The statistical fragility of arthroplasty versus fixation for femoral neck fractures: a systematic review of randomised controlled trials

Abstract

Background:

In the setting of femoral neck fractures, hip arthroplasty and internal fixation are considered as treatment interventions depending on the patient’s age and fracture characteristics. In this study, we utilised the fragility index (FI), reverse fragility index (rFI), and fragility quotient (FQ) to evaluate the robustness of statistical outcomes reported in randomised controlled trials (RCTs) evaluating arthroplasty versus fixation in the treatment of femoral neck fractures.

Methods:

Pubmed, Embase, and MEDLINE were queried from 2010 to present for RCTs reporting dichotomous outcomes where intervention arms were stratified as arthroplasty versus fixation for femoral neck fractures. The FI and rFI were calculated by outcome event reversals until the significance of the outcome was altered for significant and nonsignificant outcomes, respectively. FQ was calculated as the FI divided by the study sample size. Subgroup analysis was performed based on type of arthroplasty versus fixation and outcome type.

Results:

985 RCTs were screened with 9 studies included in the analysis comprising 30 total outcomes. The median FI across all outcomes was 5 (IQR 2–10.75) with an associated FQ of 0.039 (IQR 0.019–0.059). For 14 significant outcomes, the median FI was 4.5 (IQR 2.5–12.5) with an associated FQ of 0.045 (IQR 0.020–0.098). Across 16 nonsignificant outcomes, the median rFI was 5.5 (IQR 2.0–8.5) with an associated FQ of 0.039 (IQR 0.017–0.049). In 11/30 (36.67%) of study outcomes, the number of patients lost to follow-up was greater than or equal to the median FI. The studies evaluating hip hemiarthroplasty versus fixation were the most fragile with a median FQ of 0.018 (IQR 0.014–0.069) across 10 outcomes from 4 RCTs. The 4 studies evaluating total hip arthroplasty versus fixation were the least fragile with a median FQ of 0.049 (0.024–0.059) across 15 outcomes. 1 study evaluated either THA or HA versus fixation and reported 5 outcomes with a median FQ of 0.039 (IQR 0.039–0.044). The RCTs assessing arthroplasty versus fixation for non-displaced fractures were found to be more fragile (median FQ 0.016) compared to the RCTs looking at displaced fractures (median FQ 0.042). By outcome type, the median FI’s were 4 (IQR 2–11), 6 (IQR 2–6), 16 (IQR 10–16), and 5 (IQR 3.5–6.5) for failure/reoperation, complications/adverse events, mortality, and “other”, respectively.

Conclusions:

The statistical findings reported in RCTs comparing arthroplasty to fixation for femoral neck fractures exhibit considerable fragility, suggesting that minor changes in patient follow-up or outcome occurrences could significantly impact results. To enhance the interpretation of comparative trials in orthopaedic trauma we advocate for the routine inclusion of FI and FQ alongside p-values.