I wonder if it is correct to use Poisson Regression with robust variance instead of logistic regression in a retrospective cohort study. Poisson Regression with robust variance (PRRV) estimates prevalence ratios and is useful in cross sectional studies, but I don´t know if it could be same or better than logistic regression in retrospective cohort studies. One must consider that prevalence ratios are easier to understand than the OR from logistic regression.