Can the Size and Duration of Clinical Trials in New Drug Applications Predict Post-Approval Safety Warnings?

Drug companies seeking regulator approval for a new drug must submit evidence regarding its efficacy and safety, based partly on the results of randomized clinical trials. The number of people who must be exposed to the new drug in these trials, and the duration of their exposure, are two dimensions that are agreed to by the regulator and company. Regulators face a trade-off: insisting on high sample size and long exposure makes it less likely that unforeseen problems will emerge after drug approval, but raises the cost to companies of getting a new drug approved. Excess rigour could result in fewer beneficial drugs being developed. Under current practice, sample size and duration requirements are moderate, but after drug approval clinicians report adverse reactions to regulators, who upon investigation may issue additional safety warnings. We study whether, conditioned on drug type, regulators sufficiently adjust the sample size and duration they require in new drug applications. Specifically, we manually record the sample size and average duration of exposure for all clinical trials underlying the 424 new drugs approved by the United States’ Food and Drug Administration (FDA) between 2008 and 2019, and the number of new FDA warnings issued after each drug is approved for a specific use. We then test whether sample size and average exposure duration pre-approval can significantly predict the number of new FDA warnings post-approval. We begin with simple predictive models using only sample size, duration, and time since approval, then add controls for approval process, then type of drug, and instrument for time and duration using the number of trials and average length of trials. We do not find that pre-approval sample size and average duration of exposure reduce the number of new post-approval warnings, suggesting the FDA is not compromising safety in its current practice.