I think you can only either generalise through statistics of total crash / fire split into two populations (with / without bladder fitted) or take a rather in-depth look at individual accidents.
If looking at a single accident you would need to consider whether the bladder tank did / would have made a difference. For example, deformation of the metal tank that lead to loss of integrity and fuel leak / fire might have turned out differently with a bladder fitted. A fuel tank penetrated by a tree branch and ripped due to forward motion leading to fuel leak / fire might well have ended up with the same fuel leak / fire. That kind of analysis is often part of the accident investigation report so past accidents (with no bladder) might also be a useful source of data.
John