The dangers of non-randomized, observational studies: experience from the COVID-19 epidemic

Abstract

Abstract In regulatory evaluations, high-quality randomized controlled trials (RCTs) are considered the gold standard for assessing the efficacy of medical interventions. However, during the COVID-19 pandemic, the urgent need for treatment options led to regulatory approvals being made based on evidence from non-randomized, observational studies. In this study we contrast results from observational studies and RCTs of six drugs to treat COVID-19 infection. Across a range of studies evaluating hydroxychloroquine, remdesivir, ivermectin, aspirin, molnupiravir and tenofovir for COVID-19, there was statistically significant evidence of benefit from non-randomized observational studies, which was then not seen in RCTs. We propose that all observational studies need to be labelled as ‘non-randomized’ in the title. This should indicate that they are not as reliable for evaluating the efficacy of a drug and should not be used independently for regulatory approval decisions.