Rejoinder: Improving the performance of Bayesian logistic regression model with overdose control in oncology dose‐finding studies

Abstract

hereafter FGK); Drs. Alexia Iasonos and Graham Wheeler 3 (hereafter IW). We appreciate all of their remarks about our proposed Bayesian logistic regression model (BLRM) with overdose and underdose control design in particular. We (and it seems our discussants) see the main contribution of our article as identifying an important statistical issue with BLRM that had not before been completely recognized and providing one fairly general statistical approach to address it. We recognize that many other dose escalation models could be developed and encourage colleagues and fellow practitioners to pursue these potentially fertile research paths as laid out by the discussants. In our rejoinder, we focus on three general issues raised in the discussions: alternative approaches, practical considerations, and model flexibility and parsimony.