Hi all,
I have a problem that I can�resolve. I�m working on a survival model to describe bovine abortions using Cox proportional hazard. To model the effect of stage of gestation on the hazard of abortion I created a TVC using stsplit, splitting my data into three periods (trimesters) as follows:
*Setting up survival analysis
stset exit, id(ev_id) origin(time enter) failure(ev)
*Splitting data at days 90 and 180 to create variable trimester (trm)
stsplit ent_tr, at(90 180)
gen trm= 0
replace trm=1 if ent_tr>89
replace trm=2 if ent_tr>179
stset exit, id(ev_id) origin(time ent_tr) failure(ev)
xi: stcox nc_st la_no i.trm*nc_st
The stcox command gives me a point estimate of the hr or coef. of the trimester effect but does not give me estimates of std err or conf. int. How can this be? I did the same thing splitting the data into two periods and it worked fine.
Cox regression -- Breslow method for ties
No. of subjects = 1112 Number of obs = 3252
No. of failures = 138
Time at risk = 242846
LR chi2(12) = 38.60
Log likelihood = -923.21063 Prob > chi2 = 0.0001
If anyone could help me out I would be very grateful,
Karl St�hl
PhD-student
Virology and Parasitology,
Dept. of Biomedical Sciences and Veterinary Public Health
Faculty of Veterinary Medicine and Animal Science
SLU
Uppsala, Sweden [email protected]
