Survival (Time-to-Event) Data
Courtesy of Dr. Donald Noah.
Survival analysis deals with time-to-event variables (eg, time until death). Survival time (or time-to-event) is a special type of continuous variable, measured as the time from enrollment until the outcome of interest. Death is often the outcome of interest, but time to other events could be studied besides death (eg, time to treatment failure, time until a patient becomes disease-free, or time to disease recurrence).
A survival curve (or time-to-event curve) plots the proportion of subjects that have not yet had the event as a function of time (the proportion being an estimate of the probability of survival at that time). Every time a subject dies (or has the event occur), the curve steps downward (ie, the proportion of subjects surviving [or that are event-free] decreases). The time at which 50% of subjects have died is the median survival time.
If a subject is lost to follow-up or otherwise drops out early—or if the outcome of interest has not occurred by the end of the study—the subject is censored (removed). Censoring must be accounted for to prevent bias in the results. Kaplan-Meier survival analysis is a popular method for estimating survival curves that formally accounts for censoring.
