Stephen C. Schoenbaum
We can greatly improve the performance of the U.S. healthcare system, but first we need to reexamine some of our long-held notions.
At a Glance
Five keys to better performance in the U.S. healthcare system are:
- Coverage of the uninsured, reduction in cost shifting, and administrative simplification
- More primary care and less specialty care
- Interoperable health IT
- Mass customizationof care
- Delivery of near perfect care
We are beginning to recognize that the U.S. health system has serious performance problems. It is possible to greatly improve our system's performance, despite some obstacles. In June 2005, The Commonwealth Fund announced the formation of a Commission on a High Performance Health System. The commission soon will begin to publish an annual scorecard on the performance of the U.S. healthcare system, set goals and targets for performance over the next several years, and ultimately recommend policies that will help our country improve its system and performance. But this work is at an early stage. This article reflects my personal opinion based on 40 years of experience as a physician and healthcare organization manager.
Current Performance of the U.S. Healthcare System
The United States has the costliest healthcare system in the world. We spend about 16 percent of our gross domestic product on health care, compared with 11 percent for the next closest country, Germany. Our expenditures per capita adjusted for cost of living were about $5,300 in 2002, compared with $2,900 for the next closest country, Canada, and our expenditures are increasing at a faster rate than those of other countries. Even so, 46 million Americans have no insurance coverage and have poorer access to and quality of care. The data are clear. The uninsured are predominantly adults (four out of five) and low income (two out of three). Uninsurance is a growing problem, and the rate of growth is greatest in the U.S. middle class. Furthermore, the quality of our insurance coverage is decreasing. More Americans are becoming underinsured, leading to an increasing problem of medical debt and also to healthcare quality problems similar to those experienced by the uninsured.
If those problems were not enough, the quality of care delivered to those who have access to it and even excellent insurance coverage has great deficiencies-what the Institute of Medicine has called a "quality chasm." RAND health services researcher Elizabeth McGlynn has shown that Americans receive only an average of 55 percent of the care recommended by experts. The figures vary by condition and were under 25 percent for patients with hip fractures. In addition, 4 percent of hospitalized Americans sustain adverse events related to their care, many of which are preventable.
About 1 percent of hospitalized Americans are victims of negligent errors or malpractice. Only 1 percent to 2 percent of these people actually sue for the malpractice they have experienced, but that is enough to have created a "malpractice crisis"-really a malpractice insurance premium crisis.
Modern health care is complex. Most patients with significant health problems see multiple providers. The transfer of information among these providers is relatively poor. In a recent Commonwealth Fund survey, 23 percent of "sicker Americans" reported that their test results or records were not available at the time of an appointment. Patients often receive conflicting advice from physicians, and often feel uninformed about important elements of their care. One in six Americans report significant communication problems with physicians; and the number is greater for minorities, especially persons for whom English is not a first language, persons with low incomes, and persons with less than a college education.
In the late 1980s and early 1990s, Americans were concerned about the costs of care. Recent concerns are not just about cost but also the efficiency of care-whether we are getting value for our money. This concern shows up in simple ways, such as duplication of tests. It also shows up when quality of care by hospital, physician, or state is compared with costs of care. Several studies have produced scatter-plots that show little, if any, relationship between quality and cost. The good news is that there are some high-quality, low-cost providers that we can study to learn how they achieve their results.
Addressing Problems and Finding Solutions
One often hears the story of the person searching for keys under the lamppost when, in fact, they are in the shadows. The problem with most of the keys for improving health care is not that we are looking in the wrong place but rather that we don't recognize a key when we see it. Keys to U.S. healthcare system performance improvement that we haven't sufficiently recognized include coverage of the uninsured and administrative simplification, more primary care and relatively less specialty care, interoperable health IT, mass customization of care, and delivery of near-perfect care.
Coverage of the uninsured, reduction of cost shifting, and administrative simplification. Various estimates indicate that covering the uninsured would cost about 5 percent more than we are currently spending, or about $80 billion to $90 billion a year. Since 15 percent of the population is uninsured, you might wonder why covering them wouldn't cost 15 percent more. One reason is that we already have a complex process of cost shifting to provide some services to the uninsured and underinsured, a process that adds administrative complexity.
Administrative costs are the most rapidly growing component, now over 10 percent, of total healthcare cost. Commercial insurers, responding to employer requests, have many different insurance products, each with different eligibility, benefits packages, deductibles, and copays. When a service is about to be rendered by a healthcare provider or vendor, eligibility needs to be assessed, and when a claim is submitted, it has to be adjudicated against the requirements of the insurance product for the particular beneficiary as well as against the contractual agreements with the specific vendor.
There is an interrelationship among the uninsured, cost shifting, and administrative cost. If everyone in the United States had coverage, if each payer was required to pay its fair share of cost adjusted for the case mix of persons covered, and if there was a limitation on the number of benefit packages that could be offered, we could save much of the money needed to cover the uninsured. The key is universal coverage; without it, we don't have a prayer of eliminating cost-shifting and making significant-enough reductions in administrative complexity.
