Interpretation of survival data in clinical trials

The results of therapeutic interventions can often be best described through the distribution of times to some event. The most obvious example of this is actual survival (time to death) and the methods used in the analysis of "time to event" data are broadly referred to as "survival analysis." However, survival analysis really refers to the evaluation of time to any definable event. The goals of survival analysis are to estimate the "time to event" distribution for a defined group of patients and to compare the experience of different groups of patients. In the case of a chronic disease like cancer, the analysis is complicated because the event of interest may not yet have taken place or the patient may have been lost to follow-up. Here we review the standard techniques used in the analysis of survival data and discuss important related issues.