odds ratio and 95% CI

[lsak]

Hello, 

I am trying to figure out the prevalence of MDROs in urban and rural communities in my state. My professors say I need the proportion with 95% CI for which I am using the following code:

proc freq data=thesis order=data;

tables RUCA_3CAT*mdro_res / chisq relrisk;
exact pchi or;

title 'RUCA MDRO RES';
run;

***ruca_3cat is the variable for where the participants live and mdro_res is for MDRO results

I am not sure I am using the right thing for this, specially as this is not giving me odds ratios. Can someone help with this? 

Thanks, 
LSAK

[StatDave]

That should certainly give you odds ratios but only if the both variables have only two levels.

[ballardw]

What does the LOG show when you run that code?

I suspect you are getting something like:

NOTE: OR estimates are computed only for 2x2 tables.

Which means that if you want confidence limits of the odds ratios then Proc Freq is not appropriate. Which if you want confidence limits on odds ratios likely points to Proc Logistic.

One suspects that a variable with the bit _3CAT in the name may refer to a variable with 3 category levels.

[Reeza]

And if you do have more than two categories, you need to use PROC LOGISTIC instead.

https://support.sas.com/resources/papers/proceedings12/427-2012.pdf

https://www.youtube.com/watch?v=9_Zh-tal7Og