Statistical fragility of randomized clinical trials pertaining to femoral neck fractures

Abstract

ObjectiveP values were frequently misused and misinterpreted, the fragility index (FI) has been utilized to evaluate the robustness of randomized controlled trials (RCTs) as a complement to p-values. This study aimed to assess the statistical robustness of RCTs for femoral neck fractures through the utilization of the FI. MethodsWe systematically reviewed PubMed, Cochrane Library, and Embase database to identify RCTs pertaining to femoral neck fractures published in the top 25 highest-impact orthopaedic journals and 4 high-impact general medical journals from January 1, 2000, to December 31, 2022. The FI was calculated for the dichotomous, categorical study outcomes in the identified RCTs using the Fisher exact test, with previously published methods. Spearman correlation analyses were used to evaluate potential associated factors associated with the FI. ResultsWe identified 10 eligible RCTs with a median total sample size of 101 (IQR, 79.5 to 174.75) and a number of patients lost to follow-up of 19.5 (IQR, 4.5 to 28). The median FI was 3.5 (IQR, 1 to 14.25), implying that reversal of outcome in only 4 patients was sufficient to alter trial significance. The FI was less than the number of patients lost to follow-up in seven (70%) RCTs. P values were negatively associated with the FI, while the number of patients lost to follow-up and patients enroled were not statistically significantly associated with the FI. ConclusionsThe RCTs pertaining to femoral neck fractures were not as statistically robust as previously thought and should be interpreted with caution. We recommend that the orthopaedic RCT report FI as a supplement for the P values to help readers draw reliable conclusions based on the fragility of the outcomes.