The Science of Health Care Reform

Another health policy window has opened; through it will stream proposals to reform the US health care system. President Obama has demanded that reform proposals improve both coverage and quality of care and make health care more affordable for all Americans. Extending coverage without worrying about costs would be relatively easy. Improving quality of care without worrying about costs might also be achievable. But extending coverage and improving quality while also making coverage more affordable will be difficult.

The first step in pursuing the President's goals is to review what science has revealed about the system. Two studies could provide the context for health care reform. The first, the RAND Health Insurance Experiment (HIE),¹ was conducted more than 3 decades ago and would cost about 1 billion dollars to replicate today. The second, actually a combination of studies conducted mainly in the United States, is partially represented in the Dartmouth Atlas.²

The RAND HIE was a population- and community-based, controlled experiment in which families from 6 sites across the country were randomized to 1 of 5 health insurance plans. There were 4 fee-for-service plans with different levels of cost sharing. Some families were randomized to free care in a health maintenance organization; their experiences were comparable with those of individuals in the fee-for-service system. The HIE's conclusions were straightforward: (1) increased cost sharing proportionally decreased health care use¹ and (2) on average, individuals with free care used about one-third more care than those in cost-sharing plans, but at the end of 5 years, they were no healthier on average than their cost-sharing counterparts.³

The Dartmouth Atlas is the outcome of almost 3 decades of work examining use of health services in various geographic areas at a population level. The conclusions are straightforward: After controlling for demographic differences, health care use varies dramatically across both major geographic regions (as large as states) and smaller regions (such as hospital service areas), and at a population level, these large variations in health care do not translate into health differences.⁴ Individuals living in regions of the country that use twice as much health care as other regions are not healthier.

The findings of these population-based studies seem to support policies that reduce service use in most geographic areas and increase what patients pay for care. Such policies would not affect population-based health outcomes. However, from the perspective of an individual patient, the story is quite different. In the HIE, the reason more care did not improve health is that providing more care did not improve the quality of care individuals received.¹ Furthermore, when patients had to pay for their own care, they reduced use of effective services in equal proportion to use of ineffective services.¹ These findings have been substantiated in more recent work.⁵

What does the individual patient perspective reveal about the Dartmouth Atlas findings? To answer this question, a series of studies in the 1990s examined the appropriateness of care in regions of the United States and the United Kingdom that had substantially different rates of overall use of health care services.^{6 - 8} The studies found that perhaps one-third of common medical and surgical procedures are either equivocal (benefit and risk to the patient are about equal) or inappropriate (the procedure will produce more harm than benefit to that patient). Although this finding is disturbing, the relationship of appropriateness assessed at the individual patient level to health care services use in a given area is far more disquieting.

For instance, the use of coronary artery bypass graft (CABG) surgery in the 1980s in the Trent region of the United Kingdom was, on a population basis, one-seventh the use rate in southern California, where research had established that a substantial proportion of the procedures were performed for equivocal or inappropriate reasons. In the United Kingdom, where there was a National Health Service, regionalization, a small number of surgeons and cardiologists performing large volumes of procedures, and a use rate that was almost one-seventh that of southern California, one might assume that all CABG surgeries would have been done for medically appropriate reasons. I attended cardiac case conferences in a major academic hospital in the Trent region and observed patients with severe left main coronary artery disease being placed on long waiting lists. However, a medical record review of patients who had undergone CABG surgery revealed that in about half the cases, the surgery was not appropriate.⁸ Because these findings were so disquieting, the chief of cardiology at one of the hospitals individually reviewed every patient record and substantiated the findings.

On a population level, financial or supply constraints can be applied to control use, but some individuals will be harmed and some will benefit. As use rates in a geographic area increase, appropriateness remains about the same—some individuals will be harmed and some will benefit.⁹

Other studies have demonstrated the disconnect between variation in service use or policy changes on one hand and quality of care on the other. The only comprehensive national study of health care quality in the United States found that quality did not vary across geographic areas in which use of services varied dramatically.^{10 - 11} In the HIE, quality of care was no better for individuals enrolled in the free plan than for those in cost-sharing plans. The effect of Medicare's Prospective Payment System on quality of hospital care was generally a wash, even though the length of hospital stay decreased.¹² Care was somewhat better in the hospital, but some patients were discharged sicker and quicker, and those patients did not do well.¹³

How can physicians change the health care system in ways that both are sensitive to the needs of individual patients and reflect population-level data? Some suggestions follow.

First, however health care is reformed, the resulting system must explicitly assess the appropriateness of any major medical or surgical procedure before it is performed in a specific patient.

Second, the assessment of appropriateness must be based on reliable information. For example, previous work has shown that 40% of patients who had CABG surgery in New York State for left main coronary artery disease did not have the disease when coronary angiograms were read correctly.¹⁴ Similarly, many appropriate surgery candidates were sent home without being offered CABG surgery because the angiograms were also inappropriately read.¹⁴

Third, the problems identified by the HIE and the Dartmouth Atlas need to be addressed by eliminating unnecessary care and wasted resources. Informal discussions with various specialists about the proportion of care they provide that does not meet their own definition of “necessary” suggests an amount ranging upward from 20%. If this is true, one way the President can begin to achieve his goals—without supply constraints or increased patient cost sharing—is by eliminating unnecessary care.

Fourth, it appears that simple interventions involving common clinical encounters may translate into large savings. For instance, most physicians order certain tests once a month, see patients once a year, and draw blood in the morning hours. What if physicians added 15% to a monthly or yearly interval and extended the frequency with which procedures are performed by 1 month or 1 year? This might produce a 1-time expenditure reduction that would help relieve some of the immediate pressure on the issue of affordability of health care. Perhaps in exchange for such changes, physicians could insist that health premiums be temporarily frozen for the population as a whole.

Fifth, perhaps it is time to address the affordability question head-on and insist that research about health care delivery focus on eliminating unnecessary care and wasted resources. There has been a substantial push to stimulate studies of comparative effectiveness. What criteria will be used to determine how federal money should be spent? What about a study only being conducted with taxpayer dollars on a new drug, device, or approach if it will replace something that is more expensive, guaranteeing up-front that what will be studied is less expensive? The study would have to determine whether what can be done less expensively will improve health or, at least, not harm patients.

What if industry used its own resources to fund independent evaluations of new drugs or devices that will be more expensive, regardless of their potential health effect? Changing the rules by which publicly funded comparative effectiveness studies are conducted might motivate the health care industry to refocus research and development on identifying drugs, devices, and tests that are better and less expensive rather than better and substantially more expensive.

Health care professionals need to help the President achieve the goals he has articulated. The goals cannot be achieved by controlling supply or increasing patient cost sharing. Such methods are too blunt when applied at an individual level: they will benefit some but harm others. It is time for physicians to commit as a profession to helping the President and Congress achieve the vision of a new health care system by improving the way medicine is practiced.