I am performing a between-treatment comparison of time to completion of a task at one visit, where the time is constrained by a specified maximum (commonly referred to as Type 1 censoring). The same task is performed at Baseline, with some observations censored by the same specified maximum. I am considering analysis using Cox model. In SAS, the model would be like MODEL TIME*Censor(1)= TRT BASELINE. Is there any way to incorporate censoring at Baseline in the model? Is there any theoretical reason to do so?