Publication bias in reproductive research.

Publication bias is defined as any tendency on the part of investigators or editors to fail to publish study results on the basis of the direction or strength of the findings. This may lead to overestimation of treatment effects in published work. Inappropriate decisions about patient management may result. We investigated what proportion of abstracts at the European Society of Human Reproduction and Embryology (ESHRE) annual meeting eventually reached full publication, what was the time to publication, and which factors might have affected publication. Among the 2691 abstracts of six ESHRE annual meetings, 151 (5.6%) reporting randomized controlled trials (RCT) were identified. Comprehensive searches of electronic databases and handsearching of the two major journals in the field yielded 79 full publications pertaining to these abstracts. Kaplan-Meier analysis estimated 56% of RCT abstracts to be eventually published in full, the median time to publication being 32.5 months. Positive outcome (i.e. significant results) did not affect the publication rate, and neither did sample size, the subject category, or the native language (English/non-English) of the country of origin. Oral presentations resulted in eventual full publication significantly more frequently (69%) than posters (42%). It is concluded that a considerable publication deficit, but not a publication bias, exists for RCT in reproductive research.