STATISTICAL FRAGILITY OF RANDOMIZED CONTROLLED TRIALS IN HIP AND KNEE ARTHROPLASTY: A SYSTEMATIC REVIEW

Randomized controlled trials (RCTs) are crucial in evidence-based medicine for evaluating healthcare interventions, yet their quality varies. One key aspect of trial robustness is the fragility index (FI), which is the number of events in one trial group that must be changed to nonevents so that a staitstically significant difference becomes nonsignificant. Surgical RCTs, including orthopaedics, are often highly fragile due to challenges in trial design, sample size, and risk of bias, raising concerns about the validity of findings. Previously, the FI could only ba calculsted for binary outcomes, but there is now a method to coaculate FI for continuous outcomes. This review aimed to evaluate the statistical fragility of hip and knee arthroplasty RCTs from the past decade using both binary and ocntinuous outocmes, and examine characteristics associated with fragility. We systematically searched Medline and Embase databases for hip and knee arthroplasty RCTs. Inclusion criteria required a 1:1 parallel design with at least one statistically significant outcome. We extracted study characteristics, sample size, and statistical measures. Statistical fragility was assessed using the dichotomous fragility index (FI) for categorical outcomes and continuous fragility index (CFI) for continuous outcomes. Multivariable linear regression was conducted to analyze associations between fragility and study characteristics. We identified a total of 16,214 records, with 140 studies meeting the inclusion criteria. Among studies with dichotomous outcomes, the median FI was 2 (interquartile range [IQR] 4), indicating high fragility. For continuous outcomes, the median CFI was 8.85 (IQR 14.4). We did not identify any significant associations between FI/CFI and study characteristics. This review identified that hip and knee arthroplasty RCTs frequently exhibit statistical fragility, especially in dichotomous outcomes where results are vulnerable to minor event changes. These findings underscore the importance of robust study designs in orthopedic RCTs, which could be strengthened by incorporating FI into sample size calculations to improve the robustness of outcomes. Enhancing trial quality is critical for informed clinical guidelines and patient care in orthopedic surgery, particularly in high-volume procedures such as hip and knee arthroplasty.