A systematic review of the relative differences in adverse events used for sample size calculations in orthopaedic trauma trials.

PURPOSE: The purpose of this study was evaluate the relative differences in adverse event rates used for sample size calculations in orthopaedic trauma randomized controlled trials (RCTs) and to compare these to those actually observed.

METHODS

A systematic search was conducted using the Embase, Medline/PubMed, and Web of Science databases between 2015 and 2025 to identify orthopaedic trauma RCTs with a dichotomous adverse event as the primary outcome. The relative differences in adverse event rates used to calculate sample size were compared to those actually observed.

RESULTS

Thirty-five trials were included. The most common fractures types involved included the hip (37.1%, n = 13) and tibia (11.4%, n = 4). The most common adverse events were infection (25.7%, n = 9) and reoperations (14.3%, n = 5). The median relative difference in adverse events used for sample size calculations was 50.0% (range 21.9-90.0%), which was greater than the median observed relative difference of 20.5% (p = 0.01). Only 22.8% (n = 8) of the trials had an observed relative difference of 50.0% or greater. The adverse event rate used for sample size calculation in the control group was higher than the observed rate (27.5% vs. 13.6%, p = 0.01) while the adverse event rate used for sample size calculation in the intervention group was similar to the observed rate (8.2% vs. 12.3%, p = 0.35).

CONCLUSION

Trials used larger relative differences and higher control group event rates for sample size calculations than were actually observed and large effect sizes were infrequently observed.

LEVEL OF EVIDENCE: Level 1, therapeutic.