Survival Analysis Using SAS®: A Practical Guide

Generating Predictions and Hazard Functions

After fitting a model with PROC LIFEREG, it’s sometimes desirable to generate predicted survival times for the observations in the data set. If you want a single point estimate for each individual, the predicted median survival time is probably the best. You can get this easily with the OUTPUT statement, as shown in the following example:

proc lifereg data=recid;
   model week*arrest(0)=fin age race wexp mar paro prio
         / dist=weibull;
   output out=a p=median std=s;
run;
proc print data=a;
   var week arrest _prob_ median s;
run;

The P= keyword in the OUTPUT statement requests percentiles. By default, PROC LIFEREG calculates the 50th percentile, that is, the median. (The word median in the OUTPUT statement is ...