Publication bias in the field of subfertility

Article type
Authors
Lensen S1, Jordon V1, Showell M1, Showell E1, Shen V1, Venetis C2, Farquhar C1
1University of Auckland, New Zealand
2University of New South Wales, Australia
Abstract
Background: Publication bias due to the failure to publish study results based on the direction or strength of study findings, can lead to a misleading interpretation about the effectiveness of an intervention and ultimately impact patient care. Publication bias has been detected in many clinical areas including subfertility.
Objectives: Is the statistical significance of the results from subfertility randomized clinical trials (RCTs) that have been presented at conferences associated with the probability of publication of these RCTs as full-text articles? A preliminary report.
Methods: Eligible abstracts were subfertility RCTs reporting one or more reproductive outcomes. The Cochrane Menstrual Disorders and Subfertility Group specialised register (MDSGSR) was searched for abstracts presented between 1 January 2007 and 31 December 2010 and a search was performed to identify full-text publications. Two authors independently screened abstracts for eligibility and extracted data. One author searched for all full-text publications not found by the other author.
Results: As of March 2015, 234 eligible articles had been identified from a total of 337 retrieved. Preliminary analysis indicates that: 36% of abstracts were oral presentations, 4% were stated as registered, 9% were stated as interim or preliminary analyses, 13% acknowledged industry funding, while the source of funding was not reported in 67% of studies.
Overall, 53% of the abstracts were found to be published as full-text articles. There was no statistically significant (SS) difference between likelihood of publication of abstracts reporting SS results compared to abstracts reporting non-SS results (OR = 1.04, 95% CI 0.59 to 1.81). Of studies not reporting SS results, 13% made a positive statement about the findings, for example describing a trend towards improved outcomes.
Conclusion: This study found no evidence to suggest that subfertility RCTs of conference abstracts that reported SS results were more or less likely to be published than studies that did not report SS results.