What can Europeans learn from Americans?

Abstract

In a wide-ranging look at many aspects of health care financing and delivery, the concepts of glasnost and perestroika are used as a framework for presenting ideas from the American system that may have value for European health care planners. These include more uniform approaches to data collection and cost reporting, patient outcome studies, evaluation of service and access standards, publication of information, quality assurance review, decentralization and independent institutions, prepaid group practice, demonstrations and experiments, and managed competition. Suggestions are offered for making health care systems on both sides of the Atlantic more manageable, efficient, and responsive.

Introduction

What can Europeans learn from Americans about the financing and organization of medical care? The obvious answer is “not much.” We Americans are spending nearly 12 percent, going on 15 percent, of gross national product (GNP) on health care, while most European countries are spending an apparently stabilized 6 to 9 percent (Division of National Cost Estimates, 1987; Francis, 1989; Schieber and Poullier, 1988). The western European democracies have achieved essentially universal coverage, but some 35 million Americans—17.5 percent of the population under 65 years of age—have no financial protection against medical expenses, public or private (Short, Monheit, and Beauregard, 1989). (Those who cannot pay may get free care from community or public hospitals, after they have paid what they can. This places an inequitable burden on the hospitals that care for the uninsured and motivates them to find ways of avoiding attracting patients who cannot pay, such as by closing emergency rooms.) Millions more have inadequate coverage that leaves them exposed to large risks or to exclusions for care of preexisting conditions. At the same time, our infant mortality rate is higher than that of most of the western European democracies, but life expectancy is about in the middle of the group. A recent public opinion poll found that only 10 percent of Americans agree with the statement “on the whole, the health care system works pretty well,” compared with 56 percent of Canadians and 27 percent of the British (Blendon and Taylor, 1989). So it would be, quite frankly, ridiculous for an American to suggest that we have achieved a satisfactory system that our European friends would be wise to emulate.

Americans like to believe that we have the world's best medical care—at least for those who are insured and can pay for it. I have some doubts. For example, it has been well established that there is a pronounced negative relationship between annual volume in a hospital and mortality for complex surgical procedures such as open-heart surgery (Luft, Bunker, and Enthoven, 1979). The curve relating death rates to annual volume for coronary artery bypass graft (CABG) surgery is still descending at 150 operations a year, indeed at several hundred (Prospective Payment Assessment Commission, 1988). That is why the California Department of Health Services and the American College of Surgeons recommend minimum annual volumes of 150 for open-heart surgery. Nevertheless, in 1986, of 103 California hospitals in which open-heart surgery was performed, 37 did fewer than 150 such procedures (Steinbrook, 1988b.) This helps to explain the high death rates from CABG surgery in some California hospitals, ranging as high as 17.6 percent in 1986 (Steinbrook, 1988a). In Des Moines, Iowa, a metropolitan area with a population of 380,000, two hospitals did kidney transplants, with 1988 volumes of 8 and 15, respectively. At the university hospital about 100 miles away, the volume was 69 cases (Iowa Department of Public Health, 1989).

A recent genre in our medical literature is called “appropriateness.” A panel of expert and generalist physicians reviews the literature and determines the indications for surgery: Under what conditions is it appropriate (i.e., clearly beneficial to the patient), equivocal, or inappropriate? Then a team reviews a large sample of medical records and classifies cases. Such studies have recently reported 32 percent of carotid endarterectomies inappropriate, another 32 percent equivocal (Winslow et al., 1988); 14 percent of CABG surgery inappropriate, another 30 percent equivocal (Winslow et al., 1988); 27 percent of hospital days inappropriate (Restuccia et al., 1984); 20 percent of pacemaker implants inappropriate, another 36 percent equivocal (Greenspan et al., 1988.), etc.

Somewhere in America might be found the world's best medical care. But the merits of that claim will not be apparent to the families of hundreds of Californians who have died of inappropriate or equivocal open-heart operations in low-volume hospitals, especially if the widows are being hounded for payment because their deceased husbands did not have insurance.

I could go on. We have much to be humble about.

The difficulties of writing this article are compounded by the fact that European health care systems and practices are not all the same. Their diversity exceeds that among and within the different States in the United States. If there are lessons, their relevance will vary considerably from one country to another. Moreover, our intellectual roots and cultures are intertwined. Americans and Europeans read many of the same books and professional journals. Most, if not all, of the ideas I discuss in this article have some European roots. So identifying some ideas leading to good things in America these days is not meant to deny their European ancestry.

One approach I considered was to recommend that Europeans learn from our mistakes, lest they repeat some of them. For example, various British writers, and lately Her Majesty's Government, have proposed offering tax breaks for the purchase of private health insurance (Working for Patients, 1989). The merits of this idea are likely to depend a great deal on exactly how it is done. But in the United States, the open-ended tax break for employer-paid health insurance has had some very negative consequences (Enthoven, 1985a). It greatly weakens the incentive of upper income people to make cost-conscious choices of health care financing plans. Considering payroll and State income taxes as well as Federal income taxes, the tax break reduces by 35 to 40 percent the marginal cost, in net after-tax dollars, of the employee's decision to choose more costly coverage. It costs the Federal budget more than $40 billion a year—an amount that grows about 10 to 15 percent per year, and an amount that substantially exceeds Federal contributions to Medicaid, the Federal-State program that pays for health care for welfare recipients. About 80 percent of the revenue loss goes to households with above-average incomes.. This subsidy offers a costly inducement to buy health insurance to many who would buy coverage anyway; at the same time, it offers little to people in low income brackets and nothing to people who have no employer-paid health insurance. There are lessons to be learned from our mistakes. The problem here is that it is hard to find Europeans who need these lessons.

Each country's health care system reflects its own history, culture, political system, and society. And incremental change is one of the most persistent themes in all of our democracies. Labour rhetoric notwithstanding, there is no prospect for the Europeans to adopt the American system or vice versa. And there is no point in discussing whose system is superior. The really interesting questions are how to identify and design politically feasible incremental changes in each country that have a reasonably good chance of making things better. Each country can get useful ideas from others about how to do this.

I like to believe that there are some things in the rich variety of American experiences that may be quite useful to some Europeans, although I recognize that other Europeans are already well informed about such developments. Even the most interesting and promising of these ideas are not uniformly or even widely applied in the United States. In this article, I am pointing to “best practice,” not average practice in America. I group these ideas under two headings: Glasnost and Perestroika.

