validated in the Jamaican population and is not a measure of clinical depression but just identifies depressive symptoms. Most women were not likely to have been depressed. Also in this study, there were unadjusted differences in baseline characteristics. However, this study contributed only 14% weight to the pooled results and removal of this study did not alter the direction or precision of the effect estimate. Not downgraded.

7 Serious indirectness: The two trials included were the only two of the six trials that compared this intervention that had data over the medium term, and only one used a validated outcome measure. Downgraded by 1.

8Bolton 2003 C‐RCT Uganda used a sex‐specific Functional Impairment Questionnaire; Rahman 2008 CRCT Pakistan used the Global Assessment of Functioning (GAF) scale, Hirani 2010 CRCT Pakistan used the General Self‐efficacy Scale; Tiwari 2010 RCT China used the Short Form‐ 12 (SF‐12) (mental and physical components).

9 Serious study limitations: Two of the four studies were at risk of bias. Bolton 2003 C‐RCT Uganda was not clear about allocation concealment and quasi‐randomisation of individuals within clusters (though randomisation of clusters) may have introduced bias. Hirani 2010 CRCT Pakistan was unclear regarding allocation concealment, there was no blinding of outcome assessment (self reported outcomes), it was unclear if baseline measures and characteristics were similar in both groups; and the report provided no information on dropouts. Downgraded by 1.

10 Very serious inconsistency: I2 statistic = 90%. The inconsistency related to the direction of effect between interventions and was unexplained. Downgraded by 2.

11 Serious imprecision: the 95% CI of the pooled estimate showed appreciable benefit for interventions (appreciable SMD = 0.5) and non‐appreciable benefit for control. Downgraded by 1.

12 Serious imprecision: the 95% CI of the pooled estimate shows non‐appreciable benefit for psychological interventions and usual care (appreciable SMD = 0.5). However, the data for this outcome were from only one trial. Downgraded by 1.

Two studies suggested that there is probably a reduction in depression symptom severity at eight to 12 months post intervention (SMD ‐0.47, 95% CI ‐0.60 to ‐0.34, moderate‐quality evidence) (Figure 5) (Baker‐H 2005 CRCT Jamaica; Rahman 2008 CRCT Pakistan).

Figure 5
Forest plot of comparison: 1 NSHW‐led psychological interventions versus usual care in treating common mental disorders in adults (RCTs), outcome: 1.6 Severity of common mental disorder symptoms (includes anxiety and depression).

Forest plot of comparison: 1 NSHW‐led psychological interventions versus usual care in treating common mental disorders in adults (RCTs), outcome: 1.6 Severity of common mental disorder symptoms (includes anxiety and depression).

One CBA study, Brown 2009 CBA Rwanda's intervention of adult mentoring of youths who were heads of households, showed no difference in depression symptom severity at two years (see Table 2). Two CBA studies performed in rural post‐conflict areas suggested it is uncertain whether LHW‐ and OPHR‐led interventions decrease the severity of common mental disorder symptoms (SMD ‐0.32, 95% CI ‐0.60 to ‐0.04, very‐low‐quality evidence) (Bass 2012 CBA Indonesia; Scholte 2011 CBA Rwanda). See Characteristics of included studies and Table 5 for more details.

Open in table viewer
Table 5. SoF: NSHW‐led interventions compared with usual care in treating common mental disorders in adults in low‐ and middle‐income countries (CBAs)

What are the effects of NSHWs conducting single interventions compared with usual care in treating common mental disorders for mental health care in low‐ and middle‐income countries? (additional CBA outcomes to comparison 1)

Patient or population: Adults with CMDs (such as depression and anxiety)
Settings: Low‐ and middle‐income countries (Indonesia, Rwanda)
Intervention: NSHWs conducting single interventions
Comparison: Usual care

Outcomes

Illustrative comparative risks* (95% CI)

Relative effect
(95% CI)

No of Participants
(studies)

Quality of the evidence
(GRADE)

Comments

Assumed risk

Corresponding risk

Usual care

NSHWs

Severity of common mental disorders ‐ short term (within 6 months)
measured using CMD rating scales1

‐

The mean severity of CMDs ‐ short term (within 6 months post intervention) in the intervention groups was
0.08 standard deviations lower
(0.25 lower to 0.09 higher)

SMD ‐0.08 (‐0.25 to 0.09)

533
(2 studies)

⊕⊝⊝⊝
very low2,3

‐

Severity of CMDs ‐ medium term (8 months)
measured using SRQ‐20

‐

The mean severity of CMDs ‐ medium term (6 months to 1 year post intervention) in the intervention groups was
0.32 standard deviations lower
(0.6 to 0.04 lower)

SMD ‐0.32

(‐0.6 to ‐0.04)

200
(1 study)

⊕⊝⊝⊝
very low4,5

‐

Functional impairment ‐ male short term (1 month)
measured using WHODAS (adapted) 11 items

‐

The mean functional impairment ‐ male short term (within 6 months of intervention) in the intervention groups was
0.32 standard deviations lower
(0.65 lower to 0.02 higher)

SMD ‐0.32 (‐0.65 to 0.02)

141
(1 study)

⊕⊝⊝⊝
very low6,7

‐

Functional impairment ‐ female short term (1 month)
measured using WHODAS (adapted) 11 items

‐

The mean functional impairment ‐ female short term (within 6 months of intervention) in the intervention groups was
0.34 standard deviations lower
(0.63 to 0.06 lower)

SMD ‐0.34 (‐0.63 to ‐0.06)

192
(1 study)

⊕⊝⊝⊝
very low6

‐

*The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).
CI: confidence interval; CMD: common mental disorders; NSHW: non‐specialist health worker; SMD: standardised mean difference; SRQ: Self Reporting Questionnaire; WHODAS: World Health Organization Disability Assessment Scale.

GRADE Working Group grades of evidence
High quality: Further research is very unlikely to change our confidence in the estimate of effect.
Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.
Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.
Very low quality: We are very uncertain about the estimate.

1Bass 2012 CBA Indonesia: Hopkins Symptom Checklist 25; Scholte 2011 CBA Rwanda: SRQ‐20.
2 Very serious risk of bias: Bass 2012 CBA Indonesia: controlled before‐and‐after study so non‐random and not concealed. Also differences in baseline outcomes for girls, and doubt about reliability of primary outcomes as tool not properly validated. Scholte 2011 CBA Rwanda: controlled before‐and‐after study so non‐randomised and no concealment. Also unclear risk for incomplete outcome data, there are baseline differences in outcomes and in characteristics not all adjusted for, and high rate of loss to follow‐up with no analysis of group lost to follow‐up. Downgraded by 2.
3 No imprecision: Non‐appreciable benefit for either intervention or control group.
4 No explanation was provided.
5 No imprecision: Appreciable and non‐appreciable benefit for intervention.
6 Very serious risk of bias: Bass 2012 CBA Indonesia: controlled before‐and‐after study so non‐random and not concealed. Also differences in baseline outcomes for girls, and doubt about reliability of primary outcomes as tool not properly validated. Downgraded by 2.
7 Serious imprecision: Appreciable benefit for intervention and non‐appreciable benefit for usual care. Downgrade by 1.

3. Functional impairment of adults with common mental disorders

Four studies assessed functional impairment of which three were LHW‐led interventions (Bolton 2003 C‐RCT Uganda; Hirani 2010 CRCT Pakistan; Rahman 2008 CRCT Pakistan), and one was social worker‐led (Tiwari 2010 RCT China). It is uncertain whether these interventions lead to a reduction in functional impairment within zero to six months of interventions (SMD ‐0.33, 95% CI ‐0.80 to 0.13, 4 studies, 1243 participants, very‐low‐quality evidence due to very serious risk of bias, inconsistency and imprecision). Findings from a CBA study assessing a similar LHW intervention suggested that it is uncertain whether this reduces functional impairment (Bass 2012 CBA Indonesia).

However, LHW‐led interventions probably reduce functional impairment of patients with common mental disorders in the medium term (12 months) (SMD ‐0.56, 95% CI ‐0.70 to ‐0.42, 1 study, 798 participants, moderate‐quality evidence). The improvement at 12 but not six months may suggest that it takes longer for functional recovery.

Comparison 2. Collaborative care model (non‐specialist health workers plus specialist) versus usual care in treating common mental disorders (including depression and anxiety) (RCTs)

Setting: we identified five studies from Chile (Araya 2003 RCT Chile; Fritsch 2007 RCT Chile; Rojas 2007 RCT Chile), India (Patel 2010 C‐RCT India), and Kenya (Jenkins 2012 C‐RCT Kenya). Both Patel (2010) and Jenkins (2012) were interventions located in a combination of urban and rural settings. The Chilean trials were conducted in deprived urban areas. All trials were conducted in government‐funded PHC facilities. The Patel trial presented combined and separate results for government‐ and privately funded facilities.

Participants: In all studies, participants were adults (over 16 (Jenkins 2012 C‐RCT Kenya) and over 17 (Patel 2010 C‐RCT India) years; over 18 years for other studies) with common mental disorders (including anxiety or depression, or both) or just depression. Araya (2003), Fritsch (2007) and Rojas (2007) included only women. Most participants were of low socioeconomic status. 

Interventions:Types of NSHWs: these collaborative care models involved existing PHC staff, including private and government PHC doctors (Araya 2003 RCT Chile; Fritsch 2007 RCT Chile; Jenkins 2012 C‐RCT Kenya; Patel 2010 C‐RCT India; Rojas 2007 RCT Chile), non‐medical professional staff (nurses, social workers, midwives) (Araya 2003 RCT Chile; Jenkins 2012 C‐RCT Kenya; Rojas 2007 RCT Chile), and LHWs (Fritsch 2007 RCT Chile; Patel 2010 C‐RCT India; Rojas 2007 RCT Chile).

Training and supervision of NSHWs: doctors received four to six hours of training in all studies (except for Jenkins (2012) where it was not specified how many hours frontline staff received). LHWs training varied from two hours to two months. Those with longer training (Patel 2010) were expected to deliver a wider range of services. In all studies, NSHWs received some supervision (weekly to monthly/ad hoc) though those in Jenkins (2012) received no supervision and had poor medication supply.

Description of interventions: collaborative care models involved a multidisciplinary team consisting of one or several NSHWs and specialists. Doctors and nurses in Jenkins (2012) diagnosed patients, provided medical treatment and follow‐up/referral as per the existing government health delivery model. Araya (2003), Rojas (2007) and Patel (2010) used a stepped care intervention where doctors prescribed antidepressants and provided usual physical care and referred if there was high suicide risk. Jenkins' (2012) PHCs had poor medication supply. LHWs and non‐medical professionals provided several services such as psychoeducation, medication adherence/follow‐up (in person or by telephone) and IPT (Araya 2003 RCT Chile; Fritsch 2007 RCT Chile; Patel 2010 C‐RCT India; Rojas 2007 RCT Chile). The intensity of these interventions varied from ad hoc (Fritsch 2007 RCT Chile; Jenkins 2012 C‐RCT Kenya; Patel 2010 C‐RCT India) to eight weekly psychoeducation sessions (Rojas 2007 RCT Chile). Comparison groups were the same settings where NSHWs did not receive training/supervision (Araya 2003 RCT Chile; Fritsch 2007 RCT Chile; Jenkins 2012 C‐RCT Kenya; Rojas 2007 RCT Chile), and same settings without the addition of a lay counsellor, and where current staff received a training manual (enhanced usual care) (Patel 2010 C‐RCT India).

Results

The primary analysis performed was of prevalence, severity and functional impairment of common mental disorders. Where trials only reported depression scores, these were combined within the common mental disorder analysis (including both anxiety and depression). Data reported at six months post intervention (if available) were chosen to represent the medium‐term time point, otherwise an earlier time point (zero to five months) was combined.

1. Prevalence of common mental disorders

Three studies reported prevalence of CMDs (CMD scores: Patel 2010 C‐RCT India; depression scores: Araya 2003 RCT Chile; Patel 2010 C‐RCT India; Rojas 2007 RCT Chile). Across all facilities (private and government), the use of NSHWs may reduce the prevalence of CMDs within two to six months (RR 0.63, 95% CI 0.44 to 0.90, 2380 participants, low quality of evidence due to serious study limitations and inconsistency (I2 = 79%; P value = 0.001) (Figure 6; summary of findings Table 2). For government facilities only (where data from Patel 2010 C‐RCT India was substituted for just the government health facilities data), the effect size was similar (RR 0.57, 95% CI 0.42 to 0.78, 1528 participants, low‐quality evidence). There is probably no reduction in prevalence at 12 months in 'all facilities' (RR 0.95, 95% CI 0.68 to 1.33, 1 study, 2009 participants, moderate‐quality evidence due to imprecision) or in government facilities alone (RR 0.72, 95% CI 0.39 to 1.34, 1 study, 1104 participants; low‐quality evidence due to very serious imprecision).

Figure 6
Forest plot of comparison: 2 Collaborative care model (NSHWs plus specialist) versus usual care in treating common mental disorders (CMD) (RCTs), outcome: 2.1 Prevalence of common mental disorders (CMDs ‐ includes anxiety and depression) (completers combined) all facilities and in public and private facilities.

Forest plot of comparison: 2 Collaborative care model (NSHWs plus specialist) versus usual care in treating common mental disorders (CMD) (RCTs), outcome: 2.1 Prevalence of common mental disorders (CMDs ‐ includes anxiety and depression) (completers combined) all facilities and in public and private facilities.

We conducted a sensitivity analysis to analyse CMD scores and depression scores separately. This revealed very similar results (depression: RR 0.61, 95% CI 0.40 to 0.94, 3 studies, 1092 participants, low‐quality evidence; CMD: RR 0.80, 95% CI 0.61 to 1.05, 1 study, 1961 participants, moderate‐quality evidence). 

2. Severity of common mental disorders

Severity of CMDs was measured in five studies (CMD scores: Jenkins 2012 C‐RCT Kenya; Patel 2010 C‐RCT India, depression scores: Araya 2003 RCT Chile; Fritsch 2007 RCT Chile; Patel 2010 C‐RCT India; Rojas 2007 RCT Chile). It is uncertain whether collaborative care reduces the severity of CMDs in the short term (two to six months) despite a statistically significant small benefit (SMD ‐0.31, 95% CI ‐0.56 to ‐0.06, 5 studies, 3604 participants, very‐low‐quality evidence due to serious study limitations, serious inconsistency (I2 = 91%; P value < 0.00001), and serious indirectness) (note that a small clinically appreciable benefit was set at SMD < 0.2) (Cohen 1988) (Table 6). Government facilities analysis shows a similar magnitude of effect (SMD ‐0.32, 95% CI ‐0.58 to ‐0.07, very‐low‐quality evidence). There is probably no medium term (12 months) reduction in CMD symptom severity (SMD ‐0.03, 95% CI ‐0.12 to 0.06, 1 study, 1905 participants, moderate‐quality evidence) (Figure 7), possibly due to recurrence of depression at this point in time.

Figure 7
Forest plot of comparison: 2 Collaborative care model (NSHWs plus specialist) versus usual care in treating common mental disorders (RCTs), outcome: 2.2 Severity of symptoms of common mental disorders (completers combined) in all facilities and in public and private facilities.

Forest plot of comparison: 2 Collaborative care model (NSHWs plus specialist) versus usual care in treating common mental disorders (RCTs), outcome: 2.2 Severity of symptoms of common mental disorders (completers combined) in all facilities and in public and private facilities.

Open in table viewer
Table 6. SoF 2: Collaborative care model (NSHWs plus specialist) compared with usual care in treating common mental disorders in adults in low‐ and middle‐income countries (RCTs)

What are the effects of a collaborative care model (NSHW plus specialist supervision) for mental health care in adults with common mental disorders low‐ and middle‐income countries? (additional outcomes to comparison 2)

Patient or population: Adults (≥ 18 years) with common mental disorders (includes anxiety or depression, or both)
Settings: Middle‐income countries (Chile, India)
Intervention: Collaborative care model (NSHW plus specialist supervision)

Comparison: Enhanced usual care

Outcomes

Illustrative comparative risks* (95% CI)

Effect estimate
(95% CI)

No of Participants
(studies)

Quality of the evidence
(GRADE)

Comments

Assumed risk

Corresponding risk

usual care

Collaborative care model

Prevalence of CMDs medium term (12 months)
measured using CIS‐R generated ICD‐10 diagnosis for CMD

See comment

See comment

RR 0.95
(0.68 to 1.33)

2009
(1 study)

⊕⊕⊝⊝
low1

Patel 2010 C‐RCT India did not reveal significant differences in the prevalence of depression with both interventions in public or private care facilities.

Severity of symptoms in CMD

short term (2‐6 months)
measured using various rating scales2

‐

The mean severity of symptoms in CMD with collaborative care was
0.31 standard deviations lower
(0.56 to 0.06 lower)

SMD ‐0.31 (‐0.56 to ‐0.06)

3604
(5 studies)

⊕⊝⊝⊝
very low3,4,5

‐

Severity of symptoms in CMD

medium term (12 months)
measured using CIS‐R rating scale

‐

The mean severity of symptoms in CMD with collaborative care was
0.03 standard deviations lower
(0.12 lower to 0.06 higher)

SMD ‐0.03 (‐0.12 to 0.06)

1905
(1 study)

⊕⊕⊕⊝
moderate6

‐

Functional impairment/disability in CMD short term (2‐6 months)
measured using various functional disability scores7

‐

The mean functional impairment/disability in CMD with collaborative care was
0.22 standard deviations lower
(0.44 to 0.01 lower)

SMD ‐0.22 (‐0.44 to ‐0.01)

3604
(5 studies)

⊕⊝⊝⊝
very low5,8,9

‐

Functional impairment/disability in CMD medium term (12 months)
measured using WHODAS II scores

‐

The mean functional impairment/disability in CMD with collaborative care was
0.02 standard deviations lower
(0.11 lower to 0.07 higher)

SMD ‐0.02 (‐0.11 to 0.07)

1905

(1 study)

⊕⊕⊕⊝
moderate6

‐

*The basis for the assumed risk is the mean control group risk across studies for pooled results and the control group risk for single studies. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).
CI: confidence interval; CIS‐D: Composite International Diagnostic‐Screener; CMD: common mental disorders; ICD: International Classification of Diseases; NSHW: non‐specialist health worker; RR: risk ratio; SMD: standardised mean difference; WHODAS: World Health Organization Disability Assessment Scale.

GRADE Working Group grades of evidence
High quality: Further research is very unlikely to change our confidence in the estimate of effect.
Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.
Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.
Very low quality: We are very uncertain about the estimate.

1 Very serious imprecision: The 95% CI for the pooled estimates indicates appreciable benefit for collaborative care and appreciable benefit for usual care. Downgraded by 2.
2Jenkins 2012 C‐RCT Kenya used the General Health Questionnaire (GHQ)‐12. Patel 2010 C‐RCT India used CIS‐R to generate ICD‐10 depression diagnoses; Fritsch 2007 RCT Chile and Araya 2003 RCT Chile used the Hamilton Depression Rating Scale (HDRS); Rojas 2007 RCT Chile used the: Edinburgh Postnatal Depression Scale (EPDS).

3 Serious study limitations: In Araya 2003 RCT Chile and possibly in Fritsch 2007 RCT Chile general practitioners (GPs) did both interventions, so there was a high risk of contamination; this would have reduced the potential benefits with collaborative care in two of the four trials in the meta‐analysis. Downgraded by 1.
4 Serious inconsistency. The I2 statistic = 91% with Araya 2003 RCT Chile clearly an outlier, contributing to this unexplained inconsistency. However, the inconsistency related to the magnitude of benefit favouring collaborative care rather than in the direction of effect. Downgraded by 1.
5 Serious indirectness: Jenkins 2012 C‐RCT Kenya used the GHQ‐12 to grade severity of symptoms; the GHQ is a screening instrument that is validated to screen for CMDs; its use to rate the severity of depression is less reliable). Downgraded by 1.
6 Serious imprecision: The 95% CI for the pooled estimates indicates no appreciable benefit for collaborative care (< 0.2) and non‐appreciable benefit for usual care. The data come from one study (Patel 2010 C‐RCT India), and therefore imprecise. Downgraded by 1.

7Jenkins 2012 C‐RCT Kenya used WHODAS II long version (36 items); Patel 2010 C‐RCT India used the WHODAS II short version (12 items); Araya 2003 RCT Chile; Fritsch 2007 RCT Chile; and Rojas 2007 RCT Chile used SF‐36 social functioning component.
8 Serious study limitations: In Araya 2003 RCT Chile and probably Fritsch 2007 RCT Chile, GPs did both intervention and control interventions so there was a high risk of contamination. Downgraded by 1.

9 Serious inconsistency. The I2 statistic = 87% with Araya 2003 RCT Chile clearly an outlier, contributing to this unexplained inconsistency. However, the inconsistency related to the magnitude of benefit favouring collaborative care rather than in the direction of effect. Downgraded by 1.

The Araya trial results were an outlier for this outcome, with a much larger effect size reported (although with the same direction of effect). This may be because it was the only trial measuring major depression (moderate to severe depression). Other trials included mild depression in their inclusion criteria. This would explain the larger effect size as there is strong evidence that baseline severity of depression is a predictor of the effectiveness of depression treatments (Kirsch 2008). In a sensitivity analysis in which Araya was excluded, the reduction in symptoms no longer showed appreciable benefit (SMD ‐0.10, 95% CI ‐0.17 to ‐0.03, 3394 participants, low‐quality evidence) and the results were consistent across studies (I2 = 0%; P value = 0.39).

We conducted a sensitivity analysis to analyse CMD scores and depression scores separately. CMD scores suggested collaborative care models probably do not result in a clinically appreciable reduction in the severity of CMDs in either the short term (two to six months) (SMD ‐0.07, 95% CI ‐0.15 to 0, 2 studies, 2889 participants, moderate‐quality evidence due to serious indirectness) or the medium term (one year). The short‐term findings are inconsistent with the above prevalence findings. Possible explanations may be that the tools used to assess severity, particularly General Health Questionnaire (GHQ)‐12 in Jenkins, may not be appropriate for assessing severity, and that the sample size is smaller in this comparison, thereby giving a less precise estimate. In addition, CMDs could include many milder symptoms of anxiety and depression whereas depression scales would identify patients with more moderate to severe symptoms. The effect of the intervention would be expected to have a greater impact on those with more symptoms (Kirsch 2008).

We could not examine the difference between outcomes for government and private facilities for the severity of CMDs due to limited data.

3. Functional impairment and disability in adults with common mental disorders

Five studies (CMD scores: Jenkins 2012 C‐RCT Kenya; Patel 2010 C‐RCT India; depression scores: Araya 2003 RCT Chile; Fritsch 2007 RCT Chile; Rojas 2007 RCT Chile) reported functional impairment and disability in adults with CMD. Collaborative care probably does not reduce functional impairment over 12 months (SMD ‐0.02, 95% CI 0.11 to 0.07, 1 study, moderate‐quality evidence).

It is uncertain whether collaborative care reduces functional impairment in CMDs at six months (SMD ‐0.22, 95% CI ‐0.44 to ‐0.01, very‐low‐quality evidence because of serious risk of bias, serious inconsistency (I2 = 87%; P value < 0.00001) and serious indirectness).

The Araya trial results were outliers for this outcome, with a much larger effect size reported (although with the same direction of effect). As above, this may because included patients had more severe symptoms and, therefore, more likely to respond to an intervention. In a sensitivity analysis in which Araya was excluded, there was no longer any appreciable clinical benefit for reducing functional impairment (SMD ‐0.05, 95% CI ‐0.12 to ‐0.02, 3394 participants) but the results were now consistent (I2 = 0%; P value = 0.40). At 12 months, there was no difference in functional impairment scores with collaborative or with usual care (SMD ‐0.02, 95% CI ‐0.12 to 0.15, 1 study, moderate‐quality evidence).

We conducted a sensitivity analysis to analyse CMD scores and depression scores separately. Depression scores were similar or no different but again showed very‐low‐quality evidence. CMD scores on their own suggested no reduction in functional impairment in people with CMDs at six months (SMD ‐0.03, 95% CI ‐0.1 to 0.04, 2889 participants, high‐quality evidence) or at 12 months (one study).

Patel's study was the only study to report disability days. This showed that, over 12 months, collaborative care probably reduces the number of days of no or reduced work in the last month by 4.43 days (MD ‐4.43 days, 95% CI ‐8.37 to ‐0.48, moderate‐quality evidence) in government facilities but seems to have no reduction in disability days in private facilities (MD 0.78 days, 95% CI ‐2.25 to 3.82).

4. Suicide attempts in adults with common mental disorders

Only one study reported suicide attempts in adults with CMDs (Patel 2010 C‐RCT India). There was no difference in suicide attempts for those diagnosed with CMDs at one year (RR 0.56, 95% CI 0.24 to 1.32, 1905 participants) and within two to six months. The quality of evidence was low due to very serious imprecision.

Comparison 3. Non‐specialist health workers versus usual care in treating maternal depression (RCTs)

This group of studies combined RCTs that were also included above as part of the 'NSHW‐led' and 'collaborative' intervention comparisons and that assessed perinatal depression outcomes.

Setting: we identified four studies, which were conducted in urban settings in Chile (Rojas 2007 RCT Chile), Jamaica (Baker‐H 2005 CRCT Jamaica), and Taiwan (Chen 2000 RCT Taiwan), and rural settings in Pakistan (Rahman 2008 CRCT Pakistan). 

Participants: the trials recruited mothers at different times from the third trimester of pregnancy (Rahman 2008 CRCT Pakistan), up to 13 months' postpartum (Baker‐H 2005 CRCT Jamaica). Participants in all of the trials were generally from lower socioeconomic backgrounds, except for Chen (2000) where there was an equal distribution of participants across all socioeconomic groups.

Interventions:NSHWs: these were mainly existing government employees or aides, including doctors, midwives and LHWs (Rojas 2007 RCT Chile), nurses (Chen 2000 RCT Taiwan), and LHWs (Baker‐H 2005 CRCT Jamaica; Rahman 2008 CRCT Pakistan). In Baker‐Henningham (2005), LHW training was much more intensive than in Rahman (2008) though in both studies LHWs also received refresher training. In Rojas (2007), the midwives only were given an eight‐hour training session (other cadres' training was not specified). In all of the trials, weekly to monthly supervision was provided, apart from Chen (2000), where this was not specified.

Description of interventions: interventions were delivered at home (Baker‐H 2005 CRCT Jamaica), in the community (Rahman 2008 CRCT Pakistan), in postnatal wards (Chen 2000 RCT Taiwan), and PHC clinics (Rojas 2007 RCT Chile). Interventions ranged from collaborative care (Rojas 2007 RCT Chile), to CBT‐like intervention (Rahman 2008 CRCT Pakistan), to general adapted counselling (Baker‐H 2005 CRCT Jamaica; Chen 2000 RCT Taiwan). They varied in intensity from four weeks (Chen 2000 RCT Taiwan), to weekly home visits over 12 months (Baker‐H 2005 CRCT Jamaica).

Comparison groups from all four studies included usual care (existing NSHWs without training).

Results
1. Severity of maternal depressive symptoms

There was high‐quality evidence that NSHW interventions improved the severity of perinatal depressive symptoms (SMD within three months: ‐0.50, 95% CI ‐0.63 to ‐0.36, 2 studies), and moderate‐quality evidence that collaborative interventions slightly improved perinatal depressive symptoms within two to six months (SMD ‐0.22, 95% CI ‐0.48 to 0.04, 1 study). LHW interventions may have slightly improved perinatal depressive symptoms at 12 months (SMD ‐0.41, 95% CI ‐0.76 to ‐0.06, 1 study, low‐quality evidence) (Table 7). A meta‐analysis including all four studies showed that these interventions may have slightly reduced the severity of perinatal depressive symptoms (SMD ‐0.42, 95% CI ‐0.58 to ‐0.26, low‐quality evidence due to very serious risk of bias). Results were similar if only the three short‐term studies were combined (SMD ‐0.42, 95% CI ‐0.65 to ‐0.20). The statistical heterogeneity was low (I2 = 29%; P value = 0.24) (Figure 8; summary of findings Table 3).

Figure 8
Forest plot of comparison: 3 NSHWs versus usual care in treating maternal depression (RCTs), outcome: 3.1 Severity of symptoms in treating maternal depression.

Forest plot of comparison: 3 NSHWs versus usual care in treating maternal depression (RCTs), outcome: 3.1 Severity of symptoms in treating maternal depression.

Open in table viewer
Table 7. SoF 3: NSHWs compared with usual care for treating maternal depression (RCTs)

What are the effects of NSHW‐led interventions for treating maternal depression in low‐ and middle‐income countries? (additional outcomes for comparison 3)

Patient or population: Adult women with maternal depression
Settings: Low‐ and middle‐income countries (Chile, Jamaica, Pakistan, Taiwan)
Intervention: NSHW‐led interventions
Comparison: Usual care

Outcomes

Illustrative comparative risks* (95% CI)

Estimate effect
(95% CI)

No of Participants
(studies)

Quality of the evidence
(GRADE)

Comments

Assumed risk

Corresponding risk

Usual care

NSHWs

Severity of symptoms of perinatal depression ‐ NSHW led‐psychological interventions

short term (0‐2 months)
measured using depression rating scales1

‐

The mean severity of symptoms of perinatal depression ‐ short term with NSHW‐led interventions was
0.5 standard deviations lower
(0.63 to 0.36 lower)

SMD ‐0.5 (‐0.63 to ‐0.36)

858
(2 studies)

⊕⊕⊕⊕
high2,3

Note that a small clinically appreciable benefit was set at SMD < 0.2, and a moderate benefit at SMD of 0.5 to 0.8 (Cohen 1988)

Severity of symptoms of perinatal depression ‐ NSHW led‐psychological interventions

medium term (12 months)
measured using a depression scale4

‐

The mean severity of symptoms of perinatal depression ‐ medium term ‐ with NSHW‐led interventions was
0.41 standard deviations lower
(0.76 to 0.06 lower)

SMD ‐0.41 (‐0.76 to ‐0.06)

125
(1 study)

⊕⊝⊝⊝
low5,6

‐

Severity of symptoms of perinatal depression ‐ collaborative care‐ short term (3 months)
depression rating scale ‐ EPDS

‐

The mean severity of symptoms of perinatal depression ‐ short term ‐ with collaborative care was
0.22 standard deviations lower
(0.48 lower to 0.04 higher)

SMD ‐0.22 (‐0.48 to 0.04)

230
(1 study)

⊕⊕⊕⊝
moderate7

‐

*The basis for the assumed risk is the mean control group risk across studies for pooled results and the control group risk for single studies. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).
CI: confidence interval; EPDS: Edinburgh Postnatal Depression Score; NSHW: non‐specialist health worker; SMD: standardised mean difference.

