What Big Data Can't Tell Us About Health Care : The New Yorker
Jean Malouin, a family doctor in Michigan, woke up one morning earlier this month to an e-mail from a Washington Post reporter, who informed her that a vast release of Medicare payment data from 2012 had identified her as the highest female biller in the country, and the seventeenth-highest over all, with more than seven million dollars in payments. Malouin was soon inundated by interview requests from reporters, and her name appeared in newspapers across the country.
Malouin is a family doctor, which is not a specialty that one typically enters hoping to get rich. Delivering primary care is seen by doctors as hard work that earns comparatively little pay, and it is a job that is only getting harder. That’s because the Affordable Care Act, with the broad ambition of containing costs while improving quality, hopes to move away from a fee-for-service model, toward one in which doctors are paid primarily for keeping their patients healthy, a responsibility that will fall largely on primary-care doctors. At this point, nobody quite knows how to make this vision a reality, but Medicare has funded various demonstration projects to test innovations in care—one of which is led by Malouin, who supervises three hundred and eighty primary-care practices that treat a million patients in Michigan. Payments for care improvement from Medicare at all these clinics are made under Malouin’s name, which is how she ended up in dozens of newspaper reports on the data dump.
Even doctors who didn’t end up making headlines like Malouin told me that they felt somewhat exposed by the release of the Medicare payments data. As one friend tweeted, “Imagine if you woke up one morning to find that every person in your profession had their income reported on the New York Times web site.” For nearly thirty-five years, the American Medical Association had worked to keep this information private, after securing a federal court injunction in 1979. Dow Jones, the parent company of the Wall Street Journal, waged a legal battle against the injunction, which was overturned by a federal judge last year.
In the march toward greater price transparency in health care, the data release represents a milestone, though perhaps one more symbolic than substantive. For those who believe that greater price transparency is the key to reining in exorbitant costs and helping patients to become more savvy “health-care consumers,” the data release is a huge victory. Indeed, the early coverage, invariably emphasizing the high spending of a small group of physicians, had a tone of triumph. According to the Times, two per cent of physicians accounted for nearly a quarter of Medicare spending. Ophthalmologists led this small group of high billers, with a large portion of their payments apparently connected to the use of an expensive treatment for macular degeneration. Charts broke down payments by specialty, showing cancer doctors in the lead, while maps of the distribution in spending confirmed long-observed geographic variations. For instance, states like Florida, Texas, and New Jersey consume a large share of Medicare resources.
The calls for price transparency, as a means of bringing down health-care costs, have certainly gained momentum in the past year. In this latest chapter, the hope is that members of the public will be empowered by their access to payments data, and will use this information to identify doctors who are behaving badly, helping to end fraud and profit-driven overuse. In fact, Medicare already conducts internal audits, and the two highest billers in 2012, an ophthalmologist and a cardiologist, both from Florida, were already under federal review. But the third-highest biller, a pathologist, directs a diagnostic company that performs tests for twenty-six other pathologists, which are all billed under his name. Similarly, an oncologist from Newport Beach, California, who billed nine million dollars, explained that all the billing at his practice, which includes five physicians, was under his name, and much of it was directed toward expensive chemotherapy drugs. (One such drug, for advanced melanoma, called ipilimumab, costs about a hundred thousand dollars for four treatments.)
This release includes more than nine million rows of numbers, encompassing more than eight hundred and eighty thousand physicians and other health-care professionals who billed under Medicare Part B—which covers care delivered in an outpatient setting—in 2012. You see their names, their addresses, the services they billed for, how much Medicare reimbursed, and many other details. There are several caveats to interpreting the payment data. First, there is a difference between what Medicare pays and what doctors earn. All doctors face overhead costs: radiation oncologists, for instance, have to pay for technicians and expensive equipment. These doctors were among the highest billers, but eighty-two per cent of their Medicare reimbursements went to covering these expenses. An analysis by the Washington Post found that Medicare paid sixty-four billion dollars to doctors in 2012, of which forty-three per cent went to office overhead, forty-one per cent went to doctor compensation, and thirteen per cent went to drug costs. Even within the Medicare system, these data provide an incomplete picture, as the reimbursements do not include payments within hospital systems (which fall under Medicare Part A) or various Medicare Advantage plans, which cover many seniors but are not included in these numbers.
