Poisson Regression Coefficient Interpretation

[Rani Mohamed]

Hi everyone,

I have a question regarding the interpretation of coefficients in a poisson regression output. I have two models, one is subset to individuals who were transferred to SNFs (model 1) and the other is subset to individuals who were transferred to HHAs (model 2). The epxosure is osteoporosis, outcome is death within 30-days, and age, sex, and race are covariates. Intuitively, I believe the interpretation for osteoporosis coeffieint is "individuals with osteoporosis, compared to those without, staying in SNFs are expected to have a rate 0.80 lesser for death within-30 days" (in Model 1). For model 2, I believe the interpretation to be ""individuals with osteoporosis, compared to those without, staying in HHAs are expected to have a rate 0.89 lesser for death within-30 days". Can anyone confirm this? I've used linear reg in STATA and interpreted the coefficients in terms of percentage points but was unsure if that would apply here. Thanks.

[Carlo Lazzaro]

Rani:
please read and act on the FAQ when posting.
Please post what you typed and what Stata gave you back. Thanks.

[Bruce Weaver]

Hello Rani Mohamed. Your subject line says you estimated a Poisson regression model. But in your post, you say that the outcome is death within 30 days, which suggests binary logistic regression. Finally, you say you have used "linear reg". So please do follow Carlo Lazzaro's advice in #2, as this will clarify for everyone exactly what you did.