I have a dataset that contains count and person time information for an event (CMG) stratified by a number of different variables (e.g. age, calendar year, etc...). I am using PROC GENMOD to construct a poisson model, using log(person-time) as an offset variable. The data was over-dispersed so I had to scale by the deviance.
How can I use the basic model (with no exposure or covariates) to calculate the unadjusted incidence rate of CMG in the entire cohort (regardless of covariate profiles)? I know what the value should be but I would like to get it from the model and also look at the associated 95% confidence interval.
Here's the basic form of the model I'm using:
PROC GENMOD DATA=work.Count_PT_data
MODEL TotalCMG = / DIST=poi LINK=log offset=LOG_PT SCALE=deviance;
RUN;
Thanks for your help! Let me know if there's anything more I can post.
Consider adding the OBSTATS option. It will generate beaucoup output, with multiple variables included, but most importantly it will give X*beta and it's confidence limits. This should be identical for every observation with the intercept-only model that you have specified. The confidence bounds will be based on the rescaled error. I really think that the value with the intercept-only model will reflect the mean exposure in the dataset. To get the estimate with exposure set to zero, you could try (WARNING--untested code. Just because I think it ought to work doesn't guarantee that it will):
PROC GENMOD DATA=work.Count_PT_data
MODEL TotalCMG = exposure/ DIST=poi LINK=log offset=LOG_PT SCALE=deviance;
LSMESTIMATE intercept 'Exposure=0' 1/at exposure=0 ilink;
RUN;
Good luck.
Steve Denham
I have done the same analysis, in the way Steve Denham explains and it worked out well.
This topic is described here aswell: 24188 - Modeling a rate and estimating rates and rate ratios (with confidence intervals)
Don't miss out on SAS Innovate - Register now for the FREE Livestream!
Can't make it to Vegas? No problem! Watch our general sessions LIVE or on-demand starting April 17th. Hear from SAS execs, best-selling author Adam Grant, Hot Ones host Sean Evans, top tech journalist Kara Swisher, AI expert Cassie Kozyrkov, and the mind-blowing dance crew iLuminate! Plus, get access to over 20 breakout sessions.
Learn the difference between classical and Bayesian statistical approaches and see a few PROC examples to perform Bayesian analysis in this video.
Find more tutorials on the SAS Users YouTube channel.