Abstracts act as short, efficient sources of new information. This intentional brevity potentially diminishes scientific reliability of described findings. The authors’ objective was to 1) determine the proportion of abstracts submitted to the American Association of Neurological Surgeons (AANS) annual meeting that subsequently are published in peer-reviewed journals, 2) assess AANS abstract publications for publication bias, and 3) assess AANS abstract publications for differing results.
The authors screened all abstracts from the annual 2012 AANS meeting and identified their corresponding full-text publication, if applicable, by searching PubMed/MEDLINE. The abstract and subsequent publication were analyzed for result type (positive or negative) and differences in results.
Overall, 49.3% of abstracts were published as papers. Many (18.1%) of these published papers differed in message from their original abstract. Publication bias exists, with positive abstracts being 40% more likely to be published than negative abstracts. The top journals in which the full-text articles were published were Journal of Neurosurgery (13.1%), Neurosurgery (7.3%), and World Neurosurgery (5.4%).
Here, the authors demonstrate that alone, abstracts are not reliable sources of information. Many abstracts ultimately remain unpublished; therefore, they do not attain a level of scientific scrutiny that merits alteration of clinical care. Furthermore, many that are published have differing results or conclusions. In addition, positive publication bias exists, as positive abstracts are more likely to be published than negative abstracts.