GRADE Working Group grades of evidence
High quality: Further research is very unlikely to change our confidence in the estimate of effect.
Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.
Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.
Very low quality: We are very uncertain about the estimate.

1Chen 2000 RCT Taiwan: Taiwanese Beck Depression Inventory; Rahman 2008 CRCT Pakistan: Hamilton Depression Rating Scale (HDRS).
2 No serious study limitations: Chen 2000 RCT Taiwan was unclear for sequence generation and allocation concealment, all were self reported outcomes, there was possible contamination and there was a high dropout rate after randomisation, with no analysis of differences between dropout versus non‐dropouts differences. However, Rahman 2008 CRCT Pakistan had no serious study limitations and contributed most of the weight to the pooled analysis. Removal of data from Chen 2000 RCT Taiwan did not appreciably change effect estimates. Not downgraded.
3 No serious imprecision: appreciable benefit seen at SMD = 0.2 for Rahman 2008 CRCT Pakistan with HDRS and 0.5 for Chen 2000 RCT Taiwan with the BDI. The 95% CI includes appreciable benefit for NSHW‐led interventions. Not downgraded.

4Baker‐H 2005 CRCT Jamaica: CES‐D.
5 Serious study limitations: Baker‐H 2005 CRCT Jamaica had unadjusted differences in baseline characteristics. Downgraded by 1.

6 Serious imprecision: The 95% CI of the effect estimate indicated appreciable and non‐appreciable benefit with NSHW‐led interventions, and the sample size was small. Downgraded by 1.
7 Serious imprecision: The 95% CI of the effect estimate indicated appreciable and non‐appreciable benefit with collaborative care, and the sample size was small. Downgraded by 1.

Comparison 4. Non‐specialist health workers versus specialist care in treating common mental disorders (controlled before‐and‐after studies)

Setting: two CBA studies compared NSHWs (primary care doctors/general practitioners (GPs)) to 'gold standard' care (psychiatrists) for pharmacotherapy. These were designed as equivalence studies and were conducted in urban settings in Argentina (Lyketsos1999CBA Argentina) and Hungary (Zambori 2002 CBA Hungary).

Participants: Adults with common mental disorders (anxiety and depression) (Zambori 2002 CBA Hungary), and major depressive disorder (Lyketsos1999CBA Argentina).

Interventions:NSHWs: GPs in Lyketsos (1999) received half a day of training and ad hoc supervision from support staff. GPs in Zambori (2002) did not receive either training or supervision in the context of the trial.

Description of interventions: the GPs provided usual care for depression (prescribing medications, supportive therapy and referring). In Lyketsos (1999), both GPs and control group psychiatrists were given a protocol for prescribing antidepressants.

Results

We could not combine any outcomes. Below is a summary of the studies.

1. Severity of depression 

It is uncertain whether GPs are equivalent to specialists in delivering pharmacotherapy for depression (MD ‐0.90, 95% CI ‐1.20 to ‐0.60, 1 study, Lyketsos1999CBA Argentina) as the quality of evidence was very low (CBA study and very serious risk of bias) (summary of findings Table 4).

2. Adverse events

It is uncertain whether GPs are equivalent to specialists when adverse events get reported (RR 0.85, 95% CI 0.67 to 1.07, 1 study, Lyketsos1999CBA Argentina) as the quality of evidence was very low (Table 8).

Open in table viewer
Table 8. SoF 4: NSHWs compared with specialists in treating depression in adults in low‐ and middle‐income countries (CBAs)

What are the effects of NSHWs compared with specialists in treating depression for mental health care in low‐ and middle‐income countries? (additional outcomes for comparison 4)

Patient or population: Adults with depression
Settings: Middle‐income countries (Hungary and Argentina)

Intervention: NSHWs providing pharmacological intervention
Comparison: Specialists providing pharmacological intervention

Outcomes

Illustrative comparative risks* (95% CI)

Relative effect
(95% CI)

No of Participants
(studies)

Quality of the evidence
(GRADE)

Comments

Assumed risk

Corresponding risk

Specialists

NSHWs

Frequency of adverse events (at 56 days)
measured using the number of mild, moderate and severe adverse events

288 per 1000

245 per 1000
(193 to 308)

RR 0.85
(0.67 to 1.07)

768
(1 study)

⊕⊝⊝⊝
very low1,2

Note that a small clinically appreciable benefit was set at SMD < 0.2, and a moderate benefit at SMD of 0.5 to 0.8 (Cohen 1988)

Number of days spent in hospital (at 1 year)

‐

The mean number of days spent in hospital at 1 year in the NSHW group was
1.79 lower
(3.59 lower to 0.01 higher)

MD ‐1.79 (‐3.59 to 0.01)

124
(1 study)

⊕⊝⊝⊝
very low3,4

‐

Number of days spent in hospital (at 2 years)

‐

The mean number of days spent in hospital at 2 years in the NSHW group was
0.02 lower
(2.59 lower to 2.55 higher)

MD ‐0.02 (‐2.59 to 2.55)

124
(1 study)

⊕⊝⊝⊝
very low3,5

‐

Number of days spent on sick leave (at 1 year)

‐

The mean number of days spent on sick leave at 1 year in the NSHW group was
3.96 lower
(15.58 lower to 7.66 higher)

MD ‐3.96 (‐15.58 to 7.66)

108
(1 study)

⊕⊝⊝⊝
very low3,5

‐

Number of days spent on sick leave (at 2 years)

‐

The mean number of days spent on sick leave at 2 years in the NSHW group was
14.63 higher
(0.76 lower to 30.02 higher)

MD 14.63 (‐0.76 to 30.02)

123
(1 study)

⊕⊝⊝⊝
very low3,6

‐

*The basis for the assumed risk is the risk in the control group. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).
CBA: controlled before‐and‐after; CI: confidence interval; MD: mean difference; NSHW: non‐specialist health worker; RR: risk ratio; SMD: standardised mean difference.

GRADE Working Group grades of evidence
High quality: Further research is very unlikely to change our confidence in the estimate of effect.
Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.
Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.
Very low quality: We are very uncertain about the estimate.

1 Very serious study limitations: Lyketsos1999CBA Argentina was a CBA study so selection bias was likely. There was a risk of contamination and outcome assessments were done by same physicians doing the intervention. Downgraded by 2.

2 Serious imprecision: The 95% CI for the pooled estimates indicates appreciable and non‐appreciable harms with the interventions. In addition, results are only from one study. Downgraded by 1.
3 Very serious study limitations: Zambori 2002 CBA Hungary was a CBA study so selection is likely; and there were significant difference in baseline outcomes and baseline characteristics. Downgraded by 2.
4 Serious imprecision: The 95% CI for the pooled estimates indicates appreciable benefit for the interventions and non‐appreciable benefit for the control. In addition, results were only from one study. Downgraded by 1.
5 Very serious imprecision: Shows appreciable benefit for both interventions. Also results were only from one study. Downgraded by 2.

6 Serious imprecision: Shows appreciable benefit for specialists and non‐appreciable benefit for NSHWs. In addition, results were only from one study. Downgraded by 1.

3. Number of days spent at hospital and on sick leave

It is uncertain whether GPs were equivalent to specialists in the number of days spent at hospital (MD ‐1.79 days, 95% CI ‐3.59 to 0.01 in favour of NSHWs) and on sick leave (MD 14.63 days, 95% CI ‐0.76 to 30.02, 1 study, Zambori 2002 CBA Hungary) as the quality of evidence was very low (very serious risk of bias and imprecision).

Comparison 5. Non‐specialist health workers/other professionals with health roles‐led psychological interventions versus usual care in delivering post‐traumatic stress disorder interventions to adults (RCTs and NRCT)

Setting: we identified three studies, where participants lived in internally displaced camps (Dybdahl 2001 RCT Bosnia; Yeomans 2010 RCT Burundi) and refugee settlements (Neuner 2008 NRCT Uganda).

Participants: adults of both sexes who were diagnosed with PTSD, or with symptoms suggesting PTSD in mothers (Dybdahl 2001 RCT Bosnia).

Interventions:NSHWs/OPHRs: in Neuner (2008), LHWs with secondary school education were trained for six weeks in two counselling techniques (NET ‐ narrative exposure therapy a psychological therapy, and general trauma counselling), which they delivered in different sessions. In Yeomans (2010), the LHWs had experience in trauma workshop facilitation (so only were given one‐day training to adapt the workshop delivery) but little formal education. In Dybdahl (2001), preschool teachers were trained during a five‐day workshop that used a range of group, role play and lecture teaching methods. There was intensive supervision in Neuner (2008) and Dybdahl (2001) (not specified in Yeomans (2010)).

Description of interventions: duration: Neuner and Yeomans interventions had four to six sessions (but at different intervals) whereas Dybdahl's intervention consisted of weekly sessions for five months (20 sessions). Content: three studies' interventions were manualised (Neuner ‐ NET, Yeomans (both arms), Dybdahl). Neuner's non‐manualised trauma counselling, Yeomans workshop with counselling and Dybdahl's interventions were similar (problem solving and coping strategies, interpersonal skills, relaxation techniques and healing through reconciling communities, psychoeducation (and childcare in Dybdahl)). Neuner's first intervention was a psychological therapy NET.

Neuner and Dybdahl's comparison groups were usual care (without any LHWs, and in Dybdahl they received free medical care). Yeomans' comparison group was usual care (with LHWs without training for this intervention).

Results
1. Prevalence of post‐traumatic stress disorder symptoms

Neuner's (2008) LHW‐led interventions may have reduced the prevalence of PTSD symptoms (NET intervention: RR 0.48, 95% CI 0.27 to 0.85; trauma counselling: RR 0.55, 95% CI 0.33 to 0.93; 1 study, low‐quality evidence) (Figure 9).

Figure 9
Forest plot of comparison: 5 NSHW‐led psychological interventions versus usual care in treating adults with post‐traumatic stress disorder (RCT and NRCT), outcome: 5.1 Prevalence of post‐traumatic stress disorder (PTSD).

Forest plot of comparison: 5 NSHW‐led psychological interventions versus usual care in treating adults with post‐traumatic stress disorder (RCT and NRCT), outcome: 5.1 Prevalence of post‐traumatic stress disorder (PTSD).

2. Severity of post‐traumatic stress disorder symptoms

We pooled the three interventions that were most similar to each other (see description above). At assessment between two and six months post‐intervention, teacher/LHW interventions may have slightly improved PTSD symptoms (SMD ‐0.36, 95% CI ‐0.67 to ‐0.05, 3 studies, 223 participants, I2 = 22%, P value = 0.02, low‐quality evidence) (summary of findings Table 5). As Neuner and Yeomans had two intervention arms, we also combined these results in four ways (Neuner NET + Yeomans no psychoeducation; Neuner NET + Yeomans psychoeducation; Neuner ‐ trauma counselling + Yeomans no psychoeducation; Neuner ‐ trauma counselling + Yeomans psychoeducation). The results were very similar, ranging from SMD ‐0.31, 95% CI ‐0.58 to ‐0.04 (Dybdahl + Neuner NET + Yeomans psychoeducation) to SMD ‐0.41, 95% CI ‐0.72 to ‐0.11 (Dybdahl + Neuner NET + Yeomans no psychoeducation) (Figure 10; Table 9).

Figure 10
Forest plot of comparison: 5 NSHW‐led psychological interventions versus usual care in treating adults with PTSD (RCT and NRCT), outcome: 5.2 Severity of PTSD symptoms (N = completers).

Forest plot of comparison: 5 NSHW‐led psychological interventions versus usual care in treating adults with PTSD (RCT and NRCT), outcome: 5.2 Severity of PTSD symptoms (N = completers).

Open in table viewer
Table 9. SoF 5: NSHW and OPHR‐led psychological interventions compared with usual care in treating adults with PTSD (NRCT and RCTs)

What are the effects of NSHWs/OPHRs compared with usual mental health care in low‐ and middle‐income countries for data from an NRCT in adults with PTSD?

Patient or population: Adults with PTSD
Settings: Low‐ and middle‐income countries (Burundi, Uganda)
Intervention: NSHWs and OPHRs delivering psychological interventions (narrative exposure therapy, trauma counselling and workshops with psychoeducation)
Comparison: Usual care

Outcomes

Illustrative comparative risks* (95% CI)

Relative effect
(95% CI)

No of Participants
(studies)

Quality of the evidence
(GRADE)

Comments

Assumed risk

Corresponding risk

Usual care

NSHWs/OPHRs

Prevalence of PTSD in LHW‐led narrative exposure therapy, medium term (9 months)
measured using the PTSD diagnostic tool ‐ DSM‐IV from CIDI

632 per 1000

303 per 1000
(171 to 537)

RR 0.48
(0.27 to 0.85)

62
(1 study)

⊕⊕⊝⊝
low1,2

‐

Prevalence of PTSD in a LHW‐led trauma counselling, medium term (mean 9 months)
measured using the PTSD diagnostic tool ‐ DSM‐IV from CIDI

632 per 1000

348 per 1000
(209 to 588)

RR 0.55
(0.33 to 0.93)

65
(1 study)

⊕⊕⊝⊝
low1,2

‐

Severity of PTSD symptoms in LHW‐led psychological intervention (narrative exposure therapy) in the short term (6 months)
measured using the PTSD symptom score ‐ Post Traumatic Stress Diagnostic Scale

‐

The mean severity of PTSD with narrative exposure therapy in the short term was
0.55 standard deviations lower
(1.08 to 0.03 lower)

SMD ‐0.55 (‐1.08 to ‐0.03)

75
(1 study)

⊕⊕⊝⊝
low1,3

‐

Severity of depression ‐ psychological intervention in LHW led workshop with psychoeducation in the short term (within 2 weeks)
measured using depression rating scale: HSCL‐25

The mean [SD] scores on the HSCL‐25 was 1.83 [0.67]

The mean severity of depression with a psychosocial intervention (with psychoeducation) in the short term was
0.07 lower

(0.36 lower to 0.22 higher)

MD ‐0.07 (‐0.36 to 0.22)

76
(1 study)

⊕⊕⊝⊝
low4,5

‐

Severity of depression ‐ psychological intervention (workshop without psychoeducation) in the short term (within 2 weeks)
measured using a depression rating scale: HSCL‐25

The mean [SD] scores on the HSCL‐25 was 1.83 [0.67]

The mean severity of depression with a psychosocial intervention (without psychoeducation) short term was

0.14 lower

(0.42 lower to 0.14 higher)

MD ‐0.14 (‐0.42 to 0.14)

75
(1 study)

⊕⊕⊝⊝
low4,5

‐

*The basis for the assumed risk is the median control group risk or mean control group risk across studies for pooled estimates and the control group risk for single studies. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).
CI: confidence interval; CIDI: Composite International Diagnostic Interview; DSM: Diagnostic and Statistical Manual of Mental Disorders; HSCL: Hopkins Symptom Checklist; LHW: lay health worker; MD: mean difference; NRCT: non‐randomised controlled trial; NSHW: non‐specialist health worker; OPHR: other professionals with health roles; PTSD: post‐traumatic stress disorder; RCT: randomised controlled trial; RR: risk ratio; SD: standard deviation; SMD: standardised mean difference.

GRADE Working Group grades of evidence
High quality: Further research is very unlikely to change our confidence in the estimate of effect.
Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.
Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.
Very low quality: We are very uncertain about the estimate.

1 Serious study limitations: Neuner 2008 NRCT Uganda no allocation concealment, randomisation has no sequence generation. High dropout rate and different between groups, different baseline characteristics and likely contamination. Downgraded by 1.

2 Serious imprecision: The 95% CI of the effect estimate indicated appreciable and non‐appreciable benefit with the intervention, and the sample size was small. Downgraded by 1.

3 Serious imprecision: The 95% CI of the effect estimates demonstrated appreciable and non‐appreciable benefit with intervention and the sample size (75 participants) was small. Downgraded by 1.

4 Serious study limitations: Yeomans 2010 RCT Burundi: unvalidated Harvard Trauma Questionnaire in the local context (only validated in Burundi) so may affect reliability of outcomes. Downgraded by 1.

5 Serious imprecision: The 95% CI of the effect estimates demonstrated non‐appreciable benefit for both intervention and usual care and the sample size was small. Downgraded by 1.

A sensitivity analysis excluding Neuner (2008) (as it uses quasi‐randomisation) showed a lower effect size and imprecision in the first comparison (SMD ‐0.22, 95% CI ‐0.54 to 0.10, 2 studies, 151 participants, I2 = 0%, P value = 0.03), with similar results for the other comparisons using the other intervention arms. A subgroup analysis excluding Dybdahl, which was teacher‐led, and therefore retaining only LHWs suggested a slightly higher magnitude of effect (SMD ‐0.47, 95% CI ‐0.90 to ‐0.05, 2 studies, 148 participants, I2 = 34%, P value = 0.03).

3. Severity of depressive symptoms

LHW‐led psychological interventions may not have reduced depression severity (SMD ‐0.07, 95% CI ‐0.36 to 0.22, 1 study, both arms had similar results, 76 participants, low‐quality evidence due to imprecision and study limitations) (Analysis 5.3).

Comparison 6. Non‐specialist health workers versus usual care in improving dementia patients' and carers' outcomes (RCTs)

Setting:  we found two studies, which were conducted in urban areas in India (Dias 2008 RCT India), and Russia (Gavrilova 2009 RCT Russia).

Participants: the interventions were directed at carers of people with dementia. The carers were generally aged between 50 and 60 years and had varying economic backgrounds.

Interventions: NSHWs:Dias 2008 RCT India used two types of LHWs (home care advisors and lay counsellors) trained intensively for one week whereas Gavrilova 2009 RCT Russia used newly qualified doctors trained for two days to deliver the intervention. The LHWs were supervised every two weeks by a specialist. The supervision provided to the doctors was not described.

Description of interventions: in both studies brief carer interventions were conducted, based on a larger 10/66 dementia initiative (Prince 2004). However, Gavrilova (2009) organised a short training package for carers only, whereas Dias (2008) implemented a collaborative care package (LHWs undertook psychoeducation, counselling and followed up on treatment effects during home visits.

Results
1. Patient outcomes

At six months post intervention, NSHW‐led carer interventions for dementia probably led to slightly improved patient outcomes (including severity of behavioural symptoms (SMD ‐0.26, 95% CI ‐0.60 to 0.08, 2 studies) (Figure 11; summary of findings Table 6), quality of life (MD ‐0.43, 95% CI ‐0.98 to 0.12, 1 study), and functional impairment (MD ‐0.24, 95% CI ‐0.67 to 0.20, 1 study) (moderate‐quality evidence) (Table 10)).

Figure 11
Forest plot of comparison: 6 NSHWs versus usual care in improving dementia patients' and carers' outcomes (RCTs), outcome: 6.1 Severity of behavioural problem (patient).

Forest plot of comparison: 6 NSHWs versus usual care in improving dementia patients' and carers' outcomes (RCTs), outcome: 6.1 Severity of behavioural problem (patient).

Open in table viewer
Table 10. SoF 6: NSHWs compared with usual care in improving dementia patients' and carers' outcomes in low‐ and middle‐income countries (RCTs)

What are the effects of NSHW‐led care in improving dementia patients' and carers' outcomes for mental health care in low‐ and middle‐income countries? (additional outcomes for comparison 6)

Patient or population: Patients with dementia and their carers
Settings: Middle‐income countries (India, Russia)
Intervention: NSHWs delivering brief intervention to carers
Comparison: Usual care

Outcomes

Illustrative comparative risks* (95% CI)

Estimate effect
(95% CI)

No of Participants
(studies)

Quality of the evidence
(GRADE)

Comments

Assumed risk

Corresponding risk

Usual care

NSHWs

Patient functional ability (at 6 months)
measured using the functional ability scale (EASI)

‐

The mean patient functional ability with this brief carer intervention was
0.24 standard deviations lower
(0.67 lower to 0.20 higher)

MD ‐0.24 (‐0.67 to 0.20)

81
(1 study)

⊕⊕⊕⊝
moderate1,2

Note that a small clinically appreciable benefit was set at SMD < 0.2, and a moderate benefit at SMD of 0.5 to 0.8 (Cohen 1988)

Patient QoL (at 6 months)
measured using the quality of life score (DEMQOL)

‐

The mean patient QoL with this brief carer intervention was
0.43 standard deviations lower
(0.98 lower to 0.12 higher)

MD ‐0.43 (‐0.98 to 0.12)

53
(1 study)

⊕⊕⊕⊝
moderate1,2,3

‐

Carer mental health status (at 6 months)
measured using
general mental health status scores4

‐

The mean carer mental health status with this brief carer intervention was
0.42 standard deviations lower
(0.76 to 0.08 lower)

SMD ‐0.42 (‐0.76 to ‐0.08)

134
(2 studies)

⊕⊕⊕⊝
moderate5

‐

Carer burden
(at 6 months)
measured using a burden scale (Zarit Burden Interview)

‐

The mean carer burden in the brief carer intervention was
0.50 standard deviations lower
(0.84 to 0.15 lower)

SMD ‐0.50 (‐0.84 to ‐0.15)

134
(2 studies)

⊕⊕⊕⊝
moderate5

‐

Carer distress (at 6 months)
measured using
the carer distress scale: (NPI‐D)

‐

The mean carer distress with this brief carer intervention was
0.47 standard deviations lower
(0.82 to 0.13 lower)

SMD ‐0.47 (‐0.82 to ‐0.13)

134
(2 studies)

⊕⊕⊕⊝
moderate5

‐

Carer QoL (at 6 months)
measured using the
QoL assessment: (WHOQOL‐BREF)

‐

The mean carer quality of life in this brief carer intervention was
0.37 standard deviations lower
(0.92 lower to 0.17 higher)

MD ‐0.37 (‐0.92 to 0.17)

53
(1 study)

⊕⊕⊕⊝
moderate1,2,3

‐

*The basis for the assumed risk is the mean control group risk across studies for pooled results and the control group risk for single studies. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).
CI: confidence interval; DEMQOL: Dementia Quality of Life; EASI: Everyday Abilities Scales for India; MD: mean difference; NPI‐D: Neuropsychiatric Inventory ‐ Dementia; NSHW: non‐specialist health worker; QoL: quality of life; RCT: randomised controlled trial; SMD: standardised mean difference; WHOQOL‐BREF: World Health Organization Quality of Life‐BREF.

GRADE Working Group grades of evidence
High quality: Further research is very unlikely to change our confidence in the estimate of effect.
Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.
Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.
Very low quality: We are very uncertain about the estimate.

1No indirectness: there is only one study therefore the generalisability of this results to other settings is compromised. However, this also resulted in imprecision in the effect estimate, and hence the quality of evidence was not further downgraded.

2 Serious imprecision. The 95% CI of the MD indicated appreciable benefits for NSHWs/intervention and non‐appreciable benefits for usual care. The sample size is also small. Downgraded by 1.

3 No study limitations. Gavrilova 2009 RCT Russia was unclear whether allocation was concealed. However, no serious baseline differences in characteristics or outcomes were seen. Not downgraded.

4Dias 2008 RCT India: General Health Questionnaire (GHQ)‐12; Gavrilova 2009 RCT Russia Self Reporting Questionnaire (SRQ)‐20.

5 Serious imprecision. The 95% CI for the mean difference indicated appreciable and non‐appreciable benefits with the interventions. Downgraded by 1.

2. Carer outcomes

NSHWs probably improved/slightly improved carer outcomes, including burden (SMD ‐0.50, 95% CI ‐0.84 to ‐0.15) (Figure 12), mental health status (SMD ‐0.42, 95% CI ‐0.76 to ‐0.08) and distress (SMD ‐0.47, 95% CI ‐0.82 to ‐0.13) (moderate‐quality evidence). NSHWs probably led to little or no difference in carer quality of life. The study authors suggested that this result, which is out of keeping with the other carer outcomes, may be due to a type 2 error because the study was not statistically powered to detect differences of this size in the quality of life outcome.

Figure 12
Forest plot of comparison: 6 NSHWs versus usual care in improving dementia patients' and carers' outcomes (RCTs), outcome: 6.5 Carer burden.

Forest plot of comparison: 6 NSHWs versus usual care in improving dementia patients' and carers' outcomes (RCTs), outcome: 6.5 Carer burden.

Comparison 7. Non‐specialist health worker‐led brief alcohol interventions versus usual care for people with alcohol‐use disorders

Setting: we found two studies from rural Thailand (Noknoy 2010 RCT Thailand), and urban Kenya (Papas 2011 RCT Kenya).

Participants: adults with hazardous use of alcohol (AUDIT score ≥ 8) from primary care settings (Thailand) and patients (AUDIT score > 3) enrolled at a human immunodeficiency virus (HIV) clinic in Kenya. Patients with alcohol dependency were excluded in Noknoy (2010).

Interventions:

NSHWs: nurses in primary care clinics (Noknoy 2010 RCT Thailand), and LHWs (Papas 2011 RCT Kenya). Training ranged from six hours (Thai nurses) to 175 hours (Kenyan LHWs). Thai nurses received no specific supervision whereas the Kenyan LHWs received 300 hours, weekly monitoring and telephone supervision in the later stages of the trial.

Description of interventions: Noknoy's (2010) intervention was less intensive (three sessions (baseline, two weeks, six weeks) ‐ 15 minutes each) than Papas's (2011) (six sessions, once a week, 90 minutes per session). Noknoy's (2010) intervention was motivational enhancement therapy (MET), Papas's (2011) was a CBT intervention.

The comparison group was usual care. In Noknoy (2010), these were existing nurses without intervention training, and in Papas (2011), these were normal staff at the HIV clinic (without the LHW).

Results
1. Amount of alcohol consumed and frequency of binge drinking

At three to six months, NSHW‐led interventions for alcohol‐use problems may reduce the amount of alcohol consumed (MD ‐1.68 drinks/day, 95% CI ‐2.79 to ‐0.57, 2 studies, low‐quality evidence) and may reduce the frequency of binge drinking (MD ‐0.50, 95% CI ‐1.14 to 0.14, 1 study, low‐quality evidence due to risk of bias and imprecision) (Figure 13; summary of findings Table 7).

Figure 13
Forest plot of comparison: 7 NSHW‐led brief alcohol interventions versus usual care for adults with alcohol‐use disorders (RCTs), outcome: 7.1 Amount of alcohol consumed (MD).

Forest plot of comparison: 7 NSHW‐led brief alcohol interventions versus usual care for adults with alcohol‐use disorders (RCTs), outcome: 7.1 Amount of alcohol consumed (MD).

2. Adverse consequences

NSHW interventions for alcohol problems may not reduce road traffic accidents (RR 0.36, 95% CI 0.12 to 1.08, 1 study, 92 participants, low‐quality evidence due to sparse data, study limitations and serious imprecision). It is uncertain whether these interventions increase withdrawal symptoms (RR 2.67, 95% CI 0.29 to 24.37, 1 study, 68 participants, very‐low‐quality evidence due to sparse data, study limitations and very serious imprecision) (Table 11).

Open in table viewer
Table 11. SoF 7: NSHW‐led brief alcohol interventions compared with usual care for adults with alcohol‐use disorders (RCTs)

What are the effects of NSHWs in delivering brief alcohol interventions in RCTs for alcohol‐use disorders? (additional outcomes for comparison 7)

Patient or population: Patients with alcohol‐use disorders
Settings: Low‐ and middle‐income countries (Thailand, Kenya)
Intervention: NSHWs in delivering brief alcohol interventions
Comparison: Usual care

Outcomes

Illustrative comparative risks* (95% CI)

Relative effect
(95% CI)

No of Participants
(studies)

Quality of the evidence
(GRADE)

Comments

Assumed risk

Corresponding risk

Usual care

NSHWs

Frequency of binge drinking (3‐6 months)
measured using the frequency of binge drinking in the past week

‐

The mean frequency of binge drinking in the intervention groups was

0.50 lower

(1.14 lower to 0.14 higher)

MD ‐0.50

(‐1.14 to 0.14)

92
(1 study)

⊕⊕⊝⊝
low1,2

‐

Adverse consequences ‐ RTAs (at 6 months)
measured using the number of RTAs

220 per 1000

79 per 1000
(26 to 238)

RR 0.36
(0.12 to 1.08)

92
(1 study)

⊕⊕⊝⊝
low1,2

‐

Adverse consequences ‐ withdrawal symptoms (at 3 months)
measured using the number of withdrawal symptoms

31 per 1000

83 per 1000
(9 to 755)

RR 2.67
(0.29 to 24.37)

68
(1 study)

⊕⊝⊝⊝
very low1,3

‐

*The basis for the assumed risk is the mean control group risk across studies for pooled data or the control group risk for individual studies. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).
CI: confidence interval; MD: mean difference; NSHW: non‐specialist health worker; RCT: randomised controlled trial; RR: risk ratio; RTA: road traffic accident.

GRADE Working Group grades of evidence
High quality: Further research is very unlikely to change our confidence in the estimate of effect.
Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.
Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.
Very low quality: We are very uncertain about the estimate.

1 Serious study limitations: Noknoy 2010 RCT Thailand: high dropout rate with no information on whether they are different to completers, no validated tools in the setting, so unreliable primary outcomes. Papas 2011 RCT Kenya: unclear about whether the non‐blinding of outcome assessors would have impacted on study. Downgraded by 1.
2 Serious imprecision: The 95% CI of the effect estimates indicates an appreciable benefit for NSHW care and non‐appreciable benefit for usual care. The sample size was also small. Downgraded by 1.
3 Very serious imprecision: The 95% CI of the effect estimate indicates appreciable benefit for NSHW care and for usual care and also the study had a very small sample size. Downgraded by 2.

