Treatment comparisons using a measurement during follow-up and a right-censored time-to-event outcome.

In clinical trials, the comparison between treatment arms of a measurement is often complicated by an event such as death or progression that can make this measurement irrelevant or difficult to perform. In this article, an approach is proposed to compare treatment arms using the net treatment benefit or the win ratio. A more favorable outcome (win) for the experimental arm is defined as the event of (1) a longer time to event for a patient in the experimental arm compared to a patient from the control arm until a prespecified restriction time ϵ or (2) a more favorable outcome with respect to the measurement during follow-up for a patient from the experimental arm and the time to event greater than ϵ for both patients. An estimator is proposed. A simulation study showed approximately no bias. The methodology is illustrated using a leukemia trial.