Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
chapter 4 THE ACADEMIC HEALTH CENTER AS MODELER: A THE PATIENT CARE ROLE At a workshop sponsored by the committee during the course of this study, Snyderman (2002) coined the term "service platform" to denote the need for a better model of care delivery capable of exploiting the new technologies and capabilities that will characterize the health care system of the future. In the words of another workshop participant, "we are trying to put a new genotype on an old phenotype." It is necessary to redesign the processes of care, but doing so will also require altering the structures that deliver care. The term "platform" has been used in other industries. In information technology, it denotes the infrastructure that permits a particular use or analysis of information. In the military, the term refers to the ways in which workforce, equipment, and organization should be arranged to produce a specific capability or response. In both of these cases, the platform is de- signed to produce a clear output. Rather than starting from the available capabilities and determining what can be done, designers first ask what needs to be done and then design the platform to deliver it. This same concept can be applied in considering how to design a new service platform for health care. Given the trends described in the Chapter 2, those who deliver health care need to ask how current models for and approaches to care can be redesigned not only to treat the illnesses of patients but also to improve the health of patients and populations. If health care is to produce a different output, the platform for delivering that output needs to be rethought. In examining the clinical care role of AHCs, the committee finds the following: 65
66 ACADEMIC HEALTH CENTERS · AHCs are major providers of specialty care, and many also provide significant levels of care to the poor and uninsured. AHCs therefore have a dual safety net role as provider of last resort for the critically ill, and for the poor and uninsured. · AHCs need to play a part in redesigning care if they are to respond to the changing demands that will arise in the coming decades and be able to deliver the improved capabilities that the system will have the potential to offer. · The clinical care setting is where the AHC research and education roles intersect. The ability to incorporate the new sciences developed by research into care delivery and successfully teach students how to practice in the evolving environment of care will depend on how effectively AHCs can adapt their clinical care settings. The first section below examines the need for new models of care. The next two sections review the contributions of AHCs to patient care and the challenges facing AHCs as they work to design better models of care. The final section presents some implications for the future. THE NEED FOR NEW MODELS OF CARE The shifts and developments that will occur in health care over the coming decades argue strongly for the creation of new approaches to the organization and delivery of care. Better models are needed for care for the chronically ill and for use of the latest information and biomedical tech- nologies to maximize both the quality of care and the cost-effectiveness. Current models of care are heavily focused on interventions for treating illness. There is evidence that better approaches are needed to improve health. Although mortality rates have declined across all age groups, these general declines mask important differences (Institute of Medicine, 2002d). For example, lung cancer and chronic respiratory disease have declined or remained stable for men but increased for women (National Center for Health Statistics, 2002). Likewise, differences by race have been identified in both the diagnosis and use of therapeutic procedures for cardiovascular care (Lurie and Buntin, 2002). Since cardiovascular disease is one of the leading causes of death in the United States, improving care for those afflicted could have a significant impact if designed and targeted properly (Wong et al., 2002). As noted earlier in this report, chronic illness is the leading cause of illness and disability among the U.S. population, yet too many of those affected do not recieve adequate treatment (Wagner et al., 1996), including guidance on lifestyle changes that can help in preventing and managing these conditions. In addition, as noted above, care delivery exhibits variations that are
