The heart of the matter: Outcome reporting bias and registration status in cardio-thoracic surgery

Background Our objective is to compare registered outcomes to published reports; to evaluate for discrepancies favoring statistically significant outcomes; to examine funding source and likelihood of outcome reporting bias; and to evaluate for any temporal trends in outcome reporting bias. Methods PubMed was searched for randomized controlled trials published between 2008 and 2015 from 4 high impact cardio-thoracic journals: European Journal of Cardio-thoracic Surgery (EJCS), The Journal of Cardiothoracic Surgery (JCS), The Journal of Thoracic and Cardiovascular Surgery (JTCS), and Annals of Cardiothoracic Surgery (ACS). Data was collected using a standardized extraction form. Results We reviewed 287 articles, of which 214 (74.6%) did not meet registration criteria. Of those 214, 94 (43.9%) were published in the EJCS, 34 (15.9%) in JCS, 86 (40.2%) in JTCS, and 0 (0%) in the ACS. Of the remaining 73 articles, 34 (46.6%) had a discrepancy between the primary outcome registered and the published outcome, and 11 of the 34 reported p-values favoring the change. We also found that 12 of the 73 registrations had updated primary outcomes from the initial report to the final report. The timing of registration was an incidental finding showing 14 (19.1%) articles retrospectively registered, 29 (39.7%) registered during patient enrollment, and 30 (41.1%) registered prospectively. Conclusion The results indicated that selective outcome reporting is prevalent in cardio-thoracic surgery journals. The more concerning issue, however, is the lack of registration or provision of registration number for randomized controlled trials within these journals.