To get more specific this often means 'suicide' which absolutely is a casualty of war given the strain of war conditions, especially given the logistics concerns and duration of tours that they are forcing right now.

> If someone on a military base with no attacks going on, got a heart attack by watching the news, probably not a war fatality.

Were they at that base as a (say) mechanic servicing aircraft engines for flights over the Persian Gulf? Are the only reasons why the planes (and thus him) are there is because of the hostilities with Iran, as otherwise they would be States-side?

There’s often more ambiguity than that. “Died of a COVID linked complication with COVID and other issue that is also linked to that complication.” You can’t spend endless resources deciding if something is or isn’t responsible.

Systems use arbitrary buckets like if someone commits suicide in less than six months from coming home after a war might count it, while excluding every case after six months. That’s going to misclassify cases in both directions but means you don’t need to track millions of people for decades or look through mental health details.

Obviously people with more data to work with can make a convincing case you’re wrong, but higher levels accuracy get expensive. In the end we know more than 1 million Americans died of COVID for example, how many more isn’t precisely known but ~1 million is plenty useful for decision making around the next pandemic.