Publication bias of randomized controlled trials in emergency medicine

Abstract

Objectives: To determine the publication status and time to publication of randomized controlled trials (RCTs) that were presented at the Society for Academic Emergency Medicine (SAEM) meetings from 1995 to 2003. The impact of positive-outcome bias, time-lag bias, and gray literature bias also was assessed.

Methods: Retrospective cohort study of RCT abstracts presented at nine SAEM scientific meetings. Electronic searches identified publications from the abstracts.

Results: Of 4,399 abstracts, 383 (8.7%; 95% confidence interval [CI] = 7.8% to 9.5%) were identified as RCTs. One hundred ninety-four (50.7%; 95% CI = 45.7% to 55.7%) were subsequently published up to May 2004. The median time to publication was 32 months (95% CI = 23 to 41), with 59% of RCT abstracts published within five years of presentation. No evidence of positive-outcome bias or time-lag bias was identified; however, changes from abstract to manuscript were found. Manuscripts were less likely to endorse the experimental intervention than were abstracts (OR, 0.2; 95% CI = 0.0 to 0.6).

Conclusions: The proportion of emergency medicine RCT abstracts published is slightly lower than that for other biomedical specialties; however, biases reported by investigators in other biomedical areas do not appear to be as problematic in emergency medicine research. Differences between conclusions from abstracts and manuscripts must be considered when employing meeting abstracts as a source of evidence for future research or for systematic reviews in emergency medicine.