Comprehensive systematic review of complex interventions facilitated by means of new software program

Article type
Authors
Leiknes K1, HØie B1, Jarosch-von-Schweder L2, Berg R1, Smedslund G1
1Dept. of Evidence Based Medicine, Norwegian Knowledge Centre for the Health Services, Oslo, Norway
2Dept. of Research and Development, AFFU - Mental Health Care, Trondheim, Norway
Abstract
Background: Broad review questions and patient driven outcome on both effectiveness and harm of complex interventions often require an extensive and integrated evidence approach. Modern software systems facilitating the heavy workload of such reviews are not many. The DistillerSR online program developed by Evidence Partners (www.evidencepartners.com) is one. It allows for complex reviewing strategy, filter applications, diverse form building, immediate data export and extraction of results, constant monitoring of workload, thereby facilitating project managing and enabling reviewers to be located at different sites. Objectives: To demonstrate the utility of software system in an ongoing comprehensive systematic review on ECT for depression and to present preliminary results. Methods: A comprehensive literature search was undertaken, including databases MEDLINE, ISI, PsycINFO, EMBASE, Cinahl, British Nursing Index, Ovid Nursing, SveMed+, Cochrane Central Register of Controlled Trials. All identified references were initially title screened by hand and thereafter uploaded into DistillerSR for further streamlining, abstract screening and full text data extraction. The workload was undertaken by two reviewers at all levels. Results: Of 4973 identified references, 910 were uploaded into DistillerSR and filtered along 2 main streams A) Qualitative or B) Quantitative. Only 4 of 62 studies along the A) Qualitative (about the lived experience of ECT) stream were found explicitly qualitative in methodology and suited for final data extraction and integration. Along the B) Quantitative stream, 720 references were further filtered into 3 arms 1) RCT 2) non RCT about cognitive function (long-term memory impairment) and 3) non RCT about maintenance/continuation ECT for further full text reviewing and data extraction. Inter-rater reliability was high, overall weighted kappa 0.72. Conclusions: Managing broad systematic reviews of complex interventions is assisted by specially adapted modern online software systems where online review team collaboration, streamlining and filtering of a large number of references is made possible.