Not Hazardous to Your Health: Proportional Hazards Modeling for Survey Data With the SURVEYPHREG Procedure
The Cox proportional hazards model is widely used in practice to estimate the effects of covariates on survival function. The SAS/STAT PHREG procedure is a long-standing workhorse for performing regression analysis for event-time data based on the proportional hazards model. However, special techniques are required when the subjects are selected through complex surveys with stratification, clustering, unequal selection probabilities, or any combination of these. This paper introduces the SURVEYPHREG procedure, a new SAS/STAT procedure for finite population inference for the Cox proportional hazards model for complex surveys.