Comparison 8. Non‐specialist health workers/other professionals with health roles versus usual care in delivering interventions for children with post‐traumatic stress disorder and depression (RCTs)

Setting: we identified eight studies, which were conducted in internally displaced people camps in Bosnia (Dybdahl 2001 RCT Bosnia), Indonesia (Tol 2008 C‐RCT Indonesia), Kosovo (Gordon 2008 RCT Kosovo), Nepal (Jordans 2010 C‐RCT Nepal), Sri Lanka (Berger2009 CRCT SriLanka; Tol 2012 C‐RCT SriLanka), and Uganda (Bolton 2007 RCT Uganda; Ertl 2011 RCT Uganda). Most studies were undertaken in post‐conflict or peri‐conflict settings, except for Berger (2009), which followed a natural disaster. The settings were rural/semi‐rural (Bolton 2007 RCT Uganda; Gordon 2008 RCT Kosovo; Jordans 2010 C‐RCT Nepal; Tol 2008 C‐RCT Indonesia), urban (Berger2009 CRCT SriLanka; Dybdahl 2001 RCT Bosnia), or urban and rural (Ertl 2011 RCT Uganda; Tol 2012 C‐RCT SriLanka).

Participants: children with PTSD diagnoses or symptoms were included. Some also had depressive and anxiety symptoms, or conduct problems, or a combination. The ages of the children varied from five to six years (Dybdahl 2001 RCT Bosnia), to adolescents aged 14 to 18 years (Bolton 2007 RCT Uganda; Gordon 2008 RCT Kosovo). One study included child soldiers aged 12 to 25 years (Ertl 2011 RCT Uganda). Most children came from low‐resource backgrounds.

Interventions:NSHWs: five studies used LHWs (of both sexes) and had manual‐based training for their respective interventions (Bolton 2007 RCT Uganda; Ertl 2011 RCT Uganda; Jordans 2010 C‐RCT Nepal; Tol 2008 C‐RCT Indonesia; Tol 2012 C‐RCT SriLanka). Supervision varied from being regular (Jordans 2010 C‐RCT Nepal; Tol 2008 C‐RCT Indonesia; Tol 2012 C‐RCT SriLanka) to intensive (e.g. case discussions of their treatment sessions and their notes) (Ertl 2011 RCT Uganda).

OPHRs: three studies used existing high school or preschool teachers (Berger2009 CRCT SriLanka; Dybdahl 2001 RCT Bosnia; Gordon 2008 RCT Kosovo), who were given an additional three‐day (Berger2009 CRCT SriLanka) to 10‐day (Gordon 2008 RCT Kosovo) intensive training by researchers. Supervision was weekly (Berger2009 CRCT SriLanka; Dybdahl 2001 RCT Bosnia), or regularly (Gordon 2008 RCT Kosovo), by mental health professionals. There was no information on training for Dybdahl (2001).

Description of interventions: all interventions were delivered to groups in schools except for two in community groups (Bolton 2007 RCT Uganda; Dybdahl 2001 RCT Bosnia), and one in child soldiers in their home (Ertl 2011 RCT Uganda). All interventions were targeted at children except Dybdahl (2001) where the target group was mothers. Group interventions varied from 12 to 20 sessions spread over five weeks to five months. Jordans (2010), Tol (2008) and Tol (2012) had the same manual‐based, classroom‐room‐based intervention (CBI). This intervention included elements of creative‐expressive therapy, co‐operative play and CBT. Berger (2009), Dybdahl (2001) and Ertl (2000) were similar psychosocial/psychological interventions (psychoeducation, group activities, coping skills training) though Ertl (2000) had two arms: NET and academic catch up. Bolton (2007) was a three‐armed trial, comparing two LHW interventions (G‐IPT and creative play) delivered to single‐sex groups. Gordon (2008) used slightly different psychosocial techniques (imaginative mind‐body techniques, meditation, etc.).

Results
1. Severity of post‐traumatic stress disorder symptoms

Because of differences in outcome measures for short‐term outcomes (MCDs could not be combined with MD), we present these outcomes separately. We followed this approach for all outcomes in this comparison.

In the short term (< six months post intervention), despite a large apparent clinical benefit (SMD ‐0.89, 95% CI ‐1.49 to ‐0.30, 3 studies (including Ertl's first intervention arm: NET ‐ a psychological therapy), 298 participants), it is uncertain whether LHWs and teachers reduce the severity of PTSD symptoms due to very‐low‐quality evidence (very serious study limitations and serious inconsistency I2 = 78%; P value = 0.003) (Figure 14; summary of findings Table 8). Results were similar if Ertl's second intervention arm (academic catch‐up ‐ assisting children with their academic activities only) was combined (SMD ‐0.85, 95% CI ‐1.52 to ‐0.19, 295 participants, I2 = 82%; P value = 0.003). In a planned subgroup analysis, interventions led by teachers were analysed separately to attempt to reduce heterogeneity (Berger2009 CRCT SriLanka; Dybdahl 2001 RCT Bosnia). However, it was still uncertain whether teacher‐led interventions may reduce the severity of PTSD symptoms (SMD ‐1.20, 95% CI ‐1.52 to ‐0.88, 2 studies, 244 participants, (I2 = 0%; P value = 0.64) because of very‐low‐quality evidence (serious study limitations and imprecision due to sparse data).

Figure 14
Forest plot of comparison: 9 NSHWs/OPHRs versus usual care in conducting interventions for children with PTSD (RCTs), outcome: 9.3 Severity of PTSD symptoms ‐ teacher‐led interventions (children) (MDs).

Forest plot of comparison: 9 NSHWs/OPHRs versus usual care in conducting interventions for children with PTSD (RCTs), outcome: 9.3 Severity of PTSD symptoms ‐ teacher‐led interventions (children) (MDs).

It is uncertain whether LHW‐led CBI reduce PTSD symptoms (MCD ‐0.56, 95% CI ‐2.82 to 1.70, very‐low‐quality evidence due to very serious risk of bias, heterogeneity (I2 = 82%; P value = 0.004) and serious imprecision). In one study (Tol 2012 C‐RCT SriLanka), PTSD symptoms improved in girls in the control group (not in the intervention group), but there was no difference for boys (Analysis 8.3).

At 11 months, one study (Ertl 2000) suggested that NET or academic catch‐up interventions probably does not reduce PTSD severity (SMD ‐0.45, 95% CI ‐0.99 to 0.10, 1 study, 53 participants, moderate‐quality evidence due to serious imprecision and sparse data) (Figure 14; Table 12).

Open in table viewer
Table 12. SoF 8: NSHWs/OPHRs compared with usual care in conducting interventions for children with post‐traumatic stress disorder and depression (RCTs)

What are the effects of NSHWs/OPHRs conducting interventions for children with PTSD from RCTs in low‐ and middle‐income countries? (additional outcomes for comparison 8)

Patient or population: Children/adolescents with PTSD and related depressive/anxiety symptoms
Settings: Low‐ and middle‐income countries (Bosnia, Indonesia, Kosovo, Nepal, Sri Lanka, Uganda)
Intervention: NSHWs/OPHRs delivering psychological and psychosocial interventions

Comparison: Usual care

Outcomes

Illustrative comparative risks* (95% CI)

Estimate effect
(95% CI)

No of Participants
(studies)

Quality of the evidence
(GRADE)

Comments

Assumed risk

Corresponding risk

Usual care

NSHWs/OPHRs

Severity of PTSD symptoms in LHW‐led classroom‐based intervention, short term (1‐6 months)
measured using the CPSS

‐

The MCD in severity of PTSD symptoms in classroom‐based LHW‐led intervention groups was
0.56 lower
(2.82 lower to 1.7 higher)

MCD ‐0.56 (‐2.82 to 1.70)

1090
(3 studies)

⊕⊝⊝⊝
very low1,2,3

‐

Severity of PTSD symptoms in narrative exposure therapy, medium term (11 months)
measured using the CAPS

‐

The mean severity of depressive symptoms in teacher/LHW‐led intervention groups was
0.45 standard deviations lower
(0.99 lower to 0.10 higher)

SMD ‐0.45

(‐0.99 to 0.10)

53

(1 study)

⊕⊕⊕⊝

moderate4,5

Note that a small clinically appreciable benefit was set at SMD < 0.2, a moderate benefit at SMD of 0.5 to 0.8, and a large benefit > 0.8 (Cohen 1988).

Severity of depressive symptoms in teacher/LHW‐led interventions, short term (2‐6 months)
measured using various depression rating scales6

‐

The mean severity of depressive symptoms in teacher/LHW‐led intervention groups was
0.23 standard deviations lower
(0.45 to 0.22 lower)

SMD ‐0.23 (‐0.45 to ‐0.22)

504
(4 studies)

⊕⊕⊝⊝
low7,8

‐

Severity of depressive symptoms in LHW‐led classroom based interventions, short term (1‐6 months)
measured using the DSRS

‐

The mean severity of depressive symptoms in classroom based) LHW‐led intervention groups was
0.18 lower
(0.33 to 0.03 lower)

MCD ‐0.18

(‐0.33 to ‐0.03)

1092
(3 studies)

⊕⊕⊝⊝
low1,8

‐

Severity of depressive symptoms in narrative exposure therapy, medium term (11 months)
measured using the MINI depression rating scale

‐

The mean severity of depressive symptoms in teacher/LHW‐led intervention groups was
0.02 standard deviations lower
(0.52 lower to 0.56 higher)

SMD ‐0.02

(‐0.52 to 0.56)

53 participants (1study)

⊕⊕⊝⊝

low5,9

‐

Severity of anxiety symptoms in LHW‐led classroom based intervention, short term (1‐6 months)
measured by SCARED

‐

The mean severity of anxiety symptoms in the intervention groups was
0.34 lower
(0.75 lower to 0.07 higher)

MD ‐0.34

(‐0.75 to 0.07)

1092
(3 studies)

⊕⊝⊝⊝
very low1,3

‐

Functional impairment in LHW/teacher‐led interventions, short term (1‐6 months)
measured by various functional impairment scales10

‐

The mean functional impairment in teacher‐led interventions was
0.61 standard deviations lower
(1.13 to 0.08 lower)

SMD ‐0.61

(‐1.13 to ‐0.08)

220
(2 studies)

⊕⊕⊕⊝
moderate11,12

‐

Functional impairment in LHW‐led classroom‐based interventions, short term (1‐6 months)
measured by various functional impairment scales13

‐

The mean functional impairment in CBI (classroom based) LHW‐led intervention groups was
0.81 lower
(1.48 to 0.13 lower)

MCD ‐0.81

(‐1.48 to ‐0.13)

1092
(3 studies)

⊕⊕⊝⊝
low1,8

‐

Functional impairment, in narrative exposure therapy, medium term (11 months)
measured using CAPS functional impairment scale

‐

The mean functional impairment in LHW‐led narrative exposure therapy was
0.69 lower
(1.25 to 0.14 lower)

SMD ‐0.69

(‐1.25 to ‐0.14)

53

(1 study)

⊕⊕⊕⊝

moderate14

‐

*The basis for the assumed risk the mean control group risk across studies for pooled results and the control group risk for single studies. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).
CAPS: Clinical Administered PTSD Scale; CBI: classroom‐room‐based intervention; CI: confidence interval; CPSS: Child Posttraumatic Stress Scale; LHW: lay health worker; MCD: mean change difference; MINI: Mini International Neuropsychiatry Interview; NSHWs: non‐specialist health worker; OPHRs: other professionals with health roles; PTSD: post‐traumatic stress disorder; SMD: standardised mean difference; SCARED: Screen for Child Anxiety Related Disorders.

GRADE Working Group grades of evidence
High quality: Further research is very unlikely to change our confidence in the estimate of effect.
Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.
Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.
Very low quality: We are very uncertain about the estimate.

1 Very serious study limitations: Tol 2012 C‐RCT SriLanka: uncertainty about random sequence generation and allocation concealment. Tol 2008 C‐RCT Indonesia: outcome assessment not blind: child self ratings with help of assessors who were not blinded to treatment condition. Jordans 2010 C‐RCT Nepal: unclear allocation concealment and non‐blinded assessment of outcomes. Downgraded by 2.
2 Very serious inconsistency: I2 statistic = 82%. The inconsistency is related to the direction of effect. Downgraded by 2.
3 Serious imprecision: The 95% CI for the pooled estimates indicates appreciable benefit for intervention group and non‐appreciable benefit for usual care. Downgraded by 1.

4 Serious imprecision: The 95% CI for the pooled estimates indicates appreciable benefit for intervention group and non‐appreciable benefit for usual care. The sample size was also small. Downgraded by 1.

5 No indirectness: There was only one study, therefore, the generalisability of this results to other settings is compromised. However, this also resulted in imprecision in the effect estimate, and hence the quality of evidence was not further downgraded.

6Berger2009 CRCT SriLanka: Becks Depression Inventory; Bolton 2007 RCT Uganda: Acholi Psychosocial Assessment Instrument; Dybdahl 2001 RCT Bosnia: Birleson's depression inventory; Ertl 2011 RCT Uganda: MINI.
7 Very serious study limitations: Berger2009 CRCT SriLanka: no allocation concealment, likely contamination and outcomes not adjusted for clustering; Dybdahl 2001 RCT Bosnia: not clear if allocation concealed; differences in baseline characteristics incomplete outcome data (denominators not provided by intervention or control for each of the tests), and likelihood of contamination as both intervention and control in same camps; Bolton 2007 RCT Uganda had unclear allocation concealment, baseline characteristics (age) were different and not adjusted for in analysis, likely risk of contamination between children in both camps. Downgraded by 2.

8 No imprecision: appreciable benefit for LHW/teacher‐led care. Not downgraded.

9 Very serious imprecision: The 95% CI for the pooled estimates indicates appreciable benefit for intervention group and appreciable benefit for usual care. The sample size is also small. Downgraded by 2.

10Berger2009 CRCT SriLanka:: Child Diagnostic Interview Schedule (CDIS); Ertl 2011 RCT Uganda: CAPS ‐ Functional Impairment Section.

11 No imprecision: appreciable benefit for LHW/teacher‐led care. Not downgraded.

12 Serious study limitations: Berger2009 CRCT SriLanka: no allocation concealment, likely contamination and outcomes not adjusted for clustering. Downgraded by 1.

13Jordans 2010 C‐RCT Nepal: Children's Functional Impairment; Tol 2008 C‐RCT Indonesia; Tol 2012 C‐RCT SriLanka: Functional impairment Score (FIS). Data were change scores and hence SMD could not be used. MD is likely to be misleading but since two of the three trials used the same scale this may not be important. Removal of data from Jordans 2010 C‐RCT Nepal did not appreciably alter effect estimates.

14 Serious imprecision: appreciable benefit for LHW/teacher‐led care. However, the sample size is small. Downgraded by 1.

Two CBA studies also assessed teacher‐led interventions for children with PTSD (aged six to 17 years) from displaced populations (Thabet 2005 CBA Palestine (short term ‐ two months); Wolmer 2005 CBA Turkey (long term ‐ three years post intervention). It is uncertain whether these interventions reduced PTSD severity (SMD ‐0.10, 95% CI ‐0.34 to 0.14, 329 participants, very‐low‐quality evidence) (Table 13).

Open in table viewer
Table 13. SoF: NSHWs/OPHRs compared with usual care in conducting interventions for children with PTSD in low‐ and middle‐income countries (CBAs)

What are the effects of NSHWs/OPHRs compared with usual care in conducting interventions for children with PTSD for mental health care in low‐ and middle‐income countries? (additional CBA outcomes for comparison 8)

Patient or population: Children with PTSD
Settings: Low‐ and middle‐income countries (Palestine, Turkey)
Intervention: Teachers delivering psychoeducational and other interventions
Comparison: Usual care

Outcomes

Illustrative comparative risks* (95% CI)

Relative effect
(95% CI)

No of Participants
(studies)

Quality of the evidence
(GRADE)

Comments

Assumed risk

Corresponding risk

Usual care

NSHWs/OPHRs

Severity of PTSD symptoms, medium and short term (2‐36 months)
measured using the PTSD symptom scale ‐ CPTSD‐RI

‐

The mean severity of PTSD symptoms in the medium and short term (combined) in the intervention groups was
0.1 standard deviations lower
(0.34 lower to 0.14 higher)

SMD ‐0.10 (‐0.34 to 0.14)

351
(2 studies)

⊕⊝⊝⊝
very low1

‐

Severity of PTSD symptoms ‐ short term (2 months)
measured using the PTSD symptom scale ‐ CPTSD‐RI

‐

The mean severity of PTSD symptoms ‐ short term (within 2 months) in the intervention groups was
0.1 standard deviations higher
(0.41 lower to 0.62 higher)

SMD 0.1 (‐0.41 to 0.62)

64
(1 study)

⊕⊝⊝⊝
very low1,2

Thabet 2005 CBA Palestine

Severity of PTSD symptoms ‐ long term (3 years)
measured using the PTSD symptom scale ‐ CPTSD‐RI

‐

The mean severity of PTSD symptoms ‐ long term ‐ in the intervention groups was
0.16 standard deviations lower
(0.43 lower to 0.12 higher)

SMD ‐0.16 (‐0.43 to 0.12)

287
(1 study)

⊕⊝⊝⊝
very low1,2

Wolmer 2005 CBA Turkey

Severity of CMDs (2‐12 months)
measured using CMD severity scores3

‐

The mean severity of CMDs in the intervention groups was
0.25 standard deviations lower
(0.46 to 0.04 lower)

SMD ‐0.25 (‐0.46 to ‐0.04)

459
(2 studies)

⊕⊝⊝⊝
very low4,5

Both Thabet 2005 CBA Palestine (st) and Loughry 2006 CBA Palestin (LT)

Severity of CMDs psychosocial intervention, short term (2 months)
measured using the Child Depression Inventory

‐

The mean severity of CMDs ‐ short term ‐ in psychosocial interventions was
0.12 standard deviations lower
(0.63 lower to 0.4 higher)

SMD ‐0.12 (‐0.63 to 0.4)

64
(1 study)

⊕⊝⊝⊝
very low4,6

Thabet 2005 CBA Palestine only

Severity of CMDs social intervention ‐ medium term (12 months)
measured using the CBCL internalising score (depression, anxiety, somatic symptoms, withdrawn)

‐

The mean severity of CMDs ‐ medium term ‐ in social interventions was
0.27 standard deviations lower
(0.5 to 0.04 lower)

SMD ‐0.27 (‐0.5 to ‐0.04)

395
(1 study)

⊕⊝⊝⊝
very low4,5

Loughry 2006 CBA Palestin only

*The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% CI) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).
CBA: controlled before‐and‐after; CBCL: Child Behaviour Checklist; CI: confidence interval; CMD: common mental disorders; NSHWs: non‐specialist health worker; OPHRs: other professionals with health roles; PTSD: post‐traumatic stress disorder; SMD: standardised mean difference.

GRADE Working Group grades of evidence
High quality: Further research is very unlikely to change our confidence in the estimate of effect.
Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.
Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.
Very low quality: We are very uncertain about the estimate.

1 No imprecision: Unappreciable benefit for both NSHW care and usual care.
2 Very serious risk of bias: both CBA studies so non‐randomised and no allocation concealment. Also Thabet 2005 CBA Palestine: differences in baseline outcomes; Wolmer 2005 CBA Turkey: unvalidated tools in this setting so uncertain reliability of outcomes.
3Thabet 2005 CBA Palestine: Child Depression Inventory; Loughry 2006 CBA Palestin: CBCL internalising score (depression, anxiety, somatic symptoms, withdrawn).
4 Very serious risk of bias: both CBAs so non‐randomised and no allocation concealment. Also Thabet 2005 CBA Palestine: differences in baseline outcomes; Loughry 2006 CBA Palestin: differences in baseline outcomes and characteristics and risk of contamination and the intervention between sites differed slightly making comparisons difficult.
5 No imprecision: Non‐appreciable, and possibly appreciable benefit for intervention group.
6 Very serious imprecision: Appreciable benefit for intervention and non‐appreciable benefit for control group.

2. Severity of depression symptoms

In the short term (< six months), interventions delivered by either teachers or LHWs may slightly reduce depressive symptoms compared with usual care (SMD ‐0.23, 95% CI ‐0.45 to ‐0.22, 4 studies, 504 participants, low‐quality evidence due to very serious study limitations) (Table 12). However, LHW‐led CBI may have led to little or no difference in the severity of depression symptoms compared with usual care (MCD ‐0.18, 95% CI ‐0.33 to ‐0.03, low‐quality evidence). In one CBA study, it was uncertain if interventions delivered by teachers reduced depressive symptoms (SMD ‐0.12, 95% CI ‐0.63 to 0.40) (Thabet 2005 CBA Palestine; very‐low‐quality evidence; Table 13).

In the medium term (11 months post intervention), LHW‐led interventions may not have reduced depressive symptoms (SMD 0.02, 95% CI ‐0.52 to 0.56, 1 study, 53 participants, low‐quality evidence due to very serious imprecision). Similarly, Loughry 2006 CBA Palestin's study, a LHW‐led intervention for displaced children with PTSD, suggested that the effects are uncertain (SMD ‐0.27, 95% CI ‐0.50 to ‐0.04, very‐low‐quality evidence).

3. Severity of anxiety symptoms

It is uncertain whether LHW‐led CBI reduced anxiety severity in children compared with usual care (MCD ‐0.34, 95% CI ‐0.75 to 0.07, 3 studies, very‐low‐quality evidence due to selection bias and imprecision). Tol 2012 C‐RCT SriLanka undertook a subgroup analysis by sex that showed there may be little or no difference for boys (MCD ‐0.63, 95% CI ‐1.23 to ‐0.03, 245 participants, low‐quality evidence).

4. Functional impairment

In the short term (< six months), LHW/teacher‐led interventions probably reduce functional impairment (SMD ‐0.61, 95% CI ‐1.13 to ‐0.08, 2 studies, 220 participants, moderate‐quality evidence due to serious study limitations) (Analysis 8.9) and LHW‐led CBI (MCD ‐0.81, 95% CI ‐1.48 to ‐0.13, 3 studies, 1092 participants) may have reduced functional impairment (low‐quality evidence due to very serious study limitations) (Analysis 8.10).

At 11 months, Ertl's LHW‐led NET group probably also reduced functional impairment (SMD ‐0.69, 95% CI ‐1.25 to ‐0.14, 1 study, 53 participants, moderate‐quality evidence due to serious imprecision).

Outcomes of studies not assigned to the above comparisons

The individual studies that could not be pooled are fully described in the Characteristics of included studies tables and their outcomes are summarised in Table 3 and Appendix 4.

These studies included the following comparisons:

  1. NSHW versus usual care (life skills training) in improving drug abuse outcomes (RCT);

  2. NSHWs versus usual care for treating schizophrenia (CBA study);

  3. NSHWs versus specialist care in treating epilepsy (equivalence trial RCT);

  4. OPHRs versus usual care in delivering a psychosocial/activities intervention for parents of children with intellectual disabilities (RCT).

Economic studies

Although literature is emerging on the effectiveness of NSHWs in delivering mental health services, very limited data are available on the unit costs and resource requirements. This is mainly due to the difficulties associated with conducting economic analyses, time lags from inputs to outcomes and many confounding variables.

Table 14 shows the data from the three included studies that reported cost effectiveness or costs in relation to the care of depression in adults and PTSD in children. These studies underline the feasibility and potential cost effectiveness of NSHWs in providing mental health care, and report costs related to absenteeism and healthcare utilisation. However, all of the studies had significant risks of bias that cast doubt on the accuracy and reliability of these data. Not all relevant alternatives and costs (such as productivity loss) were considered or reported, some costs relied on estimates, future costs were not discounted properly and chosen time horizons were less than one year in Araya.

Open in table viewer
Table 14. Summary of costs and resource use from included studies

Author/year

Type of economic evaluation

Study population

Intervention

Economic results

Araya 2003 RCT Chile

Cost‐effectiveness analysis

Women with depression

Collaborative intervention (doctors, non‐medical professionals supervised by psychiatrist) with stepped care, multicomponent programme compared with usual care in depressed women in Chile

Incremental cost per person for improved care was USD37.6 more than usual care. Unit cost to obtain 1 additional depression‐free day was USD0.75

Jordans 2011

Cost analysis

Children with PTSD (7‐15 years)

LHW‐led multilayered package (including classroom‐based intervention, non‐therapeutic resilience groups, psychoeducation and counselling) (data extracted from Sri Lanka and Indonesia as related to Tol 2008 C‐RCT Indonesia and Tol 2012 C‐RCT SriLanka)

Mean cost per user of total package:

Indonesia: USD21.77 (59% of which is human resources cost). Sri Lanka: USD8.85 (56% of which is human resources cost)

Zambori 2002 CBA Hungary

Cost analysis

Patients with anxiety and mood disorders

Primary physicians versus psychiatrists in prescribing sertraline in Hungary

Absenteeism reduced from 15.7 to 6.8 days and costs of non‐psychiatric prescriptions decreased from USD138 to USD91.8 per year. Laboratory costs ranged from USD6.4 to USD11.5

LHW: lay health worker; PTSD: post‐traumatic stress disorder.

Discussion

Summary of main results

This review identified 38 RCTs and NRCTs and CBA studies evaluating the effectiveness of NSHWs delivering care for MNS disorders in seven LICs and 15 middle‐income countries. Twenty‐two studies used LHWs, and most addressed depression or PTSD. The diversity of included studies limited meta‐analysis to outcomes for eight comparisons. All analyses presented below compare interventions versus usual care.

The review showed that the use of NSHWs, compared with usual healthcare services:

  • may increase the number of adults who recover from depression or anxiety (or both) two to six months after treatment (low‐quality evidence). At seven to 12 months, LHW‐led psychological interventions probably reduced common mental disorder (anxiety and depression) symptoms and functional impairment, but collaborative care interventions (a multidisciplinary team that included one or several NSHWs and specialists) showed little or no effect over the same time period. It is unclear why this effect was lost by 12 months for collaborative care and this may be because of depression recurrence and because of the relatively short duration of the intervention. The intervention may need to carry on longer, even if just as case management, to detect early signs of relapse. There is also insufficient evidence, due to sparse data, to favour LHW‐led psychological interventions over collaborative care at this time;

  • may slightly reduce symptoms for mothers with perinatal depression symptoms (low‐quality evidence);

  • may slightly reduce the prevalence and the symptoms of adults with PTSD over six months (low‐quality evidence);

  • probably slightly improves the symptoms of people with dementia (moderate‐quality evidence);

  • probably improves/slightly improves the mental well‐being, burden and distress of carers of people with dementia (moderate‐quality evidence);

  • may decrease the amount of alcohol consumed by people with alcohol‐use disorders (low‐quality evidence).

In children experiencing PTSD, teachers and LHWs:

  • probably reduce functional impairment of PTSD‐affected children at six and 12 months following the intervention (moderate‐quality evidence);

  • may have little or no effect on depressive or conduct symptoms (low‐quality evidence);

  • it is uncertain whether LHWs or teachers reduce PTSD symptoms over six months among children (very‐low‐quality evidence).

The three studies measuring costs suggested that NSHW interventions may be cost effective for depression and PTSD, but there is insufficient evidence to draw firm conclusions. For other outcomes (including the equivalence CBA studies for NSHWs versus specialists in treating depression), the evidence is insufficient to draw conclusions regarding the effects of NSHWs. There is also insufficient evidence to determine which NSHW training or intervention strategies are likely to be most effective.

Overall completeness and applicability of evidence

This review aimed to assess the effectiveness of NSHWs in delivering care to people with MNS disorders in order to provide guidance to health policy makers in LMICs. Several issues need to be considered when making judgements about the applicability of these findings to large‐scale programmes.

Factors related to the type and role of non‐specialist health workers

The included studies reported using many different types of NSHWs/OPHRs (some of whom were existing cadres within health services while others were additionally trained resources), particularly for common mental disorders and PTSD. However, there were few studies in each comparison and often information on details of the intervention and training were inadequate. We were, therefore, not able to explore the effects of interventions according to different NSHW characteristics (including selection, training, support, incentives or remuneration). We were also not able to explore the independent effect of NSHWs when they were part of complex interventions (such as collaborative care) or the effect of the intensity of the NSHW‐led interventions. This information would help guide policymakers to tailor the type of NSHWs and their roles within scaled up programmes appropriately.

Furthermore, the review provides limited data on the effects of task‐shifting to NSHWs. Most studies considered NSHWs or OPHRs as an add‐on to usual care. Only three studies (Li 1989 RCT China for epilepsy, and Lyketsos1999CBA Argentina; Zambori 2002 CBA Hungary for depression) compared these cadres versus specialists, but these studies were of low quality and data for most outcomes could not be pooled. We, therefore, cannot be certain if task‐shifting (with appropriate supervision) to non‐specialists leads to equivalent quality of care or results in terms of appropriate care. Furthermore, very few studies measured adverse effects or unintended consequences of NSHW‐led care ‐ such effects could impact on the appropriateness and quality of care, and could lead to patient harm.

Interventions

Comparisons of studies were possible by MNS disorder and by broad types of interventions (such as drug treatment and psychological interventions), as well as who delivered them. However, again there were too few studies and substantial intervention variation within these categories, so it was not possible to draw strong conclusions on what type of intervention was most effective in relation to specific mental health disorders.

None of the included studies addressed the impact of delivering mental health care on other elements of NSHWs' healthcare roles (e.g. the impact of a mental health intervention on a PHC doctors' other tasks such as diabetes, or on their working pattern, such as consultation times). One study assessed the impact of a depression intervention on the number of days spent in hospital (i.e. both a patient outcome and a health service outcome) (Zambori 2002 CBA Hungary), but more studies looking at these indirect outcomes or unintended consequences are needed.

Programme delivery

Several issues need to be considered in applying these findings to healthcare delivery systems.

First, these are interventions delivered in a research setting where NSHWs are more likely to have been carefully selected; project leaders are more motivated; remuneration may be more available because of research funding; and training, supervision and monitoring are generally much more intensive. These conditions may not be replicable at scale or may not be as effective at scale.

Second, the types of study design chosen here were not appropriate or sufficient to inform judgements regarding the sustainability of programmes; alternative study designs, such as longitudinal studies, economic evaluations and qualitative studies, are needed for this.

Third, the elements necessary for assessing the applicability of interventions need to be considered in each setting where decisions on task‐sharing or task‐shifting are being made (Lavis 2009). These elements include the extent to which these real‐life settings resemble those of included studies, such as on‐the‐ground constraints, health service arrangements, differences in baseline conditions, presence of specific groups who might benefit from the intervention and the availability of routine data.

Fourth, it is important to know the financial burden of such interventions. Few studies reported cost data, which makes it difficult to draw any conclusions on this question.

