[Correspondence] The importance of patient selection in stroke monitoring research – Authors' reply

By the very nature of having eligibility criteria to define the study population, all randomised controlled trials are complicated by some degree of selection bias. However, Atsushi Shiraishi and colleagues highlight the potential for an extension of this issue in differential selection bias across the randomised groups to have influenced the results in the stepped-wedge cluster-randomised design of the OPTIMISTmain study.1