How on earth did they get ethical approval to run this study?
"So, do you want to enter into a 50/50 lottery, where we will either try to treat you, or let you die in peace? You don't get a choice of which, by the way."
Even if I was a cancer patient with an extremely poor outlook, I would probably have some pretty strong preference in the question of whether I wanted chemo, or end-of-life care.
I agree, it seems unlikely that you could get an IRB sign off any study that wasn't simply post-death statistic study.
You /could/ try to control that by try to separate the groups into "type of cancer" and "stage", so that you're comparing like-to-like when looking at mortality. But even then there are problems ("stages" are subjective, and there's evidence that they're slowly shifting so that old stage three would now be considered stage four), different Drs may be more aggressive at directing people to chemo, etc.
I think there's also basic issues of what we're comparing, e.g. are you undergoing chemo in an attempt to live a bit longer, or as an attempt to not die. Because at the "you're going to die within 3-6 weeks" you're going to get radically different survival rates for chemo than "you're going to die in 3-6 years". Which makes me wonder how you would deal with "EoL" chemo that does cause the cancer to die back?
I'm having trouble wording it, but what I'm curious about is what happens when you have a patient that
1. Is dying, and given 3 months to live. Can choose between chemo and hospice
2. Chemo successfully pushes the cancer back
3. Cancer comes back 5 years later
4. Patient chooses hospice
5. Patient "dies in hospice" 3 months later.
Which group should the patient be placed in?
This is why I'm glad nothing I do impacts peoples health/safety/etc
Because otherwise, there's a million statistical biases that can creep into that outcome.