Glasnost

Glasnost, of course, is generally understood to mean “openness,” and thus published information. Beginning with this general definition, we can then expand it further to include meaningful evaluated management information, especially in health services research. Some European health care systems—at least those I have visited—struck me by their lack Of relevant management information and evaluation studies based on such information. It appears to be a fair generalization that many European health care systems have not developed and put into use the tools of management information and control that any modern industrial enterprise would consider necessary to plan and manage efficiently. Nor do they take much advantage of their opportunities for research. Volvo, Mercedes, and BMW would not be selling nearly as many cars in California as they do if they attempted to conduct their operations with so little information. Nor has our health care system taken advantage of such opportunities on a wide scale.

Until recently, few policymakers have considered efficiency to be a relevant or appropriate goal for the health care system. In the western European and North American democracies, social policy was initially preoccupied with equity, with extending equal financial protection and access to health care services to most or all of the population. In more recent years, as health care expenditures have grown rapidly as a share of GNP, limiting the growth of spending has become the great preoccupation. But until recently, the efficiency with which resources were used has rarely been addressed in any fundamental way. The creation of institutions that would systematically motivate efficient behavior by providers has received even less attention. Efficiency in the use of resources has not been a part of the culture of our medical professions.

Moreover, the problems in defining and obtaining meaningful information about efficiency in medical and hospital care are exceptionally complex and subtle. Many simple measures, such as a hospital's cost per bed-day or in-hospital mortality unadjusted for medical risk, can be quite misleading. Average cost per bed-day can be reduced by needlessly prolonging hospital stays. The patients of the best surgeon in the country may have a high mortality rate because the sickest patients are referred to him or her. So, development of a really satisfactory system of management information will be a formidable intellectual task. Moreover, before the advent of modern information technology, the collection and processing of the types of data I discuss would have been prohibitively costly.

So, my purpose is not to criticize anyone for the lack of management information. There are good reasons why the “information revolution” in medical care did not happen sooner. But Europeans now have great opportunities to take major steps forward by implementing nationally some of the best ideas being developed in America. European health care systems are more organized than ours, and people do not move around as much, so it should be much easier to keep track of what happens to most patients. Europeans could now take advantage of powerful, flexible, and economical information technology to achieve truly valuable systems of medical and financial information for planning, management, analysis, and evaluation. First, I suggest some opportunities for institutionalized production of information; then I identify research opportunities.

Uniform hospital discharge data reports

Apparently uniform reporting to a central authority of all hospital inpatient cases is not mandatory in most European countries. For example, in Sweden I learned that there are discharge reports but that not all hospitals report, nothing compels them to report, and the reports are not very detailed. It is therefore difficult to compare efficiency by hospital or region without reliable summary reports on many aspects of hospital operations (e.g., per capita admission rates by age and sex, procedure rates per capita, death rates by procedure).

One good place to begin such information production would be with a system of mandatory reporting to a national data bank of all hospital discharges, including the following information:

• Personal identification.

• Date of birth.

• Sex.

• Residence.

• Hospital identification.

• Dates of admission and discharge.

• Identification of attending physician and operating physician, if there was a procedure.

• Diagnoses.

• Procedures and dates.

• Disposition of patient (i.e., alive or dead, discharged to home or to an institution).

This list comes from the Uniform Hospital Discharge Data Set (UHDDS), which must be reported for all care paid for by Medicare and Medicaid (two large Government programs that pay for care for the aged, disabled, and welfare recipients) and for all hospital cases in some States, such as California and Maryland (U.S. Department of Health and Human Services, 1985). In addition to these data, it may well be that additional information, such as some key diagnostic measurements, should also be included. The UHDDS has served as a foundation for developing diagnosis-related groups (DRGs), comprehensive longitudinal records, risk-adjusted measures of outcomes, outcomes management, utilization review, and peer review, which are discussed later in this article. For these developments to take place, it has been necessary that these data, with individual patient identification removed, be available for use by health services researchers.

Mandatory reporting alone is not enough to produce good data. The data have to be put to significant uses for the people who prepare it, so that they will be motivated to make it accurate. Both doctors and medical records technicians must be involved in coding. American experience suggests that there is a substantial potential for error (or at least disagreement) among people who prepare discharge reports. The Institute of Medicine of the National Academy of Sciences did a study in which specially trained medical records technicians prepared new abstracts from hospital records for discharges in 1974 and then compared them with reports already submitted by leading private abstracting services. The study found that the old and new abstracts agreed on principal diagnoses in 65 percent of cases and on procedures in 72 percent of cases (Institute of Medicine, 1977). Therefore, a data commission or board is needed to provide leadership in a continuing effort to improve the coding of information, including; clarifying definitions, ruling on disagreements, and requiring audits to check on accuracy. There will be some hospitals that will be reluctant to report on a timely basis. So there must be some real penalty that actually can and will be applied, such as the nonpayment of State subsidies, if timely reports are not submitted. In the American Medicare program, the hospital is not paid for a case until it has submitted a satisfactory discharge report signed by the attending physician.

National uniform hospital cost accounting

I have asked a number of people in Britain, France the Netherlands, and Sweden to tell me how the average costs per case for particular types of cases compared among hospitals. I was told that such information was not available. (American hospitals can all say what they charge for various types of cases, but few can say what their costs are, and many of their managers do not know the difference.) If European hospital managers had such information, they could analyze and compare medical and management practices in different hospitals to identify the best, i.e., the most cost-effective, practices. Regional and county managements could use the same information as a guide for resource allocation, as, for example, in deciding which services to expand. In the United States, there can be quite wide variations in the charges and apparent costs among hospitals for similar cases, e.g., more than a threefold variation in median charges for CABG surgery among Los Angeles hospitals in 1986 (Steinbrook, 1988a). At least one American management consulting firm has developed a successful business by working with groups of hospitals to identify “best demonstrated practice” in each department (Johnson, 1983). They found variations averaging 40 percent in the cost to treat the same kind of case, and thus substantial opportunities for savings.

The European health care systems could open up significant opportunities for efficiency enhancement by developing and implementing systems of hospital cost accounting capable of producing cost reports for “intermediate products” (such as laboratory tests and X-rays) and “final products” (individual patient cases). With such a system, it would be possible to compare cost per test and cost per case (e.g., CABG surgery in different hospitals) to pinpoint just how and where less costly hospitals save money. As an example of such a system, Sweden's Uppsala Academic Hospital has contracted with Transition Systems, Inc., for installation of a cost-accounting system developed at the New England Medical Center in Boston.

