In time-to-event analysis using Cox regression, what is the role of censoring?

• A. It allows inclusion of individuals without events by accounting for partial follow-up
• B. It biases estimates and should be avoided
• C. It indicates that all individuals eventually have the event
• D. It is only applicable to randomized trials

Answer: (A)

Censoring in time-to-event analysis represents incomplete information about a participant’s time to the event. In Cox regression, right-censoring occurs when a person’s follow-up ends before the event happens (due to study finish, loss to follow-up, or withdrawal). Rather than discarding these individuals, Cox regression uses the information available up to the point of censoring. They contribute to the risk set up to their censoring time, which allows us to estimate hazard ratios while properly accounting for those who did not experience the event during the study.

This approach enables inclusion of individuals without events and avoids bias from ignoring those with partial follow-up, provided censoring is non-informative (the reason for censoring is not related to the risk of the event after accounting for covariates). The other statements misstate the role: censoring does not bias estimates when handled correctly, it does not imply everyone will eventually have the event, and it applies to both observational studies and trials, not just randomized ones.