Statistical Power of Randomized Controlled Trials in Trauma Surgery

Our purpose was to conduct a bibliometric study investigating the prevalence of underpowered randomized controlled trials (RCTs) in trauma surgery.

A medical librarian conducted a search of RCTs in trauma published from 2000 to 2021. Data extracted included study type, sample size calculation, and power analyses. Post hoc calculations were performed using a power of 80% and an alpha level of 0.05. A CONSORT checklist was then tabulated from each study as well as a fragility index for studies with statistical significance.

In total 187 RCTs from multiple continents and 60 journals were examined. A total of 133 (71%) were found to have "positive" findings consistent with their hypothesis. When evaluating their methods, 51.3% of articles did not report how they calculated their intended sample size. Of those that did, 25 (27%) did not meet their target enrollment. When examining post hoc power, 46%, 57%, and 65% were adequately powered to detect small, medium, and large effect sizes, respectively. Only 11% of RCTs had complete adherence with CONSORT reporting guidelines and the average CONSORT score was 19 out of 25. For positive superiority trials with binary outcomes, the fragility index median (interquartile range) was 2 (2 to 8).

A concerningly large proportion of recently published RCTs in trauma surgery do not report a priori sample size calculations, do not meet enrollment targets, and are not adequately powered to detect even large effect sizes. There exists opportunity for improvement of trauma surgery study design, conduct, and reporting.