Austin Frakt for The Upshot (my own highlights):
A great deal of the decrease in deaths from heart attacks over the past two decades can be attributed to specific medical technologies like stents and drugs that break open arterial blood clots. But a study by health economists at Harvard, M.I.T., Columbia and the University of Chicago showed that heart attack survival gains from patients selecting better hospitals were significant, about half as large as those from breakthrough technologies. […]
Rather than clinical quality, which is hard to perceive, patients may be more directly attuned to how satisfied they, or their friends and family, are with care. That’s something they can more immediately experience and is more readily shared.
Fortunately, most studies show that patient satisfaction and clinical measures of quality are aligned. For example, patient satisfaction is associated with lower rates of hospital readmissions, heart attack mortality and other heart attack outcomes, as well as better surgical quality.
This story is a striking illustration that simple heuristics as “following the satisfaction of other people” can be surprisingly good (and cheap) at processing information. In other words, relevantly processing information does not necessarily require sophisticated individuals with alleged high cognitive skills – like homo economicus.
It’s also (and to me, very) important to note that these heuristics are collective heuristics: it’s not a single individual who processes information for the whole group. Rather, it’s virtually all the members of the group who do a little bit of information processing, and what’s actually relevant is the aggregated result. A patient with a bad experience in a given hospital is statistical noise: mistakes can happen ; a lot of patients telling bad things about a given hospital probably means that this hospital is actually not that good: a lot of mistakes means something’s not going well.