Standardized mean differences (SMDs; Hedges' g) were calculated from means and standard deviations (mean difference/pooled SD) of muscle strength data using Review Manager software (Version 5.2; The Cochrane Collaboration, London, UK). The SMDs of 0.2, 0.5, and 0.8 were considered small, moderate, and large, respectively. 9 The percentage difference in strength was also calculated to provide a further indication of the relative difference in strength between OA and control participants [(OA strength -control strength)/control strength × 100]. Results were grouped according to the type of strength measurement (eg, isometric or isokinetic) or hip action performed (eg, hip flexion or extension). Data were pooled for multiple studies in a meta-analysis within each group using a random-effects model. When studies reported on multiple grades of OA classification (eg, Kellgren-Lawrence grade 2 or 3) without reporting on muscle strength of the whole sample combined, only the more established OA grading was used in the meta-analysis, to avoid repeated assessment of control-group data. When studies measured isokinetic strength at multiple speeds, the slower speed was used, as slower speeds are considered more reliable.35,41 When a study reported results of left and right knees of participants with bilateral knee OA, only the side with the most conservative SMD was presented, to avoid repeated inclusion of control-group data.10 Statistical heterogeneity across pooled studies was assessed using the I 2 statistic, in which a value of 25%, 50%, or 75% was considered a low, moderate, or high level of heterogeneity, respectively.
23
Assessment of Quality of Body of EvidenceA modified version of the Grading of Recommendations Assessment, Development and Evaluation (GRADE) approach was used to evaluate the quality of evidence in each meta-analysis. 20,35,43 The GRADE tool has been modified to apply to observational, nonexperimental data. 35,43 Each meta-analysis was graded using the following predefined criteria: (1) inconsistency (downgraded if I 2 was 50% or greater), (2) Screened by title and abstract, n = 1737Screened by title and abstract, n = 1737Full-text articles assessed for eligibility, n = 102Full-text articles assessed for eligibility, n = 102Hip strength in knee OA compared to control, n = 5Hip strength in knee OA compared to control, n = 5Hip strength as a risk factor for knee OA development, n = 0Hip strength as a risk factor for knee OA development, n = 0