Is it reasonable to have 1.6% men? If not, then you have a sampling bias issue. The other issue is sample size. If you have 125 surveys, and 2 of them were men, then there is not much you can conclude about the men. It might be better to indicate in the methods that there were few men (as expected) and with such a small sample these data were removed from subsequent analysis. I might put in a short results paragraph describing these unusual cases. You have more options if the number of surveys from men can be considered a valid sample.