I have not suggested we need a single payer or a government-run delivery system. Almost all developed countries have universal coverage, the United States being the great exception.
Some other countries have a single payer, and a few have a government-run delivery system (e.g., the United Kingdom). Others have a single payer and a largely privately run delivery system (e.g., Canada, which has a single payer in each province). Yet others have multiple payers and largely privately run delivery systems (e.g., France and Germany). Massachusetts is about to experiment with a private, multipayer approach that could provide a model for the United States.
Universal coverage also would offer our society benefits, other than direct savings in healthcare expenditures: One important attribute of a healthcare system is its ability to help people live long and healthy lives. Deaths that might have been prevented by better health care run about 10 percent lower in countries that have universal coverage, whether or not they have a single payer, than in the United States.
Primary care versus specialty care. Is more specialty care better care? Obviously both primary and specialty care are important. No developed country has only one. Nonetheless, the body of evidence that has been compiled by experts on primary care versus specialty care-such as by Barbara Starfield, MD, at Johns Hopkins Bloomberg School of Public Health-overwhelmingly indicates that countries with better developed primary care systems have better health outcomes and lower costs. In this country, the work of Elliott Fisher, MD, and John Wennberg, MD, at Dartmouth Medical School, has shown that areas in which there is a greater supply of physicians-which really means specialist physicians, given our preponderance of specialists-have up to 30 percent greater costs with no evidence of better outcomes.
In our healthcare system, specialists use subspecialists, who use sub-subspecialists. The care keeps getting more complex, and more room for error is introduced. Although some have been worried about an impending shortage of specialists in the United States, we truly have an increasing shortage of primary care physicians.
A key to improving our system is developing more adult primary care capacity. That will require improving the lifestyle and relative compensation of primary care physicians for adults. The general principle, as pointed out by Allan Goroll, MD, a former president of the Massachusetts Medical Society and general internist, is that primary care physicians would receive more money for providing care to a smaller number of patients, but, in turn, the physician and his/her staff would be expected to meet or exceed a variety of publicly reported performance requirements. Overall, given the evidence from other countries, better relative compensation of primary care physicians likely would lead to less complex care, better outcomes, and lower costs. This is a testable hypothesis, but it isn't being tested.
Interoperable health IT. Health care is starting to be "wired," but we're a long way from realizing the potential of a computerized infrastructure for improving health care. Information transfer is a basic building block of all healthcare delivery, an essential ingredient of effective referrals between physicians and care delivery settings. Excellent care requires integration of information from the patient, pharmaceutical use, laboratory test and procedure results, and the expert opinions of various professionals.
Although the use of computers for supporting clinical care processes is accelerating, it is happening most rapidly among large physician groups, and most physicians do not practice in large groups. It is essential to create incentives for small groups and solo-practicing physicians to purchase and use computers. A study by Robert Miller, a health economist at the University of California San Francisco, has shown that most small physician practices can achieve a substantial ROI by adopting electronic medical records just from an improved ability to bill insurers more accurately for services the physician has rendered, decreased office expenses relating to maintaining paper records, and improved practice throughput. Even if most physicians agreed with this analysis and purchased electronic medical records systems, we would have taken only a first step to achieving the needed integration of information. For this integration to occur, not only do all physicians, hospitals, and vendors have to use computers in their work, but also these computers must be able to communicate with each other. They must be interoperable.
Although it is possible for large healthcare organizations, e.g., Kaiser Permanente, to achieve interoperability within their organization, most medical care in this country is not delivered by such organizations. It will take a major investment to achieve interoperability in the overall healthcare system-estimated by Rainu Kaushal, MD, and her associates to be on the order of $150 billion initially with an annual maintenance cost of $50 billion. This is an enormous sum, but still a small fraction of our total healthcare expenditure of $1.9 trillion as of 2004.
It is hard to prove in advance that investing in interoperability will pay for itself. The concept of interoperable health IT is analogous to developing a national highway system. The arguments for developing our system of interstate highways weren't "proven," but "made sense" as a way of addressing the limitations of the local and more limited federal highways that had been built earlier. Ultimately, a huge investment was made. Thus far, despite much hoopla in Washington, D.C., about the importance of health IT, the federal government is discussing investing only about $100 million to $200 million per year in the area. This amount will not buy interoperability and will not lead to the type of information transfer that will ensure better, more efficient health care.
Mass customization. For many years, we have crafted care individually for each patient. This approach, based on the idea that each patient is different, is faulty. Care that is important for most patients with a specific condition, e.g., diabetes, is not being delivered, and sometimes unnecessary, even harmful, care is delivered. One way to deliver better care is to work with groups of patients who share some common characteristics, not just individuals. These groups may be patients with a medical condition in common, similar demographic characteristics, and need for a similar service, such as a type of medical visit.
