Reporting bias in completed epilepsy intervention trials: A cross-sectional analysis

To explore the evidence of reporting bias among completed epilepsy intervention trials (EITs) and compliance of applicable EITs to Food and Drug Administration Amendments Act (FDAAA).

We included consecutive EITs registered as completed on ClinicalTrials.gov from 2008 to 2015. Descriptive data was collected including study type, study phase, funding source, primary completion date, and result reporting date. Time to result reporting was analyzed using Kaplan-Meier estimates for two time periods (2008-2011 and 2012-2015). PubMed, Web of Science, and Google scholar databases were manually searched for publication details.

Overall, 95/126 EITs (75%) reported, while remaining 31/126 (25%) did not report their results. Time to reporting was significantly lower for trials completed during 2012-2015 (16.5 months; 95% CI: 13.60-19.40; p = .002; Cohen's d = 0.68) as compared to the trials completed during 2008-2011 (25.9 months; 95% CI: 21.56-30.22). 72/126 trials were conducted in at least one U.S. center. 56/72 (78%) of the trials met the FDAAA criteria, while only 19/56 (34%) reported within the mandated one-year time frame.

The lack of reporting of nearly one-quarter of completed epilepsy intervention trials suggests existence of reporting bias. As such, it should be considered an important criterion for determining risk of bias in epilepsy systematic reviews.