Despite these caveats, members of the public were encouraged to use these data to make more informed decisions about where to seek care. Indeed, in a press briefing on Wednesday, Jonathan Blum, an administrator at the Centers for Medicare & Medicaid Services, said, “We look forward to making this important, new information available so that consumers, Medicare and other payers can get the best value for their health-care dollar.” The suggestion that these data can allow you truly to comparison-shop, however, is misleading. These data do not tell us anything about the value of care. By definition, the value of health care cannot be measured in dollars spent; it’s about what you get for those dollars, and the Medicare data, however useful, offer little new information of that sort. These numbers tell us, for example, that dermatologists receive higher reimbursements than pediatricians, that cardiologists in Oregon get paid less than their counterparts in New York, and that performing procedures pays better than talking to patients. But they cannot tell us whether doctors provided good care, because being a good doctor sometimes means doing everything, and sometimes it means doing nothing at all.
Let’s say you find that your ophthalmologist performed fewer cataract surgeries than average. It could be because he lacks experience. But it also could be because most of his patients have Medicare Advantage plans that are not included in these data. What about your vascular surgeon, who billed Medicare more than a million dollars last year? He thinks that you need a stent to open a blocked artery in your leg. Do his high numbers indicate a tendency to perform unnecessary surgeries? Maybe. But it is just as likely that his apparently high billing numbers reflect the fact that he performs procedures in his office—covered under Medicare Part B—whereas most of his peers perform similar procedures in hospitals, where their payments aren’t included in these data. And your orthopedist, who performed nearly five hundred hip replacements last year? Surely his high volume suggests that he knows what he is doing? Of all the conclusions that you can make from these data, that high procedural volume signals that better quality is the one with the most empirical backing. But even that is just one signal amid a great deal of noise.
This is not to suggest that the information released earlier this month will not play a role in making it easier for individuals to determine where they can get good care. When I spoke to Jonathan Kolstad, a professor of health-care management at Wharton, he noted that medicine has lagged far behind other industries in giving consumers data to inform their decisions—often because privacy concerns raise significant barriers. This release is a partial step, Kolstad said, but the message “is not that we should be releasing less data. It’s that we should be releasing even more.”
Amitabh Chandra, a health economist at Harvard, noted that the release of these data may be most useful not to the public or health researchers but to private insurers. These firms keep their own data, but the Medicare dataset is far more vast than any one insurer’s figures. Insurers, Chandra said, may be able to mine these data to build smarter networks that exclude high-cost providers and include high-performing ones. This type of tiered networking, on a grand scale, could actually improve the efficiency of our delivery system. It is this version of transparency-driven tiering, Chandra believes, that could assist in our cost-containment efforts.
“I think that we all agree that we have to do something about bending the cost curve,” Chandra said. “But I think we have deluded ourselves into thinking that a challenge as big as bending the cost curve can be collapsed into something as simple as transparency.” He likened the notion that transparency alone could solve the problem to imagining that we could stop global warming by driving hybrids. “We are always drawn to these tantalizing simple arguments. Transparency is just one of the ways we are seeing such aspirational thinking in health care.”
The potential danger in this data dump does not come from the new information it provides but from the old story it risks reinforcing. The existing narrative of American health care goes something like this: greedy physicians perform procedures that patients don’t need and enrich themselves in the process, which is why a third of health-care spending goes to unnecessary care. Now that members of the public can see where their tax dollars are going, they are empowered to rid the system of duplicitous doctors. Indeed, as Blum, the Medicare administrator, emphasized to the press, “We know that there’s waste in the system. We know that there’s fraud in the system. We want the public’s help to review the physician-payment data and report suspected wrongdoing.”