Approaches to managing groups of patients can be considered forms of "mass customization" of medical care delivery, and examples are beginning to emerge. "Advanced access" is a type of patient-visit scheduling system that provides a patient with a guaranteed appointment on the same day the patient asks for it. When accompanied by reshaping the visit around the patient, many unnecessary steps and waits can be eliminated. The Primary Care Development Corporation in New York has helped a clinic implement same-day appointments and reduce total visit time from 148 minutes to 50 minutes. This clinic serves a low-income population. Persons who come for appointments no longer need to miss as much work. In addition, the clinic has benefited from a marked decrease in the broken-appointment rate and the ability to increase its throughput. A win-win.
Disease management is one type of mass customization that involves implementation of evidence-based guidelines. Another type is use of decision-support software in a physician's office. Decision support integrates clinical information entered into the patient's electronic medical record with evidence about what to do for similar patients and provides the physician with just-in-time reminders and prompts.
It all seems so obvious, but relatively little has been done on these approaches. Why? First, we have no national or coordinated process for generating clinical guidelines, and more often than not, the development of a guideline, when attempted by an organization such as a medical specialty society, points up our lack of evidence for critical aspects of care. This, in turn, reflects our underinvestment in the practical research that would provide the evidence. Needed evidence can come prospectively, from clinical trials, or retrospectively. We are all aware that sometimes problems show up long after new treatments are being used in large groups of patients, many of whom differ from the people in the original clinical trial. There is a connection between having interoperable health IT and providing the basic evidence that can be a building block for mass customization approaches to improving care. By collecting information on patients' underlying conditions, treatments, and outcomes, we could greatly extend our evidence about how to manage patients most effectively and efficiently. This would require big computerized databases, which are not available now except in a few large healthcare delivery organizations.
Perfect care. Many persons in healthcare delivery seem to hold the notion that care is so complex that complications and errors are to be expected-in short, that medical care is unavoidably dangerous. But "the perfect" is not always the enemy of "the good." There are enough examples in which near-perfect or much, much better care has been achieved to allow complacency about our current rates of error and adverse events.
About 20 years ago at Harvard Community Health Plan, Donald Berwick, MD, then our vice president for quality of care measurement, found that only 70 percent of women who had a positive screening test for cancer of the cervix, a Pap smear, had a follow-up in a six-month period.
We had an automated medical record (going back to 1971, before personal computers), and my staff were able to program the system so that when a positive test result was entered from our electronic laboratory system, it started an electronic tickler and prompting system that got turned off only when a follow-up visit or test occurred. Within less than a year, more than 99 percent of women who had a positive Pap smear had had a follow-up within six months. We then got agreement among our gynecologists about the appropriate follow-up for different types of abnormal Pap smears, and about a year later, we programmed our laboratory system so that when the original abnormal Pap smear result was reported to the physician, the report also told the physician what the most appropriate next step was. Thus, we were able to mass-customize, achieve near- perfect care, and avoid potential malpractice suits.
Also in the mid-1980s, anesthesiologists led by Ellison Pierce, MD, who subsequently founded the Anesthesia Patient Safety Foundation, implemented a set of guidelines that markedly reduced anesthesia-related deaths to less than 3 per million exposures. Anesthesiologists were motivated to implement these guidelines by their high malpractice premiums in the 1980s. Not surprisingly, premiums for anesthesiologists are now much less than for many other specialists.
In the past few years, Richard Shannon, MD, at Allegheny General Hospital, has shown that we do not need to accept the national average rate of central-line-associated bloodstream infections. By examining the factors associated with the occurrence of these infections in his hospital and implementing a set of countermeasures deriving from those factors, he reduced the occurrence of CLABs in his hospital by 90 percent in 90 days and has shown that the hospital saves money for each infection prevented.
These examples indicate not only that we can provide much more effective care, but also that there are likely to be many instances where more effective care for the patient also yields substantial benefits for the provider.
A High-Performance Health System
Some steps that can lead to better performance in the U.S. health system are already being taken. For example, increasing numbers of measures of performance are being developed, which is how we now know several aspects of our absolute and relative performance. Some of these measures are being tied to public reporting, pay for performance, or both. We no longer think it is enough for a physician to have completed training and be "set" for a lifetime career in medicine. Specialty boards have been developing processes to ensure that physicians maintain their certification in a specialty, and others are beginning to talk about maintenance of licensure, as well. There also is increasing, if uneven, computerization of clinical care in physicians' offices and in hospitals.
Nonetheless, some major issues stand in our way, primarily because we have fixed opinions and have not opened ourselves up to change. To the extent that we feel that it is OK for 16 percent of our population not to have coverage, we underemphasize the importance of primary care, we do not interconnect our computers and information, we continue to think that health care is produced for one patient at a time rather than a population, and we believe error and adverse events are bound to happen, we will fall far short of high performance. Much of what it will take to improve our system requires federal leadership as well as an insistence on high performance in every healthcare delivery organization. That will not happen until we open ourselves up to reexamining some of our long-held notions and to change. If you haven't started that process, I hope you will do so now.
Stephen C. Schoenbaum, MD, is executive vice president for programs at The Commonwealth Fund and executive director of its Commission on a High Performance Health System, New York.
Publication Date: Saturday, July 01, 2006