Modeling Survival Data with Competing Risk Events using SAS Macros

Competing Risk (CR) event is one that a patient may experience (other than the event of interest) which can prevent the event of interest from occurring. A naive use of classical survival analysis methods in the presence of CR leads to a bias by overestimating the probability of disease incidence. Hence it is important to account for CR events when estimating disease incidence. This can be done using a cumulative incidence function (CIF) calculated by appropriately accounting for the presence of CR events. A nonparametric approach to implement above is possible using SAS in-built macro ‘%CIF’. This approach however, cannot directly model the effect of covariates or prognostic factors on CIF. A model-based approach proposed by Gray and Fine (1999) can overcome this problem. Present paper demonstrates both the approaches to deal with survival data in presence of CR through examples using SAS® macros.