Quality of the evidence

The review included 38 studies covering a wide range of interventions and settings. For studies included in meta‐analyses, the evidence for most outcomes was of low to moderate quality. Risk of bias assessments highlighted concerns regarding insufficient information on sequence generation and allocation concealment; differences in baseline outcome measurements; the reliability of primary outcome measures; and a failure to address incomplete outcome data, particularly safety data, adequately. Several studies were small and were probably underpowered.

Where meta‐analysis was possible, the results were fairly consistent in showing improvements in favour of NSHW interventions, although for some interventions and outcomes there were important variations in the reported effects that could not be explained.

Some studies assessed large numbers of outcomes, increasing the probability of finding statistically significant differences for some outcomes by chance. Furthermore, the diversity of the psychometric and other outcome measures used made the interpretation of statistically pooled outcome data difficult.

In the update of this review, we will consider RCTs and cluster RCTs only, as we found few NRCTs and CBA studies and no ITS studies. Those NRCTs and CBA studies that were included did not contribute significant additional data to the review.

Potential biases in the review process

NSHWs, and in particularly LHWs, are still currently poorly indexed in the literature. Though we tried covering a broad range of different synonyms for these health workers, it is possible that some studies have been missed. In addition, NSHWs and LHWs do not have standard widely accepted definitions, so some readers may disagree with these definitions or how this review has aggregated different health workers together.

There were too few studies for each comparison to assess publication bias through assessment of asymmetry. However, because many studies reported non‐statistically significant results, publication bias is probably unlikely.

Many meta‐analyses were performed; therefore, some of the findings may be due to chance. Many pooled results were statistically and clinically heterogeneous, mainly because of the small number of studies and the breadth of geographical, health worker and patient characteristics ‐ these results, therefore, need to be interpreted with caution.

Furthermore, we did not record whether, for NRCTs, the study restricted participant selection or demonstrated balance or matching between intervention and control groups on prognostic factors, or a combination of these. An imbalance of these may act as confounders (such as age, sex, socioeconomic status). However, most of the findings were reported from RCTs, so this is unlikely to have a major impact on the interpretation of our findings.

A further limitation was that trials that did not conduct an ITT analysis were generally not re‐analysed or their missing data was not imputed (except for one analysis were we were able to source data: NSHW‐led psychological interventions for depression ‐ prevalence of depression). Doing so may have impacted on the estimates of effect.

Agreements and disagreements with other studies or reviews

Several reviews in primary or community mental health care have been conducted but none have focused exclusively on the effectiveness of mental healthcare delivery by a non‐specialist workforce. Reviews have covered alternatives to inpatient care but with a focus on specialist outreach services such as specialist child community services (Shepperd 2009), or community‐based rehabilitation (without specifying the workforce) (Robertson 2012). Other studies addressed resource use and primary care provider behaviour with the addition of a mental health resource at primary care level, but did not assess the effect on patient outcomes (Harkness 2009). Certain reviews compared interventions themselves rather than the provider (Abas 2003; Huntley 2012; Wiley‐Exley 2007).

Seven reviews incorporated aspects of interventions that were included in this review (Boer 2005; Bower 2006; Huntley 2012; Parker 2008; Rahman 2013; Tol 2011; Woltmann 2012). Details of agreements and disagreements with these reviews are presented in Table 15.

Open in table viewer
Table 15. Agreements and disagreements with related reviews

Author/year

Summary of review

Agreements

Disagreements/differences

Parker 2008

Reviewed consultation liaison in primary care ‐ HICs

‐

Our review process did not find any consultation liaison in primary care in LMICs so results cannot be compared

Boer 2005

Reviewed paraprofessionals in delivering psychological interventions for anxiety and depression (HIC only)

Included studies were from HICs only, but support our findings that non‐professional care is generally equivalent to professional care (this review's equivalent of specialist care), and that non‐professional care is better than usual care

Some of their paraprofessionals would have been classified as specialist health workers in our review

Bower 2006

Reviewed the effect of collaborative care models on antidepressant use.

All included studies were from HICs except for

Araya 2003 RCT Chile

Bower found improvement of antidepressant use, particularly in studies where the case manager had a mental health background, where there was adequate supervision and where there was systematic identification of patients (rather than waiting for a referral)

We were not able to assess, as did Bower, whether lengths of training, supervision or other intervention characteristics modified these outcomes because only 5 studies were included in this comparison

Woltmann 2012

Review on collaborative care/chronic care management

They also found a statistically significant effect on reduction in depression severity among the 14 HIC studies that were included in the meta‐analysis (SMD 0.31, 95% CI 0.16 to 0.47) (Araya and Patel's studies were included in the narrative review but did not qualify for their meta‐analysis). The authors suggested that collaborative care is of moderate benefit; however, Woltmann has estimated a more conservative value of SMD > 0.5 to show moderate benefit (from the analysis of scales and how to interpret their SMDs). Our meta‐analyses of collaborative care models suggested similar improvements in symptoms and recovery from depression or CMDs  (same direction of effect, and similar magnitude)

Woltman's chronic care management had a stricter definition to our collaborative care definition

Huntley 2012

Reviewed the effect of CBT and group CBT

Huntley also found that LHW‐led psychological interventions are effective in the short and medium term in reducing symptoms of depression

Huntley described the effect of CBT and group CBT (rather than the effect of NSHWs)

Tol 2011

Systematic review on mental health interventions in humanitarian settings

Tol found similar results to our review for school‐based interventions for children with PTSD (i.e. no significant benefit) (an extra study was included in this comparison, which we had excluded as it did not meet our NSHW/OPHR definitions). This review went further and found a statistically significant benefit for improving internalising symptoms (SMD ‐0.34, 95% CI ‐0.40 to ‐0.09). For adults, a potential benefit of interventions was also seen

This review differed from ours in that it included studies of both NSHWs/OPHRs and specialists, according to our definitions

Rahman 2013

Systematic review on interventions for common perinatal mental disorders in women in LMICs

This was similar but a more in‐depth review of our perinatal depression pooled comparison, which also looked at LHW‐led interventions for mothers with perinatal depression. Their final pooled outcome was similar in magnitude and direction to ours for our perinatal depression category (SMD ‐0.38, 95% CI ‐0.56 to ‐0.21) vs, our findings (SMD ‐0.42, 95% CI ‐0.58 to ‐0.26)

This review differed from ours in that its study's inclusion criteria were broader as it included studies that measured maternal (all perinatal disorders) or child (or both) outcomes even if the intervention was not primarily targeted at these groups. It also reported child outcomes, which ours did not

CBT: cognitive behavioural therapy; CI: confidence interval; CMD: common mental disorders; HIC: high‐income country; LHW: lay health worker; LMIC: low‐ and medium‐income countries; NSHW: non‐specialist health worker; OPHR: other professionals with health roles; PTSD: post‐traumatic stress disorder; SMD: standardised mean difference.

Economic studies

Appendix 3 describes other relevant economic studies that were not included in this review. The findings of these studies are similar to those of the three studies (Araya 2003 RCT Chile; Jordans 2011; Zambori 2002 CBA Hungary) included in this review, that is, that NSHW interventions seem cost‐effective, and that these findings are difficult to generalise due to the different healthcare systems in various countries.

Study flow diagram.
Figures and Tables -
Figure 1

Study flow diagram.

Navigate to figure in ReviewOpen in new tab

Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies.
Figures and Tables -
Figure 2

Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies.

Navigate to figure in ReviewOpen in new tab

Risk of bias summary: review authors' judgements about each risk of bias item for each included study.
Figures and Tables -
Figure 3

Risk of bias summary: review authors' judgements about each risk of bias item for each included study.

Navigate to figure in ReviewOpen in new tab

Forest plot of comparison: 1 NSHW‐led psychological interventions versus usual care in treating CMDs in adults (RCTs), outcome: 1.1 Prevalence of depression (adults) (completers).
Figures and Tables -
Figure 4

Forest plot of comparison: 1 NSHW‐led psychological interventions versus usual care in treating CMDs in adults (RCTs), outcome: 1.1 Prevalence of depression (adults) (completers).

Navigate to figure in ReviewOpen in new tab

Forest plot of comparison: 1 NSHW‐led psychological interventions versus usual care in treating common mental disorders in adults (RCTs), outcome: 1.6 Severity of common mental disorder symptoms (includes anxiety and depression).
Figures and Tables -
Figure 5

Forest plot of comparison: 1 NSHW‐led psychological interventions versus usual care in treating common mental disorders in adults (RCTs), outcome: 1.6 Severity of common mental disorder symptoms (includes anxiety and depression).

Navigate to figure in ReviewOpen in new tab

Forest plot of comparison: 2 Collaborative care model (NSHWs plus specialist) versus usual care in treating common mental disorders (CMD) (RCTs), outcome: 2.1 Prevalence of common mental disorders (CMDs ‐ includes anxiety and depression) (completers combined) all facilities and in public and private facilities.
Figures and Tables -
Figure 6

Forest plot of comparison: 2 Collaborative care model (NSHWs plus specialist) versus usual care in treating common mental disorders (CMD) (RCTs), outcome: 2.1 Prevalence of common mental disorders (CMDs ‐ includes anxiety and depression) (completers combined) all facilities and in public and private facilities.

Navigate to figure in ReviewOpen in new tab

Forest plot of comparison: 2 Collaborative care model (NSHWs plus specialist) versus usual care in treating common mental disorders (RCTs), outcome: 2.2 Severity of symptoms of common mental disorders (completers combined) in all facilities and in public and private facilities.
Figures and Tables -
Figure 7

Forest plot of comparison: 2 Collaborative care model (NSHWs plus specialist) versus usual care in treating common mental disorders (RCTs), outcome: 2.2 Severity of symptoms of common mental disorders (completers combined) in all facilities and in public and private facilities.

Navigate to figure in ReviewOpen in new tab

Forest plot of comparison: 3 NSHWs versus usual care in treating maternal depression (RCTs), outcome: 3.1 Severity of symptoms in treating maternal depression.
Figures and Tables -
Figure 8

Forest plot of comparison: 3 NSHWs versus usual care in treating maternal depression (RCTs), outcome: 3.1 Severity of symptoms in treating maternal depression.

Navigate to figure in ReviewOpen in new tab

Forest plot of comparison: 5 NSHW‐led psychological interventions versus usual care in treating adults with post‐traumatic stress disorder (RCT and NRCT), outcome: 5.1 Prevalence of post‐traumatic stress disorder (PTSD).
Figures and Tables -
Figure 9

Forest plot of comparison: 5 NSHW‐led psychological interventions versus usual care in treating adults with post‐traumatic stress disorder (RCT and NRCT), outcome: 5.1 Prevalence of post‐traumatic stress disorder (PTSD).

Navigate to figure in ReviewOpen in new tab

Forest plot of comparison: 5 NSHW‐led psychological interventions versus usual care in treating adults with PTSD (RCT and NRCT), outcome: 5.2 Severity of PTSD symptoms (N = completers).
Figures and Tables -
Figure 10

Forest plot of comparison: 5 NSHW‐led psychological interventions versus usual care in treating adults with PTSD (RCT and NRCT), outcome: 5.2 Severity of PTSD symptoms (N = completers).

Navigate to figure in ReviewOpen in new tab

Forest plot of comparison: 6 NSHWs versus usual care in improving dementia patients' and carers' outcomes (RCTs), outcome: 6.1 Severity of behavioural problem (patient).
Figures and Tables -
Figure 11

Forest plot of comparison: 6 NSHWs versus usual care in improving dementia patients' and carers' outcomes (RCTs), outcome: 6.1 Severity of behavioural problem (patient).

Navigate to figure in ReviewOpen in new tab

Forest plot of comparison: 6 NSHWs versus usual care in improving dementia patients' and carers' outcomes (RCTs), outcome: 6.5 Carer burden.
Figures and Tables -
Figure 12

Forest plot of comparison: 6 NSHWs versus usual care in improving dementia patients' and carers' outcomes (RCTs), outcome: 6.5 Carer burden.

Navigate to figure in ReviewOpen in new tab

Forest plot of comparison: 7 NSHW‐led brief alcohol interventions versus usual care for adults with alcohol‐use disorders (RCTs), outcome: 7.1 Amount of alcohol consumed (MD).
Figures and Tables -
Figure 13

Forest plot of comparison: 7 NSHW‐led brief alcohol interventions versus usual care for adults with alcohol‐use disorders (RCTs), outcome: 7.1 Amount of alcohol consumed (MD).

Navigate to figure in ReviewOpen in new tab

Forest plot of comparison: 9 NSHWs/OPHRs versus usual care in conducting interventions for children with PTSD (RCTs), outcome: 9.3 Severity of PTSD symptoms ‐ teacher‐led interventions (children) (MDs).
Figures and Tables -
Figure 14

Forest plot of comparison: 9 NSHWs/OPHRs versus usual care in conducting interventions for children with PTSD (RCTs), outcome: 9.3 Severity of PTSD symptoms ‐ teacher‐led interventions (children) (MDs).

Navigate to figure in ReviewOpen in new tab

Comparison 1 NSHW‐led psychological interventions versus usual care in treating common mental disorders in adults (RCTs), Outcome 1 Prevalence of depression (completers).
Figures and Tables -
Analysis 1.1

Comparison 1 NSHW‐led psychological interventions versus usual care in treating common mental disorders in adults (RCTs), Outcome 1 Prevalence of depression (completers).

Navigate to figure in ReviewOpen in new tab

Comparison 1 NSHW‐led psychological interventions versus usual care in treating common mental disorders in adults (RCTs), Outcome 2 Prevalence of depression (ITT sensitivity analysis ‐ assumption non‐completers depressed).
Figures and Tables -
Analysis 1.2

Comparison 1 NSHW‐led psychological interventions versus usual care in treating common mental disorders in adults (RCTs), Outcome 2 Prevalence of depression (ITT sensitivity analysis ‐ assumption non‐completers depressed).

Navigate to figure in ReviewOpen in new tab

Comparison 1 NSHW‐led psychological interventions versus usual care in treating common mental disorders in adults (RCTs), Outcome 3 Prevalence of depression (ITT sensitivity analysis ‐ assumption non‐completers not depressed).
Figures and Tables -
Analysis 1.3

Comparison 1 NSHW‐led psychological interventions versus usual care in treating common mental disorders in adults (RCTs), Outcome 3 Prevalence of depression (ITT sensitivity analysis ‐ assumption non‐completers not depressed).

Navigate to figure in ReviewOpen in new tab

Comparison 1 NSHW‐led psychological interventions versus usual care in treating common mental disorders in adults (RCTs), Outcome 4 Prevalence of depression (ITT sensitivity analysis ‐ worse‐case scenario intervention group depressed; control group not depressed).
Figures and Tables -
Analysis 1.4

Comparison 1 NSHW‐led psychological interventions versus usual care in treating common mental disorders in adults (RCTs), Outcome 4 Prevalence of depression (ITT sensitivity analysis ‐ worse‐case scenario intervention group depressed; control group not depressed).

Navigate to figure in ReviewOpen in new tab

Comparison 1 NSHW‐led psychological interventions versus usual care in treating common mental disorders in adults (RCTs), Outcome 5 Prevalence of depression (ITT sensitivity analysis ‐ best‐case scenario: intervention group not depressed; control group all depressed).
Figures and Tables -
Analysis 1.5

Comparison 1 NSHW‐led psychological interventions versus usual care in treating common mental disorders in adults (RCTs), Outcome 5 Prevalence of depression (ITT sensitivity analysis ‐ best‐case scenario: intervention group not depressed; control group all depressed).

Navigate to figure in ReviewOpen in new tab

Comparison 1 NSHW‐led psychological interventions versus usual care in treating common mental disorders in adults (RCTs), Outcome 6 Severity of common mental disorder symptoms (includes anxiety and depression).
Figures and Tables -
Analysis 1.6

Comparison 1 NSHW‐led psychological interventions versus usual care in treating common mental disorders in adults (RCTs), Outcome 6 Severity of common mental disorder symptoms (includes anxiety and depression).

Navigate to figure in ReviewOpen in new tab

Comparison 1 NSHW‐led psychological interventions versus usual care in treating common mental disorders in adults (RCTs), Outcome 7 Functional impairment/disability in common mental disorders.
Figures and Tables -
Analysis 1.7

Comparison 1 NSHW‐led psychological interventions versus usual care in treating common mental disorders in adults (RCTs), Outcome 7 Functional impairment/disability in common mental disorders.

Navigate to figure in ReviewOpen in new tab

Comparison 2 Collaborative care model (NSHWs plus specialist) versus usual care in treating common mental disorders (RCTs), Outcome 1 Prevalence of common mental disorders (CMDs ‐ includes anxiety and depression) (completers‐combined) all facilities and in public and private facilities.
Figures and Tables -
Analysis 2.1

Comparison 2 Collaborative care model (NSHWs plus specialist) versus usual care in treating common mental disorders (RCTs), Outcome 1 Prevalence of common mental disorders (CMDs ‐ includes anxiety and depression) (completers‐combined) all facilities and in public and private facilities.

Navigate to figure in ReviewOpen in new tab

Comparison 2 Collaborative care model (NSHWs plus specialist) versus usual care in treating common mental disorders (RCTs), Outcome 2 Severity of symptoms of CMDs (completers‐combined) in all facilities and in public and private facilities.
Figures and Tables -
Analysis 2.2

Comparison 2 Collaborative care model (NSHWs plus specialist) versus usual care in treating common mental disorders (RCTs), Outcome 2 Severity of symptoms of CMDs (completers‐combined) in all facilities and in public and private facilities.

Navigate to figure in ReviewOpen in new tab

Comparison 2 Collaborative care model (NSHWs plus specialist) versus usual care in treating common mental disorders (RCTs), Outcome 3 Functional impairment/disability in CMD (completers‐ combined) all facilities and in public and private facilities (SMD).
Figures and Tables -
Analysis 2.3

Comparison 2 Collaborative care model (NSHWs plus specialist) versus usual care in treating common mental disorders (RCTs), Outcome 3 Functional impairment/disability in CMD (completers‐ combined) all facilities and in public and private facilities (SMD).

Navigate to figure in ReviewOpen in new tab

Comparison 2 Collaborative care model (NSHWs plus specialist) versus usual care in treating common mental disorders (RCTs), Outcome 4 Suicide attempt for those with CMDs all facilities and in public/private facilities (completers).
Figures and Tables -
Analysis 2.4

Comparison 2 Collaborative care model (NSHWs plus specialist) versus usual care in treating common mental disorders (RCTs), Outcome 4 Suicide attempt for those with CMDs all facilities and in public/private facilities (completers).

Navigate to figure in ReviewOpen in new tab

Comparison 2 Collaborative care model (NSHWs plus specialist) versus usual care in treating common mental disorders (RCTs), Outcome 5 Prevalence of CMDs (only Patel ‐ sensitivity analysis (SA)) (completers) all facilities and in public and private facilities.
Figures and Tables -
Analysis 2.5

Comparison 2 Collaborative care model (NSHWs plus specialist) versus usual care in treating common mental disorders (RCTs), Outcome 5 Prevalence of CMDs (only Patel ‐ sensitivity analysis (SA)) (completers) all facilities and in public and private facilities.

Navigate to figure in ReviewOpen in new tab

Comparison 2 Collaborative care model (NSHWs plus specialist) versus usual care in treating common mental disorders (RCTs), Outcome 6 Severity of symptoms in CMD (only Patel and Jenkins (SA)) in all facilities and in public and private facilities.
Figures and Tables -
Analysis 2.6

Comparison 2 Collaborative care model (NSHWs plus specialist) versus usual care in treating common mental disorders (RCTs), Outcome 6 Severity of symptoms in CMD (only Patel and Jenkins (SA)) in all facilities and in public and private facilities.

Navigate to figure in ReviewOpen in new tab

Comparison 2 Collaborative care model (NSHWs plus specialist) versus usual care in treating common mental disorders (RCTs), Outcome 7 Prevalence of depression (completers) (SA) all facilities and in public and private facilities.
Figures and Tables -
Analysis 2.7

Comparison 2 Collaborative care model (NSHWs plus specialist) versus usual care in treating common mental disorders (RCTs), Outcome 7 Prevalence of depression (completers) (SA) all facilities and in public and private facilities.

Navigate to figure in ReviewOpen in new tab

Comparison 2 Collaborative care model (NSHWs plus specialist) versus usual care in treating common mental disorders (RCTs), Outcome 8 Severity of symptoms of depression (SA) in all facilities and in public and private facilities.
Figures and Tables -
Analysis 2.8

Comparison 2 Collaborative care model (NSHWs plus specialist) versus usual care in treating common mental disorders (RCTs), Outcome 8 Severity of symptoms of depression (SA) in all facilities and in public and private facilities.

Navigate to figure in ReviewOpen in new tab

Comparison 2 Collaborative care model (NSHWs plus specialist) versus usual care in treating common mental disorders (RCTs), Outcome 9 Functional impairment/disability in CMD (SA) all facilities and in public and private facilities (SMD).
Figures and Tables -
Analysis 2.9

Comparison 2 Collaborative care model (NSHWs plus specialist) versus usual care in treating common mental disorders (RCTs), Outcome 9 Functional impairment/disability in CMD (SA) all facilities and in public and private facilities (SMD).

Navigate to figure in ReviewOpen in new tab

Comparison 2 Collaborative care model (NSHWs plus specialist) versus usual care in treating common mental disorders (RCTs), Outcome 10 Functional impairment/disability in CMD (SA) all facilities and in public and private facilities (MD).
Figures and Tables -
Analysis 2.10

Comparison 2 Collaborative care model (NSHWs plus specialist) versus usual care in treating common mental disorders (RCTs), Outcome 10 Functional impairment/disability in CMD (SA) all facilities and in public and private facilities (MD).

Navigate to figure in ReviewOpen in new tab

Comparison 2 Collaborative care model (NSHWs plus specialist) versus usual care in treating common mental disorders (RCTs), Outcome 11 Functional impairment/disability in depression (SA) all facilities and in public and private facilities.
Figures and Tables -
Analysis 2.11

Comparison 2 Collaborative care model (NSHWs plus specialist) versus usual care in treating common mental disorders (RCTs), Outcome 11 Functional impairment/disability in depression (SA) all facilities and in public and private facilities.

Navigate to figure in ReviewOpen in new tab

Comparison 3 NSHWs versus usual care in treating maternal depression (RCTs), Outcome 1 Severity of symptoms in treating maternal depression.
Figures and Tables -
Analysis 3.1

Comparison 3 NSHWs versus usual care in treating maternal depression (RCTs), Outcome 1 Severity of symptoms in treating maternal depression.

Navigate to figure in ReviewOpen in new tab

Comparison 4 NSHWs versus specialists in treating depression in adults (controlled before‐and‐after studies), Outcome 1 Severity of depression short term (2 months post intervention).
Figures and Tables -
Analysis 4.1

Comparison 4 NSHWs versus specialists in treating depression in adults (controlled before‐and‐after studies), Outcome 1 Severity of depression short term (2 months post intervention).

Navigate to figure in ReviewOpen in new tab

Comparison 4 NSHWs versus specialists in treating depression in adults (controlled before‐and‐after studies), Outcome 2 Frequency of adverse events.
Figures and Tables -
Analysis 4.2

Comparison 4 NSHWs versus specialists in treating depression in adults (controlled before‐and‐after studies), Outcome 2 Frequency of adverse events.

Navigate to figure in ReviewOpen in new tab

Comparison 4 NSHWs versus specialists in treating depression in adults (controlled before‐and‐after studies), Outcome 3 Number of days spent in hospital.
Figures and Tables -
Analysis 4.3

Comparison 4 NSHWs versus specialists in treating depression in adults (controlled before‐and‐after studies), Outcome 3 Number of days spent in hospital.

Navigate to figure in ReviewOpen in new tab

Comparison 4 NSHWs versus specialists in treating depression in adults (controlled before‐and‐after studies), Outcome 4 Number of days spent on sick leave.
Figures and Tables -
Analysis 4.4

Comparison 4 NSHWs versus specialists in treating depression in adults (controlled before‐and‐after studies), Outcome 4 Number of days spent on sick leave.

Navigate to figure in ReviewOpen in new tab

Comparison 5 NSHW‐led psychological interventions versus usual care in treating adults with post‐traumatic stress disorder (RCT and NRCT), Outcome 1 Prevalence of post‐traumatic stress disorder (PTSD).
Figures and Tables -
Analysis 5.1

Comparison 5 NSHW‐led psychological interventions versus usual care in treating adults with post‐traumatic stress disorder (RCT and NRCT), Outcome 1 Prevalence of post‐traumatic stress disorder (PTSD).

Navigate to figure in ReviewOpen in new tab

Comparison 5 NSHW‐led psychological interventions versus usual care in treating adults with post‐traumatic stress disorder (RCT and NRCT), Outcome 2 Severity of PTSD symptoms.
Figures and Tables -
Analysis 5.2

Comparison 5 NSHW‐led psychological interventions versus usual care in treating adults with post‐traumatic stress disorder (RCT and NRCT), Outcome 2 Severity of PTSD symptoms.

Navigate to figure in ReviewOpen in new tab

Comparison 5 NSHW‐led psychological interventions versus usual care in treating adults with post‐traumatic stress disorder (RCT and NRCT), Outcome 3 Severity of depression.
Figures and Tables -
Analysis 5.3

Comparison 5 NSHW‐led psychological interventions versus usual care in treating adults with post‐traumatic stress disorder (RCT and NRCT), Outcome 3 Severity of depression.

Navigate to figure in ReviewOpen in new tab

Comparison 6 NSHWs versus usual care in improving dementia patients' and carers' outcomes (RCTs), Outcome 1 Severity of behavioural problem (patient).
Figures and Tables -
Analysis 6.1

Comparison 6 NSHWs versus usual care in improving dementia patients' and carers' outcomes (RCTs), Outcome 1 Severity of behavioural problem (patient).

Navigate to figure in ReviewOpen in new tab

Comparison 6 NSHWs versus usual care in improving dementia patients' and carers' outcomes (RCTs), Outcome 2 Patient functional ability.
Figures and Tables -
Analysis 6.2

Comparison 6 NSHWs versus usual care in improving dementia patients' and carers' outcomes (RCTs), Outcome 2 Patient functional ability.

Navigate to figure in ReviewOpen in new tab

Comparison 6 NSHWs versus usual care in improving dementia patients' and carers' outcomes (RCTs), Outcome 3 Patient quality of life.
Figures and Tables -
Analysis 6.3

Comparison 6 NSHWs versus usual care in improving dementia patients' and carers' outcomes (RCTs), Outcome 3 Patient quality of life.

Navigate to figure in ReviewOpen in new tab

Comparison 6 NSHWs versus usual care in improving dementia patients' and carers' outcomes (RCTs), Outcome 4 Carer mental health status.
Figures and Tables -
Analysis 6.4

Comparison 6 NSHWs versus usual care in improving dementia patients' and carers' outcomes (RCTs), Outcome 4 Carer mental health status.

Navigate to figure in ReviewOpen in new tab

Comparison 6 NSHWs versus usual care in improving dementia patients' and carers' outcomes (RCTs), Outcome 5 Carer burden.
Figures and Tables -
Analysis 6.5

Comparison 6 NSHWs versus usual care in improving dementia patients' and carers' outcomes (RCTs), Outcome 5 Carer burden.

Navigate to figure in ReviewOpen in new tab

Comparison 6 NSHWs versus usual care in improving dementia patients' and carers' outcomes (RCTs), Outcome 6 Carer distress.
Figures and Tables -
Analysis 6.6

Comparison 6 NSHWs versus usual care in improving dementia patients' and carers' outcomes (RCTs), Outcome 6 Carer distress.

Navigate to figure in ReviewOpen in new tab

Comparison 6 NSHWs versus usual care in improving dementia patients' and carers' outcomes (RCTs), Outcome 7 Carer quality of life.
Figures and Tables -
Analysis 6.7

Comparison 6 NSHWs versus usual care in improving dementia patients' and carers' outcomes (RCTs), Outcome 7 Carer quality of life.

Navigate to figure in ReviewOpen in new tab

Comparison 7 NSHW‐led brief alcohol interventions versus usual care for adults with alcohol‐use disorders (RCTs), Outcome 1 Amount of alcohol consumed (MD).
Figures and Tables -
Analysis 7.1

Comparison 7 NSHW‐led brief alcohol interventions versus usual care for adults with alcohol‐use disorders (RCTs), Outcome 1 Amount of alcohol consumed (MD).

Navigate to figure in ReviewOpen in new tab

Comparison 7 NSHW‐led brief alcohol interventions versus usual care for adults with alcohol‐use disorders (RCTs), Outcome 2 Frequency of binge drinking.
Figures and Tables -
Analysis 7.2

Comparison 7 NSHW‐led brief alcohol interventions versus usual care for adults with alcohol‐use disorders (RCTs), Outcome 2 Frequency of binge drinking.

Navigate to figure in ReviewOpen in new tab

Comparison 7 NSHW‐led brief alcohol interventions versus usual care for adults with alcohol‐use disorders (RCTs), Outcome 3 Adverse consequences.
Figures and Tables -
Analysis 7.3

Comparison 7 NSHW‐led brief alcohol interventions versus usual care for adults with alcohol‐use disorders (RCTs), Outcome 3 Adverse consequences.

Navigate to figure in ReviewOpen in new tab

Comparison 8 NSHWs/OPHRs versus usual care in conducting interventions for children with post‐traumatic stress and depression (RCTs), Outcome 1 Severity of PTSD symptoms ‐ teacher/LHW‐led interventions (SMDs).
Figures and Tables -
Analysis 8.1

Comparison 8 NSHWs/OPHRs versus usual care in conducting interventions for children with post‐traumatic stress and depression (RCTs), Outcome 1 Severity of PTSD symptoms ‐ teacher/LHW‐led interventions (SMDs).

Navigate to figure in ReviewOpen in new tab

Comparison 8 NSHWs/OPHRs versus usual care in conducting interventions for children with post‐traumatic stress and depression (RCTs), Outcome 2 Severity of PTSD symptoms ‐ classroom‐based LHW interventions (MCDs).
Figures and Tables -
Analysis 8.2

Comparison 8 NSHWs/OPHRs versus usual care in conducting interventions for children with post‐traumatic stress and depression (RCTs), Outcome 2 Severity of PTSD symptoms ‐ classroom‐based LHW interventions (MCDs).

