Frequency and determinants of nonpublication of research in the stroke literature.

BACKGROUND AND PURPOSE Selective nonpublication will yield publication bias and a published literature imperfectly representative of the full range of scientific findings. We evaluated the proportion of research abstracts presented at the leading United States research meeting in stroke, the International Stroke Conference (ISC), which were subsequently published as full-length articles and investigated the factors associated with full manuscript publication. METHODS Features of all abstracts presented at the annual ISC meeting in February 2000 were analyzed. Search of the National Library of Medicine PubMed database and written communication with abstract authors was performed to determine conversion of abstracts to fully published manuscripts over the subsequent 5 years. RESULTS Among the 353 abstracts presented at the 2000 International Stroke Conference, 108 were oral presentations and 245 posters. Overall, 202/353 (62.3%) resulted in full-length publications, with a median time to publication of 15 months. In multivariate analysis, factors increasing likelihood of full-length publication were: platform rather than poster presentations (odds ratio [OR] 3.0, 95% CI, 1.6 to 5.5), authors with a university affiliation (OR 2.2, 95% CI, 1.2 to 4.1), and European region of origin (OR 2.2, 95% CI, 1.1 to 4.4), whereas topic concerning community/risk factors decreased the likelihood of publication (OR 0.3, 95% CI, 0.16 to 0.74). Positive results, multicenter collaboration and industry sponsorship did not affect publication rate. CONCLUSIONS Approximately 1 of every 3 abstracts presented at an international stroke meeting was not published as a full manuscript within 5 years. Poster abstracts were less likely to be published in full manuscript form than oral presentations.