Yoeri Bemelmans

Chapter 7 100 Thirty-one studies were conducted in the United States [2,3,4,5, 19, 21,22,23,24, 26,27,28, 30, 31, 33, 35, 38,39,40,41,42,43, 45,46,47,48,49,50,51,52,53], four in the Netherlands [25, 29, 36, 38], three in Canada [6, 32, 34], two in France [20, 44] and one in Denmark [12] (Fig. 1). All studies were published between 2005 and 2019. Of the included studies, there was 1 RCT [24], 5 prospective observational studies [5, 6, 28, 36, 51], 1 prospective observational studywith amatched inpatient cohort [43], 16 retrospective observational studies [5, 6, 29, 37, 52], 1 prospective observational study with a matched inpatient cohort [44], 16 retrospective observational studies [20, 21, 27, 30, 35, 36, 38,39,40,41, 43, 44, 47, 49, 51, 53], 16 retrospective observational studies with amatched inpatient cohort [2,3,4, 12, 19, 22, 23, 28, 31,32,33,34, 42, 45, 46, 50], and 2 qualitative studies [25, 26]. Of the included studies, 18 used large national databases (e.g. national private insurance databases or national registry databases) [2,3,4, 19, 21, 28, 31,32,33, 39, 40, 42, 43, 46, 48,49,50,51]. The setting of the outpatient pathways varied from ambulatory surgical centers [22, 26, 41, 45] or hip/knee centers [5, 38], secondary and tertiary care hospitals [6, 12, 20, 23,24,25, 27, 29, 30, 34, 36, 37, 44, 48, 52, 53], or a combination of settings [35]. Ten studies described OJA following total hip arthroplasty (THA) [5, 24, 27, 31, 34, 40, 41, 44, 51, 53], 13 following total knee arthroplasty (TKA) [3, 6, 20, 21, 28, 30, 32, 37, 39, 47,48,49, 52], 6 following unicompartmental knee arthroplasty (UKA) [2, 22, 25, 29, 35, 46] and 11 studies presented results on both hip and (partial) knee arthroplasty [4, 12, 19, 23, 26, 33, 36, 38, 42, 43, 45, 50]. In total, the complete sample of studied patients consisted of 40.758 outpatients (OS=8.358; SOS=32.400) and 1.212.370 inpatients. A summary of the study characteristics and patient demographics is presented in Table 1. Risk of bias For all observational studies, more than 75% of the studies had a moderate or high risk of bias (Fig. 2). The included RCT was of high-quality regarding the low risk-of-bias judgment (Fig. 3). Blinding of participants, to prevent performance bias, was not possible because of the nature of the studied objective. Figure 2. Overall risk of bias for non-randomized trials, with use of the ROBINS-I tool.