Dude, if I could clone you 100x I would.
You're asking reasonable questions. I honestly don't know how to explain the specific point you're raising about mortality difference. Is it real? I absolutely think it's possible. Which is why I tend to ignore papers like this. I don't care that it's not prospective or randomized. I care that people are putting so much faith in an imperfect source (all databases) and then assuming they can math their way to the truth.
By building a spider web of complicated statistics and "finding" these potentially concerning conclusions, which may or may not be real - there is now a new, very high opportunity cost for biomedical science as we must all debate and/or study this until it's supported or refuted...which might never happen. This would then lead me down the path of too many things are being published because of misaligned incentives, but I'll save that tangent for later.
Surgery is absolutely a good choice in the correct patient. I know for a fact I am more pro-surgery than some Urologists in my area. It drives my patients nuts because, as we all know, some people just want to be told what to do. They hate when I say "there's no wrong choice". And then they say "yeah but what would YOU do if you were me?" and so on.
I'll stop dunking on Coops and stick with the paper. It's a good illustration of my concerns:
View attachment 378039
This database began when Bill Clinton was President. Not only that, it was "pre-Lewinsky" Clinton.
Time-based confounders are a fan favorite, so I'll stop there.
Next, we have self-reported variables and comorbidities were assessed using the "Charlson comorbidity index". As with time, patient self-reporting is a fan favorite and beat to death. But I don't even know what this Charlson thing is. I do know, with absolute certainty, it's not perfect, whatever it is. And I know it has its own confounders internally, and then introducing it in this database causes confounders and...blah blah.
Then we hit the stomping ground of treatment decisions were left up to individual practices across the country. I've trained/practiced medicine in 5 states over the years. Nothing is done the same anywhere, and all it takes is one doc leaving or one doc joining and it's not even consistent internally. This is also a dead horse.
But, at least I can be somewhat novel with this next item:
For many years I moonlighted with a hospice organization. I have signed dozens to hundreds of death certificates. I literally don't know, I never kept track.
For those that haven't done it, death certificates are annoying. They say in black and white "YOU CANNOT JUST WRITE HEART STOPPED, IDIOT".
View attachment 378040
So here's what I would usually do:
First, I was caring for them as a hospice patient. Which means I don't need to go down an insane "chart biopsy" rabbit hole to give them appropriate care. If they have chest pain, I'm not booting up a 10-item differential and calling Montana to get records because the patient lived there for a year in 2002.
So I would always write, as the final terminal event, "cardiopulmonary arrest" because...well that's literally how you die.
But obviously that's against the rules. I can't just write that and call it a day. I need to write out the "chain of events".
Sometimes these patients would be 92 and have a chart longer than Homer's Odyssey. So I would find the admitting H&P or the last comprehensive note before they entered hospice and find what the most likely/most recent terminal illness was, like "pancreas cancer". So that's what I would write, then I would sign it, then I would go on with my day.
Now, I'm a cancer doctor. I have a comprehensive understanding of the natural history of many malignancies. If I saw someone was diagnosed with pancreatic cancer, I know how aggressive that is, and I know it probably is the true reason.
But...prostate cancer? "Most men die with prostate cancer, not of prostate cancer". There's a lot of men on hospice, or just at the end of life in general, with prostate cancer.
Unless it was metastatic prostate cancer, I would almost never write "prostate cancer" in the chain of events. I know how indolent it is. Usually, the H&P note would be something like "past medical history notable for prostate cancer, 1 month ago experienced ground level fall with hip fracture, BIBA and was evaluated by Ortho in ED, determined not to be a surgical candidate, admitted for pain management where he developed aspiration pneumonia and AMS..."
Reading that, I would write "traumatic fracture, pneumonia, cardiopulmonary arrest". I wouldn't write prostate cancer because it has nothing to do with the death.
But...the people most commonly signing death certificates are not Radiation Oncologists. They're often not cancer doctors at all. They're usually in the realm of general internal medicine.
Give a hospitalist that same H&P, and their death certificate would probably read "prostate cancer, pneumonia, cardiopulmonary arrest".
So my closest guess as to what's happening with this CAPSURE study? It's that. It's how the database was designed to record PCSM, and who was filling out the death certificates.
Which is, of course, something we can't "prove".
My point being:
Surgery and radiation are both good options for prostate cancer.