Dear listers.
I'm having trouble adjusting a prevalence with logistic regression. Or
rather, I'm having trouble adjusting it to that "time point" or
covariates that I want. An example:
We have two time points, A and B. The prevalence of asthma among females
is 10%, and among males 20% at both time points. However, the gender
composition has changed from 50% females in A to 10% females in B. Thus,
the crude prevalence has changed. However, if we adjust the prevalence
to the gender composition at time point A, then the adjusted prevalence
at A should be the same as the crude prevalence, and the adjusted
prevalence at B should be the same as the prevalence at B. I can't seem
to do this. Working example:
clear
input time sex asthma freq
0 0 0 450
0 0 1 50
0 1 0 400
0 1 1 100
1 0 0 810
1 0 1 90
1 1 0 80
1 1 1 20
end
expand freq
*A:
logistic asthma time
adjust , by(time) pr format(%16.15f)
logistic asthma time sex
*C:
adjust sex=0.5, by(time) pr format(%16.15f)
*D:
adjust sex=0.570485, by(time) pr format(%16.15f)
In A, we get the right crude prevalences. In B, the prevalences are
equal between the time points, which is right. But the prevalences are
strange - which population is it adjusted to ? Through trial and error,
I found that the adjustment in D produces what I would call the correct
adjusted prevalence (both prevalences equal to the crude prevalence at
time point A).
Any help would be greatly appreciated.
Yours sincerely,
Jan Brogger, MD, Institute of Medicine, University of Bergen,Norway
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/