Underreporting of Secondary Endpoints in Randomized Trials: Cross-sectional, Observational Study

OBJECTIVE: To determine if underreporting of secondary endpoints in randomized controlled trials occurs, using surgical site infection (SSI) as an example.

BACKGROUND: SSI is a commonly measured endpoint in surgical trials and can act as a proxy marker for primary and secondary endpoint assessments across trials in a range of medical specialties.

METHODS: Cross-sectional observational study of randomized trials including patients undergoing gastrointestinal surgery published in a representative selection of general medical and general surgical journals. Studies were included if SSI assessment was a prespecified endpoint. Adjusted binary logistic regression was used to identify factors associated with a high rate of SSI detection (≥10%).

RESULTS: From 216 trials including 45,633 patients, the pooled SSI rate was 7.7% (3519/45,633), which was significantly higher when assessed as a primary endpoint (12.6%, 1993/15,861, 49 studies) vs as a secondary endpoint (5.1%, 1526/29,772, 167 studies, P < 0.001). When assessed as a secondary outcome, standardized definitions and formal clinical reviews were used significantly less often. When adjusted for surgical contamination and methodological confounders, secondary assessment was associated with reduced SSI detection compared with primary assessment (adjusted odds ratio 0.24, 95% confidence interval 0.08-0.69, P = 0.008).

CONCLUSIONS: Secondary endpoint assessment of SSI in randomized trials was associated with significantly reduced rigor and subsequent detection rates compared with assessment as a primary endpoint. Trial investigators should ensure that primary and secondary endpoint assessments are equally robust. PRISMA guidelines should be updated to promote the conduct of meta-analysis based only on primary outcomes from randomized controlled trials.