Methods
Review Format
This review has followed the Evidence for Policy and Practice Information and Co-ordinating Centre's (EPPI-Centre) Methods for Conducting a Systematic Review. The EPPI format allows for a greater variety of study designs to be included, as well as a mixed-methods synthesis when appropriate whereas traditional systematic reviews (e.g. Cochrane method) include only Randomised Controlled Trials (RCTs) and are interested in combining numerical data in the form of a meta-analysis for data synthesis. The EPPI-Centre focuses on social science and public policy reviews, and like other systematic review institutes, requires accountability, rigour and explicit methods in conducting a review. There is no review protocol of this review.
A narrative synthesis is used to present and analyse findings. As this review includes both experimental and observational studies with large heterogeneity in interventions and measures, a statistical analysis is inappropriate. When reviews include studies that cannot be combined statistically but are still undertaken with the same amount of rigour and quality, they are classified as qualitative systematic reviews.
Search Strategy
Due to the diversity of potential interventions, populations, study types and outcomes, a multi-stage search strategy was developed to identify relevant publications. The search terminology was adapted, with permission, from Lewin et al., which included the methodological component of Cochrane's Effective Practice and Organisation of Care Group (EPOC) guidelines, combined with free text and key terms. The Lewin et al. search terminology was then adjusted to be more relevant to this study's research question by adding both key terms and free text terms relating to the topic that were identified by various scoping searches, see Additional file 1. Alternative names for Community Health Workers, see Additional file 2, identified through various literature sources, were added.
As recommended by the EPPI-Centre the strategy attempted to balance sensitivity with specificity in its results; however, due to the large amount of heterogeneity in inclusion criteria the strategy was quite sensitive (i.e. produced large quantities of articles irrelevant to the topic). The finalized search strategy was then modified to fit with the different electronic databases' nomenclature. The following databases were searched from June 8 – 11, 2012: CINAHL, Embase, Ovid Nursing Database, PubMed, Scopus, Web of Science and POPLINE. In addition, the Google and Google Scholar search engines, as well as The World Health Organization's website were searched for relevant articles on June 11, 2012. Reference lists from other related systematic reviews were also searched as well as the references from articles that were identified for inclusion in the review. Article screening was conducted in several stages by one reviewer with the assistance of a second reviewer in the final stage.
Inclusion and Exclusion Criteria
Target beneficiaries include women pregnant to 42 days post termination, children under 5 and primary caregivers of children under 5 who have the ability to influence the child's health.
Interventions were required to be preventive and have been delivered at the household level by Community Health Workers, using the WHO Study Group definition of "…members of the communities where they work, should be selected by the communities, should be answerable to the communities for their activities, should be supported by the health system but not necessarily a part of its organization, and have shorter training than professional workers".
No restrictions on outcome or study design were included. Due to the varying types of anticipated study designs, no restrictions were imposed on the control or comparison group. Only studies conducted in low- and middle-income countries, as identified by The World Bank at time of study initiation, and only those articles published from 1990 to present were included. This time period was chosen to coincide with the re-emergence of the popularity of CHW programmes, to be consistent with the MDG timeframe and for scoping feasibility due do this study's resource restriction. To limit bias in both intervention areas and research publication sites, there was no language restriction. Non-English papers' abstracts were reviewed provided they were available in English and assessed for inclusion. The decision was taken to translate non-English papers, however this was unnecessary as none fit the inclusion criteria upon abstract review.
Articles were excluded if they did not meet the inclusion criteria or: if interventions were not clear or in studies with multiple intervention techniques where it was not possible to separate out specific preventive intervention outcomes; if the description of CHWs was insufficient or their role in the intervention was ill-defined; and if multiple health cadres were responsible for the intervention's implementation and the CHW's specific role could not be discerned, see Additional file 3.
Study Quality Assessment
Due to the scope of study designs included in this review, which may affect quality rating, and the lack of a meta-analysis, no studies were excluded based on the quality assessment. Studies were empirically rated using Effective Public Health Practice Project (EPHPP)'s Quality Assessment Tool for Quantitative Studies which uses a generic scale to evaluate a range of study designs and has been independently evaluated and judged as suitable for use in article appraisal for systematic reviews.
Data Extraction and Synthesis
Data was extracted systematically using a pre-formulated tool consisting of: setting, study design, population, intervention, control/comparison, group allocation methods, outcomes and quality rating, see Additional file 4. Data was then synthesised qualitatively by combining studies with similarities in interventions. Articles were subsequently coded descriptively by the reviewer to synthesise the results and draw conclusions.