Prevalence risk and 95% confidence interval for ordinal variables

[westbestern]

I'm using PROC FREQ to calculate associations between potential confounders, like age, and retinopathy. When I use PROC FREQ on a variable that has more than 2 groups, it won't show me prevalence ratio or 95% CI like it does with a two-level variable such as sex. For example, I want to use PROC FREQ to find the association between the age ordinal variable (age 30-60=1, age 61-70=2, >70=3) and retinopathy (0=no, 1=yes)

I did PROC FREQ age*retinopathy/no col nopercent all; run;

It gives me the chi-square but no CI or relative risk. How do I fix this?

[sbxkoenk]

Hello,

I would adopt a modelling approach, with PROC GENMOD for example.

See here (I haven't read the paper ! , just Googled it in order to try to give you an example) :

Proper Estimation of Relative Risk Using PROC GENMOD in Population Studies
Kechen Zhao, University of Southern California, Los Angeles, California

https://www.lexjansen.com/wuss/2013/81_Paper.pdf

Good luck,

Koen