You are, again, failing to note that it’s a right skewed distribution. The mean is not what you care about.
I think the skewed distribution is less relevant than the sample group. There are limits to what we can expect from medicine. We don't have improved survival after decapitation, but that doesn't mean surgery has been at a standstill.
It would be interesting to look up survival rates for earlier stages. I expect the difference would be more substantial.
It seems relevant, as a patient to ask, on average, how much longer will I live if I take this drug?
If you run a year long trial where you flip a coin and immediately kill or save people based on the results, your average survival will be 6 months