The data is in Andersen-Gill (long) format for longitudinal/survival analysis. Say you're trying to fit the following model in SAS for a discrete-time logistic regression survival analysis with a complimentary log-log link:

proc logistic data=dataset;

class estrogen testosterone time / param=glm;

model CANCER (event='1')= ESTROGEN TESTOSTERONE TIME

/ noint link=cloglog CL;

run;

If the coefficient for estrogen was say b=0.50, how would you interpret that in English with respect to the relative risk/OR/HR for cancer with a CLOGLOG link? I'm not familiar with the cloglog link function for this data structure.

Thanks community!