Harmen Beurmanjer

23 Review on the GHB using Population 2 and controlled studies were excluded. We excluded controlled studies due to the possible influence of each study’s inclusion and exclusion criteria on the generalisability of the population. Studies focusing on pharmaceutical GHB use (e.g. narcolepsy, alcohol addiction) and mechanistic studies (pharmacological and biological effects of GHB) were also excluded. Furthermore, we excluded studies in which GHB-use was a small minority of the studied population and/or without description of demographic characteristics. Finally, studies concerning involuntary ingestion (e.g. when taken as a rape-drug) were excluded. Using the software Rayyan (Ouzzani, Hammady, Fedorowicz & Elmagarmid, 2016) two reviewers (EJ and HB) independently assessed the inclusion or exclusion based on titles and abstracts. Disagreements were resolved via discussion and consensus between the two reviewers. Data analysis Tables were used to summarize all studies, including the study aim, design, methods, population (including demographics), results and additional comments on the included studies. It was expected that study design, setting, population, and reported outcomes varied significantly, given the large variation in user groups of GHB. Therefore, we decided a priori not to perform meta-analyses. Results Study selection Details of the search strategy and results are shown in Fig. 1. The literature search (March 2018) resulted in 2847 citations and 1417 unique references after de-duplicating from consulted databases. Update of the search (September 2019) resulted in 372 new unique references. After reviewing titles and abstracts, we kept 80 articles to read in full-text. Primary reasons for exclusion based on abstract alone were related to the population studied (e.g. focusing on pharmaceutical GHB use), the publication type (comments, review, single case studies, mechanistic studies), and non-English language (except Dutch). Based on consensus between the two reviewers, 60 of the 80 articles were included. Six articles describing two individual studies were excluded as they were controlled studies; for one controlled study the participants were already included in this review as they were also part of an observational study, the other controlled study was completely excluded. The other 14 articles were excluded due to lack of description of demographic characteristics of the sample. In the 60 included articles, 51 unique samples were described