Robin Hanson on Health Care
He sent this email to me and to others:
Imagine someone claimed that casinos produce, not just entertainment, but also money. I would reply that while some people have indeed walked away from casinos with more money than they arrived with, it is very rare for anyone to be able to reasonably expect this result. There may well be a few such people, but there are severe barriers to creating regular social practices wherein large groups of people can reasonably expect to make money from casinos. We have data suggesting such barriers exist, and we have reasonable theories of what could cause such barriers.
Regarding medicine (the stuff doctors do), my claims are similar…
1. We have many multivariate regressions on health that include measures of medical consumption or spending, and these consistently give zero coefficients. Thus when people choose more medicine as a result of variations in price, local practice, wealth, doctor concentration, location, or the other common causes of medicine variations, they seem to get equal measures of helpful and harmful medicine.
2. If medicine is on average helping health (relative to the counterfactual of no medicine), it must be that when these common causes of medical variation induce people to choose less medicine, people selectively eliminate the more harmful medicine, relative to the helpful medicine.
3. Our best single datum on the health effect of medicine is the RAND experiment, which confirms a zero effect. This experiment also strongly suggests that any selective elimination of harmful medicine when prices rise is not mediated by doctor judgements of severity of diagnosis or medical appropriateness. So selective patient choice seems the most plausible explanation for such an effect, if it exists.
4. The model that says doctors read the literature on controlled clinical trials and then use those insights to on average selectively choose helpful over harmful medicine fits poorly with the above evidence.
5. While it may be possible for an honest thoughtful person to carefully read the medical literature and produce a better than random distinction between harmful and helpful medicine, we have yet to observe social institutions of any substantial size that rely on such judgements to selectively choose helpful over harmful medicine when choosing medical variation for real patients. Many social institutions claim to be successfully making such selective judgements, but we have at best only weak evidence that any actually succeed.
6. We can see plausible social mechanisms that could explain the difficulty of creating such institutions. For example, fear of death and showing-that-you-care incentives might induce high levels of unwarranted trust in doctors, whose overconfidence and financial interests could consistently lead them to reject outside quality judgements.
7. If we were to tax medicine instead of subsidizing it, it is pretty clear that we would not be less healthy due to getting less medicine now. This policy might, however, induce less medicine to be developed for future use, which would harm us if such medicine is on average helpful, though we have at best only weak evidence for this claim.
I disagree with Hanson’s interpretation of finding 1. Instead, I carry around in my head a model of health care having big benefits and big waste.
I use the made-up example of a $100,000 procedure that saves the life an infant, combined with $7 million of wasted health care procedures on others. The benefits of saving the infant’s life outweigh the costs of all the waste, but you would not necessarily find in a cross-section regression a relationship between more spending and better outcomes. Thus, this hypothetical example is consistent with both David Cutler’s claims of huge benefits of medical care over time and the results that Hanson cites of estimates of benefits in cross-section that are close to zero.
Another way to put this is that I do not take it as literally true that the average benefit of medicine is zero. I do not think that harmful medicine is equal in magnitude to beneficial medicine. Instead, I think that the average benefit is positive but not large relative to cost. There is some beneficial medicine, somewhat less harmful medicine, and a whole lot of wasted medicine.
A lot of the waste is difficult to eliminate, because of statistical issues. A test for cancer may have only a 3 percent chance of extending someone’s life (because they may not have cancer or you may not be able to cure it), but it still may be cost effective in an expected value sense, or in some risk-adjusted sense. Still, 97 percent of the time, ex post it is a waste.
I agree that we have not developed institutions to measure the effectiveness of medical protocols more accurately. But as I just pointed out, this preference for ignorance is not unique to health care.
I do not think we have to go so far as to tax health care to improve the incentive to consume it wisely. Simply removing the insulation that we provide via third-party payments would be a start.