Research publication in pediatric surgery: a cross-sectional study of papers presented at the Canadian Association of Pediatric Surgeons and the American Pediatric Surgery Association.

Introduction: Several studies have examined predictors of publication of research presented in scientific meetings in different disciplines. A tendency toward publishing studies with positive results has been described as "publication bias." Our objective was to determine the proportion of the studies that were published, time to publication, and factors that could predict publication in pediatric surgery.

Methods: The abstract books of the Canadian Association of Pediatric Surgeons and the American Pediatric Surgery Association meetings for 2001 to 2002 were reviewed. Data were gathered regarding the methodology and characteristics of each study. Case reports and editorials were excluded. A Medline search was then conducted to determine the publication status. Analysis using univariate and multivariate techniques was undertaken, comparing the difference between published and unpublished studies.

Results: Two hundred seven abstracts were reviewed. Of the 183 abstracts included, 118 (64.5%) were published. Most studies were published 1 year after presentation (93.2%). Presentation in the American Pediatric Surgery Association meeting and research originating from North America and reporting statistically significant results were significantly associated with subsequent publication on univariate analysis. The presence of statistically significant results was the only factor associated with successful publication on multivariate analysis (odds ratio, 3.3; confidence interval, 1.5-7.7).

Conclusion: The strong association between successful publication and the presence of statistically significant results point to the strong possibility of publication bias affecting decisions made about publishing research in the pediatric surgery.