Hello Manar,

0. 43 cases isn't a big sample, so there can obviously be a lot of variation among the possible permutation outcomes.

1. "+inf" implies that at least 2.5% of the permutations of the data yielded a _very_ large computed OR (or an undefined OR, due to zero denominator), such that the upper CI limit cannot be assigned a specific numeric value.

3. Median (unbiased) estimate uses a different algorithm than ordinary maximum likelihood (for "ordinary" logistic regression models). Have a look at this reference for a deeper dive: https://www.jstor.org/stable/2685158

Good luck with your work.

David Morse Thank you so much.

So is it better to use Firth’s penalized logistic regression instead?

Hello again Manar,

WIth only two IVs, I don't know that it would make a lot of difference, but you could try. Personally, I'd prefer to know that the exact method indicated that the estimate of (conditional/adjusted) OR for IV #2 isn't very precise, given your data set.

Good luck with your work.

Thank you Dr. David.