Publication bias in papers presented to the Australian Orthopaedic Association Annual Scientific Meeting

Background: The selective publication of articles based on factors, such as positive outcome, statistical significance and study size is known as publication bias. If publication bias is present, any clinical decision based on a review of the published work will also be biased. Publication bias has been shown in various specialties, based on review of publication rates for abstracts presented at major scientific meetings. This study was conducted to investigate publication bias in orthopaedics. Methods: Abstracts presented at the 1998 Australian Orthopaedic Association Annual Scientific Meeting were reviewed independently by two reviewers. Details of sample size, study setting, country of origin, outcome and study type were recorded for each abstract. Publication within 5 years was ascertained by electronic searching of Medline and Embase databases and direct author contact. Logistic regression analysis was used to identify predictors of publication. Results: The overall publication rate was 31%. Publication was more likely if the study was a laboratory study, rather than a clinical study (odds ratio (OR), 3.45; 95% confidence interval (CI) 1.69-7.01, P < 0.001). Sample size, country of origin, study type, statistical significance and positive outcome were not significantly associated with publication. Conclusion: According to this study, laboratory studies were significantly more likely to be published than clinical studies. In contrast to previous studies, publication bias due to the selective publication of papers with a positive outcome or those reporting statistical significance was not found.