Estimate two risk ratios with a predictor that has three levels

[hongjie76]

Hi,

I try to use proc genmod to estimate two risk ratios because the predictor of smokes has three levels (never, current, and former smokers). Would you please tell me how to do that? Here is the SAS syntax I made. Thank you in advance.

proc genmod ;
class seqn smokes /param=ref ref=first;
model lung_cancer=smokes /dist=poi link=log;
repeated subject=seqn;
run;

[StatDave · Accepted solution]

See this note on estimating risk ratios (relative risks). As shown there, you can use PROC LOGISTIC with LSMEANS followed by the NLMeans macro or PROC GENMOD, again with an LSMEANS statement. Since the latter seems to be the approach you are taking, you just need to add an LSMEANS statement with the DIFF, EXP, and CL options (assuming you want confidence intervals). The LSMEANS statement will handle however many levels are in the predictor. But you will need to change PARAM=REF to PARAM=GLM to use the LSMEANS statement.

[hongjie76]

Many thanks. It works. Hongjie