I assume the "per 100 women" part suffices to regress towards the mean