Survival Analysis II
- 11 Downloads
Survival analysis modeling is integral to clinical trial analysis, as even in well-designed randomized trials where the primary inference is to be based on fundamental quantities such as estimated survival distributions and nonparametric tests, survival models offer additional insights and succinct treatment effect summaries. The ubiquitous Cox proportional hazards model has numerous variations and extensions to fit specific analytic needs and has become a mainstay of biomedical and clinical trial data analysis. However, other models and treatment effect metrics are increasingly available and should be adopted in cases where model assumptions are not met.
A natural extension of survival analysis pertains to the case where multiple potential causes of failure may be in effect. When these causes of failure are mutually exclusive, then competing risks observations are encountered, while in other cases, there may be multiple failures per individual. Methods that address these extensions of time to event data are needed to (a) assess of value of treatment in the presence of events that may preclude observation of the disease process of interest, (b) evaluate risks and benefits of treatment in a way that reflects patient experience, and (c) provide tools for study of factors related to different failure types and model more complex multi-event failure processes.
KeywordsSurvival modeling Hazard regression Proportional hazards Competing risks Cause-specific hazard Subdistribution hazard Cumulative incidence
- Rubinstein LV, Gail MH, Santner TJ (1981) Planning the duration of a comparative clinical trial with loss to follow-up and a period of continued observation. J Chron Dis 34(9–10):469–479Google Scholar
- Mantel N (1966) Evaluation of survival data and two rank order statistics in its consideration. Cancer Chemother Rep 50:163–170Google Scholar