I agree with your points entirely. Also, these types of 'outcomes' statistical studies that use large database info (e.g. the Medicare database; combined registries) often do not tell the whole story. There is a ton of biological variability among people, and there may be subgroups of people who respond better than others. God forbid we let beancounters define the future of medicine.
Great point about differences among subgroups. Indeed, as the saying goes, the devil is in the details. The sort of useful information gleaned from those different outcomes will be lost if the “don’t even bother” mentality prevails.