Clinical trials often collect and assess survival (or time to an event) data from patients. The data can be used to compare two or more groups for the rate at which an event of interest occurs and the total number of events. Survival data does not follow the typical distribution of many other types of data as they are non-negative and depend upon the rate at which the observed events occur. They are subject to censoring (i.e. incomplete data), which can occur for various reasons. Survival analysis is an area of statistics designed for modeling this type of data. In medical research, it is often used to analyze the time to disease remission, progression, or death for patients’ cohorts or compare different treatments within a clinical trial.
It is possible to observe an event for every patient in a clinical trial. Thus survival analysis techniques are required to help fill-in the missing data. Censoring occurs in case the subjects or participants drop out of a study for reasons independent of survival or fail to follow-up. It includes administrative censoring, which occurs when subject events do not occur before the cessation of follow-up. The models that are designed to analyze recurrent events data, consider event frequency and the likelihood of further events occurring given the number of previous events. It is also possible to include terms within models that adjust for time-dependent covariates or suspected unobserved heterogeneity among subjects.