Presentation versus publication of primary results in clinical trials

Background. Presentation of study results at scientific meetings is a common practice. A significant fraction of those full length publication despite the common belief that presentation speeds publication.; Objectives. This research was undertaken (1) to investigate the effect of presentation of primary results on their subsequent publication, i.e., whether presentation speeds or delays publication and (2) to describe the policies of clinical trials groups regarding presentations versus publications.; Methods. This dissertation consisted of two components: (1) a survey of investigators identified via MEDLINE, and (2) a followup of 291 HIV/AIDS trials from the National Institute of Allergy and Infectious Diseases registry. For the MEDLINE survey, a 17-item questionnaire was mailed to the authors of 601 selected articles. In both components, dates for trial completion (i.e., publication dataset finalization or end of patient followup), presentation and publication of results were collected. Publication rates were calculated using survival analysis techniques. Publication hazard ratios (HR) and confidence intervals (CI) were calculated using multivariate Cox proportional hazards regression models to account for confounding variables.; Results. Authors representing 383 (63.7%) trials responded to the MEDLINE survey. The median time from trial completion to first submission of the results paper was 10 months and the median time to publication was 23 months. Trials absent prior presentation of primary results were 3 8% [HR = 1.38, 95% CI: 1.06–1.81] more likely to be published than trials in which results were presented before submission for publication.; In the NIAID registry, primary results from 167 (57.4%) of the 291 HIV/AIDS trials had been published as of March 2000. Trials absent prior presentation of primary results were 60% [HR = 1.60, 95% CI: 1.04–2.46] more likely to be published than those presented before submission for publication.; Conclusions. Contrary to the common belief, presentation before submission of the results paper delayed publication about 6 to 10 months when median time to publication was compared between trials in which results were presented before submission and those in which results were either not presented before publication or presented only after submission for publication.