We consider a comparison of Kaplan–Meier curves from clinical trials in which there may be a delayed treatment effect. Any such delay takes us outside the umbrella of a proportional hazards structure ...