Surveylogistic: output issues

[SAS93]

I’m using SAS to look at older ACEs data from the 2012 BRFSS dataset. Right now, I’m just trying to match the estimates I have from an official report done by the state health department I’m looking at. They have the ACEs divided into categories of: 0, 1, 2, 3, and 4+ (out of 8 total ACE types they look at).

I’ve successfully gotten the prevalence estimates of each of the ACEs and the individual categories. What I’m having issues with is the surveylogistic procedure; they used the categories to look at outcomes of various things like employment, graduation rates, and smoking habits.
I can’t match the ORs. Sometimes I can get a relatively close pattern/trend, but never close to the ORs themselves.

I’ve tried recoding the variables & categories various ways, but it doesn’t get me any closer to a more reasonable output. Any ideas of what might be going wrong and how to approach fixing it?

[ballardw]

---

@SAS93 wrote:
I’m using SAS to look at older ACEs data from the 2012 BRFSS dataset. Right now, I’m just trying to match the estimates I have from an official report done by the state health department I’m looking at. They have the ACEs divided into categories of: 0, 1, 2, 3, and 4+ (out of 8 total ACE types they look at).

I’ve successfully gotten the prevalence estimates of each of the ACEs and the individual categories. What I’m having issues with is the surveylogistic procedure; they used the categories to look at outcomes of various things like employment, graduation rates, and smoking habits.
I can’t match the ORs. Sometimes I can get a relatively close pattern/trend, but never close to the ORs themselves.

I’ve tried recoding the variables & categories various ways, but it doesn’t get me any closer to a more reasonable output. Any ideas of what might be going wrong and how to approach fixing it?

---

Did "they" share the code used?

Are you sure that you are using the same data set as "they"?  If you use the National available data set it is not the same as available to state BRFSS coordinators and you can get different results. Plus the state health department may have had other considerations for inclusion/exclusion of data for a specific analysis.

And as a minimum you might want to share the code you used as there are options that are typically used for BRFSS that you may not have used such as NOMCAR.