AHC AS A MODELER: THE PATIENT CARE ROLE 67 unrelated to type or severity of illness or patient preferences (Wennberg, 2002). The risk of hospitalization at all of Boston's teaching hospitals is higher than that at YaleNew Haven Hospital, even after adjusting for age, sex, race, illness, and the price of medical care (Center for the Evaluative Clinical Sciences at Dartmouth Medical School, 1999a). Among referral areas that contain at least one medical school, the age-, sex-, race-, and illness-adjusted discharge rate for medical conditions per 1,000 Medicare enrollees ranges from 285 in Jackson, Mississippi, to 165 in Salt Lake City (Center for the Evaluative Clinical Sciences at Dartmouth Medical School, 1999b). Medicare beneficiaries who live in regions that exhibit higher spend- ing levels receive about 60 percent more care than residents in lower- spending regions but do not show better quality of care, outcomes, or satisfaction with care (Fisher et al., 2003). Some variation would be ex- pected and desirable to reflect patient preferences and customized care. However, the levels of variation among all institutions across the country (not just AHCs) suggest that much of clinical practice remains empirical and not necessarily driven by science (Wennberg, 2002). In general, current approaches to care are reactive, involving treating patients when they present with symptoms through a series of sporadic interventions that are predominantly physician directed. To address the changing needs of people and exploit technological advances, 21st century health care will need to be more proactive, interactive with patients, and evidence based (Snyderman and Saito, 2000). The prevalence of chronic illness will demand better approaches to care, and the new technologies will enable the prediction and prevention of disease, especially through an un- derstanding of genetic susceptibility and behavioral risks and the benefits of their modification. Examples of better models of care are beginning to emerge. These models target particular populations, such as those with chronic illness, the frail elderly, the poor, the uninsured, and those with specific conditions. The examples described here are offered as illustrative examples to demon- strate that better approaches are possible. The Chronic Care Model is designed to improve coordination and collaboration in care for chronically ill populations. The model is charac- terized by (1) a protocol or plan containing an explicit description of what is to be done for individual patients, as well as for groups of patients with specific clinical features; (2) a redesign of practice to include regular patient contact, collection of data on health and disease status, and efforts to address patients' psychosocial needs; (3) a strong focus on patient informa- tion and self-management, including support for behavioral and lifestyle changes to improve outcomes of care; (4) the availability of specialized expertise for practitioners managing care; and (5) good information about patients, their care, and the outcomes of care, including the use of registries
68 ACADEMIC HEALTH CENTERS and reminder systems to support care plans (Wagner et al., 1996). The model emphasizes self-management, care planning with an interdisciplinary team, ongoing assessment and follow-up, and linkages with community programs (e.g., exercise programs) (Bodenheimer et al., 2002; Institute of Medicine, 2001b; Lurie and Buntin, 2002). Although care for the chroni- cally ill demands greater coordination and communication along the con- tinuum of care settings, the current system is characterized by fragmenta- tion and poor coordination. For instance, it has been estimated that fewer than half of patients with hypertension, depression, diabetes, and asthma are receiving appropriate treatment (Rundall et al., 2002). A second model, focused on the frail elderly, is the Program for All- Inclusive Care for the Elderly, or PACE. This model is designed to provide and coordinate all needed preventive, primary, acute, and long-term care for the frail elderly, with the aim of optimizing health and functioning while permitting participants in the program to continue to live in the community (Program of All Inclusive Care for the Elderly, 2002). The program was developed during the 1970s, was tested beginning in the 1980s, and was established as a permanent model in Medicare under the Balanced Budget Act of 1997. Evaluations of PACE programs have found that participants have better functional status, receive more primary care and preventive services, and experience fewer days in the hospital despite having greater morbidity and disability than other elderly populations, although programs exhibited considerable variation (Burton et al., 2002; Wieland et al., 2000; Mukamel et al., 1998). A number of AHCs, including Johns Hopkins, Mount Sinai, and the University of Pennsylvania, sponsor PACE programs, (Program of All Inclusive Care for the Elderly, 2002). Better models of care can also result in improved care for poor and uninsured populations. The University of New Mexico, for example, cre- ated a type of managed care arrangement for the uninsured. This model emphasized primary care, a continuous patientphysician relationship, and the priority of preventive care (Kaufman et al., 2000). Results observed over a 2-year period revealed that ambulatory visits (including