Mini-Reviews For Iterative Scoping Review Protocol Development

Introduction: While several reviews have been published about psychosocial interventions in LMIC countries, little is known about the characteristics of psychosocial, group-directed interventions led by community health workers in the SAARC countries (India, Nepal, Bangladesh, Afghanistan, Pakistan, Sri Lanka, Bhutan, tMaldives). For our forthcoming scoping review on this topic, we tested our protocol on the results of a quick search of one publicly available resource (the National Institute of Health’s PubMed service).The review protocol trial was used to (i) gain further clarity on the inclusion and exclusion criteria; (ii) develop an easy to use extraction scheme for full text extraction within the software tool Covidence; (iii) refine the research questions; (iv) generate seed papers for the main search.
Methods: Medline-PubMed was searched through the web interface on June 17, 2018 by MW. The database was chosen because it is easily accessible, and represents the most up to date variant of Medline. The search queries used were:(i) (("mental health"[All Fields] AND "community"[All Fields]) OR "community mental health"[All Fields]) AND ("<country>"[MeSH Terms] OR "<country>"[All Fields])(ii) “mental health"[All Fields] AND "task shifting"[All Fields] Query (i) was run separately for each of the SAARC countries. MW excluded clearly irrelevant abstracts manually before exporting to reference software (Zotero) to focus on core inclusion and exclusion criteria. Results of Query (i) and Query (ii) were merged using Zotero. We included study protocols, implementation studies, and pre/post study designs as well as randomised controlled trials. We also included editorials and position papers, in case they might contain brief descriptions of interventions. The rapid PubMed search was supplemented by a search of the grey literature through mailshot. 
Results: 152 abstracts were entered into Covidence. Three (two percent) were excluded as duplicates. Of the remaining 149 (ninety-eight percent) studies, sixty (forty percent) were excluded as irrelevant through abstract screening, and eighty-nine (sixty percent) full texts were reviewed. Seventy-one (eighty-one percent) of those were excluded, leaving eighteen (nineteen percent) studies for full extraction. Thirty-one (forty-five percent) of the excluded papers did not describe a group intervention, twenty-one (thirty percent) did not describe an intervention, nine (ten percent) did not include information about who was treated by whom, where, and how, six (nine percent) were not from SAARC, two (four percent) were delivered by a health professional, one (one percent) included a paediatric population, and one (one percent) was an undetected duplicate. Two reviewers assessed each abstract and full paper, and conflicts were resolved by discussion. As a result, we revised our criteria to exclude interventions that are delivered by trainee health professionals, that focused on families, and that describe training of mental health workers. Abstract inclusion now requires that at least one specific intervention is mentioned. Five papers from different venues were selected as seed papers. 
Conclusion: The fast pre-review was useful for developing better guidelines for abstract screening and clearer exclusion criteria. Conflict of Interests: KM is the author of papers potentially included in the review