Data derivation
For the EMA, for each approved drug, we saved the scientific discussion, label, and/or public assessment report listed in the database. Figure 1 shows a schematic overview of our method. To find candidate documents, we searched for the following terms [5, 7-10]: adaptive design, group sequential design, seamless design, adaptations to the sample size, sample size re-estimation, adaptations to the patient population, enrichment, adaptations to the treatment arm selection, adaptive dose-ranging, adaptive dose-finding, hypothesis adaptive design, biomarker adaptive design, adaptive treatment switching design, multiple adaptive design, umbrella design, basket design, platform design, adaptations to patient allocation, covariate-adaptive treatment assignment, adaptations to endpoint selection, adaptive randomization method, and drop-loser design . Once at least 1 of these terms occurred in a document, the associated submission package was scanned to determine whether it included terms listed above. These approvals were incorporated into the candidate list.