Navigate to figure in ReviewOpen in new tab

Comparison 8 NSHWs/OPHRs versus usual care in conducting interventions for children with post‐traumatic stress and depression (RCTs), Outcome 3 Severity of PTSD symptoms ‐ classroom‐based LHW interventions ‐ boys/girls.
Figures and Tables -
Analysis 8.3

Comparison 8 NSHWs/OPHRs versus usual care in conducting interventions for children with post‐traumatic stress and depression (RCTs), Outcome 3 Severity of PTSD symptoms ‐ classroom‐based LHW interventions ‐ boys/girls.

Navigate to figure in ReviewOpen in new tab

Comparison 8 NSHWs/OPHRs versus usual care in conducting interventions for children with post‐traumatic stress and depression (RCTs), Outcome 4 Severity of depressive symptoms ‐ teacher/LHW‐led interventions (SMDs).
Figures and Tables -
Analysis 8.4

Comparison 8 NSHWs/OPHRs versus usual care in conducting interventions for children with post‐traumatic stress and depression (RCTs), Outcome 4 Severity of depressive symptoms ‐ teacher/LHW‐led interventions (SMDs).

Navigate to figure in ReviewOpen in new tab

Comparison 8 NSHWs/OPHRs versus usual care in conducting interventions for children with post‐traumatic stress and depression (RCTs), Outcome 5 Severity of depressive symptoms ‐ classroom‐based LHW interventions (MCDs).
Figures and Tables -
Analysis 8.5

Comparison 8 NSHWs/OPHRs versus usual care in conducting interventions for children with post‐traumatic stress and depression (RCTs), Outcome 5 Severity of depressive symptoms ‐ classroom‐based LHW interventions (MCDs).

Navigate to figure in ReviewOpen in new tab

Comparison 8 NSHWs/OPHRs versus usual care in conducting interventions for children with post‐traumatic stress and depression (RCTs), Outcome 6 Severity of depressive symptoms (MCDs) Tol 2012 boys/girls.
Figures and Tables -
Analysis 8.6

Comparison 8 NSHWs/OPHRs versus usual care in conducting interventions for children with post‐traumatic stress and depression (RCTs), Outcome 6 Severity of depressive symptoms (MCDs) Tol 2012 boys/girls.

Navigate to figure in ReviewOpen in new tab

Comparison 8 NSHWs/OPHRs versus usual care in conducting interventions for children with post‐traumatic stress and depression (RCTs), Outcome 7 Severity of anxiety symptoms ‐ classroom‐based intervention (within 6 months post intervention).
Figures and Tables -
Analysis 8.7

Comparison 8 NSHWs/OPHRs versus usual care in conducting interventions for children with post‐traumatic stress and depression (RCTs), Outcome 7 Severity of anxiety symptoms ‐ classroom‐based intervention (within 6 months post intervention).

Navigate to figure in ReviewOpen in new tab

Comparison 8 NSHWs/OPHRs versus usual care in conducting interventions for children with post‐traumatic stress and depression (RCTs), Outcome 8 Severity of anxiety symptoms ‐ classroom‐based intervention ‐ boys/girls.
Figures and Tables -
Analysis 8.8

Comparison 8 NSHWs/OPHRs versus usual care in conducting interventions for children with post‐traumatic stress and depression (RCTs), Outcome 8 Severity of anxiety symptoms ‐ classroom‐based intervention ‐ boys/girls.

Navigate to figure in ReviewOpen in new tab

Comparison 8 NSHWs/OPHRs versus usual care in conducting interventions for children with post‐traumatic stress and depression (RCTs), Outcome 9 Functional impairment teacher/LHW‐led interventions.
Figures and Tables -
Analysis 8.9

Comparison 8 NSHWs/OPHRs versus usual care in conducting interventions for children with post‐traumatic stress and depression (RCTs), Outcome 9 Functional impairment teacher/LHW‐led interventions.

Navigate to figure in ReviewOpen in new tab

Comparison 8 NSHWs/OPHRs versus usual care in conducting interventions for children with post‐traumatic stress and depression (RCTs), Outcome 10 Functional impairment LHW‐led ‐ classroom‐based intervention.
Figures and Tables -
Analysis 8.10

Comparison 8 NSHWs/OPHRs versus usual care in conducting interventions for children with post‐traumatic stress and depression (RCTs), Outcome 10 Functional impairment LHW‐led ‐ classroom‐based intervention.

Navigate to figure in ReviewOpen in new tab

Comparison 8 NSHWs/OPHRs versus usual care in conducting interventions for children with post‐traumatic stress and depression (RCTs), Outcome 11 Functional impairment ‐ classroom‐based LHW intervention ‐ boys/girls.
Figures and Tables -
Analysis 8.11

Comparison 8 NSHWs/OPHRs versus usual care in conducting interventions for children with post‐traumatic stress and depression (RCTs), Outcome 11 Functional impairment ‐ classroom‐based LHW intervention ‐ boys/girls.

Navigate to figure in ReviewOpen in new tab
Summary of findings for the main comparison. NSHW‐led psychological interventions compared with usual care in treating depression in adults in low‐ and middle‐income countries (RCTs)

What are the effects of NSHW‐led psychological interventions for treating depression in adults in low‐ and middle‐income countries?

Patient or population: Adults with depression
Settings: Low‐ and middle‐income countries (Taiwan, Pakistan, Uganda)
Intervention: NSHWs conducting psychological interventions
Comparison: Usual care

Outcomes

Illustrative comparative risks* (95% CI)

Effect estimate
(95% CI)

No of Participants
(studies)

Quality of the evidence
(GRADE)

Comments

Assumed risk

Corresponding risk

Usual care

NSHWs

Prevalence of depression (adults), short term (0‐8 weeks)
measured using various depression rating scales1

300 per 1000

91 per 1000

RR 0.30
(0.14 to 0.64)

1082
(3 studies)

⊕⊕⊝⊝
low2,3

‐

*The basis for the assumed risk is the mean control group risk across studies for pooled results and the control group risk for single studies. The corresponding risk (and its 95% CI) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).
CI: confidence interval; DSM: Diagnostic and Statistical Manual of Mental Disorders; NSHW: non‐specialist health worker; RCT: randomised controlled trial; RR: risk ratio.

GRADE Working Group grades of evidence
High quality: Further research is very unlikely to change our confidence in the estimate of effect.
Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.
Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.
Very low quality: We are very uncertain about the estimate.

1Bolton 2003 C‐RCT Uganda: DSM‐IV criteria A, C and E; Rahman 2008 CRCT Pakistan: Hamilton Depression Rating scale; Chen 2000 RCT Taiwan: Taiwanese Beck Depression Inventory.

2 Serious study limitations: Two of the three studies were at risk of bias. Bolton 2003 C‐RCT Uganda was judged unclear for allocation concealment, and quasi‐randomisation of individuals within clusters (though randomisation was in clusters) could have introduced bias; Chen 2000 RCT Taiwan was unclear for sequence generation and allocation concealment, all outcomes were self reported, there was possible contamination and the dropout rate after randomisation was high, with no analysis of differences in dropouts versus non‐dropouts. These two studies contributed 62% of the weight in the pooled analysis. Downgraded by 1.

3Serious inconsistency: I2 was 81%. However, the inconsistency related to the magnitude of benefit favouring collaborative care rather than in the direction of effect. Downgraded by 1.

Figures and Tables -
Summary of findings for the main comparison. NSHW‐led psychological interventions compared with usual care in treating depression in adults in low‐ and middle‐income countries (RCTs)
Navigate to table in Review
Summary of findings 2. Collaborative care model (NSHWs plus specialist) compared with usual care in treating common mental disorders in adults in low‐ and middle‐income countries (RCTs)

What are the effects of a collaborative care model (NSHW plus specialist supervision) for mental health care in adults with common mental disorders low‐ and middle‐income countries?

Patient or population: Adults (≥ 18 years) with CMDs (includes anxiety or depression, or both)
Settings: Middle‐income countries (Chile, India)
Intervention: Collaborative care model (NSHW plus specialist supervision)

Comparison: Enhanced usual care

Outcomes

Illustrative comparative risks* (95% CI)

Effect estimate
(95% CI)

No of Participants
(studies)

Quality of the evidence
(GRADE)

Comments

Assumed risk

Corresponding risk

Usual care

Collaborative care model

Prevalence of CMDs, short term (2‐6 months)
measured using various CMD/depression rating scales1

205 per 1000

140 per 1000

RR 0.63
(0.44 to 0.90)

2380
(3 studies)

⊕⊕⊝⊝
low2,3

In Patel 2010 C‐RCT India; collaborative care reduced the prevalence of CMDs at 6 months in a subgroup of people treated at public health facilities (RR 0.57, 95% CI 0.42 to 0.78; 1528 participants). This effect was not seen in people treated at private facilities (RR 1.12, 95% CI 0.68 to 1.84; 823 participants)

*The basis for the assumed risk is the mean control group risk across studies for pooled results and the control group risk for single studies. The corresponding risk (and its 95% CI) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).
CI: confidence interval; CIS: Clinical Interview Schedule; CMD: common mental disorder; EPDS: Edinburgh Postnatal Depression Scale; GP: general practitioner; HDRS: Hamilton Depression Rating Scale; ICD: International Classification of Diseases; NSHW: non‐specialist health worker; RCT: randomised controlled trial; RR: risk ratio.

GRADE Working Group grades of evidence
High quality: Further research is very unlikely to change our confidence in the estimate of effect.
Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.
Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.
Very low quality: We are very uncertain about the estimate.

1Araya 2003 RCT Chile: HDRS; Patel 2010 C‐RCT India: CIS‐R generated ICD‐10 diagnosis for CMD; Rojas 2007 RCT Chile: EPDS with a 6‐point reduction in score indicating recovery.

2Serious study limitations: In Araya 2003 RCT Chile, GPs provided both intervention and control treatments, so there was a high risk of contamination. Downgraded by 1.

3Serious inconsistency: I2 was 79% with Araya 2003 RCT Chile clearly an outlier, contributing to this unexplained inconsistency. However, the inconsistency related to the magnitude of benefit favouring collaborative care rather than in the direction of effect. Downgraded by 1.

Figures and Tables -
Summary of findings 2. Collaborative care model (NSHWs plus specialist) compared with usual care in treating common mental disorders in adults in low‐ and middle‐income countries (RCTs)
Navigate to table in Review
Summary of findings 3. NSHWs compared with usual care for treating maternal depression (RCTs)

What are the effects of NSHW‐led interventions for treating maternal depression in low‐ and middle‐income countries?

Patient or population: Adult women with maternal depression
Settings: Low‐ and middle‐income countries (Chile, Jamaica, Pakistan, Taiwan)
Intervention: NSHW‐led interventions
Comparison: Usual care

Outcomes

Illustrative comparative risks* (95% CI)

Estimate effect
(95% CI)

No of Participants
(studies)

Quality of the evidence
(GRADE)

Comments

Assumed risk

Corresponding risk

Usual care

NSHWs

Severity of symptoms of perinatal depression, (short and medium term: 0‐12 months)
measured using various depression rating scales1

‐

The mean severity of symptoms of perinatal depression ‐ medium term with NSHW‐led interventions was
0.42 standard deviations lower
(0.58 to 0.26 lower)

SMD ‐0.42 (‐0.58 to ‐0.26)

1213
(4 studies)

⊕⊕⊝⊝
low2,3

Note that a small clinically appreciable benefit was set at SMD < 0.2, and a moderate benefit at SMD of 0.5 to 0.8 (Cohen 1988)

*The basis for the assumed risk is the mean control group risk across studies for pooled results and the control group risk for single studies. The corresponding risk (and its 95% CI) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).
BDI: Becks Depression Inventory; CES‐D: Center for Epidemiologic Studies Depression Scale; CI: confidence interval; EPDS: Edinburgh Postnatal Depression Scale; HDRS: Hamilton Depression Rating Scale; NSHW: non‐specialist health worker; RCT: randomised controlled trial; SMD: standardised mean difference.

GRADE Working Group grades of evidence
High quality: Further research is very unlikely to change our confidence in the estimate of effect.
Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.
Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.
Very low quality: We are very uncertain about the estimate.

1Baker‐H 2005 CRCT Jamaica CES‐D; Chen 2000 RCT Taiwan Taiwanese BDI; Rahman 2008 CRCT Pakistan: HDRS; Rojas 2007 RCT Chile: EPDS.

2 Serious study limitations: Baker‐H 2005 CRCT Jamaica; Chen 2000 RCT Taiwan has study limitations and together contributed 24% weight to the pooled estimates. Removal of these trials altered the results to favour NSHW‐led interventions strongly. Downgraded by 1.

3 Serious imprecision: The 95% CI of the SMD indicated appreciable and non‐appreciable benefit for NSHW‐led interventions. Downgarded by 1.

Figures and Tables -
Summary of findings 3. NSHWs compared with usual care for treating maternal depression (RCTs)
Navigate to table in Review
Summary of findings 4. NSHWs compared with specialists in treating depression in adults in low‐ and middle‐income countries (CBAs)

What are the effects of NSHWs compared with specialists in treating depression for mental health care in low‐ and middle‐income countries?

Patient or population: Adults with depression
Settings: Middle‐income countries (Hungary and Argentina)

Intervention: NSHWs providing pharmacological intervention
Comparison: Specialists providing pharmacological intervention

Outcomes

Illustrative comparative risks* (95% CI)

Relative effect
(95% CI)

No of Participants
(studies)

Quality of the evidence
(GRADE)

Comments

Assumed risk

Corresponding risk

Specialists

NSHWs

Severity of depression, short term (0‐56 days)
measured using HDRS
Follow‐up: 56 days

The mean score (SD) on the HDRS was 9.6 (2.1)

The mean severity of depression ‐ short term (2 months post intervention) in the NSHW group was
0.9 lower
(1.2 to 0.6 lower)

MD ‐0.90 (‐1.20 to ‐0.60)

768
(1 study)

⊕⊝⊝⊝
very low1,2

Note that a small clinically appreciable benefit was set at SMD < 0.2, and a moderate benefit at SMD of 0.5 to 0.8 (Cohen 1988)

*The basis for the assumed risk is the risk in the control group. The corresponding risk (and its 95% CI) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).
CBA: controlled before‐and‐after; CI: confidence interval; HDRS: Hamilton Depression Rating Scale; MD: mean difference; NSHW: non‐specialist health worker; RR: risk ratio; SD: standard difference; SMD: standardised mean difference.

GRADE Working Group grades of evidence
High quality: Further research is very unlikely to change our confidence in the estimate of effect.
Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.
Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.
Very low quality: We are very uncertain about the estimate.

1 Very serious study limitations: Lyketsos1999CBA Argentina was a CBA study so selection bias was likely. There was a risk of contamination and outcome assessments were done by same physicians doing the intervention. Downgraded by 2.

2 Serious imprecision: The MD on the HDRS was < 1 point and this is not clinically a meaningful difference on the HDRS; and the 95% CI of the MD indicated only non‐appreciable benefits with NSHW intervention versus specialist intervention. However, the data came from only one study, so estimate is imprecise. Downgraded by 1.

Figures and Tables -
Summary of findings 4. NSHWs compared with specialists in treating depression in adults in low‐ and middle‐income countries (CBAs)
Navigate to table in Review
Summary of findings 5. NSHW‐led psychological interventions compared with usual care in treating adults with PTSD (NRCT)

What are the effects of NSHWs compared with usual mental health care in low‐ and middle‐income countries for data from an NRCT in adults with PTSD?

Patient or population: Adults with PTSD

Settings: Low‐ and middle‐income countries (Bosnia, Burundi, Uganda)
Intervention: NSHWs and OPHRs delivering psychological interventions (narrative exposure therapy, trauma counselling and workshops with psychoeducation)
Comparison: Usual care

Outcomes

Illustrative comparative risks* (95% CI)

Relative effect
(95% CI)

No of Participants
(studies)

Quality of the evidence
(GRADE)

Comments

Assumed risk

Corresponding risk

Usual care

NSHWs/OPHRs

Severity of PTSD symptoms in LHW/teacher‐led psychological interventions (trauma counselling, workshop with psychoeducation, mother intervention) in the short term (2 weeks to 6 months)
measured using various PTSD symptom scales1

The mean severity of PTSD with psychological interventions in the short term (within 6 months post‐intervention) was
0.36 standard deviations lower
(0.67 to 0.05 lower)

SMD ‐0.36 (‐0.67 to ‐0.05)

223
(3 studies)

⊕⊕⊝⊝
low2,3

*The basis for the assumed risk is the median control group risk or mean control group risk across studies for pooled estimates and the control group risk for single studies. The corresponding risk (and its 95% CI) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).
CI: confidence interval; LHW: lay health workers; NRCT: non‐randomised controlled trial; NSHW: non‐specialist health worker; OPHR: other professionals with health roles; PTSD: post‐traumatic stress disorder; SMD: standardised mean difference.

GRADE Working Group grades of evidence
High quality: Further research is very unlikely to change our confidence in the estimate of effect.
Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.
Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.
Very low quality: We are very uncertain about the estimate.

1Neuner 2008 NRCT Uganda: Post‐traumatic Stress Diagnostic Scale; Yeomans 2010 RCT Burundi: Harvard Trauma Questionnaire; Dybdahl 2001 RCT Bosnia: Impact of Events Scale.

2Serious study limitations: Neuner 2008 NRCT Uganda no allocation concealment, randomisation had no sequence generation. High dropout rate and different between groups, different baseline characteristics and likely contamination; Yeomans 2010 RCT Burundi: unvalidated Harvard Trauma Questionnaire in the local context (only validated in Burundi) so may affect reliability of outcomes. Dybdahl 2001 RCT Bosnia: incomplete outcome reporting, Impact of Events Scale not previously validated in this setting. Downgraded by 1.
3Serious imprecision: The 95% CI of the effect estimates demonstrated appreciable and non‐appreciable benefit with NSHW care. Downgraded by 1.

Figures and Tables -
Summary of findings 5. NSHW‐led psychological interventions compared with usual care in treating adults with PTSD (NRCT)
Navigate to table in Review
Summary of findings 6. NSHWs compared with usual care in improving dementia patients' and carers' outcomes in low‐ and middle‐income countries (RCTs)

What are the effects of NSHW‐led care in improving dementia patients' and carers' outcomes for mental health care in low‐ and middle‐income countries?

Patient or population: People with dementia and their carers
Settings: Middle‐income countries (India, Russia)
Intervention: NSHWs delivering brief intervention
Comparison: Usual care

Outcomes

Illustrative comparative risks* (95% CI)

Estimate effect
(95% CI)

No of Participants
(studies)

Quality of the evidence
(GRADE)

Comments

Assumed risk

Corresponding risk

Usual care

NSHWs

Severity of patient behavioural problems, short term (6 months)
measured using the behavioural symptom scale (NPI‐S)

The mean severity of patient behavioural problems with this brief carer intervention was
0.26 standard deviations lower
(0.60 lower to 0.08 higher)

SMD ‐0.26 (‐0.60 to 0.08)

134
(2 studies)

⊕⊕⊕⊝
moderate1,2

Note that a small clinically appreciable benefit was set at SMD < 0.2, and a moderate benefit at SMD of 0.5‐0.8 (Cohen 1988)

*The basis for the assumed risk is the mean control group risk across studies for pooled results and the control group risk for single studies. The corresponding risk (and its 95% CI) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).
CI: confidence interval; NPI‐S: Neuropsychiatric Inventory ‐ Severity; NSHW: non‐specialist health worker; RCT: randomised controlled trial; SMD: standardised mean difference.

GRADE Working Group grades of evidence
High quality: Further research is very unlikely to change our confidence in the estimate of effect.
Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.
Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.
Very low quality: We are very uncertain about the estimate.

1 No serious study limitations: Gavrilova 2009 RCT Russia was unclear whether allocation concealed. Dias 2008 RCT India was at low risk of bias and contributed > 60% of the weight to the pooled estimates. Removal of the former study did not alter the results. Not downgraded.
2 Serious imprecision: The 95% CI for the pooled estimates indicates appreciable benefit for NSHW care and non‐appreciable benefit for usual care. Downgraded by 1.

Figures and Tables -
Summary of findings 6. NSHWs compared with usual care in improving dementia patients' and carers' outcomes in low‐ and middle‐income countries (RCTs)
Navigate to table in Review
Summary of findings 7. NSHW‐led brief alcohol interventions compared with usual care for adults with alcohol‐use disorders (RCTs)

What are the effects of NSHWs in delivering brief alcohol interventions in RCTs for alcohol‐use disorders?

Patient or population: People with alcohol‐use disorders
Settings: Low‐ and middle‐income countries (Thailand, Kenya)
Intervention: NSHWs in delivering brief alcohol interventions
Comparison: Usual care

Outcomes

Illustrative comparative risks* (95% CI)

Relative effect
(95% CI)

No of Participants
(studies)

Quality of the evidence
(GRADE)

Comments

Assumed risk

Corresponding risk

Usual care

NSHWs

Amount of alcohol consumed, short term (3‐6 months)
measured using the number of drinks/drinking day (in past week to 30 days)

The mean amount of alcohol consumed in the intervention groups was 1.68 lower (2.79 lower to 0.57 lower)

MD ‐1.68 (‐2.79 to ‐0.57)

167
(2 studies)

⊕⊕⊝⊝
low1,2

*The basis for the assumed risk is the mean control group risk across studies for pooled data or the control group risk for individual studies. The corresponding risk (and its 95% CI) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).
CI: confidence interval; MD: mean difference; NSHW: non‐specialist health worker; RCT: randomised controlled trial; RR: risk ratio.

GRADE Working Group grades of evidence
High quality: Further research is very unlikely to change our confidence in the estimate of effect.
Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.
Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.
Very low quality: We are very uncertain about the estimate.

1 Serious study limitations: Noknoy 2010 RCT Thailand: high dropout rate with no information on whether they are different to completers, no validated tools in the setting, so unreliable primary outcomes. Papas 2011 RCT Kenya: unclear about whether the non‐blinding of outcome assessors would have impacted on study. Downgraded by 1.
2 Serious imprecision: The 95% CI of the MD in number of drinks indicates marginal benefit and no appreciable benefit with interventions. The sample size was also low. Downgraded by 1.

Figures and Tables -
Summary of findings 7. NSHW‐led brief alcohol interventions compared with usual care for adults with alcohol‐use disorders (RCTs)
Navigate to table in Review
Summary of findings 8. NSHWs/OPHRs compared with usual care in conducting interventions for children with post‐traumatic stress disorder and depression (RCTs)

What are the effects of NSHWs/OPHRs conducting interventions for children with PTSD from RCTs in low‐ and middle‐income countries?

Patient or population: Children/adolescents with PTSD and related depressive/anxiety symptoms
Settings: Low‐ and middle‐income countries (Bosnia, Kosovo, Sri Lanka)
Intervention: NSHWs/OPHRs delivering psychological and psychosocial interventions

Comparison: Usual care

Outcomes

Illustrative comparative risks* (95% CI)

Estimate effect
(95% CI)

No of Participants
(studies)

Quality of the evidence
(GRADE)

Comments

Assumed risk

Corresponding risk

Usual care

NSHWs/OPHRs

Severity of PTSD symptoms in LHW/teacher‐led interventions, short term (1‐6 months)
measured using various PTSD severity of symptom scales1

The mean severity of PTSD symptoms in children in teacher‐led intervention groups was
1.2 standard deviations lower
(1.52 to 0.88 lower)

SMD ‐0.89 (‐1.49 to ‐0.30)

298
(3 studies)

⊕⊝⊝⊝
very low2,3

Note that a small clinically appreciable benefit was set at SMD < 0.2, a moderate benefit at SMD of 0.5‐0.8, and a large benefit > 0.8 (Cohen 1988)

*The basis for the assumed risk the mean control group risk across studies for pooled results and the control group risk for single studies. The corresponding risk (and its 95% CI) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).
CI: confidence interval; LHW: lay health workers; NSHW: non‐specialist health worker; OPHR: other professionals with health roles; PTSD: post‐traumatic stress disorder; RCT: randomised controlled trial; SMD: standardised mean difference; UCLA: University of California at Los Angeles.

GRADE Working Group grades of evidence
High quality: Further research is very unlikely to change our confidence in the estimate of effect.
Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.
Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.
Very low quality: We are very uncertain about the estimate.

1Berger2009 CRCT SriLanka: UCLA PTSD scale; Gordon 2008 RCT Kosovo: Harvard Trauma Questionnaire; Ertl 2011 RCT Uganda: Clinician‐administered PTSD scale (CAPS).
2 Very serious study limitations: Gordon 2008 RCT Kosovo no allocation concealment, also likely contamination, and no blinding of outcome assessments; Berger2009 CRCT SriLanka no allocation concealment, likely contamination and outcomes not adjusted for clustering. Two of the three trials are at risk of bias and contribute to > 60% weight to the pooled results. Downgraded by 2.
3 Serious inconsistency: I2 = 78%. The inconsistency is not related to the direction of effect. Downgraded by 1.

Figures and Tables -
Summary of findings 8. NSHWs/OPHRs compared with usual care in conducting interventions for children with post‐traumatic stress disorder and depression (RCTs)
Navigate to table in Review
Table 1. Definitions

Adult

Patients who were ≥ 18 years old. However, if some studies had an age range from, for example, 16 years upwards and the majority of participants are over 18 years, we included these study participants as adults.

Children and adolescents

Children (from birth to 18 years) were considered as a separate group of participants as they have 1. different patterns of psychopathology/mental disorders; 2. different help‐seeking behaviours that would, therefore, require different interventions, in different settings (e.g. schools) and a different approach to careworker interventions (such as teacher‐led interventions).

Mental, neurological and substance‐abuse (MNS) disorders

This review included MNS disorders as defined by any criteria within included papers. For the purpose of subgroup analysis, we subcategorised these disorders using the International Classification of Diseases (ICD)‐10 criteria for mental and behavioural disorders and epilepsy in adults (the related ICD‐10 code is listed in brackets). These categories are most likely to be used in LMIC mental health service delivery, and are based on Patel's classification (Patel 2003c), and the World Health Organization (WHO) MNS disorder categorisation (WHO 2008)

1. Common mental disorders

Mild to moderate mood (affective) disorders (F32‐38)

Neurotic, stress‐related and somatoform disorders  (F40‐49)

Behavioural syndromes associated with physiological disturbances and physical factors (F50‐59)

2. Severe mental disorders

Schizophrenia, schizotypal and delusional disorders (F20‐F29)

Bipolar affective disorder (F31)

Severe depressive episode with/without psychosis (F32.2, F32.3)

3. Neuropsychiatric disorders

Organic, including symptomatic, mental disorders (includes dementia) (F1‐9)

Mental retardation (F70‐79)

Epilepsy (G40)

4. Disorders caused by substance abuse

Mental and behavioural disorders due to psychoactive substance use (F10‐19)

5. Mental disorders specifically related to childhood/development

Conduct disorders

Developmental disorders

Eating disorders

Pervasive developmental disorders

The diagnosis could be made in clinical practice or in the context of the trial.

First level care, primary care and community

First level of contact with formal health services were community‐based interventions or  primary care interventions (or both), on their own or attached to hospital settings, provided they had no specialist input apart from supervision (modified from Wiley‐Exley 2007). This would include individuals with mental illness living in the community and programmes in outpatient clinics or primary care practices. This would not include programmes in hospitals unless the programmes in the hospitals were providing care to outpatients (i.e. generalists in outpatient departments).

Community: as mentioned above detection of mental disorders in all age groups were often done outside the health facility, for example through school, training and other community settings. Therefore, we considered interventions outside the health sector.

Low‐ and middle‐income country (LMIC)

Any country that has ever been an LMIC, as defined by the World Bank lists of LMICs

Non‐specialist health workers (NSHWs)

Health workers who were not specialised in MNS disorders or have not received in‐depth professional specialist training in this clinical area. These included doctors, nurses, auxiliary nurses, lay health workers, as well as allied health personnel such as social workers, occupational therapists. This category did not include professional specialist health workers such as psychiatrists, neurologists, psychiatric nurses or mental health social workers. For inclusion, NSHWs received some training in MNS disorders (in either the control or the intervention group), but this would not constitute a professional category. The authors made a judgement of what constitutes 'some training'. Examples of 'some training' may be an undergraduate module or a short course in mental health.

Other professionals with health roles (OPHRs)

People who were involved as community‐level workers but were not within the health sector, as many people, particularly adolescents and young adults, have low contact with health workers. This category included teachers/trainers/support workers from schools and colleges, and other volunteers or workers within community‐based networks or non‐governmental organisations. These OPHRs have an important role particularly in the promotion of mental health and detection of mental disorders (Patel 2007c; Patel 2008a; WHO 2003a)

We excluded studies that looked at informal care provided by family members or extended members only to members of his or her own family (i.e. who were unavailable to other members of the community) from this review. As previously highlighted in Lewin;s Cochrane review, "these interventions are qualitatively different from other LHW [lay health worker] interventions included in this review given that parents or spouses have an established close relationship with those receiving care which could affect the process and effects of the intervention" (Lewin 2010).

Clinical interventions

1. Detection (recognition and diagnosis) of illness, including screening

2. Acute interventions: drug treatment, non‐drug treatment/care (such as specific psychological therapies, or interventions with psychosocial components like counselling, psychoeducation, coping skills, etc.), referral

3. Follow‐up, rehabilitation

Service interventions

These include change in staffing, or change in mechanism of mental health service delivery (e.g. extension of mental health services through camps and such other outreach services, mobile vans, etc.).

