I'm reading an article where the AOR was computed after adjusting for confounders; when I calculated the OR it was lower than the AOR presented in the article. Why did the AOR increase? I thought it would be lower after adjustment


1 Answer 1


When you adjust anything can happen. For an example consider the following very simple data-set.

    r   n  size treat
1  81  87 small     a
2 234 270 small     b
3 192 263 large     a
4  55  80 large     b

taken from http://www.bmj.com/content/292/6524/879, an article comparing two treatment methods on two different sizes of renal stones where there were $r$ successes out of $n$ people treated.

The log odds ratio for small compared to large unadjusted is 1.070 and for treatment b compared to a is 0.2899. If you adjust each for the other we get 1.2606 and -0.3572 respectively. So small is now further from unity after adjustment and treat is larger in magnitude but with a different sign.

For further details of this example see this Wikipedia article


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.