In the European nonmarket systems, I believe there is a case for a uniform national system, at least down to the level of cost per case by type of case, despite the preferences of many hospital administrators to be free to develop their own systems. In a market system, people use prices as indicators of the costs of goods and services they are thinking of buying. It does not matter whether all producers use the same system of cost accounting, because their customers will compare quoted prices. But in the nonmarket systems, such as in the United Kingdom and Scandinavia, there are few or no prices. Cost comparisons must be based on cost information. In this case, there are two reasons to prefer uniform national systems. First, without a truly uniform system, every proposed cost comparison is likely to bog down in detailed arguments about why one hospital's data are not comparable with another's. People who do not want to be compared can prevent comparisons by raising issues of accounting definition. Second, such systems are costly to design and implement. It would be more economical for each country to use a single system. American experience suggests great resistance to such uniformity. Attempts by the U.S. Government to require uniform cost reporting in the early 1980s failed in part because hospitals consider detailed cost information to be trade secrets in our competitive, pluralistic system. As in the case of discharge data reports, auditing and supervision by an accounting principles board would be needed to put life into this idea.

Diagnosis-related groups

In the 1970s, a team at Yale University developed a system for describing a hospital's production called diagnosis-related groups (DRGs) (Fetter et al., 1980). In 1983, the Medicare program adopted the prospective payment system (PPS), based on DRGs. All inpatient cases are classified in one or another of about 470 DRGs that are relatively homogeneous with respect to resource use, and hospitals are paid a fixed price per case, depending primarily on the assigned DRG. Medicare DRGs are now updated each year, based on the latest available information. PPS has not solved all of Medicare's cost problems. The physician fee and outpatient care part of Medicare remains open-ended and out of control. But PPS has had great success in slowing the growth of real inpatient cost per beneficiary. From 1980 to 1983, real Medicare inpatient costs per beneficiary rose 6.8 percent per year; from 1983 to 1984, they rose 2.7 percent.

Kaiser Permanente, our largest nongovernmental medical care organization, uses DRGs as a management tool. Hospital administrators in their Southern California Region are evaluated on the basis of their ability to control their cost per case, with DRGs used to measure case mix. Hospital administrators with costs per case above the average have been directed to bring their costs down to the level of costs in the low-cost hospitals. I have been told that this innovation has led administrators in the high-cost hospitals to become very interested in how the low-cost hospitals achieve their favorable results, and that cost differences have been narrowed considerably.

DRGs are being studied actively in Europe. The most promising use for DRGs in Europe that I can see is as an indicator of a hospital's total inpatient workload or output, to serve as a denominator in a calculation of cost per case. Although it is not perfect, it is the best available indicator of hospital inpatient case load. There are continuing unresolved issues about differences in severity of illness within DRGs, and research is under way to produce a more refined system. Medicare has experienced some “DRG creep,” that is, a change in reported case mix for what appears to be in fact the same case mix. But these problems have proven to be relatively minor. Again, regular audits are needed.

Like Kaiser Permanente, European health care systems might evaluate and compensate hospital managers in part on the basis of their ability to control and reduce growth in cost per case, using DRGs. Europeans would need to develop their own sets of DRGs, based on their own medical practices. To get an adequate sample size, the smaller countries would need to form groupings.

Research is now under way in the United States to develop systems for long-term care and ambulatory care that would be somewhat similar in purpose.

Studies of medical practice variations

John E. Wennberg, M.D., professor of medicine at Dartmouth Medical School, has pioneered in the study of geographic variations in medical practice patterns. In an early study of variations in incidence of surgery in different hospital service areas in Vermont, he found a greater-than-eightfold variation in the per capita incidence of tonsillectomy and adenoidectomy from the lowest to the highest areas (Gittelsohn and Wennberg, 1977). Nonphysicians used to think that there were well-established scientifically based standards for medical practice. Wennberg's studies made us aware that this was not the case. As Wennberg has effectively illustrated with data, there is great uncertainty and differing opinion associated with much of medical practice. And there is a widespread lack of scientific data, especially on the quantitative aspects of medical decisionmaking.

In addition, Wennberg found that feeding data back to doctors led the high users of some procedures to cut back (Wennberg et al., 1977).

Wennberg has teamed up with Europeans to study variations in common surgical procedures in New England, England, and Norway. Similar degrees of variability in surgery rates were found in England and Norway as were found in New England (McPherson et al., 1982). McPherson and colleagues at Oxford have done similar studies (McPherson et al., 1981). With a national uniform hospital discharge report, it should be possible for each country to prepare regular reports of age-sex standardized per capita rates of hospitalization by DRG and procedure, by district, department, or county of patient origin. Such reports, when fed back to doctors, would help “outliers” to see where they are. These data might complement cost-per-case reports. A “low-use” district might justify higher costs per case in certain diagnoses, because fewer patients are hospitalized there than in other districts and only when they are sicker. These data could be used to target for further study areas of high medical uncertainty affecting large numbers of patients.

Comprehensive longitudinal patient records

One of the large handicaps under which American physicians work is a lack of longitudinal data on outcomes of care. Unfortunately, most surgical patients can be followed systematically only to the hospital door. Registries are kept for some patients in some institutions, but these are quite limited in scope. Continuous comprehensive records exist for long-term members of some health maintenance organizations, but usually these are not electronically stored and easily retrievable. Patients in some controlled clinical trials are followed for years. Dr. Wennberg has recently linked Medicare inpatient and outpatient records and Social Security records (which record survival) for Medicare beneficiaries in New England. This has enabled him to follow histories of surgical patients over an 8-year period to see what happens to them. For example, a recent study of patients who had undergone transurethral resection of the prostate found a considerably greater incidence of mortality, complications, and reoperation than previous professional consensus held (Wennberg et al., 1988). Similar records exist in the Provinces in Canada.

The development of standardized longitudinal records has been inhibited in the United States by the decentralized and pluralistic nature of the American health care system. Nobody is in charge to direct such a development. Americans regularly change insurance carriers and providers as they move, change jobs, or merely exercise their choices. Medicare may offer us our most promising data source, because practically all Americans are enrolled in Medicare at age 65 and remain in it for the rest of their lives. Because European health care systems are more homogeneous and people do not move around as much, it should be far more feasible for Europeans to keep track of each patient's medical history in a standardized way. Wennberg's work shows that problems such as preserving confidentiality can be managed and that analysis of such longitudinal data—possibly supplemented by followup questionnaires and other studies—can provide very important information about the outcomes of different treatments.

Risk-adjusted measures of outcomes

An important and promising new development in the United States has been called risk-adjusted measures of outcomes (RAMO) by Dr. Mark Blumberg of the Kaiser Permanente Medical Care Program (Blumberg, 1986). The steps in this process as he describes it are as follows:

• Select a study population.

• Select a clinical care subject (e.g., a procedure or event).

• Select appropriate measures of outcome.

• Identify independent variables that measure risk of adverse outcome (e.g., birthweight, age, presence of multiple diagnoses).

• Develop a technique to estimate expected risk of adverse outcome (e.g., multiple regression, recursive partitioning).

• Estimate the probability of adverse outcome for each case.

• Compare the actual number of adverse outcomes with the expected number for each provider.

