Presentation in relation to publication of results from clinical trials.

BACKGROUND: Results from clinical trials are typically disseminated first by presentation at scientific meetings. An important question has to do with the role of presentation in improving the quality of manuscripts submitted to the journals as well as the effect of presentation in speeding, or delaying subsequent publication. The aim of this research is focused on presentation practices of trialists to examine their effect on the timing of publications of clinical trial results. METHODS: Six hundred and one (601) trials published in 1996 and 1997 were identified via MEDLINE using medical subject heading "clinical trials" or the occurrence of the term in the text and by limiting to publication type "clinical trial". Authors of those trials were surveyed to determine prior presentation history for the identified trials. RESULTS: Among the 601 trials identified, complete responses to questionnaires were obtained for 379 (63%) trials. The median time from completion to first submission of the primary results manuscript was 11 months and the median time from completion to publication was 25 months for the 220 trials involving presentation prior to submission for publication. The corresponding median times from completion to first submission and publication for the subset of trials not involving presentation prior to the submission were 8 and 19 months (159 trials), respectively. The adjusted relative hazard for publication for trials involving presentation prior to first submission was 0.55 versus trials not involving presentation prior to first submission (95% confidence interval, 0.44 to 0.69). CONCLUSION: Despite the importance of dissemination of results prior to publication, investigators should carefully weigh a potential gain in quality against a potential for delay in submission of the primary results manuscript by presentation at scientific meetings. The findings of our study suggest that presentation prior to submission may increase time to publication. Inclusion of presentation dates in clinical trial registers should be considered to allow future studies investigating presentation and publication practices.

Treatment effects monitoring committees and early stopping in large clinical trials.

BACKGROUND: Treatment effects monitoring is a process carried out over the course of a trial to determine whether it should continue unaltered. The purpose of monitoring is to protect persons enrolled from harm, caused either by exposure to an ineffective or harmful treatment or failure to provide a better treatment. The aim of this paper is to characterize treatment effects monitoring committees (TEMCs) as extant in published trials and to examine the effect of their presence on the premature stopping of a trial. METHODS: Trials published in 1990-1995 in the Annals of Internal Medicine, Archives of Internal Medicine, British Medical Journal, Journal of the American Medical Association, Lancet, New England Journal of Medicine or Controlled Clinical Trials were identified via MEDLINE (4279 publications). Abstracts were screened to include only trials with parallel treatment designs and sample sizes of > or = 200, reducing the set to 661 papers. Those papers were then read, reducing the set to 562 after exclusion of papers not reporting trial results. The results that follow come from a review of these 562 published papers and from the responses to two questionnaires. The first was mailed to the corresponding author, and the second to the chair of the monitoring body or other contact as provided by the author in response to the first questionnaire. RESULTS: Less than half (48%) of the 562 trials had TEMCs. Factors having a positive univariate relationship with the presence of a TEMC include National Institutes of Health (NIH) sponsorship, larger sample size, multicentered, longer data collection and follow-up, and mortality as an outcome. Most of the early stops occurred in trials with TEMCs (66 out of 78). The odds ratios for early stopping for trials with TEMCs versus those without TEMCs was 4.4 (CI: 2.3-8.5). CONCLUSIONS: The evidence suggests that early stopping is associated with the presence of a TEMC, but a substantial number of trial reports do not mention the presence of a TEMC even when one was used to stop a trial early.