Figures and Tables -
Table 1. Definitions
Navigate to table in Review
Table 2. Risk of bias economic studies ‐ CHEC list criteria

Study

Risk of bias issues

Araya 2003 RCT Chile

‐ time horizon < 1 year

‐ a societal perspective would have been more appropriate

‐ not all relevant costs reported

‐ not all relevant outcomes included (only ambulatory, not hospital)

‐ no discounting

Jordans 2010 C‐RCT Nepal

‐ no discounting

‐ no sensitivity analysis

‐ not all important variables listed

‐ no discussion of ethical/distributional issues

Zambori 2002 CBA Hungary

‐ the competing alternatives were not described

‐ time horizon at 1 year was not appropriate (needs to be longer)

‐ not all relevant outcomes assessed (e.g. effect of treatment on severity, number of healthcare visits to psychiatrist)

‐ outcomes not measured appropriately (self reporting meant low response; standard prices used may not reflect actual prices)

‐ outcomes not valued (only the short‐term outcome)

‐ no sensitivity analysis

‐ conclusions do not all follow from results
Figures and Tables -
Table 2. Risk of bias economic studies ‐ CHEC list criteria
Navigate to table in Review
Table 3. Outcomes of studies not assigned to meta‐analyses

Study, and outcomes measured and tools

Intervention data [no. of participants]

Control data

Measure of effect (95% CI)

P value

Authors' conclusions

Brown 2009 CBA Rwanda(depression in youth)

Mentoring programme by LHW

Usual care

‐

‐

‐

Severity of depression at 2 years (mean) measured using CID‐S

Mean

[no. of participants]

23.27

[347]

Mean

[no. of participants]

23.28

[345]

‐

0.99

Reduction in intervention group but not in control group (at baseline higher score in intervention group). However, the score indicates continuing levels of depression in both groups

Levels of marginalisation at 2 years (mean) measured using a non‐validated marginalisation scale

3.35

3.13

‐

‐

Improved scores in intervention group, which are no different to control group

Levels of grief at 2 years (mean) measured using a non‐validated 7 point grief scale

3.42

3.38

‐

‐

Baseline lower levels of grief in the control group. No change at the end of the intervention though grief increased in control group and remained stable in the intervention group

Li 1989 RCT China(epilepsy ‐ adults and children)

Village doctors

Psychiatrists

‐

‐

‐

Effective epilepsy control with phenobarbital after 3 months

No. seizures/month

[no. of participants]

12

[20]

No. seizures/month

[no. of participants]

11

[20]

‐

‐

‐

Total number of adverse events after 3 months

No. events

[no. of participants]

19

[20]

No. events

[no. of participants]

39

[20]

‐

‐

‐

Paranthaman2010CBAMalaysi(people with schizophrenia and their carers)

Medical assistants/nurses

Usual care

MD (95% CI)

Pvalue

Authors' conclusions

Carer burden (activities in daily living) (mean) at 6 months. Measured using the Family Burden Interview schedule

Mean (SD)

[no. of participants]

9.41 (3.99)

[54]

Mean (SD)

[no. of participants]

8.93 (4.47)

0.48 (‐1.11 to 2.07)

0.55

Mostly there are similar scores between control and intervention groups.

Carer assistance in daily living severity ‐ ADL at 6 months
measured using the Family Burden Interview Schedule

‐

‐

0.83 (‐0.94 to 2.60)

‐

‐

Re‐admission rates

No. (events)

[no. of participants]

3

[54]

No. (events)

[no. of participants]

5

[55]

‐

0.47

‐

Defaulting from follow‐up

No. (events)

[no. of participants]

6

[54]

No. (events)

[no. of participants]

14

[55]

‐

0.03

important improvement in follow‐up rate for intervention group

Shin 2009 RCT Vietnam(children with intellectual disabilities)

Teacher‐led portage programme (OPHRs)

Usual care

MD (95% CI)

P value

Authors' conclusions

Functional impairment (motor skills) at 6 months (similar at 12 months) measured using the Vineland Adaptive Behaviour Scales

Mean (SD)

[no. of participants]

47.6 (16.8)

[16]

Mean (SD)

[no. of participants]

49 (15.4)

[14]

‐1.40 (‐12.93 to 10.13)

0.81

No significant difference for any mental outcomes but some improvement for motor and personal care outcomes if looked at time x effect interaction)

Functional impairment (social skills) at 6 months (similar at 12 months) measured using the Vineland Adaptive Behaviour Scales

47.1 (15.5)

[16]

46.3 (18.3)

[14]

0.80 (‐11.51 to 13.11)

0.93

‐

Behavioural changes at 6 months (similar at 12 months) measured using the Vineland Adaptive Behaviour Scales

55.6 (10.5)

[16]

55.7 (10)

[14]

‐0.10 (‐7.44 to 7.24)

0.98

‐

Sutcliffe2009RCT Thailand(people with drug abuse disorder)

Peer educator‐led psychoeducation (LHWs)

Usual care (life skills training)

RR/MD (95% CI)

P value

Authors' conclusions

Methamphetamine use at 6 months (similar results at 3, 9 and 12 months)

No.

[no. of participants]

272

[442]

No.

[no. of participants]

267

[440]

RR 1.01 (0.91 to 1.13)

0.79

Randomised peer education, social network intervention and control (social skills training) are both associated with reductions in methamphetamine use and increases in condom use over 12 months among a sample of young Thai people

Recovery of depressive symptoms at 12 months (index patient) measured using CES‐D score

Mean (SD)

[no. of participants]

15.7 (9.7)

[209]

Mean (SD)

[no. of participants]

17.9 (9.3)

[206]

MD ‐2.20 (‐4.03 to ‐0.37)

‐

The effect was strongly observed amount intervention index participants compared with both control and network participants

Recovery of depressive symptoms at 12 months (index and network patient combined) measured using CES‐D score

[no. of participants]

[495]

[no. of participants]

[488]

MD ‐1.05 [‐3.20 to 1.11]

‐

Contrary to expectation, mea and in CES‐D score change did not substantially differ between intervention network participants and control network participants. Thus, there is no evidence that the differential intervention effect on depression diffuses to network members

Prevalence of depression at 12 months (index patient) measured using CES‐D score

Events (No.)

[no. of participants]

57

[209]

Events (No.)

[no. of participants]

70

[206]

RR 0.80 (0.60 to 1.07)

‐

‐

Prevalence of depression at 12 months (index and network patient combined) measured using CES‐D score

[no. of participants]

[495]

[no. of participants]

[488]

RR 0.88 (0.73 to 1.06)

‐

‐

Hirani 2010 CRCT Pakistan(adults with depression, economic skills building intervention arm)

NSHW‐led economics skill building

n = 9

Usual care

n = 8

SMD (95% CI)

‐

Comment: these are presented as SMDs (calculated in RevMan, to compare with other SMDs in comparison 1.6 and 1.7)

Severity of depressive symptoms measured using Becks Depression Inventory II

Mean (SD)

20.1 (11.3)

Mean (SD)

27.63 (9.1)

SMD ‐0.69 (‐1.73 to 0.35)

‐

This study documents improved self efficacy and employment for women enrolled in economic skill‐building compared with general counselling and to control.

Functional impairment measured using the General Self‐Efficacy scale

28.7 (6.2)

21.63 (3.8)

SMD ‐1.29 (‐2.41 to ‐0.16)

‐

‐

CES‐D: Center for Epidemiological Studies Depression scale; CID‐S: Composite International Diagnostic‐Screener; CI: confidence interval; LHW: lay health workers; MD: mean difference; No.: number; OPHR: other professionals with health roles; RR: risk ratio; SD: standard deviation; SMD: standardised mean difference.

Figures and Tables -
Table 3. Outcomes of studies not assigned to meta‐analyses
Navigate to table in Review
Table 4. SoF 1: NSHW‐led psychological interventions compared with usual care in treating common mental disorders in adults in low‐ and middle‐income countries (RCTs)

What are the effects of NSHW‐led psychological interventions for treating common mental disorders in adults in low‐ and middle‐ income countries? (additional outcomes to comparison 1)

Patient or population: Adults with common mental disorders (depression or anxiety, or both)
Settings: Low‐ and middle‐income countries (China, Jamaica, Pakistan, Taiwan, Uganda)
Intervention: NSHWs conducting psychological interventions
Comparison: Usual care

Outcomes

Illustrative comparative risks* (95% CI)

Effect estimate
(95% CI)

No of Participants
(studies)

Quality of the evidence
(GRADE)

Comments

Assumed risk

Corresponding risk

Usual care

NSHWs

Severity of CMD symptoms ‐ all interventions

short term (0‐6 months)
measured using various depression rating scales1

‐

The mean severity of CMD symptoms ‐ NSHW interventions short term was
0.75 standard deviations lower
(1.29 to 0.21 lower)

SMD ‐0.75 (‐1.29 to ‐0.21)

1470
(6 studies)

⊕⊝⊝⊝
very low2,3,4

‐

Severity of CMD symptoms ‐ all interventions

medium term (12 months)
measured using various CMD rating scales5

‐

The mean severity of CMD symptoms ‐ NSHW interventions medium term was
0.47 standard deviations lower
(0.60 to 0.34 lower)

SMD ‐0.47 (‐0.60 to ‐0.34)

923

(2 studies)

⊕⊕⊕⊝
moderate6,7

‐

Functional impairment/disability in adults with CMD ‐ NSHW interventions

short term (2‐6 months)
measured using various functional impairment scales8

‐

The mean functional impairment of adults with CMD ‐ NSHW interventions short term was
0.33 standard deviations lower

(0.80 lower to 0.13 higher)

SMD ‐0.33 (‐0.80 to 0.13)

1243
(4 studies)

⊕⊝⊝⊝
very low9,10,11

‐

Functional impairment/disability in depression/CMD (adults) ‐ NSHW interventions

medium term (2‐6 months)
measured using the Global Assessment of Functioning scale

‐

The mean functional impairment of adults with CMD ‐ NSHW interventions medium term was
0.56 standard deviations lower
(0.70 to 0.42 lower)

SMD ‐0.56 (‐0.70 to ‐0.42)

798
(1 study)

⊕⊕⊕⊝
moderate12

‐

*The basis for the assumed risk is the mean control group risk across studies for pooled results and the control group risk for single studies. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).
CMD: common mental disorders; CI: confidence interval; NSHW: non‐specialist health worker; SMD: standardised mean difference.

GRADE Working Group grades of evidence
High quality: Further research is very unlikely to change our confidence in the estimate of effect.
Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.
Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.
Very low quality: We are very uncertain about the estimate.

1Chen 2000 RCT Taiwan, Hirani 2010 CRCT Pakistan and Tiwari 2010 RCT China used the Beck's Depression Inventory; Rahman 2008 CRCT Pakistan used the Hamilton Depression Rating Scale; Ali 2003 RCT Pakistan used the AKUADS; Bolton 2003 C‐RCT Uganda used the Hopkins Symptom Checklist (HSCL).

2 Serious study limitations: Two of the six trials in this analysis were judged at high risk of bias and one was unclear about possible risk of bias. Chen 2000 RCT Taiwan had unclear sequence generation and allocation concealment, all were self reported outcomes, there was possible contamination and there was a high dropout rate after randomisation, with no analysis of dropout versus non‐dropout differences; Hirani 2010 CRCT Pakistan was unclear regarding allocation concealment, there was no blinding of outcome assessment (self reported outcomes), it was unclear if baseline measures and characteristics were similar in both groups; and the report provided no information on dropouts. Bolton 2003 C‐RCT Uganda was not clear about allocation concealment and quasi randomisation of individuals within clusters (though randomisation of clusters) may have introduced bias. The three trials contributed 45% of the weight in the pooled analysis. Downgraded by 1.

3Serious inconsistency: I2 statistic = 94%. However, the inconsistency related to the magnitude of benefit favouring NSHW interventions rather than in the direction of effect. Downgraded by 1.
4 Serious imprecision: The 95% CI for the pooled estimates indicates appreciable benefit and non‐appreciable benefit for collaborative care (appreciable SMD = ≥ 0.5; non‐appreciable benefit ≤ 0.2). Downgraded by 1.

5Baker‐H 2005 CRCT Jamaica used the CED‐S; Rahman 2008 CRCT Pakistan used the Hamilton Depression Rating Scale.

6 No serious study limitations: The CES‐D used in Baker‐H 2005 CRCT Jamaica is not validated in the Jamaican population and is not a measure of clinical depression but just identifies depressive symptoms. Most women were not likely to have been depressed. Also in this study, there were unadjusted differences in baseline characteristics. However, this study contributed only 14% weight to the pooled results and removal of this study did not alter the direction or precision of the effect estimate. Not downgraded.

7 Serious indirectness: The two trials included were the only two of the six trials that compared this intervention that had data over the medium term, and only one used a validated outcome measure. Downgraded by 1.

8Bolton 2003 C‐RCT Uganda used a sex‐specific Functional Impairment Questionnaire; Rahman 2008 CRCT Pakistan used the Global Assessment of Functioning (GAF) scale, Hirani 2010 CRCT Pakistan used the General Self‐efficacy Scale; Tiwari 2010 RCT China used the Short Form‐ 12 (SF‐12) (mental and physical components).

9 Serious study limitations: Two of the four studies were at risk of bias. Bolton 2003 C‐RCT Uganda was not clear about allocation concealment and quasi‐randomisation of individuals within clusters (though randomisation of clusters) may have introduced bias. Hirani 2010 CRCT Pakistan was unclear regarding allocation concealment, there was no blinding of outcome assessment (self reported outcomes), it was unclear if baseline measures and characteristics were similar in both groups; and the report provided no information on dropouts. Downgraded by 1.

10 Very serious inconsistency: I2 statistic = 90%. The inconsistency related to the direction of effect between interventions and was unexplained. Downgraded by 2.

11 Serious imprecision: the 95% CI of the pooled estimate showed appreciable benefit for interventions (appreciable SMD = 0.5) and non‐appreciable benefit for control. Downgraded by 1.

12 Serious imprecision: the 95% CI of the pooled estimate shows non‐appreciable benefit for psychological interventions and usual care (appreciable SMD = 0.5). However, the data for this outcome were from only one trial. Downgraded by 1.

Figures and Tables -
Table 4. SoF 1: NSHW‐led psychological interventions compared with usual care in treating common mental disorders in adults in low‐ and middle‐income countries (RCTs)
Navigate to table in Review
Table 5. SoF: NSHW‐led interventions compared with usual care in treating common mental disorders in adults in low‐ and middle‐income countries (CBAs)

What are the effects of NSHWs conducting single interventions compared with usual care in treating common mental disorders for mental health care in low‐ and middle‐income countries? (additional CBA outcomes to comparison 1)

Patient or population: Adults with CMDs (such as depression and anxiety)
Settings: Low‐ and middle‐income countries (Indonesia, Rwanda)
Intervention: NSHWs conducting single interventions
Comparison: Usual care

Outcomes

Illustrative comparative risks* (95% CI)

Relative effect
(95% CI)

No of Participants
(studies)

Quality of the evidence
(GRADE)

Comments

Assumed risk

Corresponding risk

Usual care

NSHWs

Severity of common mental disorders ‐ short term (within 6 months)
measured using CMD rating scales1

‐

The mean severity of CMDs ‐ short term (within 6 months post intervention) in the intervention groups was
0.08 standard deviations lower
(0.25 lower to 0.09 higher)

SMD ‐0.08 (‐0.25 to 0.09)

533
(2 studies)

⊕⊝⊝⊝
very low2,3

‐

Severity of CMDs ‐ medium term (8 months)
measured using SRQ‐20

‐

The mean severity of CMDs ‐ medium term (6 months to 1 year post intervention) in the intervention groups was
0.32 standard deviations lower
(0.6 to 0.04 lower)

SMD ‐0.32

(‐0.6 to ‐0.04)

200
(1 study)

⊕⊝⊝⊝
very low4,5

‐

Functional impairment ‐ male short term (1 month)
measured using WHODAS (adapted) 11 items

‐

The mean functional impairment ‐ male short term (within 6 months of intervention) in the intervention groups was
0.32 standard deviations lower
(0.65 lower to 0.02 higher)

SMD ‐0.32 (‐0.65 to 0.02)

141
(1 study)

⊕⊝⊝⊝
very low6,7

‐

Functional impairment ‐ female short term (1 month)
measured using WHODAS (adapted) 11 items

‐

The mean functional impairment ‐ female short term (within 6 months of intervention) in the intervention groups was
0.34 standard deviations lower
(0.63 to 0.06 lower)

SMD ‐0.34 (‐0.63 to ‐0.06)

192
(1 study)

⊕⊝⊝⊝
very low6

‐

*The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).
CI: confidence interval; CMD: common mental disorders; NSHW: non‐specialist health worker; SMD: standardised mean difference; SRQ: Self Reporting Questionnaire; WHODAS: World Health Organization Disability Assessment Scale.

GRADE Working Group grades of evidence
High quality: Further research is very unlikely to change our confidence in the estimate of effect.
Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.
Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.
Very low quality: We are very uncertain about the estimate.

1Bass 2012 CBA Indonesia: Hopkins Symptom Checklist 25; Scholte 2011 CBA Rwanda: SRQ‐20.
2 Very serious risk of bias: Bass 2012 CBA Indonesia: controlled before‐and‐after study so non‐random and not concealed. Also differences in baseline outcomes for girls, and doubt about reliability of primary outcomes as tool not properly validated. Scholte 2011 CBA Rwanda: controlled before‐and‐after study so non‐randomised and no concealment. Also unclear risk for incomplete outcome data, there are baseline differences in outcomes and in characteristics not all adjusted for, and high rate of loss to follow‐up with no analysis of group lost to follow‐up. Downgraded by 2.
3 No imprecision: Non‐appreciable benefit for either intervention or control group.
4 No explanation was provided.
5 No imprecision: Appreciable and non‐appreciable benefit for intervention.
6 Very serious risk of bias: Bass 2012 CBA Indonesia: controlled before‐and‐after study so non‐random and not concealed. Also differences in baseline outcomes for girls, and doubt about reliability of primary outcomes as tool not properly validated. Downgraded by 2.
7 Serious imprecision: Appreciable benefit for intervention and non‐appreciable benefit for usual care. Downgrade by 1.

Figures and Tables -
Table 5. SoF: NSHW‐led interventions compared with usual care in treating common mental disorders in adults in low‐ and middle‐income countries (CBAs)
Navigate to table in Review
Table 6. SoF 2: Collaborative care model (NSHWs plus specialist) compared with usual care in treating common mental disorders in adults in low‐ and middle‐income countries (RCTs)

What are the effects of a collaborative care model (NSHW plus specialist supervision) for mental health care in adults with common mental disorders low‐ and middle‐income countries? (additional outcomes to comparison 2)

Patient or population: Adults (≥ 18 years) with common mental disorders (includes anxiety or depression, or both)
Settings: Middle‐income countries (Chile, India)
Intervention: Collaborative care model (NSHW plus specialist supervision)

Comparison: Enhanced usual care

Outcomes

Illustrative comparative risks* (95% CI)

Effect estimate
(95% CI)

No of Participants
(studies)

Quality of the evidence
(GRADE)

Comments

Assumed risk

Corresponding risk

usual care

Collaborative care model

Prevalence of CMDs medium term (12 months)
measured using CIS‐R generated ICD‐10 diagnosis for CMD

See comment

See comment

RR 0.95
(0.68 to 1.33)

2009
(1 study)

⊕⊕⊝⊝
low1

Patel 2010 C‐RCT India did not reveal significant differences in the prevalence of depression with both interventions in public or private care facilities.

Severity of symptoms in CMD

short term (2‐6 months)
measured using various rating scales2

‐

The mean severity of symptoms in CMD with collaborative care was
0.31 standard deviations lower
(0.56 to 0.06 lower)

SMD ‐0.31 (‐0.56 to ‐0.06)

3604
(5 studies)

⊕⊝⊝⊝
very low3,4,5

‐

Severity of symptoms in CMD

medium term (12 months)
measured using CIS‐R rating scale

‐

The mean severity of symptoms in CMD with collaborative care was
0.03 standard deviations lower
(0.12 lower to 0.06 higher)

SMD ‐0.03 (‐0.12 to 0.06)

1905
(1 study)

⊕⊕⊕⊝
moderate6

‐

Functional impairment/disability in CMD short term (2‐6 months)
measured using various functional disability scores7

‐

The mean functional impairment/disability in CMD with collaborative care was
0.22 standard deviations lower
(0.44 to 0.01 lower)

SMD ‐0.22 (‐0.44 to ‐0.01)

3604
(5 studies)

⊕⊝⊝⊝
very low5,8,9

‐

Functional impairment/disability in CMD medium term (12 months)
measured using WHODAS II scores

‐

The mean functional impairment/disability in CMD with collaborative care was
0.02 standard deviations lower
(0.11 lower to 0.07 higher)

SMD ‐0.02 (‐0.11 to 0.07)

1905

(1 study)

⊕⊕⊕⊝
moderate6

‐

*The basis for the assumed risk is the mean control group risk across studies for pooled results and the control group risk for single studies. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).
CI: confidence interval; CIS‐D: Composite International Diagnostic‐Screener; CMD: common mental disorders; ICD: International Classification of Diseases; NSHW: non‐specialist health worker; RR: risk ratio; SMD: standardised mean difference; WHODAS: World Health Organization Disability Assessment Scale.

GRADE Working Group grades of evidence
High quality: Further research is very unlikely to change our confidence in the estimate of effect.
Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.
Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.
Very low quality: We are very uncertain about the estimate.

1 Very serious imprecision: The 95% CI for the pooled estimates indicates appreciable benefit for collaborative care and appreciable benefit for usual care. Downgraded by 2.
2Jenkins 2012 C‐RCT Kenya used the General Health Questionnaire (GHQ)‐12. Patel 2010 C‐RCT India used CIS‐R to generate ICD‐10 depression diagnoses; Fritsch 2007 RCT Chile and Araya 2003 RCT Chile used the Hamilton Depression Rating Scale (HDRS); Rojas 2007 RCT Chile used the: Edinburgh Postnatal Depression Scale (EPDS).

3 Serious study limitations: In Araya 2003 RCT Chile and possibly in Fritsch 2007 RCT Chile general practitioners (GPs) did both interventions, so there was a high risk of contamination; this would have reduced the potential benefits with collaborative care in two of the four trials in the meta‐analysis. Downgraded by 1.
4 Serious inconsistency. The I2 statistic = 91% with Araya 2003 RCT Chile clearly an outlier, contributing to this unexplained inconsistency. However, the inconsistency related to the magnitude of benefit favouring collaborative care rather than in the direction of effect. Downgraded by 1.
5 Serious indirectness: Jenkins 2012 C‐RCT Kenya used the GHQ‐12 to grade severity of symptoms; the GHQ is a screening instrument that is validated to screen for CMDs; its use to rate the severity of depression is less reliable). Downgraded by 1.
6 Serious imprecision: The 95% CI for the pooled estimates indicates no appreciable benefit for collaborative care (< 0.2) and non‐appreciable benefit for usual care. The data come from one study (Patel 2010 C‐RCT India), and therefore imprecise. Downgraded by 1.

7Jenkins 2012 C‐RCT Kenya used WHODAS II long version (36 items); Patel 2010 C‐RCT India used the WHODAS II short version (12 items); Araya 2003 RCT Chile; Fritsch 2007 RCT Chile; and Rojas 2007 RCT Chile used SF‐36 social functioning component.
8 Serious study limitations: In Araya 2003 RCT Chile and probably Fritsch 2007 RCT Chile, GPs did both intervention and control interventions so there was a high risk of contamination. Downgraded by 1.

9 Serious inconsistency. The I2 statistic = 87% with Araya 2003 RCT Chile clearly an outlier, contributing to this unexplained inconsistency. However, the inconsistency related to the magnitude of benefit favouring collaborative care rather than in the direction of effect. Downgraded by 1.

Figures and Tables -
Table 6. SoF 2: Collaborative care model (NSHWs plus specialist) compared with usual care in treating common mental disorders in adults in low‐ and middle‐income countries (RCTs)
Navigate to table in Review
Table 7. SoF 3: NSHWs compared with usual care for treating maternal depression (RCTs)

What are the effects of NSHW‐led interventions for treating maternal depression in low‐ and middle‐income countries? (additional outcomes for comparison 3)

Patient or population: Adult women with maternal depression
Settings: Low‐ and middle‐income countries (Chile, Jamaica, Pakistan, Taiwan)
Intervention: NSHW‐led interventions
Comparison: Usual care

Outcomes

Illustrative comparative risks* (95% CI)

Estimate effect
(95% CI)

No of Participants
(studies)

Quality of the evidence
(GRADE)

Comments

Assumed risk

Corresponding risk

Usual care

NSHWs

Severity of symptoms of perinatal depression ‐ NSHW led‐psychological interventions

short term (0‐2 months)
measured using depression rating scales1

‐

The mean severity of symptoms of perinatal depression ‐ short term with NSHW‐led interventions was
0.5 standard deviations lower
(0.63 to 0.36 lower)

SMD ‐0.5 (‐0.63 to ‐0.36)

858
(2 studies)

⊕⊕⊕⊕
high2,3

Note that a small clinically appreciable benefit was set at SMD < 0.2, and a moderate benefit at SMD of 0.5 to 0.8 (Cohen 1988)

Severity of symptoms of perinatal depression ‐ NSHW led‐psychological interventions

medium term (12 months)
measured using a depression scale4

‐

The mean severity of symptoms of perinatal depression ‐ medium term ‐ with NSHW‐led interventions was
0.41 standard deviations lower
(0.76 to 0.06 lower)

SMD ‐0.41 (‐0.76 to ‐0.06)

125
(1 study)

⊕⊝⊝⊝
low5,6

‐

Severity of symptoms of perinatal depression ‐ collaborative care‐ short term (3 months)
depression rating scale ‐ EPDS

‐

The mean severity of symptoms of perinatal depression ‐ short term ‐ with collaborative care was
0.22 standard deviations lower
(0.48 lower to 0.04 higher)

SMD ‐0.22 (‐0.48 to 0.04)

230
(1 study)

⊕⊕⊕⊝
moderate7

‐

*The basis for the assumed risk is the mean control group risk across studies for pooled results and the control group risk for single studies. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).
CI: confidence interval; EPDS: Edinburgh Postnatal Depression Score; NSHW: non‐specialist health worker; SMD: standardised mean difference.

GRADE Working Group grades of evidence
High quality: Further research is very unlikely to change our confidence in the estimate of effect.
Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.
Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.
Very low quality: We are very uncertain about the estimate.

1Chen 2000 RCT Taiwan: Taiwanese Beck Depression Inventory; Rahman 2008 CRCT Pakistan: Hamilton Depression Rating Scale (HDRS).
2 No serious study limitations: Chen 2000 RCT Taiwan was unclear for sequence generation and allocation concealment, all were self reported outcomes, there was possible contamination and there was a high dropout rate after randomisation, with no analysis of differences between dropout versus non‐dropouts differences. However, Rahman 2008 CRCT Pakistan had no serious study limitations and contributed most of the weight to the pooled analysis. Removal of data from Chen 2000 RCT Taiwan did not appreciably change effect estimates. Not downgraded.
3 No serious imprecision: appreciable benefit seen at SMD = 0.2 for Rahman 2008 CRCT Pakistan with HDRS and 0.5 for Chen 2000 RCT Taiwan with the BDI. The 95% CI includes appreciable benefit for NSHW‐led interventions. Not downgraded.

4Baker‐H 2005 CRCT Jamaica: CES‐D.
5 Serious study limitations: Baker‐H 2005 CRCT Jamaica had unadjusted differences in baseline characteristics. Downgraded by 1.

6 Serious imprecision: The 95% CI of the effect estimate indicated appreciable and non‐appreciable benefit with NSHW‐led interventions, and the sample size was small. Downgraded by 1.
7 Serious imprecision: The 95% CI of the effect estimate indicated appreciable and non‐appreciable benefit with collaborative care, and the sample size was small. Downgraded by 1.

Figures and Tables -
Table 7. SoF 3: NSHWs compared with usual care for treating maternal depression (RCTs)
Navigate to table in Review
Table 8. SoF 4: NSHWs compared with specialists in treating depression in adults in low‐ and middle‐income countries (CBAs)

What are the effects of NSHWs compared with specialists in treating depression for mental health care in low‐ and middle‐income countries? (additional outcomes for comparison 4)

Patient or population: Adults with depression
Settings: Middle‐income countries (Hungary and Argentina)

Intervention: NSHWs providing pharmacological intervention
Comparison: Specialists providing pharmacological intervention

Outcomes

Illustrative comparative risks* (95% CI)

Relative effect
(95% CI)

No of Participants
(studies)

Quality of the evidence
(GRADE)

Comments

Assumed risk

Corresponding risk

Specialists

NSHWs

Frequency of adverse events (at 56 days)
measured using the number of mild, moderate and severe adverse events

288 per 1000

245 per 1000
(193 to 308)

RR 0.85
(0.67 to 1.07)

768
(1 study)

⊕⊝⊝⊝
very low1,2

Note that a small clinically appreciable benefit was set at SMD < 0.2, and a moderate benefit at SMD of 0.5 to 0.8 (Cohen 1988)

Number of days spent in hospital (at 1 year)

‐

The mean number of days spent in hospital at 1 year in the NSHW group was
1.79 lower
(3.59 lower to 0.01 higher)

MD ‐1.79 (‐3.59 to 0.01)

124
(1 study)

⊕⊝⊝⊝
very low3,4

‐

Number of days spent in hospital (at 2 years)

‐

The mean number of days spent in hospital at 2 years in the NSHW group was
0.02 lower
(2.59 lower to 2.55 higher)

MD ‐0.02 (‐2.59 to 2.55)

124
(1 study)

⊕⊝⊝⊝
very low3,5

‐

Number of days spent on sick leave (at 1 year)

‐

The mean number of days spent on sick leave at 1 year in the NSHW group was
3.96 lower
(15.58 lower to 7.66 higher)

MD ‐3.96 (‐15.58 to 7.66)

108
(1 study)

⊕⊝⊝⊝
very low3,5

‐

Number of days spent on sick leave (at 2 years)

‐

The mean number of days spent on sick leave at 2 years in the NSHW group was
14.63 higher
(0.76 lower to 30.02 higher)

MD 14.63 (‐0.76 to 30.02)

123
(1 study)

⊕⊝⊝⊝
very low3,6

‐

*The basis for the assumed risk is the risk in the control group. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).
CBA: controlled before‐and‐after; CI: confidence interval; MD: mean difference; NSHW: non‐specialist health worker; RR: risk ratio; SMD: standardised mean difference.

GRADE Working Group grades of evidence
High quality: Further research is very unlikely to change our confidence in the estimate of effect.
Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.
Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.
Very low quality: We are very uncertain about the estimate.

1 Very serious study limitations: Lyketsos1999CBA Argentina was a CBA study so selection bias was likely. There was a risk of contamination and outcome assessments were done by same physicians doing the intervention. Downgraded by 2.