• Where there are significant differences, investigate them.

The first example of a risk-adjusted measure of outcomes was R. L. Williams' study of perinatal mortality in California (Williams et al., 1980). The Williams study is now an annual report that compares actual with expected perinatal mortality for every hospital in California. Blumberg has recently analyzed mortality from elective surgery in Maryland (Blumberg, 1988). And the Health Care Financing Administration, which administers the Medicare program, is reporting risk-adjusted mortality by hospital for Medicare patients.

This research is still in its infancy. As Blumberg emphasizes, there are many difficult problems to be overcome, including data accuracy, development of good risk-adjustment models, identification of appropriate outcome measures, and overcoming statistical bias in estimation. Publication of RAMO studies in the United States has been criticized by some physicians on the grounds that “it could be misleading.” But analysis of such data, interpreted by people using informed judgment, is really all we have to go on in evaluating outcomes of care. There is no other scientific way of evaluating the quality of care. In the United States, providers have resisted publication of any data that could link results with specific providers. But growth in health care expenditures has forced government and employers to take cost-cutting measures. In response, providers have argued that cost containment would threaten the quality of care. This, in turn, has led government and employers to start measuring the quality of care directly and to take action to correct poor quality care. When significant variations in risk-adjusted outcomes are identified, they should be investigated. In California, risk-adjusted mortality rates for CABG surgery in 1986 ranged all the way from 1.0 percent to 17.6 (Steinbrook, 1988a). The methods used by the best hospitals should be considered for adoption by the worst hospitals. Prospective patients should have a right to such information.

European health care systems ought to designate at least one center in each country for RAMO and embark on a systematic well-funded research and development program to monitor outcomes of care. American experience shows this could be done.

Outcomes management

Dr. Paul Ellwood, chairman of the influential health policy research institute InterStudy, has recently proposed a bold concept he calls “outcomes management … a common patient-understood language of health outcomes; a national data base containing information and analysis on clinical, financial, and health outcomes that estimates … the relation between medical interventions and health outcomes … and an opportunity for each decisionmaker to have access to the analyses that are relevant to the choices they must make” (Ellwood, 1988). InterStudy is now working with participating medical centers to implement outcomes management by defining the common data set. The proposed data base will include patient description, diagnostic information, therapies, periodic reports by the patient on quality of life, specific medical results, and complications peculiar to the patient's illness or therapy (InterStudy, 1988).

Dr. William Roper, until recently head of the Health Care Financing Administration, and associates responded to Ellwood's proposal with an “effectiveness initiative” …” a four-step process involving monitoring, analysis of variations, assessment of interventions, and feedback and education. In Step 1, monitoring, an ongoing universal data base composed of all Medicare claims is used to characterize the health status of the population involved, … monitor the outcomes of various interventions, … and screen for emerging beneficial or adverse trends. … In Step 2, the goal is to describe and define variations in medical care, in terms of both practice patterns and outcomes. Such studies may be population-based … or may examine the effect of certain interventions. … In Step 3, interventions are assessed. … Step 4 concerns feedback and education” (Roper et al., 1988).

This is an important idea, not yet an achievement. As with the other ideas I have mentioned, this one has European antecedents. Florence Nightingale first proposed this more than 100 years ago, so this is not particularly an American idea. If successful, this initiative could open up large and valuable sources of data regarding what does and does not work and for whom. This could lead to substantial improvements in medical practice. As mentioned earlier, progress in the United States has been inhibited by the diversity, independence, and pluralism of our health care financing and delivery arrangements. Europeans would have an easier time of it, because of their unified comprehensive health care system, and should pursue outcomes management aggressively.

Service and access standards

British and Swedish people complain about access to doctors and about insensitivity of the health care system to reasonable patient demands. Saltman and von Otter (1987) summarized the Swedish problems in these terms: “… non-medical characteristics of service delivery often respond more to the internal interests of the provider organizations rather than valid concerns of the patient … the continued rationing by queue of certain elective surgical procedures … inability to accommodate fundamental differences in treatment preferences … long waiting room times, inconvenient appointment hours, … complicated regulations regarding delivery sites, poorly coordinated services, and so forth.” Dr. David Owen has written of the British situation, “The public concern about NHS [National Health Service] is expressed by 'waiting': waiting for an appointment; waiting then in hospitals or in surgeries for the doctor; waiting to come into hospital; waiting at home for the promised visit. Those who work in the NHS, particularly doctors, have grown to accept too easily that waiting is inevitable.” (Owen, 1988).

Poor service to patients is not an inevitable part of medical care, even in large institutional settings. All of our health care systems could learn important lessons from the best companies in service industries such as hotels, restaurants, and airlines. In the United States, large multispecialty group practices have had to work hard on the design and operation of their systems to improve patient access to compete effectively with solo and small group practices. Kaiser Permanente has experimented with detached primary care clinics of various sizes and with primary care panel systems. They have found that waiting times can be reduced and patient satisfaction improved by implementing procedures designed on the basis of management engineering and operations research studies. For example, appointment scheduling has been improved through use of a computerized “airline reservation” type of system. Access to doctors on the same day that the request is made has been improved by reserving a number of places in each doctor's schedule for same-day appointments; the actual number reserved is equal to the statistically estimated demand for that day. Knowing that Monday morning is a time of exceptionally heavy telephone demand, the organization cross-trains some personnel to answer telephones on Monday mornings, while performing other duties the rest of the time. Improved systems of electronic storage and retrieval of records offer great potential for saving doctors' and patients' time.

NHS regions and Swedish county governments could contract with independent market research organizations to measure patient preferences regarding different combinations of service aspects of the health care system. Based on the results, they could develop and publish service and access standards, create systems of measurement of performance in relation to those standards, and regularly publish the results. Examples of such standards might be along the following lines:

• Patients should not have to wait more than 3 months for elective surgery.

• Ninety-five percent of telephone calls to primary care centers should be answered by the sixth ring.

• Primary care centers should be open and staffed a certain number of evening and weekend hours.

• Waits for appointments (excluding routine physical and eye examinations) should not exceed 3 weeks.

• Ninety-five percent of in-office waits to see the doctor should be less than 30 minutes from the scheduled time.

These data must be interpreted with judgment. Waiting lists can be manipulated by providers to strengthen their case for more resources. But these data can be used to assess relative service efficiency in different centers. Comparative performance can be assessed, and poor performers can be encouraged to adopt the practices of the best performers.

