Survival Analysis Using SAS

NONPROPORTIONALITY VIA STRATIFICATION

Another approach to nonproportionality is stratification, a technique that is most useful when the covariate that interacts with time is both categorical and not of direct interest. Consider the myelomatosis data described in Chapter 2 and analyzed in Chapter 3. For that data, the primary interest was in the effect of the treatment indicator TREAT. There was also a covariate called RENAL, an indicator of whether renal functioning was impaired at the time of randomization to treatment. In Chapter 3, we found that RENAL was strongly related to survival time, but the effect of TREAT was not significant. Suppose we believe (or suspect) that the effect of RENAL varies with time since randomization. Alternatively, ...