ABSTRACT
The primary objective of this study was to compare survival on the experimental treatment (CT + RT) to that on the standard treatment (RT only). Under the proportional hazards assumption, we can write a model for these data as
~(tlz) = ~o(t) exp(Oz) where ~o( t) represents the baseline hazard at time t, Z is a treatment indicator, and 0 is the log hazard ratio. The null hypothesis of no treatment difference is Ho: 0 = 0; the alternative hypothesis that the two treatments differ with respect to survival (two-sided) is HI: 0 =1= O. If we let Ll be the hazard ratio [i.e., Ll = exp(O)], we can write Ho: Ll = 1 and HI: Ll =1= 1.