Measuring patient satisfaction

As a part of the Health Insurance Experiment, Allyson Davies and John Ware at the RAND Corporation developed a patient satisfaction questionnaire to evaluate the impact of different health care financing arrangements on patient satisfaction (Davies and Ware, 1988). Some American employers are now polling their employees about their perceptions of the quality of their health care and feeding back the results to the health care organizations that serve them. This is intended to identify needs and motivate improvements in service and care. To provide useful information, the questions should be focused quite sharply on specific aspects of service delivery. For example, one employer asks whether employees have experienced a wound infection after surgery or a medical problem at the end of a stay in the hospital that they didn't have before they entered the hospital. Researchers know that the answers depend on how the questions are framed. So it makes sense for the questionnaires to be designed and administered by organizations that are independent of the health care system and that have a consumer point of view. Patients themselves are a great potential source of information about the quality of care and service they receive. The practice of obtaining such information and using it is not yet well developed in America. But I believe it is potentially important, and I include it for the sake of completeness of the glasnost story.

Publication of information

European voters and public policymakers would be helped in their decisions if the results of all this data gathering and analysis could be interpreted and published in a way that would be accessible to them. For example, it would be very helpful if newspapers would make the investment to develop a corps of a few journalists with the special background needed to understand and responsibly interpret data on the health care system for the general public. This type of reporting might be done by physicians with some postgraduate education in quantitative management tools.

For example, the Los Angeles Times has employed Robert Steinbrook, M.D., as a medical writer. Here are some examples of headlines and lead sentences from Dr. Steinbrook's recent articles: “Care for Newborns Varies, Studies of Hospitals Show … California's hospitals vary widely in their ability to provide quality medical care to newborn babies, according to a sophisticated hospital-by-hospital analysis of perinatal death-rate data by researchers at the University of California, Santa Barbara.” (Steinbrook, 1987.) This article reports the results of R. L. Williams' RAMO study for the years 1980-84. “Heart Surgery Death Rates Found High in 1 in 6 Hospitals ... Nearly one-sixth of California hospitals with heart surgery programs had significantly high death rates for heart bypass patients in 1986, according to a Times analysis of data covering all such operations in the state” (Steinbrook, 1988a). This article reported the results of a study actually commissioned by the Times and performed with the assistance of three leading academic health services researchers at the University of California. A third, “U.S. Issues Data About Hospitals' Death Rates,” described a risk-adjusted mortality study of Medicare beneficiaries published by the Health Care Financing Administration (Steinbrook and Rosenblatt, 1987). This article published mortality rates for California hospitals significantly above and below average, for all Medicare patients, and for patients with several specific diagnoses. The names of the hospitals were published, and the sky did not fall in. Nor have the patients fled the poorly performing hospitals, which is a disappointment to those of us who believe informed consumer choice is potentially a powerful force for good. If Europeans are looking for incentives to improve efficiency and effectiveness in their health care systems, it seems reasonable to suppose that the professional pride of doctors and managers would motivate many of them to take energetic and imaginative action to avoid appearing on the list of the worst departments or hospitals.

Utilization review

In the 1970s, our Congress created professional standards review organizations (PSROs) to review the use of services in the Medicare and Medicaid programs. These were local nonprofit cooperatives of doctors in each of about 200 health service areas. Studies in the late 1970s showed that these organizations were ineffective in reducing Medicare utilization (Ginsburg and Koretz, 1979). This was not surprising. There was no incentive for PSROs to be effective. A dollar saved in Medicare in California, at the expense of California doctors and hospitals, was a dollar returned to Washington.

In the 1980s, PSROs were replaced by peer review organizations (PROs). These are independent organizations in each State that contract with the Health Care Financing Administration to review the quality and appropriateness of care. These organizations compete to win and keep PRO contracts, so they have a real incentive to produce results. They use statistical “screens” to identify problem areas meriting detailed examination, and they use experienced physicians in the appropriate specialty to evaluate the care given. We do not have broadly based studies evaluating the effectiveness of the PROs.

Our many private sector insurance carriers engage in a great deal of utilization control and review activities, mostly for inpatient hospital care. They engage in preadmission review and authorization for nonemergency admissions, concurrent review, and discharge planning. There is little controlled evaluation of all this activity. One controlled study reports that a Blue Cross utilization review program reduced hospital admissions by 12.3 percent, inpatient days by 8.0 percent, and hospital expenses by 11.9 percent (Feldstein, Wickizer, and Wheeler, 1988). We have no documented evidence of effects on quality. In any case, inpatient hospital admissions and days have been declining markedly in this decade. For example, total admissions for people under 65 years of age fell about 9.0 percent from 1984 to 1986.

The utilization review approach to quality and economy of care in the United States attempts to correct the deficiencies in a system the basic incentives of which do not motivate quality and economy to begin with. This approach has several fundamental defects. First, it looks for outliers, “bad apples” that can be identified, punished, and removed. There is no doubt that we have bad apples that ought to be removed. But this approach contributes to an atmosphere of fear, defensiveness, and resentment among physicians, and this atmosphere may be counterproductive in the quest for better quality. By definition, outliers are a small minority. And this month's outlier may be next month's average performer. Removal of outliers will not do much to improve average performance. A second defect in the utilization review approach is that it is too costly, if not impossible, to detect and control the behavior of doctors who are motivated to defeat the utilization control system. Such controls may have a useful effect on inpatient care, but ambulatory care is another matter. The indications for care are too numerous, too uncertain, and too changeable for a police force of reasonable size to be able to keep up. Some system of auditing and real public accountability is needed. But negative restraints in the face of inappropriate incentives seem unlikely to be nearly as effective as positive incentives to do the right thing to begin with. What we all need are systems of organization of care that include evaluation and feedback as a positive incentive to motivate continuous improvement in average performance. For the most part, American systems of utilization review and control are symptoms of the fact that we have not yet achieved that desirable state. We all need to think carefully how this can best be done.

Perestroika

Decentralization and independent institutions

American health care may suffer from an excess of pluralism, diversity, and innovation, without an effective market system to encourage the high-quality economical providers while driving out the low-quality and costly providers. But European health care systems, either of the public or highly regulated private variety, often appear frozen, resistant to innovation and change in financing or organization of delivery.

This is not surprising. There are several reasons public systems in Europe or the United States are especially resistant to change. First, there is what Charles Schultze has called the rule of “Do no direct harm. ... we cannot be seen to cause harm to anyone as the direct consequence of collective actions” (Schultze, 1977). Thus we find it extraordinarily difficult to close an unneeded public hospital or military base. Second, politicians are understandably risk-averse. Most of the innovations people think of prove not to be good ideas, despite the positive connotation of the word. But this can often be discovered only in actual practice. So if politicians try something, the odds are it will fail and they will be blamed. If it succeeds, the rewards are usually quite limited. In the private sector, people can take risks with their own money. In the public sector, the risk-reward ratio often does not favor innovation. And third, most public sector services are monopolies.

