A new SAS macro for flexible parametric survival modeling: applications to clinical trials and surveillance data

Author(s): Ron Dewar & Iftekhar Khan

Survival analysis is often performed using the Cox proportional hazards model. Parametric models are useful in several applications, including health economic evaluation, cancer surveillance and event prediction. Flexible parametric models extend standard parametric models (e.g., Weibull) to increase the flexibility of the shape of the hazard function. We present a new SAS® macro for implementing flexible parametric models with a similar functionality to that of Stata®, with examples using data from cancer surveillance and clinical trials. Results from SAS® were identical with similar computational time to Stata®. The flexible parametric approach to modeling survival data is shown to be superior to standard parametric methods. This SAS® macro will facilitate an increase in the use of flexible parametric models.