Use a Scalpel, Not a Meat Cleaver: Robin Hanson’s essay has some good features, some bad features, and some undeveloped themes. Let me pick up on each.
First the good. Hanson is correct that a lot of medical spending doesn’t add much value. Research by Elliott Fisher and colleagues at Dartmouth shows that Medicare spending varies by a factor of two across parts of the country, without comparable health benefits. Fisher et al. estimate that if the high spending areas were brought to the level of the lower spending areas (or more accurately, the 25th percentile of the distribution of spending), we could save 25 to 30 percent of Medicare spending. No one doubts that the same is true about non-Medicare spending as well.
On top of this excessive provision of medical care is administrative waste. Insurance administration is 14 percent of premiums in the United States, compared to less than 5 percent in countries with a single payer and perhaps 10 percent in countries with mixed public-private systems. A good share of this additional administrative expense is for screening groups to determine who is healthy and sick, and for marketing. It is not for computerization, practice reminders, or other valuable services. Thus, we could happily eliminate it. In addition, there are further administrative expenses in physicians’ offices and hospitals. Dealing with multiple insurers adds considerably to expense, not to mention headaches. Woolhandler and Himmelstein estimate that total administrative expenses are 31 percent of health care in the United States, compared to half that in Canada. Without a single payer-system we could not get to Canadian levels, but 10 percent savings is not unreasonable.
If one takes the 25 percent of care that needn’t be provided and 10 percent in unnecessary administrative expense, that’s 35 percent of the nation’s medical bill that could be eliminated without loss. Allow for further savings from information technology, reduced errors, investment in disease management, or generation of comparative effectiveness information, and the savings could approach 50 percent. The potential savings are as high as Hanson guesses.
As a sociological observation, I am surprised by Hanson’s argument that this hasn’t been much noted. The work of the Dartmouth team has received enormous media attention, including front page coverage in the New York Times, for example. My book suggests large possible savings as well. And the knowledge that non-medical factors are important for health has been amply documented in many contexts. If Hanson wants to add his agreement to this array of research, I’m all for it.
Medical care is currently 16 percent of GDP. In 1975, it was half that amount. Reading Hanson’s essay, one might conclude that we were better off in 1975 than today. This is wrong, and it gets to a key part of why Hanson’s analysis is too simplistic. The most important reason why medical costs increase over time is because we develop new ways of treating patients and provide that care to ever more people.
Consider the most expensive part of medical care: care for people with cardiovascular disease. In 1950, a person with a heart attack received bed rest and morphine (to dull the pain). That was how Dwight Eisenhower was treated when he had a heart attack in 1955. This therapy is not very expensive, but it is also not very effective. Today, such a person receives clot-busting drugs and other medications, and intensive interventions such as bypass surgery or angioplasty. These technologies are certainly costly. Spending in the few months after a heart attack is about $25,000 per patient. And yet the care provides enormous benefits. Mortality in the aftermath of a heart attack has fallen by three-quarters since the 1950s. The average person aged 45 will live 3 years longer than he used to solely because medical care for cardiovascular disease has improved. Virtually every study of medical innovation suggests that changes in the nature of medical care over time are clearly worth the cost.
If you don’t want to take this on faith, take a quick quiz: would you rather get today’s medical care at today’s prices, or be given $5,000 per year but only get medical care at the 1975 level — doctors trained at that level, hospitals with equipment from that level, and drugs available then? The $5,000 is the increase in real per person medical spending in the past three decades, so this is asking you to choose today’s care at today’s prices or 1975’s care at 1975’s prices. I have yet to encounter more than a handful of people who want the standard of care in 1975, even with the money back. My quasi-market test always confirms that the care is worth it, taken as a whole.
Reconciling this finding with the fact that there is a lot of waste is not hard conceptually. We provide a lot of valuable care and a lot of care with little value. Over time, the technology profile is dominated by the valuable care. But like the fat man at the buffet, we cannot stop ourselves from too many helpings.
The problem in medical care is how to separate the good from the bad. What can we do to maintain the services that are very effective but get rid of the waste? Hanson’s essay reads as if a meat cleaver is appropriate (“most any way to implement such a cut would likely have big gains…”). For a delicate operation, though, I prefer a scalpel.
Broadly speaking, there are two approaches to eliminating waste in medical care. The first is the demand-side approach: give patients more information about what is effective, raise the price that they pay for using care, and rely on informed choices to produce efficient outcomes. The alternative is the supply-side approach: invest in information technology, monitor what physicians do, and pay providers more for better care than for less good care.
As befits a Cato Institute argument, Hanson implicitly favors the demand-side approach. I suspect this is wrong. To spark discussion, let me state my sense of the literature fairly strongly (too strongly, perhaps): we have no evidence that consumers facing higher cost sharing will make the right medical care decisions. Indeed, the evidence suggests the opposite.
It is very clear that medical care consumers respond to prices of medical services. The Rand experiment shows this, as do countless experiments since then (for a review, see my chapter with Richard Zeckauser in the Handbook of Health Economics.) But consumers appear to cut back indiscriminately. Consider prescription medications. When cost sharing was raised in the Rand experiment, people stopped taking anti-hypertensive drugs. When firms raise the price of one medication in a class, some consumers switch to cheaper drugs in that class, but others stop taking the medication entirely (see here and here and here). Although this is not known with absolute certainly, the people who stop taking the medication do not appear to be those who benefit from it the least. Rather, price sensitivity seems to vary in the population, just as family history of cholesterol and hypertension do. It is not unlike saving for retirement. Some people save and some do not, and the two groups are not particularly associated with need.
In light of these findings, the demand-side approach is wrong for many, perhaps most of the population. Rather, I see no alternative to thinking clearly, systematically, and expansively about the potential role for supply-side policies to improve the delivery of medical care. Carefully targeted evaluations of what is done and how to pay for it better are the fundamental way that we can eliminate the waste in medical care but still retain the valuable core.
Comments