those to the emergency room), hospital discharges, and hospital days decreased. Among the subset of high users, outpatient and specialty visits increased, hospital discharges and days decreased, and there was no change in the number of emergency room visits. Hospital revenues increased as well because the lost volume was replaced by paying patients. Meeting the needs of this popula- tion also required attention to social support services, such as transporta- tion, translation, and other types of referrals (e.g., to literacy programs). Prior to this program, the emphasis was on providing inpatient and spe- cialty services for primary care problems, which did not meet the needs of this group (Kaufman et al., 2000). The Diabetes Control and Complications Trial represents the develop-
AHC AS A MODELER: THE PATIENT CARE ROLE 69 ment of an intensive care strategy to care for people with diabetes. Spon- sored by the National Institute of Diabetes and Digestive and Kidney Dis- eases from 1983 to 1993 and conducted in multiple sites, the study showed that the onset and progression of complications from diabetes could be slowed with intensive clinical management that included not only testing of blood glucose levels four or more times a day, four daily insulin injections (or use of an insulin pump), and adjustment of insulin doses according to food intake and exercise, but also a diet and exercise plan and monthly visits to a health care team that included a physician, nurse educator, dietitian, and behavioral therapist (National Institute of Diabetes and Di- gestive and Kidney Diseases, 2002). Effective management required a clear and explicit clinical plan, strong patient involvement, and the support of an integrated care team. This approach is similar to disease management mod- els that emphasize a systematic approach to care, employ interdisciplinary teams to deliver care, use practice guidelines and protocols appropriate to the target population, and can potentially include services across the entire continuum of care (Blumenthal and Buntin, 1998). A report of the Institute of Medicine (2001b)--Crossing the Quality Chasm: A New Health System for the 21st Century--describes the need for redesigning care delivery in several areas. First, the processes used to deliver care need to be made more reliable and to make better use of information technologies to automate clinical information and improve communica- tions. Second, clinicians must be provided with the new knowledge they will need to translate the evidence base into practice and manage the result- ing changes. Third, interdisciplinary teams must be created and maintained; to this end, it will be necessary to overcome training, structural, and finan- cial barriers that can hinder team functioning. Fourth, care needs to be better coordinated across patient conditions, services, and settings and over time; coordination with community resources or the public health system is particularly difficult to achieve. Finally, performance and outcome mea- sures for improvement and accountability need to be incorporated into the daily work of health care organizations so they can continually evaluate and improve the care delivered. Several characteristics are common across the models described above. First, each encompasses an interdisciplinary approach. As discussed earlier, interdisciplinary teams are needed in health care, in part because of the increased complexity of care. Treatment for many conditions is so complex that the knowledge of multiple practitioners--including various medical specialists as well as other clinicians, such as therapists or nutritionists--is needed to manage a single condition. In addition, there has been a signifi- cant increase in the number of nonphysician clinicians. In the early 1900s, physicians represented one of every three health care workers (Aiken, 2001). Today this figure is about one in ten (counting health practitioners involved
70 ACADEMIC HEALTH CENTERS in direct care, and excluding managers and support personnel) (Bureau of Labor Statistics, 2001). As the mix of health care workers diversifies, they must increasingly work in teams to deliver care. Characteristics of effective teams include appropriate size and composition, good communication pro- cesses, clarity in team tasks, and an environment in which the team can acquire needed resources (Institute of Medicine, 2001b). Ineffective teams can inadvertently cause errors if, for example, there are too many hand-offs that are not well planned or executed properly. The field of aviation is often cited as a model for the training and attention given to developing effective teams (Institute of Medicine, 2000). A second common characteristic of the models described above is a patient-centered focus. Patient-centered care is defined as "health care that is closely congruent with and responsive to patients' wants, needs and preferences" (Laine and Davidoff, 