On the whole, we have benefited from our Federal system of government. Health care finance and regulation is a mixed Federal-State responsibility. Californians can try many things that appeal to them without being blocked by New Yorkers, who are culturally quite different from Californians. Unitary states in Europe give proposed innovations an “all or none” character.

In the spirit of 1992 (when the trade barriers come down), the different countries of Europe can fill the role of the separate American States. Europeans should build on their practice of studying and learning from each other's experiences, while avoiding legislation that would force uniformity. And within their own countries, Europeans would do well to think more seriously about decentralization, to accommodate more diverse preferences, and to create a climate more tolerant of experimentation.

In the United States, we benefit greatly from the existence of independent nonprofit institutions in the fields of health, education, and social welfare. Indeed, most of our famous universities and hospitals are in that category. These institutions depend on a variety of sources of support, including payments from those they serve, tax-deductible contributions, and grants and contracts from foundations and governments. The element of consumer and provider choice is important. All this creates a framework that fosters diversity and innovation. For example, in health care, we have benefited greatly from the existence of independent nonprofit prepaid group practices (as discussed in the next section). Doctors in the traditional sector tried hard to stop them, including using the power of the State. We never would have had this important innovation, if health care had been entirely provided or controlled by the public sector. The public sector is inherently the protector of the status quo. The established interests have all the power. Our independent nonprofit institutions are usually more socially responsible and long-term oriented than the for-profit sector, but less rigid than the public sector.

Of course, independent (nongovernmental) institutions in health care and finance also exist in Europe. The sickness funds of Belgium, the Netherlands, and the Federal Republic of Germany are independent nonprofit institutions, as are many of their hospitals. Britain has independent provident associations as well as independent hospitals, both nonprofit and investor-owned. The challenge for European societies is to find ways to expand the roles of independent institutions to take advantage of their flexibility and potential for innovation, without sacrificing the social goal of universal access. For example, the British Government is now proposing to transform NHS hospitals into self-governing NHS Trusts, potentially a very productive step in the direction of greater decentralization and greater tolerance of innovation (Working for Patients, 1989).

Prepaid group practice

There has been a great deal of European interest in American multispecialty prepaid group practice, (e.g., Kaiser Permanente, Harvard Community Health Plan, and Group Health Cooperative of Puget Sound). I use the term “prepaid group practice” rather than “health maintenance organization” (HMO), because the latter is quite nonspecific and is also used to describe what amounts to insurance arrangements with little actual organization and management of care. There has been a great deal of research and documentation of the performance of these organizations (Luft, 1981; Manning et al., 1984).

Prepaid group practices combine multispecialty group practice and periodic, per capita payment set in advance in a competitive marketplace. The patients always have an annual choice of health plan, so the prepaid group practice has some incentive to solve patients' medical problems while holding down the cost—in short, to give value for money. This feature probably makes prepaid group practice unique, and therefore understandably an object of considerable interest. Their incentive to seek efficiency in the United States is often attenuated by a lack of serious competitors and by employer practices and features of our tax laws that subsidize employees' choice of more versus less costly health care arrangements. Nevertheless, these organizations have developed a number of characteristics worthy of study and emulation by others.

Prepaid group practices have attracted the loyalty, commitment, and responsible participation in management of their doctors. They have managed to bridge the cultures of medicine and management. Doctors and managers work together in an atmosphere of mutual respect. Management principles are applied to matters of quality and economy of care. The opportunity for continuous quality improvement is enhanced by the fact that the doctors are full-time salaried members of the organization, not independent operators with no organizational commitment. In prepaid group practices, making the correct diagnosis promptly and treating the patient without causing complications are rewarded. (In the fee-for-service system, failure to make a prompt diagnosis results in more visits and more money for the doctor.) These organizations have been leaders in systematic quality measurement and control. They match resources used to the needs of the population served, including numbers and types of doctors. Thus, in each specialty, doctors have full schedules seeing and treating patients whose problems fit their specialty. This is good for proficiency and economy. Doctors can make a good living at a low cost per case because they have lots of cases, and they are not under economic pressure to do procedures that are not really indicated. All this is in marked contrast to our fee-for-service solo practice system, which now has an excess of doctors and no effective way of aligning numbers of doctors to patients' needs. The prepaid group practices concentrate specialized services in regional centers to assure economies of scale and experience. They have pioneered the use of treatment settings less costly than inpatient hospital care: outpatient day surgery, many other treatments on an outpatient basis, and home nursing. They have orderly processes for technology assessment and organized responses to changes in technology. (Doctors in fee-for-service solo practices have powerful incentives to deny the validity of new information that is negative about their “bread and butter” procedures. A large multispecialty group can assist the doctors to retrain in other procedures.) Also, these organizations have innovated efficient use of paramedical personnel, such as nurse practitioners.

A unique feature of prepaid group practice is systematic regular professional interaction of generalist and specialist physicians. With relative ease, the generalist can call on the specialist for consultation in which they can examine the patient and discuss the treatment together. This contributes to the professional education and stimulation of the generalist and keeps the generalist's perspective of the whole patient in the picture when the specialist becomes involved. The generalist need not fear “losing the patient” when he or she makes a referral, and the specialist need not fear a loss of business from assisting the generalist to care for the patient. Professional checks and balances help to moderate single specialty points of view.

Some of these features can be found in some European health care systems, but not in others. For example, with respect to regional concentration of specialized services, Kaiser Permanente probably resembles the British and Swedish systems more than the typical American setting.

Some European countries may find it advantageous to attempt to create similar organizations. For example, Launois et al. (1985) have proposed an adaptation of the idea as an experiment in France. The recent proposal of the British Government to create some budget-holding general practitioner (GP) group practices draws some inspiration from the same idea. Alternatively, many Europeans would do well to examine prepaid group practices for detailed ideas on how to improve efficiency.

None of our countries will achieve a truly satisfactory health care system until we find a way to create internal incentives and dynamism for quality, economy, and good customer service. The model of prepaid group practice in a competitive environment comes as close to that as we have seen.

Demonstrations, pilot projects, and experiments

In the United States, we have gained a great deal of useful information from demonstration projects and social experiments in health care and other fields. The Office of Research and Demonstrations of the Health Care Financing Administration directs and supports more than 300 research, evaluation, and demonstration projects related to the management, organization, and finance of Medicare and Medicaid, our public health care financing programs for the aged and the poor (Health Care Financing Administration, 1989). And other agencies such as the National Center for Health Services Research and Health Care Technology Assessment sponsor and conduct many more. Faculty members from leading research universities and institutes participate in the research designs, and generally a high standard of research design is achieved. Some examples follow:

Medicare and health maintenance organizations

Until 1985, care for Medicare beneficiaries was paid for on the basis of fee-for-service and cost reimbursement (or DRGs), even if the beneficiary got his or her care from an HMO. In the 1970s, there were legislative proposals to pay HMOs on a per capita basis, but no action was taken until the late 1970s, when a new law was proposed, providing for Medicare per capita prepayment of HMOs. The Health Care Financing Administration (HCFA) contracted with four HMOs to test the proposed payment method. The test was a success (Greenlick, Lamb, and Carpenter, 1983). Many fears expressed by the critics were shown to be unfounded. A new law was enacted in 1982 to implement the results of the experiment, and the law went into effect in 1985. Subsequently, 1 million Medicare beneficiaries joined HMOs on a “risk-basis” capitation contract. Now HCFA is sponsoring a dozen followup studies of refinements to the Medicare HMO payment methodology.

