Conclusion

PROC LIFEREG provides effective methods for regression analysis of censored survival data, especially data with left censoring or interval censoring. These methods are somewhat less robust than the more widely used Cox regression analysis performed by PROC PHREG but, in most cases, the results produced by the two approaches are very similar. Moreover, unlike PROC PHREG, PROC LIFEREG makes it possible to test certain hypotheses about the shape of the hazard function.

The biggest limitation of the PROC LIFEREG models is the inability to incorporate time-dependent covariates, although you can accomplish this to some degree with the piecewise exponential model. We now turn to PROC PHREG, which excels at this particular task.