Conclusion

PROC LIFETEST is a useful procedure for preliminary analysis of survival data and for testing simple hypotheses about differences in survival across groups. For experimental studies, PROC LIFETEST gives results that are analogous to a one-way analysis of variance. But the procedure is not adequate for two-factor designs because there is no way to test for interactions. Similarly, while the TEST statement in PROC LIFETEST may be useful for screening large numbers of quantitative covariates, it is not adequate for examining the effects of variables controlling for other covariates. In most cases, therefore, you will need to move to the estimation of regression models with PROC LIFEREG or PROC PHREG.

It is also important to recognize that survival curves and their associated hazard functions can be misleading when the sample is heterogeneous. As explained in Chapter 8, uncontrolled heterogeneity tends to make hazard functions look as though they are declining, even when there is no real decline for any individual in the sample. One way to reduce the effect of heterogeneity is to estimate survivor functions on residuals from regression models, as shown in Chapter 4. Alternatively, you can estimate and plot baseline survivor functions after fitting Cox regression models, as shown in Chapter 5. Also in Chapter 4, we’ll see how to use PROC LIFETEST to decide among the several parametric models that can be estimated with PROC LIFEREG.