Enhancing Systematic Review Methods by Incorporating Unpublished Drug Trials

Systematic reviews often omit unpublished data due to time constraints, unclear data integration methodologies, and uncertainties about the value of including these data. This can lead to inflated efficacy estimates and underestimated harms or burdens, as unpublished studies often document negative efficacy outcomes and/or harms. This brief methods project demonstrated the integration of U.S. Food and Drug Administration (FDA) drug approval packages (i.e., FDA reviews) into three systematic reviews focusing on psychoactive drugs—paroxetine (Paxil®) for adults with post-traumatic stress disorder (PTSD), escitalopram for major depressive disorder (MDD) in adolescents, and aripiprazole for bipolar disorder in adults. We searched the FDA reviews and matched identified trials with those in the Agency for Healthcare Research and Quality’s Effective Health Care Program systematic reviews. Then, we conducted meta-analyses combining unpublished and published studies. Our analysis identified important discrepancies in effect sizes, with meta-analyses of both published and unpublished data for paroxetine showing smaller effect sizes than those using only published trials. Our findings suggest a potential overestimation of paroxetine efficacy in published literature. Incorporating FDA reviews could enhance the accuracy of effect estimates in systematic reviews. However, using unpublished data requires careful consideration due to the intensive resources required. Future research should expand to other therapeutic areas and include more sources of unpublished data to increase the robustness of systematic reviews.