Queries about Logistic regression -- 2

Hi, I am looking out for some advise about using nominal independent variable for binary logistic regression that has 5 or 6 IVs. Sample size is 240.

For a categorical independent variable (binary/3 categories) that is skewed and there are low frequencies in one or two categories ( below 30) when tabulated across the binary dependent variable. Can such a variable be used if further grouping/transformation is not possible? Are there any rules wrt to the minimum of no of observations needed per category of a nominal variable for use in logistic regression? Pleaae give a peer reviewed reference.