31 Review of social network intervention studies 2 following terms were used (including synonyms and closely related words) as index terms or free-text words: ‘psychiatric patients’ and ‘social network’ or ‘social support’, and ‘intervention’, and ‘randomized controlled trials’, see Appendix A for the complete search strategy. We additionally searched Google Scholar and the Cochrane Library using the same keywords, to broadly examine the field of literature, locate key articles, and inspect previously conducted systematic reviews. Finally, reference lists of relevant articles were inspected for additional citations. After removal of duplicates, the application Rayyan was used for the selection of studies (Ouzzani, Hammady, Fedorowicz, & Elmagarmid, 2016). Both titles and abstracts were screened for relevance according to a predefined and piloted screening protocol. Next, full texts were screened to determine final eligibility, which was verified by three reviewers. Inclusion criteria Studies were included in the meta-analysis if they met four criteria. First, studies had to examine the effectiveness of interventions designed to enhance the social network (e.g., size, quality, and social relationships) or social support of participants in their natural environment. Interventions were eligible if social network-related goals were explicitly mentioned in the sections outlining the intervention strategies. Second, studies had to focus on patients with psychiatric disorders according to DSM-IV-TR or DSM-5 criteria, or psychiatric patients inmental healthcare settings. Third, samples had to containparticipants aged 16 years or older. Lastly, studies had to be randomized controlled trials or quasi-experimental designs. Intervention groups had to be compared to a control group not including a social network intervention. Both published and unpublished trials were eligible. Background articles, reviews, conference papers, poster or presentation abstracts, as well as articles only accessible in languages other than English or Dutch were excluded (see Appendix A, Table 1, for the list of excluded studies). Data extraction Data was extracted in duplicate by two reviewers using a predefined and piloted coding instrument, see Appendix A. We extracted outcome, intervention, sample, and study characteristics that were expected to moderate effects sizes. A broad range of quantitative outcomes at multiple timepoints were extracted. As outcomes were measured in many different ways, for example various instruments and questionnaires were used and concepts were described differently, we created overall outcome categories and more specific subcategories and pooled the single effect sizes in these (sub)categories. We coded single outcomes into one primary outcome category (i.e., social network characteristics), consisting of multiple subcategories, and six secondary outcome
RkJQdWJsaXNoZXIy MjY0ODMw