a question about how to do age-adjustment in SAS

[liiiiiiiib]

dear friends,

I want to  age-standardize the prevalence of different groups divided by agegroup, but I don't know how to implement it in SAS system. And some questions are still remaining. First, can I age-standardize the prevalence of different groups divided by agegroup? and second, how to write the code? here is my code ,but it seems that the prevalence is not age-standardized or only the total prevanlence is adjusted and others isn't. 

proc format;
VALUE riagendr 1 = 'Male'
2 = 'Female';
VALUE agegroup 1 = '18-44'
2 = '45-64'
3 = '65+';
value weightcat 1="bmi 0-24.9"
2="bmi 25-29.9"
3="bmi 30-34.9"
4="bmi 35+";
VALUE waistgroup 1="Abdominal obesity"
0="no Abdominal obesity";
run;

PROC SURVEYREG data=diffcomcat nomcar;
STRATA sdmvstra;CLUSTER sdmvpsu;CLASS agegroup;WEIGHT wtmec12yr;DOMAIN sub;
MODEL comcat = agegroup /noint solution vadjust=none;
ESTIMATE 'total' agegroup 0.5305 0.2992 0.1703;
TITLE 'Age-standardized prevalence of persons 18 years and older with comcat-age';
ods OUTPUT estimates=comcatageadj_prev_total;
run;

help please!!!!!!!

[sbxkoenk]

Hello,

I have not read below paper, but does it help you?

Adding the Feature of Age Adjustment for Survey-related Procedures in SAS

-- Age Adjusting Prevalence Estimates from Population Based Surveys
Zhiwei Zeng, County of Los Angeles, Los Angeles, CA

https://www.lexjansen.com/wuss/2007/CodersCorner/COD_Zheng_AddingTheFeature.pdf

It is a paper from the SAS WUSS (WESTERN USERS OF SAS SOFTWARE) 2007 conference.

Always a good idea to check :
https://www.lexjansen.com/

when searching for SAS programs on a topic.

Cheers,

Koen

[liiiiiiiib]

thank you for your help, but it doesn't solve my problem. Owing to my poor Statistical knowledge, I wonder whether it's ok to do age-adjustment in groups devided by age. 

[StatDave]

Standardization of rates (for count response) and risks (for binary responses) can be done in PROC STDRATE. You can find examples in the STDRATE documentation. However, it is not appropriate for sample survey data if that is the nature of your data.