The point wasn't to highlight the distinction between low-quality life and high-quality life per se. (That's a separate issue, as epoxyhockey said.) The point was only that the badness of an accidental death, and the cost we might be willing to imposed to reduce them, should probably reflect that counterfactual where no mistake occurs. QALYs lost are one way to measure the badness of a death, but he could have also just said "expected days of life lost".
Of course, epoxyhockey's would likely still disagree.
On the one hand, I could see arguing that a surgeon's error is independent of the patient. To the extend that's true, it would be important to count errors whether or not the patient is close to death.
But on the other hand, maybe the surgeon is performing a risky, error-prone procedure precisely because the patient is close to death. To the extent that's true, it might be misleading to compare this surgeon's record with someone else's.
I wonder if some of the resistance we hear from doctors about measuring/enforcing these things, is because of difficulties like this that are obvious to them and really screw them over?
Medical error. Just because the patient was going to die sooner than later, does not erase the fact that a medical error was committed.
Quality of life is a completely different topic, in my opinion, and I am sure we would find more common ground in that discussion.