0
$\begingroup$

I am currently using SAS 9.4 to carry out a retrospective analysis of patients who switched from drug A to drug B. I am interested in assessing the hospitalization counts before and after switch to drug B. Baseline drug A treatment was for 3 months, and follow-up time on drug B was for 1 year. The patient cohort served as their own comparator group. How do I account for the differences in treatment periods (3 months of drug A vs 1 year of drug B) when comparing incidence rates using Poisson GLMM models? Is there a way I could account for the differences in time periods in the model? Is there another way I could compare the incidence rates?

$\endgroup$
0
$\begingroup$

You should use $\log(\text{exposure time)$ as an offset in the (log link) poisson regression. That way you are comparing rates and not expected counts. For details see How is a Poisson rate regression equal to a Poisson regression with corresponding offset term? (and its links and references.) You do this the same way with a random effects model (I suppose you plan to include patient-level random intercepts?) and for a pure glm with only fixed effects.

| cite | improve this answer | |
$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.