1
$\begingroup$

I have a binary outcome (Y) and a continuous exposure (X). I am running a logistic regression such that my effect estimate is per one unit increase in X. However, I would like to obtain an odds ratio per IQR increase in X. How do I need to transform my data to do this?

As an analogue, I know that if I wanted to get an odds ratio per standard deviation increase in X I would need to use the z-score of X instead of the raw values.

Improve this question
$\endgroup$
2
  • $\begingroup$ So far I'm just manually multiplying the OR and confidence limits by the IQR (EXP(LOG(estimate)*IQR)). That works but I'm sure there's a more efficient way. $\endgroup$
    – pakalla
    Commented Feb 28, 2020 at 23:51
  • 1
    $\begingroup$ Why would transforming the data help to answer this question? $\endgroup$
    – Glen_b
    Commented Feb 29, 2020 at 3:22

1 Answer 1

Reset to default
1
$\begingroup$

You can just divide that variable by its IQR. The reason Z-scoring a variable allows you to interpret its coefficient as a change in the outcome corresponding to a one-standard-deviation change in the predictor is that you are dividing by the standard deviation. The same logic applies regardless of the measure you divide by. Note this will not change the fit of the model or the value of any other estimates.

Improve this answer
$\endgroup$

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.