I would think when making an approval submission to the FDA with an all cause mortality endpoint they'd also break out the data for clearly unrelated events. All cause mortality should be the gold standard. A drug may cause negative effects unrelated to the primary focus of the trial.

From a logical standpoint including mortality that could not be caused by the treatment unfairly skews the results. If all patients improved in a trial but then an earthquake collapses the hospital they're in and all die does that mean the drug has no efficacy?