Single-arm clinical trials that supported FDA accelerated approvals have modest effect sizes and were at high risk of bias

•Single-arm trials that informed treatment approval by the FDA-AA have modest effect sizes.•The lower bound of the 95% CI of 53% of approved treatments crossed the line of null effect.•The effect sizes were not larger in the studies that were followed by the FDA-required confirmatory randomized trials. To describe effect sizes of single-arm clinical trials that supported AA approvals. We reviewed all the single-arm approvals granted by the FDA-AA pathway between June1992 to December2020. Two independent reviewers identified single-arm studies and extracted data from FDA Full-Medical Reviews. We performed a meta-analysis to estimate the effect sizes and compared it between studies that met post-approval FDA requirements for RCTs with those that did not. From the total of 254 approvals, single arm clinical trials describing effects of 54 drugs for 72 clinical indications were evaluated. The effect size estimated was OR:2.22(CI95%:1.76-2.81) [relative risk (RR) = 1.63(95CI% 1.38-1.92)]; 53% of treatments had a lower 95% CI bound crossing the null effect. Effect size did not differ between the treatments that met the FDA requirement for conducting post-approval RCTs. Treatment effects observed in the FDA AA single-arm studies was modest and can be to ascribed to bias.