Familial history is likely not a good excuse since there would be a very broad population even in a single branch.
Behavioral factors are still very much relevant because you can use it for a public health policy, if smoking and drinking is worse in the military than in the general population then it’s something that you can definitely try to address.
Another factor that can play here is that servicemen have probably better healthcare than the average population especially when it comes to preventative healthcare and checkups.
So whilst this study isn’t enough to identify causation it’s enough for correlation which is quite often as important when it comes to public health policies.