Personally, I gag when I see “cardiopulmonary arrest” listed as a cause of death. In every setting where I’ve been involved with death certificates (research, then a stint in hospital administration, then med school and beyond) we were always told it was not an acceptable cause of death. If the ME got involved, that was almost guaranteed to get the certificate sent back (but I think the ME just gave our hospital a hard time in general, for unknown reasons predating my tenure – they’s end anything back that wasn’t black medium point ballpoint, if they spotted it: no blue-black ink, fine points, fountain pens, felt tips… )
It’s a running joke in the medical records world. It just means “his heart stopped beating and he stopped breathing”–pretty much a definition of death, not a cause. The Valentines Day Massacre wasn’t "natural causes"just because “it’s only natural to die when you’ve been shot in the head that many times with a Tommy gun.”
Yet it was the single most common ‘cause of death’ on certificates up to the early 90s (when I last had cause to check) and probably still is. Old habits die hard, I guess. I’ll also concede that it’s often tough to coherently summarize the cause of death in a aged patient with multiple serious ailments–but it’s rewarding to make that effort.
In every setting I’ve been in, the “cause of death” is certified (signed) by a physician, coroner (though many rural coroners have no medical training at all) or equivalent. A hospice nurse may write a cause of death, but generally it’s held to be officially declared by whoever signs the death certificate. It’s also not static. It’s not completely rare for the findings to be revised; there’s just rarely any driving cause to do so.
Autopsies are much less common than they used to be. Hospitals often do only as many as they need to to maintain certification (and may take a hit, accepting a shorter cert rather than meet the ideal standards). This is a pity in my book, because traditionally (again, back when I was in touch with such things) a fair fraction of all autopsies found something unexpected, which often may have affected treatment. If you don’t look, you’ll never know what you might have missed – and though that info often wouldn’t have changed this patient’s outcome appreciably, the feedback can improve your care of others.
IMHO, those unexpected (and not always relevant) findings are precisely why fewer autopsies were done as society became increasingly litigious. It’s often said that insurance won’t pay for autopsies, but in my lifetime, the cost was routinely absorbed as a cost of training or quality assurance. By the time I went to med school, that had ended around here. (Let’s just say the routine autopsy dies of natural causes – it’s only natural that almost no one will want them, if you start charging the family thousands of dollars for them, and Grandpa is, after all, already dead).