Study results are often presented as abstracts at scientific conferences before publication as full articles in peer-reviewed journals. Given the current emphasis on evidence-based decision-making, it is vital that the peer-reviewed literature represents as broad and un-biased a selection of studies as possible. While the proportion of abstracts published as full papers in the peer-reviewed literature has been extensively studied in human healthcare, no such studies have been published in the field of food safety. The goal of this study was to estimate the proportion published and average time to publication for conference abstracts involving studies of pre-harvest or abattoir interventions to reduce foodborne pathogens. Abstracts were obtained by hand-searching available proceedings between 1995 and 2004 from 10 conferences. Included abstracts were limited to those detailing non-observational, controlled in vivo trials where outcome(s) were measured in livestock, carcasses or eggs. Data on abstract type (<or=500 words, >500 words), species, intervention, study type, sample size, number housed together and outcomes were recorded. Four databases (Agricola, CAB, Web of Science, Scholar's Portal) were searched for published papers corresponding to the conference abstracts using author and intervention/pathogen terms. Time to publication and overall median time to publication were estimated. Chi-squared, logistic regression and survival analyses were used to test for significant differences in proportion published and time to publication between variable levels. Of the 149 abstracts identified, 68 (45.6%) were published in peer-reviewed journals within 4 years. The median time to publication was 13.5 months (range: 0, 72). Abstracts shorter than 1 page were significantly more likely to be published (OR=2.2, 95% CI=1.0, 4.8), and abstracts involving pork or pigs were significantly less likely to be published that those involving poultry (OR=0.4: 0.2, 0.8). Abstracts reporting at least one positive outcome were more likely to be published (OR=2.608: 1.097, 6.196) and were published faster (HR=2.3: 1.1, 4.7). Time to publication decreased with the number of positive outcomes reported (HR=1.1: 1.0, 1.3). Sample size could only be determined for 46% of abstracts, with a median sample size of 9 (range 1-378), and housing was sufficiently described to determine sample size in 35% of pre-slaughter studies. The potential effects of this bias on systematic reviews and uses of interventions could be significant, and thus improvements may be warranted in the proportion of conference abstracts resulting in papers in the peer-reviewed literature.
(c) 2010 Elsevier B.V. All rights reserved.