Validity of repeated-measures analyses of in vitro arthroplasty kinematics and kinetics

In vitro models of arthroplasty enable pre-clinical testing and inform clinical decision making. Repeated-measures comparisons maximise resource efficiency, but their validity without testing order randomisation is not known. This study aimed to identify if there were any large testing order effects for cadaveric models of knee and hip arthroplasty. First, the effect of testing order on total knee arthroplasty (TKA) biomechanics was assessed. Extension moments for TKAs (N = 3) implanted into the native knee (TKA-only) were compared to a dataset of TKAs (N = 24) tested after different combinations of partial knee arthroplasty (TKA-last). The effect of repeatedly testing the same knee five times over 36 h on patellofemoral and tibiofemoral kinematics was also quantified. Second, the effect of testing order on capsular ligament function after total hip arthroplasty (THA) was assessed. Randomisation was removed from a previously published dataset to create increasing and decreasing head size groups, which were compared with t-tests. All three TKA-only extension moments fell within the 95% CI of the TKA-last knees across the full range of knee flexion/extension. Repeated testing resulted in root-mean-squared kinematics errors within 1 mm, 1°, or < 5% of total range of motion. Following THA, smaller head-size resulted in greater laxity in both the increasing (p = 0.01) and decreasing (p < 0.001) groups. Testing order did not have large effects on either knee or hip arthroplasty biomechanics measured with in vitro cadaveric models.