Health Care Reform

Four leading experts on health care quality argue that competition to provide the best care will prove a more rational and cost-effective basis for health care reform than price competition alone. Further, they contend, the tools needed to help keep the reforms on track are all within reach. These include process-of-care and outcome measures, a method of comparing quality across health plans, guidelines for helping doctors decide when a given procedure or test is appropriate and necessary, and even, in the not-too-distant future, a consumer health handbook. Interviewed for the article were:

Robert H. Brook, director of the RAND Health Sciences Program, the nation's largest private program of health policy research. He is also a professor of medicine and health services at UCLA and director of the Clinical Scholars Program there.

Katherine L. Kahn, RAND senior staff member and associate professor of medicine at UCLA. Dr. Kahn has led recent studies on refining measurements of quality of care, the management and outcomes of childbirth, and the quality of nursing care.

Kenneth B. Wells, RAND senior staff member and professor of psychiatry, UCLA School of Medicine. His recent studies include the course of depression in adult outpatients, depression in patients with hypertension, diabetes or recent heart attacks, and the quality of hospital care for depressed elderly patients. An earlier study documenting the disabling effects of depression received nationwide attention.

Elizabeth McGlynn, who holds a doctoral degree in policy analysis from the RAND Graduate School, is a researcher in the Department of Social Policy at RAND. She is currently working with a consortium of 11 HMOs to develop measures of quality of care in competing health plans. Recent studies include the quality of inpatient care for elderly persons who have had a recent heart attack, the appropriateness of the use of hysterectomy, and the quality of prenatal care.

If the Clinton administration has its way, Americans will have greater choice among health plans in the future and a commensurately greater responsibility for choosing wisely. At present, however, the average person is better informed about--and thus better able to weigh--the complicated cost-benefit trade-offs involved in buying a new car than in making the most elementary decisions about health care. That is why a key provision of the President's health reform proposal calls for the development and broad public dissemination of information about the quality of care provided by the nation's hospitals and health professionals.

Consumers, of course, are not the only group with a vital stake in quality-of-care data. In a post-reform world, such information will be needed by businesses to guide their choice of employee health plans, by hospitals and physicians to improve their services, by health plan administrators to decide which doctors and hospitals to contract with, and by government regulators to monitor and evaluate the effects of the reforms.

But critics are skeptical of the notion of consumer guides to health care quality. They point out that current indicators of quality are crude and frequently misleading--hospital mortality rates, for example--and no systematic method exists for assessing and comparing quality across competing health plans.

"That's true," observes Dr. Robert H. Brook, head of RAND's health sciences program, "but we're much closer to achieving those goals than most people realize."

Brook, a veteran of RAND's massive, decade-spanning Health Insurance Experiment in the 1970s, says that much has changed in the 20-odd years since the last major overhaul of the health care system was being contemplated. "The government asked us to find out what impact national health insurance would have on medical costs and people's health, but no one had ever figured out how to measure those things. We had to invent the instruments as we went along."

One of the nation's top experts on quality-of-care issues, Brook has a reassuring message for health reformers in the Administration and Congress. "The research community has made enormous strides in quality assessment since then and is in a far better position to ensure that the reforms work as they are intended--whether the system that emerges is something close to the Clinton proposal or to one of the congressional alternatives."

Today, the shelves contain many proven research tools and techniques--instruments for determining when a given procedure is appropriate and necessary; measures of quality of care in hospitals; measures of patient functioning and well-being before, during and after treatment; and even a prototype method for comparing quality across different systems of health care. The remaining steps involve developing more and better instruments, and adapting, refining and simplifying existing ones for practical, day-to-day use. That task will take time and a financial investment that, while substantial, is small when placed in the context of total health spending.

Cutting Costs Not Quality

A nice example of the importance of quality measures in tracking the effects of reforms comes from a RAND evaluation of the prospective payment system, or PPS, a radical reform of Medicare reimbursement policies introduced in the early 1980s. Then, as now, the medical community feared that the government's determination to curb costs would end up hurting the quality of hospital care. But study director Katherine L. Kahn, a research physician, says that this is not an inevitable outcome, even when the cost-containment measures are stringent.

With PPS, Washington served notice that it would no longer simply pay whatever Medicare bills hospitals and doctors presented. Instead, providers would be paid a preset amount according to a complicated formula based on the patient's diagnosis.

