Fragility of randomized clinical trials comparing surgical approaches for inguinal hernia repair: a systematic review.

INTRODUCTION

Randomized clinical trials (RCTs) remain the reference standard for estimating causal effects of surgical interventions when appropriately designed and conducted. However, P values alone do not describe how sensitive a dichotomous statistical-significance classification is to small changes in outcome counts. The Fragility Index (FI) and Reverse Fragility Index (RFI) are post hoc descriptive metrics that quantify the number of outcome-status changes required to alter statistical significance. This study aimed to evaluate FI/RFI for inguinal hernia repair (IHR) RCT findings, assess individual complication endpoints, and examine associations with sample size and conflict-of-interest reporting.

METHODS

Embase, PubMed, LILACS, and ClinicalTrials.gov were searched through June 2025 for RCTs comparing open, laparoscopic, or robotic inguinal hernia repair with dichotomous primary outcomes. FI, RFI, and sample-size-adjusted quotients were calculated from reconstructed 2 × 2 tables using two-sided Fisher exact tests. Individual complications and associations with trial characteristics were analyzed.

RESULTS

Forty-two RCTs comprising 14,505 patients and 44 primary comparisons were included; all eligible comparisons evaluated laparoscopic versus open repair. Twelve comparisons were significant, with a median FI of 4 (IQR, 3-5.25); 4 (33.3%) had an FI of 3 or less. Among 31 nonsignificant comparisons, the median RFI was 5 (IQR, 3.5-6.5). All 4 significant individual-complication results had an FI of 3 or less (median, 1). Significance classification differed between reported and recalculated P values in 4 of 78 extracted comparisons. FI was not significantly correlated with sample size.

CONCLUSIONS

Randomized trials remain central to comparative evidence in inguinal hernia repair. FI and RFI provide complementary post hoc context regarding how many outcome-status changes would alter a dichotomous significance classification and should be interpreted alongside prespecified trial design, effect estimates, confidence intervals, P values, sample size, missing data, and clinical relevance.