Yoeri Bemelmans

Chapter 7 98 Risk of bias in individual studies Two reviewers independently assessed the risk of bias of each of the included studies. The Cochrane Risk of Bias tool [15] (version 2 of the Cochrane risk-of-bias tool for randomized trials) was used to assess the risk of bias in randomized trials. The risk of bias for non-randomized studies was assessed with use of the ROBINS-I tool [16]. Any discrepancies between reviewers were discussed and, if necessary, an agreement was reached by the adjudication of a third author. Data synthesis All outcome measures were analysed using RevMan 5 [17]. Continuous data were described asmean differences with 95%confidence intervals (CI) or standardizedmean differences. Dichotomous data were described as odds ratios (OR). Where possible, the data were pooled in ameta-analysis. To quantify statistical heterogeneity, the Cochran’s Q test and I2 statistic (the proportion of the total variance explained by heterogeneity) were conducted. I2 values of > 75% were interpreted as high heterogeneity [18]. Subanalyses were performed for the two defined outpatient groups (experimental): (1) OS versus (2) SOS, compared with inpatient pathways (control) to account for differences in “outpatient” definitions as described above. Subanalyses were also performed for THA, TKA and UKA to account for fundamental differences between the three types of arthroplasty. In case meta-analysis was eligible, forest and funnel plots were presented. Ethical considerations This review is a non-Medical Research Involving Human Subjects Act study. Results This systematic electronic search identified 8.718 references. No additional records were identified by cross-checking reference lists. The search and selection process are shown in the PRISMA flowdiagram (Fig. 1). After removal of duplicates, 4.538 references remained for screening on title and abstract. In total, 52 full texts were assessed for eligibility. After review, 41 studies met the inclusion criteria and were included in the systematic review. Twenty-six studies were allocated to the OS group [2,3,4,5,6, 12, 19,2 0,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38] and 15 to the SOS group [39,40,41 ,42,43,44,45,46,47,48,49,50,51,52,53] for subanalyses.