This put tremendous pressure on hospitals to keep hospital stays short and raised concerns that elderly patients would suffer and perhaps die needlessly if they were sent home too soon. The new payment scheme dramatically reduced the number of hospital days charged to Medicare, Kahn's team found, and though some patients were being discharged before they were stable, the majority received good care and came to no harm as a consequence of the shorter hospitalizations. The researchers were also able to show that the adverse effects of the reforms were nearly offset by a long-term improvement in hospital care and to raise a warning flag concerning the premature discharge of unstable patients.

"We wouldn't have been able to do that if we had not first developed methods of measuring process of care. By process, I mean all the things that doctors and nurses do for patients on a daily basis--from ordering tests and procedures to prescribing and administering medication to monitoring blood pressure and taking temperatures," Kahn says.

The ability to measure quality of care will be key to keeping reforms on track.

Kahn acknowledges that developing process-of-care measures is an expensive proposition because of the time put in by researchers collecting and analyzing detailed data from thousands of individual medical records. But it also answers questions that other approaches cannot touch. "We can track the treatment given to a patient from the moment of admission to discharge and beyond. We know how sick the patient is at admission, what kind of services are provided throughout the hospitalization, the condition at discharge, and the long-term outcome."

From a research perspective, she insists, the approach is a bargain. "We end up with a multipurpose data set that can be used to address clinical problems, methods problems and policy problems."

Process measures represent a breakthrough in the field of quality assessment, points out Dr. Kenneth B. Wells, a psychiatrist who has led numerous RAND studies of mental health issues.

"Health researchers disagree about how to define and assess quality, with many insisting it's better and cheaper to measure `outcomes'--changes as a result of treatment, for example, in the patient's ability to perform routine tasks like climbing stairs, shopping, going to work, and so on. Kahn's team and others at RAND have demonstrated that it is also possible to develop valid measures of process of care for a wide range of conditions and diseases--hip fracture, stroke, heart attack, pneumonia, congestive heart failure. These measures are better at detecting differences in quality than are outcome measures. Even the treatment of mental illness, which is universally regarded as the softest of sciences, can be measured this way."

No "One Right Way"

Brook agrees that process measures have opened new doors for research and that RAND work is helping to dispel many myths about the definition and measurement of quality, but he does not want to see the research community get bogged down in technical wrangles over which techniques are better. He points out that Wells employed outcome measures in his own pathbreaking study of serious depression, the first to demonstrate the heavy burden of illness borne by people with that condition. "It's not a matter of there being one right way to measure quality. A number of promising approaches are being pursued and a good deal of synergy has been created. Researchers here and at other institutions are using what they learn from each other to improve their own tools and methods."

Health policy analyst Elizabeth McGlynn concurs, adding that a balanced approach to quality assessment should include a mix of process and outcome measurements. "Both have strengths and weaknesses. Outcome measures have the advantage of allowing for a `bottom line' assessment of how well hospitals, doctors and health plans are doing, but it can take a long time to track important outcomes--five years for breast cancer survival rates, for example. Process measures give quicker answers--an important advantage in treatment--and can identify instances of poor care with fewer cases than are needed for outcome measures. On the other hand, these rather technical measures of differences among health plans may have little meaning for the patient unless they can be shown to predict differences in health or quality of life."

In work with a consortium of health maintenance organizations (HMOs), McGlynn and her colleagues have shown how evaluating process of care can complement the current focus on outcomes measurement. In a study of prenatal care in six managed care plans, the researchers found no differences among the plans in adverse pregnancy outcomes--such as low-birthweight babies. One reason they did not was that adverse outcomes are relatively rare events and large sample sizes are required to detect clinically meaningful differences among health plans--for example, 1,800 births per plan are needed to conclude that a two-percentage-point difference in low-birthweight rates is statistically significant.

Needed: A way of comparing quality across competing health plans.

But when they investigated whether there were differences in specific processes of care that were known to improve a range of outcomes, they found real differences in quality among the plans. Criteria the researchers examined included compliance with the use of routine screening tests, the provision of other routine care to all patients, and the provision of care for specific problems or complications of pregnancy. On the use of routine screening tests, the best HMO had an average compliance rate of 92.6 percent compared with 63.9 percent in the worst plan. These differences in process indicate that individuals enrolled in plans with lower compliance are at higher risk for poor outcomes. The research team was able to determine this by reviewing 100 charts per plan rather than the 1,800 that would have been needed for an outcomes study.

McGlynn and her colleagues are also working to develop a means of evaluating quality across competing health care systems and plans--one that would allow fair comparisons between HMOs and private, fee-for-service physicians, for example, or between expensive plans and lower-cost ones.

