I'm a bit late to the party, but this is fantastic! Is there a faster way to run PHREG when testing the proportionality using time dependent covariates?  I find this is VERY time consuming in SAS, it would be great if there was a way to speed up the calculations. e.g.  proc phreg data=uis; model time*censor(0) = age race treat site agesite age treatt; aget = age*log(time); treatt = treat*log(time); run;   ... View more