Can an odd ratio be as low as $< 0.001$? With a 95% CI of $< 0.001$ to $> 999.999$? Any idea what could be wrong?
(Sorry about the french output, will gladly provide more info if needed)
Thanks a lot!

 A: This will almost definitely be due to having zero events (individuals with the outcome) in the level revenu_final == 0. Thus the odds in that group are zero, and that will create issues with the estimation of the odds ratio.
The odds ratio is basically being printed as zero (the point estimate printed as less than 0.001) and the confidence interval is basically from zero to infinity... so in other words, can't really say much about this group from the data you have to hand!
It can be helpful to look back at the parameter estimates table (i.e. the log odds-ratios for these dummy variables; this is somewhat dependent on how the model is parameterised) and you will see that the log-OR for this parameter is a very large negative number and the standard error will be massive too.
A: It is possible for the maximum likelihood estimate of the log odds ratio to be infinite or negative infinite (the simplest case is a 2x2 table of binary $X$ by $Y$ with a zero cell).  If the MLE of the log odds ratio is negative infinite, the odds ratio is zero.
