The problem with a retrospective approach is that it is likely to elicit a biased response.
Any groups of people who feel strongly that want their views known (pro or anti) will respond as a higher proportion than others.
And, of course, such a retrospective approach can get no responses from those who have died. A prospective study will get responses up until death which may not be perfect in terms of inferring if the intervention was related to their deaths, is still far better than a retrospective approach where nothing would be known.
I think this is why Cass etc have suggested and support a prospective trial.