Another important quality issue in health reform turns on the question of what constitutes appropriate care, and how doctors can be sure that they are providing it. As early as the mid-1980s, RAND researchers found inexplicably large geographic variations in the use of many common medical and surgical procedures. For example, coronary artery bypass graft was performed three times more often in one part of the country than another, hiatus hernia repair six times, and one type of treatment for hemorrhoids 26 times more frequently.

In an effort to understand what accounted for the differences, the research team selected three procedures--carotid endarterectomy, the surgical removal of plaque from the main arteries supplying the brain; coronary angiography, an X-ray and dye technique for observing arteries inside the heart; and gastrointestinal endoscopy, an examination of the digestive organs via a fiber-optic tube. They found significant overuse of all three procedures, ranging from 17 percent of angiographies to 33 percent of endarterectomies.

Reasoning that reducing inappropriate care would not only protect patients from unnecessary risk but save the system a great deal of money as well, the researchers next turned their attention to the painstaking task of developing "appropriateness" criteria for a dozen of the costliest and most-used procedures. These criteria, when translated into "practice guidelines" accessible through interactive computer programs, ultimately could be used by doctors to determine the clinical conditions under which a given procedure is appropriate and/or necessary, inappropriate or of doubtful value for a particular patient.

The method RAND pioneered to achieve the appropriateness ratings is a scientifically exacting process involving a summary of the scientific evidence in the literature concerning the procedure's effectiveness and risks; a structured set of all indications (there are typically several hundred) for which the procedure has ever been performed or considered; and finally, ratings by a multispecialty panel of physicians on the appropriateness of performing the procedure for given indications.

Brook estimates that appropriateness guidelines could be developed within three years for the 75 to 100 most-used procedures at a cost of about $150 million a year.

Developing tools for measuring quality at the level of the health plan, the hospital and the individual physician--information that could be used to produce a consumer health handbook--will require an additional investment on the order of $50 million a year, according to Brook.

The work done at RAND suggests that the government can get more for the money it puts into research on quality issues by focusing on those areas of medicine in which the greatest good or harm can be done. Candidates include coronary artery bypass surgery and the care of patients suffering heart attack or heart failure--areas in which quality of care makes a profound difference. Studies have shown that outcomes in such cases depend heavily on who performs the procedure or in which hospital the patient is cared for. Five extra deaths per 100 people operated on for coronary artery disease is not an unexpected finding.

Likewise, in a national study RAND found that for heart attack patients over 65, the death rate within 30 days of hospital admission is 24 percent if admitted to a hospital in the top quarter of the quality distribution compared with a death rate of 30 percent for a hospital in the bottom quarter.

This suggests that perhaps one-quarter of hospital deaths from pneumonia, heart attack, stroke, or heart failure might be prevented if quality of care in American hospitals was better.

If one clear credo has emerged over the years of studying health policy issues at RAND, says Brook, it is that "competition to provide the best care will prove a more rational and cost-effective basis for reform than price competition alone." Thus, he applauds the attention to quality contained in the Administration's plan and in the other health reform plans that have been advanced.

"President Clinton deserves tremendous credit for galvanizing the country to tackle health care reform," Brook says. "He's given us a wonderful opportunity to fix the system--and this time, we have the tools to get it right."

References

The Effects of the DRG-Based Prospective Payment System on Quality of Care for Hospitalized Medicare Patients: Executive Summary, Katherine L. Kahn, David Draper, Emmett B. Keeler, et al., RAND/R-3930-HCFA, 1991.

The Functioning and Well-Being of Depressed Patients: Results from the Medical Outcomes Study, Kenneth B. Wells, Anita Stewart, Ron D. Hays, Audrey Burnam, William H. Rogers, et al., RAND/N-3037-RWJ/HJK/PMT, August 1989.

"Quality Measures for Prenatal Care," Paul J. Murata, Elizabeth A. McGlynn, et al., Archives of Family Medicine, January 1994, Vol. 3, pp. 42-49.

"Does Inappropriate Use Explain Geographic Variations in the Use of Health Care Services? A Study of Three Procedures," Mark R. Chassin, Jacqueline Kosecoff, R. E. Park, et al., The Journal of the American Medical Association, Vol. 258, January 13, 1987, pp. 2533-3537.

"Quality of Care: Do We Care?" Robert H. Brook, Annals of Internal Medicine, Vol. 115, 1991, pp. 486-490.

"Maintaining Hospital Quality: The Need for International Cooperation," Robert H. Brook, The Journal of the American Medical Association, Vol. 270, 1993, pp. 985-987.

This article originally appeared in the Winter 1993-1994 issue of the RAND Research Review.

To search for a RAND publication by author, title, or keyword, click here.

RAND Review Home Page