Just another suggestion !!
Check the Gary King's web page (http://gking.harvard.edu),
specifically the software link at the botton of the page
He worked in the implementation of the rare events logistic regression
methods for Stata and GAUSS statistical packages suggested in King and
(1999a,b) for generating approximately unbiased and lower-variance
estimates of logit coefficients and their variance-covariance matrix
by correcting for small samples and rare events.
I've been checking the Stata command and you can obtain the robust
estandart estimation and it may be an aproximation for your