Time and change: Using survival analysis in clinical assessment and treatment evaluation

Survival analysis is a set of statistical techniques that is useful for modeling the types of changes commonly encountered in clinical assessment and treatment evaluation. This article provides a practical, comprehensive, mathematically sound yet nontechnical introduction to survival methods. After discussing study design and data, a complete example data set from a fictitious study of alcohol relapse patterns is introduced to illustrate commonly used survival analysis procedures, including the life-table method, the Kaplan-Meier procedure, the Cox proportional hazards model, and fully parametric survival models. These methods are used to describe the general survival and hazard functions, compare survival probabilities for groups of patients, develop multivariate hazard models, model the shape of the hazard function over time, and use diagnostic tools to check statistical assumptions. Complete SAS and SPSS programs are included in Appendix B.