THE FATE OF ABSTRACTS SUBMITTED TO A CANCER MEETING - FACTORS WHICH INFLUENCE PRESENTATION AND SUBSEQUENT PUBLICATION

Abstracts that are published in the proceedings of meetings receive minimal peer-review, but may be referenced or used to make decisions about management of patients. We have studied factors which influence the probability of acceptance for presentation, and of subsequent publication of articles, from abstracts included in the Proceedings of the American Society of Clinical Oncology (ASCO). From a random sample of 197 abstracts submitted to the 1984 meeting, 81 were accepted for presentation and a Cancerline computer search revealed 103 papers that were published subsequently in peer-reviewed journals. Communication with authors of the remaining abstracts led to identification of 12 additional articles that had been published. Major reasons for non-publication were insufficient priority or lack of time, funds or other resources. Abstracts which reported 'positive' results were more likely to be presented than those reporting 'negative' results (60% vs. 35%, p = 0.03) and to lead to subsequent publication (74% vs. 32%, p = 0.0001). Of the 81 abstracts in our sample that were selected for presentation at the meeting, 63 (78%) led to publications, compared to 45% (52/116) of those not selected (p = 0.00001). There were no significant differences in the frequency of citation of abstracts that did or did not lead to subsequent publications. We made detailed comparisons of abstracts and subsequent papers for 18 randomized phase III trials. For 15 studies (83%), there was good correlation between the conclusions of the article and of the abstract. Although many abstracts reflect the final conclusions of the studies that they describe, the failure to publish articles based on about 40% of submitted abstracts should warrant caution in their citation or use in making clinical decisions. This article derives from an abstract submitted to the 1990 ASCO meeting.