Full publication of abstracts submitted at a major international conference on drug addiction : a longitudinal study

Article type
Authors
Vecchi S, Belleudi V, Amato L, Davoli M
Abstract
Background: Data from the literature show that about 45% of abstracts presented at biomedical meetings will be published in full.

Some studies have indicated that full publication is associated with the direction of the study (publication bias). No study has looked into the occurrence of publication bias in the field of addiction.

Objectives: To investigate whether the direction of results is associated with the publication of abstracts presented at a major international scientific meeting on addiction.

Methods: We selected all randomized controlled trials (RCTs) and controlled clinical trials (CCTs) presented at the United States Annual Meeting of the College on Problems of Drug Dependence 1993-2000, evaluating interventions for prevention, rehabilitation and treatment of drug addiction. Subsequent publication in peer-reviewed journals was searched for in MEDLINE and EMBASE 1993 - October 2005 with no language limitations.

Results: Four hundred and thirty-nine abstracts met inclusion criteria; 230 (66.7%) were subsequently published in peer-reviewed journals. Trials presented at the meeting showed: statistically significant positive results 55.8%, 74.3% published; no statistical differences 29.8%, 64.9% published; no quantitative results 11.6%, 39.2% published; negative results 2.7%, 50% published. Median interval between presentation and publication: positive studies, 28 months; no differences, 31; no results, 26; negative results, 5. Compared to studies with positive results, relative risk (RR) of subsequent publication, using a Cox model, was: 0.80 (CI 95% 0.62 - 1.04) for no differences, RR 0.77 (IC 95% 0.34-1.73) for negative results, and RR 0.41 (CI 95% 0.26-0.65) for no quantitative results.

Conclusions: Our study shows weak evidence of publication bias; the very low prevalence of studies with negative results limits the power to detect statistically significant differences, and could indicate a possible bias in submission of abstracts from authors or acceptance of abstracts at the meeting. The very short lag between presentation of negative results to the meeting and their publication would suggest that negative results are presented when already submitted for publication.