2 Serious imprecision: The 95% CI for the pooled estimates indicates appreciable and non‐appreciable harms with the interventions. In addition, results are only from one study. Downgraded by 1.
3 Very serious study limitations: Zambori 2002 CBA Hungary was a CBA study so selection is likely; and there were significant difference in baseline outcomes and baseline characteristics. Downgraded by 2.
4 Serious imprecision: The 95% CI for the pooled estimates indicates appreciable benefit for the interventions and non‐appreciable benefit for the control. In addition, results were only from one study. Downgraded by 1.
5 Very serious imprecision: Shows appreciable benefit for both interventions. Also results were only from one study. Downgraded by 2.

6 Serious imprecision: Shows appreciable benefit for specialists and non‐appreciable benefit for NSHWs. In addition, results were only from one study. Downgraded by 1.

Figures and Tables -
Table 8. SoF 4: NSHWs compared with specialists in treating depression in adults in low‐ and middle‐income countries (CBAs)
Navigate to table in Review
Table 9. SoF 5: NSHW and OPHR‐led psychological interventions compared with usual care in treating adults with PTSD (NRCT and RCTs)

What are the effects of NSHWs/OPHRs compared with usual mental health care in low‐ and middle‐income countries for data from an NRCT in adults with PTSD?

Patient or population: Adults with PTSD
Settings: Low‐ and middle‐income countries (Burundi, Uganda)
Intervention: NSHWs and OPHRs delivering psychological interventions (narrative exposure therapy, trauma counselling and workshops with psychoeducation)
Comparison: Usual care

Outcomes

Illustrative comparative risks* (95% CI)

Relative effect
(95% CI)

No of Participants
(studies)

Quality of the evidence
(GRADE)

Comments

Assumed risk

Corresponding risk

Usual care

NSHWs/OPHRs

Prevalence of PTSD in LHW‐led narrative exposure therapy, medium term (9 months)
measured using the PTSD diagnostic tool ‐ DSM‐IV from CIDI

632 per 1000

303 per 1000
(171 to 537)

RR 0.48
(0.27 to 0.85)

62
(1 study)

⊕⊕⊝⊝
low1,2

‐

Prevalence of PTSD in a LHW‐led trauma counselling, medium term (mean 9 months)
measured using the PTSD diagnostic tool ‐ DSM‐IV from CIDI

632 per 1000

348 per 1000
(209 to 588)

RR 0.55
(0.33 to 0.93)

65
(1 study)

⊕⊕⊝⊝
low1,2

‐

Severity of PTSD symptoms in LHW‐led psychological intervention (narrative exposure therapy) in the short term (6 months)
measured using the PTSD symptom score ‐ Post Traumatic Stress Diagnostic Scale

‐

The mean severity of PTSD with narrative exposure therapy in the short term was
0.55 standard deviations lower
(1.08 to 0.03 lower)

SMD ‐0.55 (‐1.08 to ‐0.03)

75
(1 study)

⊕⊕⊝⊝
low1,3

‐

Severity of depression ‐ psychological intervention in LHW led workshop with psychoeducation in the short term (within 2 weeks)
measured using depression rating scale: HSCL‐25

The mean [SD] scores on the HSCL‐25 was 1.83 [0.67]

The mean severity of depression with a psychosocial intervention (with psychoeducation) in the short term was
0.07 lower

(0.36 lower to 0.22 higher)

MD ‐0.07 (‐0.36 to 0.22)

76
(1 study)

⊕⊕⊝⊝
low4,5

‐

Severity of depression ‐ psychological intervention (workshop without psychoeducation) in the short term (within 2 weeks)
measured using a depression rating scale: HSCL‐25

The mean [SD] scores on the HSCL‐25 was 1.83 [0.67]

The mean severity of depression with a psychosocial intervention (without psychoeducation) short term was

0.14 lower

(0.42 lower to 0.14 higher)

MD ‐0.14 (‐0.42 to 0.14)

75
(1 study)

⊕⊕⊝⊝
low4,5

‐

*The basis for the assumed risk is the median control group risk or mean control group risk across studies for pooled estimates and the control group risk for single studies. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).
CI: confidence interval; CIDI: Composite International Diagnostic Interview; DSM: Diagnostic and Statistical Manual of Mental Disorders; HSCL: Hopkins Symptom Checklist; LHW: lay health worker; MD: mean difference; NRCT: non‐randomised controlled trial; NSHW: non‐specialist health worker; OPHR: other professionals with health roles; PTSD: post‐traumatic stress disorder; RCT: randomised controlled trial; RR: risk ratio; SD: standard deviation; SMD: standardised mean difference.

GRADE Working Group grades of evidence
High quality: Further research is very unlikely to change our confidence in the estimate of effect.
Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.
Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.
Very low quality: We are very uncertain about the estimate.

1 Serious study limitations: Neuner 2008 NRCT Uganda no allocation concealment, randomisation has no sequence generation. High dropout rate and different between groups, different baseline characteristics and likely contamination. Downgraded by 1.

2 Serious imprecision: The 95% CI of the effect estimate indicated appreciable and non‐appreciable benefit with the intervention, and the sample size was small. Downgraded by 1.

3 Serious imprecision: The 95% CI of the effect estimates demonstrated appreciable and non‐appreciable benefit with intervention and the sample size (75 participants) was small. Downgraded by 1.

4 Serious study limitations: Yeomans 2010 RCT Burundi: unvalidated Harvard Trauma Questionnaire in the local context (only validated in Burundi) so may affect reliability of outcomes. Downgraded by 1.

5 Serious imprecision: The 95% CI of the effect estimates demonstrated non‐appreciable benefit for both intervention and usual care and the sample size was small. Downgraded by 1.

Figures and Tables -
Table 9. SoF 5: NSHW and OPHR‐led psychological interventions compared with usual care in treating adults with PTSD (NRCT and RCTs)
Navigate to table in Review
Table 10. SoF 6: NSHWs compared with usual care in improving dementia patients' and carers' outcomes in low‐ and middle‐income countries (RCTs)

What are the effects of NSHW‐led care in improving dementia patients' and carers' outcomes for mental health care in low‐ and middle‐income countries? (additional outcomes for comparison 6)

Patient or population: Patients with dementia and their carers
Settings: Middle‐income countries (India, Russia)
Intervention: NSHWs delivering brief intervention to carers
Comparison: Usual care

Outcomes

Illustrative comparative risks* (95% CI)

Estimate effect
(95% CI)

No of Participants
(studies)

Quality of the evidence
(GRADE)

Comments

Assumed risk

Corresponding risk

Usual care

NSHWs

Patient functional ability (at 6 months)
measured using the functional ability scale (EASI)

‐

The mean patient functional ability with this brief carer intervention was
0.24 standard deviations lower
(0.67 lower to 0.20 higher)

MD ‐0.24 (‐0.67 to 0.20)

81
(1 study)

⊕⊕⊕⊝
moderate1,2

Note that a small clinically appreciable benefit was set at SMD < 0.2, and a moderate benefit at SMD of 0.5 to 0.8 (Cohen 1988)

Patient QoL (at 6 months)
measured using the quality of life score (DEMQOL)

‐

The mean patient QoL with this brief carer intervention was
0.43 standard deviations lower
(0.98 lower to 0.12 higher)

MD ‐0.43 (‐0.98 to 0.12)

53
(1 study)

⊕⊕⊕⊝
moderate1,2,3

‐

Carer mental health status (at 6 months)
measured using
general mental health status scores4

‐

The mean carer mental health status with this brief carer intervention was
0.42 standard deviations lower
(0.76 to 0.08 lower)

SMD ‐0.42 (‐0.76 to ‐0.08)

134
(2 studies)

⊕⊕⊕⊝
moderate5

‐

Carer burden
(at 6 months)
measured using a burden scale (Zarit Burden Interview)

‐

The mean carer burden in the brief carer intervention was
0.50 standard deviations lower
(0.84 to 0.15 lower)

SMD ‐0.50 (‐0.84 to ‐0.15)

134
(2 studies)

⊕⊕⊕⊝
moderate5

‐

Carer distress (at 6 months)
measured using
the carer distress scale: (NPI‐D)

‐

The mean carer distress with this brief carer intervention was
0.47 standard deviations lower
(0.82 to 0.13 lower)

SMD ‐0.47 (‐0.82 to ‐0.13)

134
(2 studies)

⊕⊕⊕⊝
moderate5

‐

Carer QoL (at 6 months)
measured using the
QoL assessment: (WHOQOL‐BREF)

‐

The mean carer quality of life in this brief carer intervention was
0.37 standard deviations lower
(0.92 lower to 0.17 higher)

MD ‐0.37 (‐0.92 to 0.17)

53
(1 study)

⊕⊕⊕⊝
moderate1,2,3

‐

*The basis for the assumed risk is the mean control group risk across studies for pooled results and the control group risk for single studies. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).
CI: confidence interval; DEMQOL: Dementia Quality of Life; EASI: Everyday Abilities Scales for India; MD: mean difference; NPI‐D: Neuropsychiatric Inventory ‐ Dementia; NSHW: non‐specialist health worker; QoL: quality of life; RCT: randomised controlled trial; SMD: standardised mean difference; WHOQOL‐BREF: World Health Organization Quality of Life‐BREF.

GRADE Working Group grades of evidence
High quality: Further research is very unlikely to change our confidence in the estimate of effect.
Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.
Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.
Very low quality: We are very uncertain about the estimate.

1No indirectness: there is only one study therefore the generalisability of this results to other settings is compromised. However, this also resulted in imprecision in the effect estimate, and hence the quality of evidence was not further downgraded.

2 Serious imprecision. The 95% CI of the MD indicated appreciable benefits for NSHWs/intervention and non‐appreciable benefits for usual care. The sample size is also small. Downgraded by 1.

3 No study limitations. Gavrilova 2009 RCT Russia was unclear whether allocation was concealed. However, no serious baseline differences in characteristics or outcomes were seen. Not downgraded.

4Dias 2008 RCT India: General Health Questionnaire (GHQ)‐12; Gavrilova 2009 RCT Russia Self Reporting Questionnaire (SRQ)‐20.

5 Serious imprecision. The 95% CI for the mean difference indicated appreciable and non‐appreciable benefits with the interventions. Downgraded by 1.

Figures and Tables -
Table 10. SoF 6: NSHWs compared with usual care in improving dementia patients' and carers' outcomes in low‐ and middle‐income countries (RCTs)
Navigate to table in Review
Table 11. SoF 7: NSHW‐led brief alcohol interventions compared with usual care for adults with alcohol‐use disorders (RCTs)

What are the effects of NSHWs in delivering brief alcohol interventions in RCTs for alcohol‐use disorders? (additional outcomes for comparison 7)

Patient or population: Patients with alcohol‐use disorders
Settings: Low‐ and middle‐income countries (Thailand, Kenya)
Intervention: NSHWs in delivering brief alcohol interventions
Comparison: Usual care

Outcomes

Illustrative comparative risks* (95% CI)

Relative effect
(95% CI)

No of Participants
(studies)

Quality of the evidence
(GRADE)

Comments

Assumed risk

Corresponding risk

Usual care

NSHWs

Frequency of binge drinking (3‐6 months)
measured using the frequency of binge drinking in the past week

‐

The mean frequency of binge drinking in the intervention groups was

0.50 lower

(1.14 lower to 0.14 higher)

MD ‐0.50

(‐1.14 to 0.14)

92
(1 study)

⊕⊕⊝⊝
low1,2

‐

Adverse consequences ‐ RTAs (at 6 months)
measured using the number of RTAs

220 per 1000

79 per 1000
(26 to 238)

RR 0.36
(0.12 to 1.08)

92
(1 study)

⊕⊕⊝⊝
low1,2

‐

Adverse consequences ‐ withdrawal symptoms (at 3 months)
measured using the number of withdrawal symptoms

31 per 1000

83 per 1000
(9 to 755)

RR 2.67
(0.29 to 24.37)

68
(1 study)

⊕⊝⊝⊝
very low1,3

‐

*The basis for the assumed risk is the mean control group risk across studies for pooled data or the control group risk for individual studies. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).
CI: confidence interval; MD: mean difference; NSHW: non‐specialist health worker; RCT: randomised controlled trial; RR: risk ratio; RTA: road traffic accident.

GRADE Working Group grades of evidence
High quality: Further research is very unlikely to change our confidence in the estimate of effect.
Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.
Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.
Very low quality: We are very uncertain about the estimate.

1 Serious study limitations: Noknoy 2010 RCT Thailand: high dropout rate with no information on whether they are different to completers, no validated tools in the setting, so unreliable primary outcomes. Papas 2011 RCT Kenya: unclear about whether the non‐blinding of outcome assessors would have impacted on study. Downgraded by 1.
2 Serious imprecision: The 95% CI of the effect estimates indicates an appreciable benefit for NSHW care and non‐appreciable benefit for usual care. The sample size was also small. Downgraded by 1.
3 Very serious imprecision: The 95% CI of the effect estimate indicates appreciable benefit for NSHW care and for usual care and also the study had a very small sample size. Downgraded by 2.

Figures and Tables -
Table 11. SoF 7: NSHW‐led brief alcohol interventions compared with usual care for adults with alcohol‐use disorders (RCTs)
Navigate to table in Review
Table 12. SoF 8: NSHWs/OPHRs compared with usual care in conducting interventions for children with post‐traumatic stress disorder and depression (RCTs)

What are the effects of NSHWs/OPHRs conducting interventions for children with PTSD from RCTs in low‐ and middle‐income countries? (additional outcomes for comparison 8)

Patient or population: Children/adolescents with PTSD and related depressive/anxiety symptoms
Settings: Low‐ and middle‐income countries (Bosnia, Indonesia, Kosovo, Nepal, Sri Lanka, Uganda)
Intervention: NSHWs/OPHRs delivering psychological and psychosocial interventions

Comparison: Usual care

Outcomes

Illustrative comparative risks* (95% CI)

Estimate effect
(95% CI)

No of Participants
(studies)

Quality of the evidence
(GRADE)

Comments

Assumed risk

Corresponding risk

Usual care

NSHWs/OPHRs

Severity of PTSD symptoms in LHW‐led classroom‐based intervention, short term (1‐6 months)
measured using the CPSS

‐

The MCD in severity of PTSD symptoms in classroom‐based LHW‐led intervention groups was
0.56 lower
(2.82 lower to 1.7 higher)

MCD ‐0.56 (‐2.82 to 1.70)

1090
(3 studies)

⊕⊝⊝⊝
very low1,2,3

‐

Severity of PTSD symptoms in narrative exposure therapy, medium term (11 months)
measured using the CAPS

‐

The mean severity of depressive symptoms in teacher/LHW‐led intervention groups was
0.45 standard deviations lower
(0.99 lower to 0.10 higher)

SMD ‐0.45

(‐0.99 to 0.10)

53

(1 study)

⊕⊕⊕⊝

moderate4,5

Note that a small clinically appreciable benefit was set at SMD < 0.2, a moderate benefit at SMD of 0.5 to 0.8, and a large benefit > 0.8 (Cohen 1988).

Severity of depressive symptoms in teacher/LHW‐led interventions, short term (2‐6 months)
measured using various depression rating scales6

‐

The mean severity of depressive symptoms in teacher/LHW‐led intervention groups was
0.23 standard deviations lower
(0.45 to 0.22 lower)

SMD ‐0.23 (‐0.45 to ‐0.22)

504
(4 studies)

⊕⊕⊝⊝
low7,8

‐

Severity of depressive symptoms in LHW‐led classroom based interventions, short term (1‐6 months)
measured using the DSRS

‐

The mean severity of depressive symptoms in classroom based) LHW‐led intervention groups was
0.18 lower
(0.33 to 0.03 lower)

MCD ‐0.18

(‐0.33 to ‐0.03)

1092
(3 studies)

⊕⊕⊝⊝
low1,8

‐

Severity of depressive symptoms in narrative exposure therapy, medium term (11 months)
measured using the MINI depression rating scale

‐

The mean severity of depressive symptoms in teacher/LHW‐led intervention groups was
0.02 standard deviations lower
(0.52 lower to 0.56 higher)

SMD ‐0.02

(‐0.52 to 0.56)

53 participants (1study)

⊕⊕⊝⊝

low5,9

‐

Severity of anxiety symptoms in LHW‐led classroom based intervention, short term (1‐6 months)
measured by SCARED

‐

The mean severity of anxiety symptoms in the intervention groups was
0.34 lower
(0.75 lower to 0.07 higher)

MD ‐0.34

(‐0.75 to 0.07)

1092
(3 studies)

⊕⊝⊝⊝
very low1,3

‐

Functional impairment in LHW/teacher‐led interventions, short term (1‐6 months)
measured by various functional impairment scales10

‐

The mean functional impairment in teacher‐led interventions was
0.61 standard deviations lower
(1.13 to 0.08 lower)

SMD ‐0.61

(‐1.13 to ‐0.08)

220
(2 studies)

⊕⊕⊕⊝
moderate11,12

‐

Functional impairment in LHW‐led classroom‐based interventions, short term (1‐6 months)
measured by various functional impairment scales13

‐

The mean functional impairment in CBI (classroom based) LHW‐led intervention groups was
0.81 lower
(1.48 to 0.13 lower)

MCD ‐0.81

(‐1.48 to ‐0.13)

1092
(3 studies)

⊕⊕⊝⊝
low1,8

‐

Functional impairment, in narrative exposure therapy, medium term (11 months)
measured using CAPS functional impairment scale

‐

The mean functional impairment in LHW‐led narrative exposure therapy was
0.69 lower
(1.25 to 0.14 lower)

SMD ‐0.69

(‐1.25 to ‐0.14)

53

(1 study)

⊕⊕⊕⊝

moderate14

‐

*The basis for the assumed risk the mean control group risk across studies for pooled results and the control group risk for single studies. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).
CAPS: Clinical Administered PTSD Scale; CBI: classroom‐room‐based intervention; CI: confidence interval; CPSS: Child Posttraumatic Stress Scale; LHW: lay health worker; MCD: mean change difference; MINI: Mini International Neuropsychiatry Interview; NSHWs: non‐specialist health worker; OPHRs: other professionals with health roles; PTSD: post‐traumatic stress disorder; SMD: standardised mean difference; SCARED: Screen for Child Anxiety Related Disorders.

GRADE Working Group grades of evidence
High quality: Further research is very unlikely to change our confidence in the estimate of effect.
Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.
Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.
Very low quality: We are very uncertain about the estimate.

1 Very serious study limitations: Tol 2012 C‐RCT SriLanka: uncertainty about random sequence generation and allocation concealment. Tol 2008 C‐RCT Indonesia: outcome assessment not blind: child self ratings with help of assessors who were not blinded to treatment condition. Jordans 2010 C‐RCT Nepal: unclear allocation concealment and non‐blinded assessment of outcomes. Downgraded by 2.
2 Very serious inconsistency: I2 statistic = 82%. The inconsistency is related to the direction of effect. Downgraded by 2.
3 Serious imprecision: The 95% CI for the pooled estimates indicates appreciable benefit for intervention group and non‐appreciable benefit for usual care. Downgraded by 1.

4 Serious imprecision: The 95% CI for the pooled estimates indicates appreciable benefit for intervention group and non‐appreciable benefit for usual care. The sample size was also small. Downgraded by 1.

5 No indirectness: There was only one study, therefore, the generalisability of this results to other settings is compromised. However, this also resulted in imprecision in the effect estimate, and hence the quality of evidence was not further downgraded.

6Berger2009 CRCT SriLanka: Becks Depression Inventory; Bolton 2007 RCT Uganda: Acholi Psychosocial Assessment Instrument; Dybdahl 2001 RCT Bosnia: Birleson's depression inventory; Ertl 2011 RCT Uganda: MINI.
7 Very serious study limitations: Berger2009 CRCT SriLanka: no allocation concealment, likely contamination and outcomes not adjusted for clustering; Dybdahl 2001 RCT Bosnia: not clear if allocation concealed; differences in baseline characteristics incomplete outcome data (denominators not provided by intervention or control for each of the tests), and likelihood of contamination as both intervention and control in same camps; Bolton 2007 RCT Uganda had unclear allocation concealment, baseline characteristics (age) were different and not adjusted for in analysis, likely risk of contamination between children in both camps. Downgraded by 2.

8 No imprecision: appreciable benefit for LHW/teacher‐led care. Not downgraded.

9 Very serious imprecision: The 95% CI for the pooled estimates indicates appreciable benefit for intervention group and appreciable benefit for usual care. The sample size is also small. Downgraded by 2.

10Berger2009 CRCT SriLanka:: Child Diagnostic Interview Schedule (CDIS); Ertl 2011 RCT Uganda: CAPS ‐ Functional Impairment Section.

11 No imprecision: appreciable benefit for LHW/teacher‐led care. Not downgraded.

12 Serious study limitations: Berger2009 CRCT SriLanka: no allocation concealment, likely contamination and outcomes not adjusted for clustering. Downgraded by 1.

13Jordans 2010 C‐RCT Nepal: Children's Functional Impairment; Tol 2008 C‐RCT Indonesia; Tol 2012 C‐RCT SriLanka: Functional impairment Score (FIS). Data were change scores and hence SMD could not be used. MD is likely to be misleading but since two of the three trials used the same scale this may not be important. Removal of data from Jordans 2010 C‐RCT Nepal did not appreciably alter effect estimates.

14 Serious imprecision: appreciable benefit for LHW/teacher‐led care. However, the sample size is small. Downgraded by 1.

Figures and Tables -
Table 12. SoF 8: NSHWs/OPHRs compared with usual care in conducting interventions for children with post‐traumatic stress disorder and depression (RCTs)
Navigate to table in Review
Table 13. SoF: NSHWs/OPHRs compared with usual care in conducting interventions for children with PTSD in low‐ and middle‐income countries (CBAs)

What are the effects of NSHWs/OPHRs compared with usual care in conducting interventions for children with PTSD for mental health care in low‐ and middle‐income countries? (additional CBA outcomes for comparison 8)

Patient or population: Children with PTSD
Settings: Low‐ and middle‐income countries (Palestine, Turkey)
Intervention: Teachers delivering psychoeducational and other interventions
Comparison: Usual care

Outcomes

Illustrative comparative risks* (95% CI)

Relative effect
(95% CI)

No of Participants
(studies)

Quality of the evidence
(GRADE)

Comments

Assumed risk

Corresponding risk

Usual care

NSHWs/OPHRs

Severity of PTSD symptoms, medium and short term (2‐36 months)
measured using the PTSD symptom scale ‐ CPTSD‐RI

‐

The mean severity of PTSD symptoms in the medium and short term (combined) in the intervention groups was
0.1 standard deviations lower
(0.34 lower to 0.14 higher)

SMD ‐0.10 (‐0.34 to 0.14)

351
(2 studies)

⊕⊝⊝⊝
very low1

‐

Severity of PTSD symptoms ‐ short term (2 months)
measured using the PTSD symptom scale ‐ CPTSD‐RI

‐

The mean severity of PTSD symptoms ‐ short term (within 2 months) in the intervention groups was
0.1 standard deviations higher
(0.41 lower to 0.62 higher)

SMD 0.1 (‐0.41 to 0.62)

64
(1 study)

⊕⊝⊝⊝
very low1,2

Thabet 2005 CBA Palestine

Severity of PTSD symptoms ‐ long term (3 years)
measured using the PTSD symptom scale ‐ CPTSD‐RI

‐

The mean severity of PTSD symptoms ‐ long term ‐ in the intervention groups was
0.16 standard deviations lower
(0.43 lower to 0.12 higher)

SMD ‐0.16 (‐0.43 to 0.12)

287
(1 study)

⊕⊝⊝⊝
very low1,2

Wolmer 2005 CBA Turkey

Severity of CMDs (2‐12 months)
measured using CMD severity scores3

‐

The mean severity of CMDs in the intervention groups was
0.25 standard deviations lower
(0.46 to 0.04 lower)

SMD ‐0.25 (‐0.46 to ‐0.04)

459
(2 studies)

⊕⊝⊝⊝
very low4,5

Both Thabet 2005 CBA Palestine (st) and Loughry 2006 CBA Palestin (LT)

Severity of CMDs psychosocial intervention, short term (2 months)
measured using the Child Depression Inventory

‐

The mean severity of CMDs ‐ short term ‐ in psychosocial interventions was
0.12 standard deviations lower
(0.63 lower to 0.4 higher)

SMD ‐0.12 (‐0.63 to 0.4)

64
(1 study)

⊕⊝⊝⊝
very low4,6

Thabet 2005 CBA Palestine only

Severity of CMDs social intervention ‐ medium term (12 months)
measured using the CBCL internalising score (depression, anxiety, somatic symptoms, withdrawn)

‐

The mean severity of CMDs ‐ medium term ‐ in social interventions was
0.27 standard deviations lower
(0.5 to 0.04 lower)

SMD ‐0.27 (‐0.5 to ‐0.04)

395
(1 study)

⊕⊝⊝⊝
very low4,5

Loughry 2006 CBA Palestin only

*The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% CI) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).
CBA: controlled before‐and‐after; CBCL: Child Behaviour Checklist; CI: confidence interval; CMD: common mental disorders; NSHWs: non‐specialist health worker; OPHRs: other professionals with health roles; PTSD: post‐traumatic stress disorder; SMD: standardised mean difference.

GRADE Working Group grades of evidence
High quality: Further research is very unlikely to change our confidence in the estimate of effect.
Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.
Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.
Very low quality: We are very uncertain about the estimate.

1 No imprecision: Unappreciable benefit for both NSHW care and usual care.
2 Very serious risk of bias: both CBA studies so non‐randomised and no allocation concealment. Also Thabet 2005 CBA Palestine: differences in baseline outcomes; Wolmer 2005 CBA Turkey: unvalidated tools in this setting so uncertain reliability of outcomes.
3Thabet 2005 CBA Palestine: Child Depression Inventory; Loughry 2006 CBA Palestin: CBCL internalising score (depression, anxiety, somatic symptoms, withdrawn).
4 Very serious risk of bias: both CBAs so non‐randomised and no allocation concealment. Also Thabet 2005 CBA Palestine: differences in baseline outcomes; Loughry 2006 CBA Palestin: differences in baseline outcomes and characteristics and risk of contamination and the intervention between sites differed slightly making comparisons difficult.
5 No imprecision: Non‐appreciable, and possibly appreciable benefit for intervention group.
6 Very serious imprecision: Appreciable benefit for intervention and non‐appreciable benefit for control group.

Figures and Tables -
Table 13. SoF: NSHWs/OPHRs compared with usual care in conducting interventions for children with PTSD in low‐ and middle‐income countries (CBAs)
Navigate to table in Review
Table 14. Summary of costs and resource use from included studies

Author/year

Type of economic evaluation

Study population

Intervention

Economic results

Araya 2003 RCT Chile

Cost‐effectiveness analysis

Women with depression

Collaborative intervention (doctors, non‐medical professionals supervised by psychiatrist) with stepped care, multicomponent programme compared with usual care in depressed women in Chile

Incremental cost per person for improved care was USD37.6 more than usual care. Unit cost to obtain 1 additional depression‐free day was USD0.75

Jordans 2011

Cost analysis

Children with PTSD (7‐15 years)

LHW‐led multilayered package (including classroom‐based intervention, non‐therapeutic resilience groups, psychoeducation and counselling) (data extracted from Sri Lanka and Indonesia as related to Tol 2008 C‐RCT Indonesia and Tol 2012 C‐RCT SriLanka)

Mean cost per user of total package:

Indonesia: USD21.77 (59% of which is human resources cost). Sri Lanka: USD8.85 (56% of which is human resources cost)

Zambori 2002 CBA Hungary

Cost analysis

Patients with anxiety and mood disorders

Primary physicians versus psychiatrists in prescribing sertraline in Hungary

Absenteeism reduced from 15.7 to 6.8 days and costs of non‐psychiatric prescriptions decreased from USD138 to USD91.8 per year. Laboratory costs ranged from USD6.4 to USD11.5

LHW: lay health worker; PTSD: post‐traumatic stress disorder.
Figures and Tables -
Table 14. Summary of costs and resource use from included studies
Navigate to table in Review
Table 15. Agreements and disagreements with related reviews

Author/year

Summary of review

Agreements

Disagreements/differences

Parker 2008

Reviewed consultation liaison in primary care ‐ HICs

‐

Our review process did not find any consultation liaison in primary care in LMICs so results cannot be compared

Boer 2005

Reviewed paraprofessionals in delivering psychological interventions for anxiety and depression (HIC only)

Included studies were from HICs only, but support our findings that non‐professional care is generally equivalent to professional care (this review's equivalent of specialist care), and that non‐professional care is better than usual care

Some of their paraprofessionals would have been classified as specialist health workers in our review

Bower 2006

Reviewed the effect of collaborative care models on antidepressant use.

All included studies were from HICs except for

Araya 2003 RCT Chile

Bower found improvement of antidepressant use, particularly in studies where the case manager had a mental health background, where there was adequate supervision and where there was systematic identification of patients (rather than waiting for a referral)

We were not able to assess, as did Bower, whether lengths of training, supervision or other intervention characteristics modified these outcomes because only 5 studies were included in this comparison

Woltmann 2012

Review on collaborative care/chronic care management

They also found a statistically significant effect on reduction in depression severity among the 14 HIC studies that were included in the meta‐analysis (SMD 0.31, 95% CI 0.16 to 0.47) (Araya and Patel's studies were included in the narrative review but did not qualify for their meta‐analysis). The authors suggested that collaborative care is of moderate benefit; however, Woltmann has estimated a more conservative value of SMD > 0.5 to show moderate benefit (from the analysis of scales and how to interpret their SMDs). Our meta‐analyses of collaborative care models suggested similar improvements in symptoms and recovery from depression or CMDs  (same direction of effect, and similar magnitude)

Woltman's chronic care management had a stricter definition to our collaborative care definition

Huntley 2012

Reviewed the effect of CBT and group CBT

Huntley also found that LHW‐led psychological interventions are effective in the short and medium term in reducing symptoms of depression

Huntley described the effect of CBT and group CBT (rather than the effect of NSHWs)

Tol 2011

Systematic review on mental health interventions in humanitarian settings

Tol found similar results to our review for school‐based interventions for children with PTSD (i.e. no significant benefit) (an extra study was included in this comparison, which we had excluded as it did not meet our NSHW/OPHR definitions). This review went further and found a statistically significant benefit for improving internalising symptoms (SMD ‐0.34, 95% CI ‐0.40 to ‐0.09). For adults, a potential benefit of interventions was also seen

This review differed from ours in that it included studies of both NSHWs/OPHRs and specialists, according to our definitions

Rahman 2013

Systematic review on interventions for common perinatal mental disorders in women in LMICs

This was similar but a more in‐depth review of our perinatal depression pooled comparison, which also looked at LHW‐led interventions for mothers with perinatal depression. Their final pooled outcome was similar in magnitude and direction to ours for our perinatal depression category (SMD ‐0.38, 95% CI ‐0.56 to ‐0.21) vs, our findings (SMD ‐0.42, 95% CI ‐0.58 to ‐0.26)

This review differed from ours in that its study's inclusion criteria were broader as it included studies that measured maternal (all perinatal disorders) or child (or both) outcomes even if the intervention was not primarily targeted at these groups. It also reported child outcomes, which ours did not

CBT: cognitive behavioural therapy; CI: confidence interval; CMD: common mental disorders; HIC: high‐income country; LHW: lay health worker; LMIC: low‐ and medium‐income countries; NSHW: non‐specialist health worker; OPHR: other professionals with health roles; PTSD: post‐traumatic stress disorder; SMD: standardised mean difference.