The health insurance experiment

Does requiring patients to pay 25 percent of their medical bills, up to an annual limit on out-of-pocket costs (as compared with free care), reduce the use of services? Is it more likely to reduce inappropriate, rather than appropriate services? Does it harm patients' health? The RAND Corporation, under a long-term contract with the Department of Health and Human Services, conducted a long-term, multisite, randomized controlled trial of alternative health insurance arrangements. They found that requiring a 25-percent coinsurance payment reduced spending by about 19 percent, compared with no coinsurance requirement and, with a few small exceptions, had no discernable effect on health (Newhouse et al., 1981; Sloss et al., 1987; Brook et al., 1983). This put to rest debates about whether coinsurance was penny-wise and pound-foolish. They compared fee-for-service with a prepaid group practice HMO and found the HMO cut total resource use by 28 percent and hospital use by 40 percent, with no significant negative effect on health (Manning et al., 1984). This was important in settling debates as to whether or not HMO economies could be explained as the consequence of favorable selection of patients.

Preferred provider insurance

Preferred provider insurance (PPI) was effectively outlawed in most of the United States until a coalition of business, labor, and the insurance industry defeated organized medicine in the California legislature in the summer of 1982. Subsequently, most of the larger States have also changed their laws to authorize PPI. The Health Care Financing Administration recently announced a demonstration project to test PPI for Medicare beneficiaries in five different cities (U.S. Department of Health and Human Services, 1989). If someone attempted to pass a law requiring all physicians serving Medicare patients to accept the Medicare-approved fee as payment in full, organized medicine would doubtless be able to block it. But they have not been able to block this demonstration. And it seems likely that if this experiment succeeds, it will be replicated on a much larger scale, at which point it may acquire a momentum of its own.

Of course, such demonstration and pilot projects are far from unknown in Europe. A recent paper by Kirkman-Liff and van de Ven (1989) describes more than 20 very interesting local demonstration projects in the Netherlands in the areas of monitoring and feedback of medical care utilization and costs, incentives for cost-effective care, community care (substituting home nursing for hospital), and coordination of care. The British National Health Service has attempted clinical budgeting experiments and is now doing pilot projects of indicative prescribing budgets for general medical practitioners. Launois, Majnoni d'Intignano, Stephan, and Rodwin (1985) proposed experimental réseaux de soins coordonnés, (networks of coordinated care) an idea inspired by American HMOs adapted to French circumstances. However, established interests in France were too entrenched to permit a potentially threatening idea to get a start, even as an experiment. (Of course, entrenched vested interests are not unknown in America.)

My general recommendation to Europeans would be to make more widespread and large-scale use of pilot and demonstration projects and to make less use of coercive decree, to foster a process of continuous incremental improvement rather than discrete “great leaps forward” ordered from the center. For example, in 1983, the Griffiths inquiry made a number of very sensible recommendations regarding NHS management in Britain, including competitive tendering by commercial contractors for catering, cleaning, and laundry services. The government attempted to implement this by decree, requiring all districts to submit programs and meet tight schedules. In 1985 I wrote that, “… it would have made far more sense to begin with a dozen pilot Districts whose managements were enthusiastic about the idea, develop and test the methods, with plenty of expert advice from private sector hospital groups, … from airlines and hotels that have much relevant experience, then push tendering to the maximum, display the benefits for all to see, then write the manuals and sample contracts, and develop the short training courses” (Enthoven, 1985b). In 1989, the British Government again proposes some promising and innovative ideas, such as NHS Hospital Trusts. But they announced tight timetables for implementation of ideas that have not been pretested and shown to be workable in practice (Working for Patients, 1989). I believe that in the long run, a phased pilot-project approach would be more effective.

Managed competition

The two best known simple conceptual models for organizing the health care economy are at opposite ends of a spectrum: the free market and the tax-supported public sector monopoly. Proponents of each like to point to the evident deficiencies of the other in support of their own preference. In fact, a free market cannot work in health insurance and health care. There are too many ways in which these markets depart from the conditions necessary for a market to produce an efficient outcome: pervasive uncertainty, great asymmetry of information, moral hazard, adverse selection, many not-truly-voluntary transactions, etc. A free market in health insurance cannot produce either equity or efficiency (Enthoven, 1988). In the United States, for the most part, we do not have a free market in health insurance at the individual level. We have mainly collective purchases by groups, in which the elements of tax subsidy and other government regulations are important. We do have roughly 40 million or so Americans who do not get their health insurance through employment-related groups or public programs Most of them are uninsured and must rely on public hospitals and clinics.

On the other hand, public sector monopolies have their problems, which any impartial observer will admit. They generally contain no serious incentives to improve efficiency. Indeed, they are likely to contain perverse incentives that punish efficiency (Enthoven, 1985b). They are unresponsive to consumer preferences regarding times and places and modalities of treatment. They are guided much more by provider preferences and convenience than consumer preferences. They ration by queues. They lack accountability.

So it is understandable that people are searching for intermediate possibilities, institutional arrangements that capture some of the advantages of markets without their disadvantages, arrangements that can motivate efficiency while safeguarding equity.

A desirable arrangement would separate the demand side from the supply side so that an independent demand side could present the desires of consumers and taxpayers to the providers, set standards, measure performance, and make choices. A desirable arrangement would allow the demand side to become well informed about the costs and the benefits produced by different providers. Thus it would allow the demand side to compel glasnost as described earlier.

A desirable arrangement would also allow choices at two levels: at the level of large group purchasers and at the level of individual choice. The large group purchaser would be able to bring to bear the information and expertise to evaluate all suppliers and exclude those with unacceptable performance; such a purchaser could also structure the market for individual choices so that consumers could make well-informed choices and so that consumers would be guided by correct signals to choose those suppliers that produce high-quality economical care. The element of consumer choice would make the system responsive to responsible consumer preferences regarding quality of care and service.

