We conducted a clinical trial comparing two treatments in which for one of the endpoints one of the groups had no events, resulting in the risk difference being the rate oberved in the comparator arm. I know that there are approaches (such as the rule of three for estimating the superior limit of the 95%CI for a single group with no events), however, I would like to know if there is any recommended approach for calculating the 95% CI of the risk difference when one of the groups has had no event. 

Thank you very much!