Figures and Tables -
Table 15. Agreements and disagreements with related reviews
Navigate to table in Review
Comparison 1. NSHW‐led psychological interventions versus usual care in treating common mental disorders in adults (RCTs)

Outcome or subgroup title

No. of studies

No. of participants

Statistical method

Effect size

1 Prevalence of depression (completers) Show forest plot

3

1082

Risk Ratio (Random, 95% CI)

0.30 [0.14, 0.64]

1.1 Short term (within 6 months post intervention)

3

1082

Risk Ratio (Random, 95% CI)

0.30 [0.14, 0.64]

2 Prevalence of depression (ITT sensitivity analysis ‐ assumption non‐completers depressed) Show forest plot

3

Risk Ratio (Random, 95% CI)

Subtotals only

2.1 Short term (within 6 months post intervention)

3

1359

Risk Ratio (Random, 95% CI)

0.61 [0.43, 0.84]

3 Prevalence of depression (ITT sensitivity analysis ‐ assumption non‐completers not depressed) Show forest plot

3

Risk Ratio (Random, 95% CI)

Subtotals only

3.1 Short term (within 6 months post intervention)

3

1359

Risk Ratio (Random, 95% CI)

0.39 [0.20, 0.78]

4 Prevalence of depression (ITT sensitivity analysis ‐ worse‐case scenario intervention group depressed; control group not depressed) Show forest plot

3

Risk Ratio (Random, 95% CI)

Subtotals only

4.1 Short term (within 6 months post intervention)

3

1359

Risk Ratio (Random, 95% CI)

1.11 [0.56, 2.21]

5 Prevalence of depression (ITT sensitivity analysis ‐ best‐case scenario: intervention group not depressed; control group all depressed) Show forest plot

3

Risk Ratio (Random, 95% CI)

Subtotals only

5.1 Short term (within 6 months post intervention)

3

1359

Risk Ratio (Random, 95% CI)

0.20 [0.09, 0.45]

6 Severity of common mental disorder symptoms (includes anxiety and depression) Show forest plot

7

Std. Mean Difference (Random, 95% CI)

Subtotals only

6.1 Short term (within 6 months post intervention)

6

1470

Std. Mean Difference (Random, 95% CI)

‐0.75 [‐1.29, ‐0.21]

6.2 Medium term (1 year)

2

923

Std. Mean Difference (Random, 95% CI)

‐0.47 [‐0.60, ‐0.34]

7 Functional impairment/disability in common mental disorders Show forest plot

4

Std. Mean Difference (Random, 95% CI)

Subtotals only

7.1 Short term (within 6 months post intervention)

4

1243

Std. Mean Difference (Random, 95% CI)

‐0.33 [‐0.80, 0.13]

7.2 Short term (advocacy empowerment physical functioning) short term (6 months post intervention)

1

200

Std. Mean Difference (Random, 95% CI)

0.08 [‐0.20, 0.36]

7.3 Medium term (8 months post intervention)

1

798

Std. Mean Difference (Random, 95% CI)

‐0.56 [‐0.70, ‐0.42]
Figures and Tables -
Comparison 1. NSHW‐led psychological interventions versus usual care in treating common mental disorders in adults (RCTs)
Navigate to table in Review
Comparison 2. Collaborative care model (NSHWs plus specialist) versus usual care in treating common mental disorders (RCTs)

Outcome or subgroup title

No. of studies

No. of participants

Statistical method

Effect size

1 Prevalence of common mental disorders (CMDs ‐ includes anxiety and depression) (completers‐combined) all facilities and in public and private facilities Show forest plot

3

Risk Ratio (Random, 95% CI)

Subtotals only

1.1 All facilities short term (within 6 months post intervention)

3

2380

Risk Ratio (Random, 95% CI)

0.63 [0.44, 0.90]

1.2 Public facilities short term (within 6 months post intervention) (subgroup)

3

1528

Risk Ratio (Random, 95% CI)

0.57 [0.42, 0.78]

1.3 Private facilities short term (within 6 months post intervention) (subgroup)

1

823

Risk Ratio (Random, 95% CI)

1.12 [0.68, 1.84]

1.4 All facilities medium term (at 1 year post intervention)

1

2009

Risk Ratio (Random, 95% CI)

0.95 [0.68, 1.33]

1.5 Public facilities medium term (at 1 year post intervention) (subgroup)

1

1104

Risk Ratio (Random, 95% CI)

0.72 [0.39, 1.34]

1.6 Private facilities at 1 year post intervention

1

801

Risk Ratio (Random, 95% CI)

1.25 [0.76, 2.06]

2 Severity of symptoms of CMDs (completers‐combined) in all facilities and in public and private facilities Show forest plot

5

Std. Mean Difference (Random, 95% CI)

Subtotals only

2.1 All facilities short term (within 6 months post intervention)

5

3604

Std. Mean Difference (Random, 95% CI)

‐0.31 [‐0.56, ‐0.06]

2.2 Public facilities short term (within 6 months post intervention) (subgroup)

5

2781

Std. Mean Difference (Random, 95% CI)

‐0.32 [‐0.58, ‐0.07]

2.3 Private facilities short term (within 6 months post intervention) (subgroup)

1

823

Std. Mean Difference (Random, 95% CI)

0.03 [‐0.11, 0.16]

2.4 All facilities medium term (at 1 year post intervention)

1

1905

Std. Mean Difference (Random, 95% CI)

‐0.03 [‐0.12, 0.06]

2.5 Public facilities medium term (at 1 year post intervention) (subgroup)

1

1104

Std. Mean Difference (Random, 95% CI)

‐0.07 [‐0.19, 0.05]

2.6 Private facilities medium term (at 1 year post intervention) (subgroup)

1

801

Std. Mean Difference (Random, 95% CI)

0.02 [‐0.11, 0.16]

3 Functional impairment/disability in CMD (completers‐ combined) all facilities and in public and private facilities (SMD) Show forest plot

5

Std. Mean Difference (Random, 95% CI)

Subtotals only

3.1 All facilities short term (within 6 months post intervention)

5

3604

Std. Mean Difference (Random, 95% CI)

‐0.22 [‐0.44, ‐0.01]

3.2 Public facilities short term (within 6 months post intervention) (subgroup)

5

2781

Std. Mean Difference (Random, 95% CI)

‐0.24 [‐0.45, ‐0.02]

3.3 Private facilities short term (within 6 months post intervention) (subgroup)

1

823

Std. Mean Difference (Random, 95% CI)

0.02 [‐0.12, 0.15]

3.4 All facilities medium term (at 1 year post intervention)

1

1905

Std. Mean Difference (Random, 95% CI)

‐0.02 [‐0.11, 0.07]

3.5 Public facilities medium term (at 1 year post intervention) (subgroup)

1

1104

Std. Mean Difference (Random, 95% CI)

‐0.05 [‐0.17, 0.07]

3.6 Private facilities medium term (at 1 year post intervention) (subgroup)

1

801

Std. Mean Difference (Random, 95% CI)

0.03 [‐0.11, 0.17]

4 Suicide attempt for those with CMDs all facilities and in public/private facilities (completers) Show forest plot

1

Risk Ratio (M‐H, Random, 95% CI)

Subtotals only

4.1 All facilities short term (6 months post intervention)

1

1961

Risk Ratio (M‐H, Random, 95% CI)

0.75 [0.29, 1.97]

4.2 Public facilities short term (6 months post intervention)

1

1138

Risk Ratio (M‐H, Random, 95% CI)

0.66 [0.32, 1.40]

4.3 Private facilities short term (6 months post intervention)

1

823

Risk Ratio (M‐H, Random, 95% CI)

0.71 [0.12, 4.22]

4.4 All facilities medium term (1 year post intervention)

1

1905

Risk Ratio (M‐H, Random, 95% CI)

0.56 [0.24, 1.32]

4.5 Public facilities medium term (1 year post intervention)

1

1104

Risk Ratio (M‐H, Random, 95% CI)

0.78 [0.18, 3.48]

4.6 Private facilities medium term (1 year post intervention)

1

801

Risk Ratio (M‐H, Random, 95% CI)

0.40 [0.11, 1.50]

5 Prevalence of CMDs (only Patel ‐ sensitivity analysis (SA)) (completers) all facilities and in public and private facilities Show forest plot

1

Risk Ratio (Random, 95% CI)

Subtotals only

5.1 All facilities short term (within 6 months post intervention)

1

1961

Risk Ratio (Random, 95% CI)

0.80 [0.61, 1.05]

5.2 Public facilities short term (within 6 months post intervention) (subgroup)

1

1109

Risk Ratio (Random, 95% CI)

0.59 [0.41, 0.85]

5.3 Private facilities at 6 months post intervention

1

823

Risk Ratio (Random, 95% CI)

1.12 [0.68, 1.84]

5.4 All facilities medium term (at 1 year post intervention)

1

2009

Risk Ratio (Random, 95% CI)

0.95 [0.68, 1.33]

5.5 Public facilities medium term (at 1 year post intervention) (subgroup)

1

1104

Risk Ratio (Random, 95% CI)

0.72 [0.39, 1.34]

5.6 Private facilities at 1 year post intervention

1

801

Risk Ratio (Random, 95% CI)

1.25 [0.76, 2.06]

6 Severity of symptoms in CMD (only Patel and Jenkins (SA)) in all facilities and in public and private facilities Show forest plot

2

Std. Mean Difference (Random, 95% CI)

Subtotals only

6.1 All facilities short term (within 6 months post intervention)

2

2889

Std. Mean Difference (Random, 95% CI)

‐0.07 [‐0.15, ‐0.00]

6.2 Public facilities short term (within 6 months post intervention) (subgroup)

2

2066

Std. Mean Difference (Random, 95% CI)

‐0.11 [‐0.21, ‐0.00]

6.3 Private facilities short term (within 6 months post intervention) (subgroup)

1

823

Std. Mean Difference (Random, 95% CI)

0.03 [‐0.11, 0.16]

6.4 All facilities medium term (at 1 year post intervention)

1

1905

Std. Mean Difference (Random, 95% CI)

‐0.03 [‐0.12, 0.06]

6.5 Public facilities medium term (at 1 year post intervention) (subgroup)

1

1104

Std. Mean Difference (Random, 95% CI)

‐0.07 [‐0.19, 0.05]

6.6 Private facilities medium term (at 1 year post intervention) (subgroup)

1

801

Std. Mean Difference (Random, 95% CI)

0.02 [‐0.11, 0.16]

7 Prevalence of depression (completers) (SA) all facilities and in public and private facilities Show forest plot

3

Risk Ratio (Random, 95% CI)

Subtotals only

7.1 All facilities short term (within 6 months post intervention)

3

1092

Risk Ratio (Random, 95% CI)

0.61 [0.40, 0.94]

7.2 Public facilities short term (within 6 months post intervention) (subgroup)

3

828

Risk Ratio (Random, 95% CI)

0.56 [0.37, 0.84]

7.3 Private facilities short term (within 6 months post intervention) (subgroup)

1

254

Risk Ratio (Random, 95% CI)

1.59 [0.40, 6.32]

7.4 All facilities medium term (1 year post intervention)

1

652

Risk Ratio (Random, 95% CI)

0.95 [0.68, 1.33]

7.5 Public facilities medium term (1 year post intervention) (subgroup)

1

398

Risk Ratio (Random, 95% CI)

0.72 [0.39, 1.34]

7.6 Private facilities medium term (at 1 year post intervention) (subgroup)

1

254

Risk Ratio (Random, 95% CI)

1.25 [0.76, 2.06]

8 Severity of symptoms of depression (SA) in all facilities and in public and private facilities Show forest plot

4

Std. Mean Difference (Random, 95% CI)

Subtotals only

8.1 All facilities short term (within 6 months post intervention)

4

1388

Std. Mean Difference (Random, 95% CI)

‐0.39 [‐0.78, 0.01]

8.2 Public facilities short term (within 6 months post intervention) (subgroup)

4

1124

Std. Mean Difference (Random, 95% CI)

‐0.41 [‐0.79, ‐0.04]

8.3 Private facilities short term (within 6 months post intervention) (subgroup)

1

254

Std. Mean Difference (Random, 95% CI)

0.16 [‐0.09, 0.41]

8.4 All facilities medium term (at 1 year post intervention)

1

652

Std. Mean Difference (Random, 95% CI)

0.11 [‐0.05, 0.26]

8.5 Public facilities medium term (at 1 year post intervention) (subgroup)

1

398

Std. Mean Difference (Random, 95% CI)

‐0.09 [‐0.29, 0.12]

8.6 Private facilities medium term (at 1 year post intervention) (subgroup)

1

254

Std. Mean Difference (Random, 95% CI)

‐0.03 [‐0.28, 0.22]

9 Functional impairment/disability in CMD (SA) all facilities and in public and private facilities (SMD) Show forest plot

2

Std. Mean Difference (Random, 95% CI)

Subtotals only

9.1 All facilities short term (within 6 months post intervention)

2

2889

Std. Mean Difference (Random, 95% CI)

‐0.03 [‐0.10, 0.04]

9.2 Public facilities short term (within 6 months post intervention) (subgroup)

2

2066

Std. Mean Difference (Random, 95% CI)

‐0.06 [‐0.15, 0.02]

9.3 Private facilities short term (within 6 months post intervention) (subgroup)

1

823

Std. Mean Difference (Random, 95% CI)

0.02 [‐0.12, 0.15]

9.4 All facilities medium term (at 1 year post intervention)

1

1905

Std. Mean Difference (Random, 95% CI)

‐0.02 [‐0.11, 0.07]

9.5 Public facilities medium term (at 1 year post intervention) (subgroup)

1

1104

Std. Mean Difference (Random, 95% CI)

‐0.05 [‐0.17, 0.07]

9.6 Private facilities medium term (at 1 year post intervention) (subgroup)

1

801

Std. Mean Difference (Random, 95% CI)

0.03 [‐0.11, 0.17]

10 Functional impairment/disability in CMD (SA) all facilities and in public and private facilities (MD) Show forest plot

2

Mean Difference (Random, 95% CI)

Subtotals only

10.1 All facilities short term (within 6 months post intervention)

2

2889

Mean Difference (Random, 95% CI)

‐0.53 [‐2.06, 1.01]

10.2 Public facilities short term (within 6 months post intervention) (subgroup)

2

2066

Mean Difference (Random, 95% CI)

‐1.24 [‐2.94, 0.46]

10.3 Private facilities short term (within 6 months post intervention) (subgroup)

1

823

Mean Difference (Random, 95% CI)

0.35 [‐2.39, 3.09]

10.4 All facilities medium term (at 1 year post intervention)

1

1905

Mean Difference (Random, 95% CI)

‐0.41 [‐2.37, 1.55]

10.5 Public facilities medium term (at 1 year post intervention) (subgroup)

1

1104

Mean Difference (Random, 95% CI)

‐1.49 [‐4.93, 1.95]

10.6 Private facilities medium term (at 1 year post intervention) (subgroup)

1

801

Mean Difference (Random, 95% CI)

0.83 [‐2.32, 3.98]

11 Functional impairment/disability in depression (SA) all facilities and in public and private facilities Show forest plot

4

Std. Mean Difference (Random, 95% CI)

Subtotals only

11.1 All facilities short term (within 6 months post intervention)

4

3144

Std. Mean Difference (Random, 95% CI)

‐0.29 [‐0.62, 0.04]

11.2 Public facilities short term (within 6 months post intervention) (subgroup)

4

2131

Std. Mean Difference (Random, 95% CI)

‐0.31 [‐0.62, 0.00]

11.3 Private facilities short term (at 6 months post intervention) (subgroup)

1

1013

Std. Mean Difference (Random, 95% CI)

0.06 [‐0.18, 0.31]

11.4 All facilities medium term (at 1 year post intervention)

1

2367

Std. Mean Difference (Random, 95% CI)

‐0.07 [‐0.23, 0.09]

11.5 Public facilities medium term (at 1 year post intervention) (subgroup)

1

1416

Std. Mean Difference (Random, 95% CI)

‐0.09 [‐0.29, 0.12]

11.6 Private facilities medium term (at 1 year post intervention) (subgroup)

1

981

Std. Mean Difference (Random, 95% CI)

‐0.02 [‐0.27, 0.23]
Figures and Tables -
Comparison 2. Collaborative care model (NSHWs plus specialist) versus usual care in treating common mental disorders (RCTs)
Navigate to table in Review
Comparison 3. NSHWs versus usual care in treating maternal depression (RCTs)

Outcome or subgroup title

No. of studies

No. of participants

Statistical method

Effect size

1 Severity of symptoms in treating maternal depression Show forest plot

4

1213

Std. Mean Difference (Random, 95% CI)

‐0.42 [‐0.58, ‐0.26]

1.1 NSHW‐led interventions short term (within 3 months post intervention)

2

858

Std. Mean Difference (Random, 95% CI)

‐0.50 [‐0.63, ‐0.36]

1.2 Collaborative care short term (at 3 months post intervention)

1

230

Std. Mean Difference (Random, 95% CI)

‐0.22 [‐0.48, 0.04]

1.3 NSHW‐led intervention medium term (at 1 year post intervention)

1

125

Std. Mean Difference (Random, 95% CI)

‐0.41 [‐0.76, ‐0.06]
Figures and Tables -
Comparison 3. NSHWs versus usual care in treating maternal depression (RCTs)
Navigate to table in Review
Comparison 4. NSHWs versus specialists in treating depression in adults (controlled before‐and‐after studies)

Outcome or subgroup title

No. of studies

No. of participants

Statistical method

Effect size

1 Severity of depression short term (2 months post intervention) Show forest plot

1

768

Mean Difference (IV, Random, 95% CI)

‐0.90 [‐1.20, ‐0.60]

2 Frequency of adverse events Show forest plot

1

768

Risk Ratio (M‐H, Random, 95% CI)

0.85 [0.67, 1.07]

3 Number of days spent in hospital Show forest plot

1

Mean Difference (IV, Random, 95% CI)

Subtotals only

3.1 Outcomes at 1 year

1

124

Mean Difference (IV, Random, 95% CI)

‐1.79 [‐3.59, 0.01]

3.2 Outcomes at 2 years

1

124

Mean Difference (IV, Random, 95% CI)

‐0.02 [‐2.59, 2.55]

4 Number of days spent on sick leave Show forest plot

1

Mean Difference (IV, Random, 95% CI)

Subtotals only

4.1 Outcome at 1 year

1

108

Mean Difference (IV, Random, 95% CI)

‐3.96 [‐15.58, 7.66]

4.2 Outcome at 2 years

1

123

Mean Difference (IV, Random, 95% CI)

14.63 [‐0.76, 30.02]
Figures and Tables -
Comparison 4. NSHWs versus specialists in treating depression in adults (controlled before‐and‐after studies)
Navigate to table in Review
Comparison 5. NSHW‐led psychological interventions versus usual care in treating adults with post‐traumatic stress disorder (RCT and NRCT)

Outcome or subgroup title

No. of studies

No. of participants

Statistical method

Effect size

1 Prevalence of post‐traumatic stress disorder (PTSD) Show forest plot

1

Risk Ratio (M‐H, Fixed, 95% CI)

Subtotals only

1.1 LHW‐led narrative exposure therapy short term (6 months post intervention)

1

62

Risk Ratio (M‐H, Fixed, 95% CI)

0.48 [0.27, 0.85]

1.2 LHW‐led trauma counselling short term (6 months post intervention)

1

65

Risk Ratio (M‐H, Fixed, 95% CI)

0.55 [0.33, 0.93]

2 Severity of PTSD symptoms Show forest plot

3

Std. Mean Difference (IV, Random, 95% CI)

Subtotals only

2.1 Short term LHW‐led counselling with PTSD psychoeducation (6 months post intervention)

3

223

Std. Mean Difference (IV, Random, 95% CI)

‐0.36 [‐0.67, ‐0.05]

2.2 Short term (Yeomans second arm) (2 weeks post intervention)

1

75

Std. Mean Difference (IV, Random, 95% CI)

‐0.44 [‐0.90, 0.02]

2.3 Short term (Neuner first arm ‐ narrative exposure therapy) (6 months post intervention)

1

75

Std. Mean Difference (IV, Random, 95% CI)

‐0.55 [‐1.08, ‐0.03]

3 Severity of depression Show forest plot

1

Mean Difference (IV, Random, 95% CI)

Subtotals only

3.1 LHW‐led workshop with psychoeducation short term (2 weeks post intervention)

1

76

Mean Difference (IV, Random, 95% CI)

‐0.07 [‐0.36, 0.22]

3.2 LHW‐led workshop without psychoeducation short term (2 weeks post intervention)

1

75

Mean Difference (IV, Random, 95% CI)

‐0.14 [‐0.42, 0.14]
Figures and Tables -
Comparison 5. NSHW‐led psychological interventions versus usual care in treating adults with post‐traumatic stress disorder (RCT and NRCT)
Navigate to table in Review
Comparison 6. NSHWs versus usual care in improving dementia patients' and carers' outcomes (RCTs)

Outcome or subgroup title

No. of studies

No. of participants

Statistical method

Effect size

1 Severity of behavioural problem (patient) Show forest plot

2

134

Std. Mean Difference (Random, 95% CI)

‐0.26 [‐0.60, 0.08]

2 Patient functional ability Show forest plot

1

81

Mean Difference (Random, 95% CI)

‐0.24 [‐0.67, 0.20]

3 Patient quality of life Show forest plot

1

53

Mean Difference (Random, 95% CI)

‐0.43 [‐0.98, 0.12]

4 Carer mental health status Show forest plot

2

134

Std. Mean Difference (Random, 95% CI)

‐0.42 [‐0.76, ‐0.08]

5 Carer burden Show forest plot

2

134

Std. Mean Difference (Random, 95% CI)

‐0.50 [‐0.84, ‐0.15]

6 Carer distress Show forest plot

2

134

Std. Mean Difference (Random, 95% CI)

‐0.47 [‐0.82, ‐0.13]

7 Carer quality of life Show forest plot

1

53

Mean Difference (Random, 95% CI)

‐0.37 [‐0.92, 0.17]
Figures and Tables -
Comparison 6. NSHWs versus usual care in improving dementia patients' and carers' outcomes (RCTs)
Navigate to table in Review
Comparison 7. NSHW‐led brief alcohol interventions versus usual care for adults with alcohol‐use disorders (RCTs)

Outcome or subgroup title

No. of studies

No. of participants

Statistical method

Effect size

1 Amount of alcohol consumed (MD) Show forest plot

2

167

Mean Difference (Random, 95% CI)

‐1.68 [‐2.79, ‐0.57]

2 Frequency of binge drinking Show forest plot

1

92

Mean Difference (IV, Random, 95% CI)

‐0.50 [‐1.14, 0.14]

3 Adverse consequences Show forest plot

2

160

Risk Ratio (M‐H, Random, 95% CI)

0.77 [0.11, 5.29]

3.1 Road traffic accidents

1

92

Risk Ratio (M‐H, Random, 95% CI)

0.36 [0.12, 1.08]

3.2 Withdrawal symptoms

1

68

Risk Ratio (M‐H, Random, 95% CI)

2.67 [0.29, 24.37]
Figures and Tables -
Comparison 7. NSHW‐led brief alcohol interventions versus usual care for adults with alcohol‐use disorders (RCTs)
Navigate to table in Review
Comparison 8. NSHWs/OPHRs versus usual care in conducting interventions for children with post‐traumatic stress and depression (RCTs)

Outcome or subgroup title

No. of studies

No. of participants

Statistical method

Effect size

1 Severity of PTSD symptoms ‐ teacher/LHW‐led interventions (SMDs) Show forest plot

3

Std. Mean Difference (Random, 95% CI)

Subtotals only

1.1 Short term (within 6 months post intervention)

3

298

Std. Mean Difference (Random, 95% CI)

‐0.89 [‐1.49, ‐0.30]

1.2 Short term (Ertl second arm) (5 months post intervention)

1

51

Std. Mean Difference (Random, 95% CI)

‐0.12 [‐0.67, 0.44]

1.3 Medium term (Ertl first arm) (11 months post intervention)

1

53

Std. Mean Difference (Random, 95% CI)

‐0.45 [‐0.99, 0.10]

1.4 Medium term (Ertl second arm) (11 months post intervention)

1

51

Std. Mean Difference (Random, 95% CI)

‐0.04 [‐0.59, 0.52]

2 Severity of PTSD symptoms ‐ classroom‐based LHW interventions (MCDs) Show forest plot

3

Mean Change Difference (Random, 95% CI)

Subtotals only

2.1 Short term (within 6 months post intervention)

3

1090

Mean Change Difference (Random, 95% CI)

‐0.56 [‐2.82, 1.70]

3 Severity of PTSD symptoms ‐ classroom‐based LHW interventions ‐ boys/girls Show forest plot

1

399

Mean Change Difference (Random, 95% CI)

1.40 [‐1.58, 4.37]

3.1 Short term (boys) (within 6 months post intervention)

1

245

Mean Change Difference (Random, 95% CI)

0.0 [‐2.02, 2.02]

3.2 Short term (girls) (within 6 months post intervention)

1

154

Mean Change Difference (Random, 95% CI)

3.05 [0.39, 5.71]

4 Severity of depressive symptoms ‐ teacher/LHW‐led interventions (SMDs) Show forest plot

4

Std. Mean Difference (Random, 95% CI)

Subtotals only

4.1 Short term (within 6 months post intervention)

4

504

Std. Mean Difference (Random, 95% CI)

‐0.23 [‐0.45, ‐0.02]

4.2 Short term (Bolton second arm) (6 months post intervention)

1

209

Std. Mean Difference (Random, 95% CI)

0.08 [‐0.20, 0.35]

4.3 Short term (Ertl second arm) (5 months post intervention)

1

51

Std. Mean Difference (Random, 95% CI)

0.03 [‐0.52, 0.58]

4.4 Medium term (Ertl first arm) (11 months post intervention)

1

53

Std. Mean Difference (Random, 95% CI)

0.02 [‐0.52, 0.56]

4.5 Medium term (Ertl second arm) (11 months post intervention)

1

51

Std. Mean Difference (Random, 95% CI)

0.17 [‐0.38, 0.72]

5 Severity of depressive symptoms ‐ classroom‐based LHW interventions (MCDs) Show forest plot

3

Mean Change Difference (Random, 95% CI)

Subtotals only

5.1 Short term (within 6 months post intervention)

3

1092

Mean Change Difference (Random, 95% CI)

‐0.18 [‐0.33, ‐0.03]

6 Severity of depressive symptoms (MCDs) Tol 2012 boys/girls Show forest plot

1

399

Mean Change Difference (Random, 95% CI)

0.27 [‐0.58, 1.12]

6.1 Short term (boys) (within 6 months post intervention)

1

245

Mean Change Difference (Random, 95% CI)

‐0.02 [‐1.18, 1.14]

6.2 Short term (girls) (within 6 months post intervention)

1

154

Mean Change Difference (Random, 95% CI)

0.61 [‐0.63, 1.85]

7 Severity of anxiety symptoms ‐ classroom‐based intervention (within 6 months post intervention) Show forest plot

3

1092

Mean Change Difference (Random, 95% CI)

‐0.34 [‐0.75, 0.07]

7.1 Short term (within 6 months post intervention)

3

1092

Mean Change Difference (Random, 95% CI)

‐0.34 [‐0.75, 0.07]

8 Severity of anxiety symptoms ‐ classroom‐based intervention ‐ boys/girls Show forest plot

1

399

Mean Change Difference (Random, 95% CI)

‐0.22 [‐1.09, 0.65]

8.1 Short term (boys) (within 6 months post intervention)

1

245

Mean Change Difference (Random, 95% CI)

‐0.63 [‐1.23, ‐0.03]

8.2 Short term (girls) (within 6 months post intervention)

1

154

Mean Change Difference (Random, 95% CI)

0.26 [‐0.53, 1.05]

9 Functional impairment teacher/LHW‐led interventions Show forest plot

2

Std. Mean Difference (Random, 95% CI)

Subtotals only

9.1 Short term (within 6 months post intervention)

2

220

Std. Mean Difference (Random, 95% CI)

‐0.61 [‐1.13, ‐0.08]

9.2 Medium term (11 months post intervention)

1

53

Std. Mean Difference (Random, 95% CI)

‐0.69 [‐1.25, ‐0.14]

10 Functional impairment LHW‐led ‐ classroom‐based intervention Show forest plot

3

Mean Change Difference (Random, 95% CI)

Subtotals only

10.1 Short term (within 6 months post intervention)

3

1092

Mean Change Difference (Random, 95% CI)

‐0.81 [‐1.48, ‐0.13]

11 Functional impairment ‐ classroom‐based LHW intervention ‐ boys/girls Show forest plot

1

399

Mean Difference (Random, 95% CI)

‐0.94 [‐1.80, ‐0.08]

11.1 Short term (boys) (within 6 months post intervention)

1

245

Mean Difference (Random, 95% CI)

‐1.19 [‐2.23, ‐0.15]

11.2 Short term (girls) (within 6 months post intervention)

1

154

Mean Difference (Random, 95% CI)

‐0.40 [‐1.93, 1.13]
Figures and Tables -
Comparison 8. NSHWs/OPHRs versus usual care in conducting interventions for children with post‐traumatic stress and depression (RCTs)
Navigate to table in Review