A desirable arrangement would thus allow the demand side some choice of supplier. It would systematically select and promote the organization and delivery of high-quality, economical, responsive care. How this goal is to be approached in any given country must reflect the cultural preferences, history, and institutional realities of that country. Useful policy proposals must represent politically feasible incremental change. A model that makes sense in one country may have little apparent relevance to another. However, insights gained in one country's experience may be usefully adapted to another.

For the United States, I have been working out and proposing a concept I call managed competition (Enthoven, 1988). Managed competition joins two ideas. First, as previously noted, we now have in the United States a rich variety of schemes that join health care financing and delivery, schemes of varying success in organizing high-quality economical care. Each, in its way, is trying to innovate to find ways to control cost without cutting quality of care or service. Second, managed competition is based on the recognition that the market for health insurance in the United States involves three types of parties: consumers, health insurers (including prepaid group practice and other arrangements), and sponsors. The sponsors are the large group purchasers: employers and the public programs such as Medicare and Medicaid. In managed competition, the sponsor's job is to structure the marketplace, to design and actively manage a process of informed, cost-conscious consumer choice, to motivate the participating health care financing and delivery schemes to produce a favorable combination of efficiency and equity. Efficiency here means value for money as seen by informed consumers. Equity means that the sick do not have to pay much more than the well for coverage and care. Perfect efficiency and equity are of course far from possible. Thus, the sponsor should manage a process of consumer choice that rewards with more subscribers those health care financing and delivery schemes that produce better quality, less costly care, and that does not reward them for selecting good risks, segmenting the market, or doing anything that does not contribute to high-quality economical care.

We have some prototypical examples of managed competition in actual operation. There is the Federal Employees Health Benefits Program, in which more than 400 health plans of various types compete to serve about 9 million Federal employees, dependents, and retirees. This program has been in operation since 1960. In recent years, it has suffered from various correctable design deficiencies that make it vulnerable to risk selection, segmentation, and other problems (Enthoven, 1989a). In California, we have a similar system for public sector employees. And many large private sector employers offer multiple choice of health plan to employees. Richard Kronick and I have recently shown how these concepts might be generalized into a model of universal health insurance for the United States (Enthoven and Kronick, 1989).

Much of managed competition as described here is specific to the American scene, where we have multiple competing health care financing and delivery schemes and strong cultural preference for such pluralism. But some Europeans have been watching this development with interest, to see if similar ideas can be adapted to their situations.

In March 1987, the Committee on the Structure and Financing of Health Care, an advisory committee set up by the Netherlands Government, chaired by Dr. W. Dekker, published a report that proposed major changes to the Dutch health care system (Ministry of Welfare, Health and Cultural Affairs, 1988). In this proposal, market forces would be used to motivate the search for efficiency, especially through better coordination of health and social services, and flexible substitution of more effective, less costly services. In the words of the Ministry of Welfare, Health and Cultural Affairs, English summary, “Market forces provide an answer to the organizational inflexibility and cumbersome operation of the health system in the Netherlands, characterized as it is by high costs, lack of choice and lack of incentives for change.” In the Dekker scheme, all citizens would become free to choose among health insurers. Insurers would be paid in two ways. First, a central fund would collect an income-related premium from all those able to pay, and it would pay insurers a risk-related premium contribution based on the characteristics of its subscribers. (This is an important sponsor function in managed competition.) Second, each insurer would charge a flat-rate premium to all insureds. Insurers would compete on this flat-rate component and would be free to negotiate selectively with providers for pay and scope of services. Insurers would no longer be obliged to contract with all providers. They could select those they considered to be efficient. And guaranteed funding for providers would be eliminated. After much public debate, the Netherlands Government indicated broad agreement with the Dekker proposals, and, in March 1988, issued a plan for their cautious and gradual implementation. This Dutch version of managed competition will give Europeans a “demonstration project” to watch much closer to home.

In January 1989, the British Government published a white paper outlining its strategy and proposals for restructuring the National Health Service (Working for Patients, 1989). Broadly speaking, one might characterize it as a strategy for separating the demand and supply sides of the market and for strengthening the ability of the demand side to make informed choices. In the government's strategy, District Health Authorities (DHAs), which are now monopoly suppliers of services to the people in their districts, are to be recast as purchasers of services on behalf of the populations they serve, which services are to be supplied competitively. That is, DHAs will be free to seek value for money outside their districts and even outside the NHS, in the private sector. Regions will actually receive their main budget allocations on the basis of population, adjusted for age, morbidity, and other demand-related factors, with the present adjustments for cross-boundary flows replaced by direct payments for services between regions. Regional targets have long been based on such a formula, but actual payments followed targets only gradually, because of fear of disrupting the supply side. In the new plan, district budgets will be based on estimated need, not influenced by the services they produce.

The government's strategy includes allowing hospitals to opt out of district control and to become independent self-governing NHS Trusts. These hospitals will be free to set their own pay, contract with their own personnel, and compete to serve several districts. A key idea is that “money follows patients.” Today, a hospital that does an excellent job of producing high-quality care efficiently, thereby reducing or eliminating its queue, is likely to attract more patients without correspondingly more money— a perverse disincentive. Under the new scheme, such a hospital will be able to contract with sending districts for a prompt payment per case. The government's plan also includes strengthening medical audit, and experimentation with the idea of large GP practices holding budgets for a broad range of services beyond primary care. All this is sometimes referred to as an “internal market” for health care, compatible with universal tax-supported provision of comprehensive care (Enthoven, 1985b; Owen, 1988).

The Swedish health care system seems even less amenable to concepts of managed competition than the British. And the geographic pattern of very large county hospitals seems almost a guarantee of territorial monopolies for inpatient care. However, introduction of some elements of managed competition is not beyond the realm of conceivable political reality. I have recommended a program of glasnost like the one described in this article, combined with a policy of rewarding with pay and promotion those physicians and managers who demonstrate superior performance (Enthoven, 1989b). Beyond this, it might be productive to consider competition within the public sector at the primary care level, along the lines proposed by Saltman and von Otter. Moreover, the Swedish Government instituted an arrangement whereby patients waiting for certain procedures could obtain care from other counties if the waiting time in their own county exceeded certain limits, with the patient's county paying the providing county, and the government throwing in a bonus. I understand that this appreciably reduced waiting times. What is needed is the political will for the Swedish people to create an institution independent of the health care providers, with the power to compel production of information and the resources and charter to take initiatives to get more informed choice into the system.

What these ideas and experiences illustrate is that intellectual discourse on health policy does not need to be limited to debates over the merits of polar opposites. Nor is it useful to argue abstractly over the merits of regulation versus competition. Every health care system is likely to have elements of both. The really interesting questions today are about the merits of marketlike incremental changes intended to make our systems more efficient and responsive to consumers. In this realm, American research and debate have produced what ought to be a good deal of interesting reading for Europeans.