Help Interpreting IRRs in Proc Genmod

[luvduvbunny]

Not sure if this is the right location to post, but I'm fairly new to SAS and I would like some help interpreting IRRs (and general data) in proc genmod.

I'm not sure what the column "exponentiated" are supposed to tell me. Does it mean that one group has higher/lower odds of an event happening?

Below is a screenshot of my data. Some background info, I'm trying to see if people with underweight, normal, overweight, and obese BMIs are more likely to see a healthcare provider

[StatDave · Accepted solution]

You should always show your PROC statements so we know exactly what you did. Those results only make sense as relative risk estimates if your model is a log-linked Poisson model with a binary (0,1-coded) response. If so, then see the "Zou's modified Poisson approach" section of this note. The exponentiated column in the differences table gives the point estimates of the relative risks. The BMI and _BMI columns indicate which levels are in the numerator and denominator, respectively, of each relative risk. So, the first estimate, 0.88, means that the probability of the response event for BMI=NORMAL is 0.88 times the probability of the event for BMI=OBESE. The exponentiated column in the preceding table shows the individual level probabilities where you can see that the NORMAL probability estimate is 2.01 and the OBESE estimate is 2.26 - the ratio being 0.88.