PROC Logisitic Question

pramenon1 — Sun, 18 Jul 2021 21:56:22 GMT

I have three variables:

cancer = 1 (they have cancer), 0 (no cancer)
death = 1 (died), 0 (did not die)
agegroup = 1, 2, 3, 4, 5 (depending on what age bracket they fall under.

I am trying to see if cancer status affects death in people of the above agegroups.

My code right now is:

PROC LOGISTIC data = dtanl.hivcancerdta order = data;
class agegroup (ref = "1");
model death01 = agegroup cancer/ expb;
ODDSRATIO agegroup / diff = all;
WHERE ALSO death01 ^= 9;
RUN;

The way I am interpreting the above model is "what is the likelihood of death in persons of a certain agegroup when controlling for cancer". Is this correct?

Re: PROC Logistic Question

PaigeMiller — Mon, 19 Jul 2021 13:01:55 GMT

Technically, the fitted model predicts the logarithm of the ODDS RATIO. From the predicted log(odds ratio), you can compute — or have PROC LOGISTIC compute — the probability of DEATH01 being 0.

topic PROC Logisitic Question in Statistical Procedures

PROC Logisitic Question

Re: PROC Logistic Question