This narrative has many problems. First, the data suggesting the extent of unnecessary care have come under widespread criticism. Economists have also pointed out that other factors, such as high administrative costs and expensive technology, play a greater role in exorbitant costs than overuse. But, even if eliminating waste would, by itself, cure our ailing health-care system, these data do not allow us to identify waste—because waste is not the same as spending a lot. Waste is not even the same as high spending that doesn’t make people healthier. Doctors do that all the time: not because we are trying to enrich ourselves but because we are trying to help our patients. What looks wasteful in retrospect may have looked like a live-saving intervention at the time it was made—and this is as true for expensive chemotherapies that fail to save a life as it is for expensive tests that don’t reveal disease.
That’s why asking the public to use this information to identify waste belies the complexity of physician decision-making. Do physicians respond to financial incentives? Yes. Should we tolerate care that offers patients no benefit? Absolutely not. But are profit motives the primary drivers of physician behavior? My own sense is that most physicians are primarily motivated by trying to do the right thing for their patients. Combing through these data, however, creates the impression that the pecuniary trumps the humane. What else can one conclude from information that only tells you how much physicians do and what they bill?
From an experimental standpoint, unpacking non-financial drivers of physician behavior is far harder than demonstrating that physicians respond to financial incentives. But Kolstad, the Wharton health economist, recently published a study suggesting that wanting to perform better was a far more powerful motivator than wanting to earn more. The paper, which was recently awarded the prestigious Arrow Award for the best study in health economics, examines the behavior of heart surgeons in Pennsylvania. Kolstad took advantage of a report-card system implemented in Pennsylvania in 2006, which created a financial incentive for surgeons to lower their mortality rates because of a need to attract patients. He compared this incentive to simply giving surgeons feedback on their performance and showing them how they compared to their peers. This feedback was four times more powerful in improving physician performance than the financial incentives.
While much more experimentation of this type is necessary in our quest to understand how to improve quality and cut costs, it is no wonder that we cling to the story that we have told. It offers heroes and villains, fosters the American ideal of the individual over any collective authority, and, above all, provides the hopeful illusion that we don’t have to confront the hardest questions that doctors and patients grapple with every day. How much are we willing to spend to save a life? Is a groundbreaking hepatitis C drug worth eighty-four thousand dollars? What about chemotherapy that costs a hundred thousand dollars and may only prolong a life by four months?How much do we value quality of life, or patient satisfaction, against cost? We know, for instance, that M.R.I.s for patients with back pain typically leave them no healthier, but they do leave them more satisfied. Is this a cost that the American taxpayer should bear?
These types of questions are burning beneath the surface of our superficial discussions about how to improve the value of health care, but the political will to address them, from both an empirical and ethical standpoint, is decidedly absent. Indeed, Medicare is prohibited from considering cost effectiveness in coverage decisions, and, even though the Affordable Care Act does emphasize the need to fund more research comparing the efficacy of various treatments, translating these findings into practice requires a collective willingness to consider costs in coverage. Because such discussions are often conflated with rationing, any attempt to do this is a political nonstarter. Perpetuating the attribution of high health-care costs to physician greed just makes addressing these critical questions more difficult.
There is nothing wrong with trying to improve the value of health care. But better value will depend as much on doing more of what’s good as it will upon doing less of what’s bad. So much of that good isn’t captured by these numbers. You don’t bill for talking to a patient about how he wants to die. There’s no code for providing reassurance rather than ordering a test. And, for all the talk about transforming our health-care system from one that treats illness to one that promotes health, no one pays you to talk to patients about how they might lead healthier lives.
Photograph by Skynesher/Vetta/Getty.
Source article:
What Big Data Can't Tell Us About Health Care : The New Yorker