1996, p. 152). It encompasses disclosure of information to and active discourse with patients; patients' participation in decision making about their care; and recognition of outcomes that include functional status, satisfaction, and quality, all of which require patient input to measure. Patient-centered care is assuming increasing importance for a variety of reasons. For one thing, chronic illness demands greater self-management by patients (Bodenheimer et al., 2002). For another, as technological advances expand treatment options, the choice of treatment should reflect patient input when possible and desired by the patient (Barry et al., 1995). There is also evidence that patients who are more involved in their care have better outcomes (Institute of Medicine, 2003b). Finally, there is some evidence that patients who share in decision making may have decreased demand for services (Wagner et al., 1995). Therefore, redesigned models of care need to recognize the patient as part of the care-giving team. Finally, all of the models described above are characterized by a broad view of health that not only reflects excellent science-based clinical care but also addresses other factors that influence health, such as exercise, nutrition counseling, and community services. The models were designed to revolve around the needs of patients to maximize their health and functioning, instead of focusing on the capabilities of a particular setting of care. CONTRIBUTIONS OF AHCS TO PATIENT CARE AHCs are recognized throughout the world for their specialty care. Although AHC hospitals represent just 3 percent of all hospitals in the United States, they house 33 percent of transplant services, 16 percent of neonatal units, and 15 percent of open-heart surgical units (see Appendix A). The provision of specialty services at AHCs also ensures standby capac-
AHC AS A MODELER: THE PATIENT CARE ROLE 71 ity that benefits the local community. Like fire departments, AHCs assure people that the services of a trauma or burn unit are available. About half of patients with rare and uncommon conditions are cared for at AHCs and major teaching hospitals. Yet such patients represent a relatively small proportion of the volume at these centers, accounting for about 13 percent of overall admissions (The Commonwealth Fund Task Force on Academic Health Centers, 2000), although individual AHCs may exhibit differing proportions of routine to specialty care. AHCs also receive a large proportion of patients who are transferred from other hospitals for all types of care (not just rare and uncommon conditions). The Common- wealth Task Force on Academic Health Centers (2000) estimates that the proportion of AHC patients who were transferred from other hospitals was more than 8 percent in 1995, up from about 5 percent 3 years earlier. Transfer patients tend to be older, to have more comorbidities, and to require more complex treatment than other patients. The AHC clinical enterprise has grown rapidly in recent years. Overall, the average daily census at AHC hospitals has declined by 2 percent be- tween 1990 and 2000, but during the same period, outpatient volume has increased by 133 percent; emergency room visits by 54 percent (see Appen- dix A); and clinical faculty, who deliver the care, by 52 percent (Jonas et al., 1990; Barzansky and Etzel, 2001). In 1990, AHCs represented 2 percent of hospitals, 7 percent of hospital beds, and 10 percent of total hospital days; by 1999, they represented 3 percent of hospitals, 10 percent of hospital beds, and 13 percent of total hospital days (see Appendix A). The market share of AHC hospitals increased during a time at which inpatient admis- sions in general were declining. For most AHCs, revenues from clinical activities support education and research activities and make it possible to care for the uninsured. Whether these historical levels of growth can be sustained into the future is unclear, however. As discussed earlier, many AHCs are also an important part of their local community's safety net. In a study of 38 communities with AHCs (Reuter, 1999), the AHCs represented about 6 percent of hospitals and 13 percent of hospital beds, yet they provided: · 36 percent of care for Medicaid AIDS patients and 34 percent of uninsured AIDS patients. · 36 percent of trauma care for Medicaid trauma cases and 36 per- cent of uninsured trauma cases. · 25 percent of care for Medicaid high-risk infants and 26 percent of care for uninsured high-risk infants. Although there has been much analysis of safety-net providers, The Commonwealth Task Force on Academic Health Centers is one of the few
72 ACADEMIC HEALTH CENTERS sources that has specifically examined AHCs (as opposed to teaching hospi- tals or hospitals generally). The task force has estimated that in 1991, AHC hospitals accounted for almost 40 percent of total charity care provided; by 1996, this proportion had grown to 44 percent (The Commonwealth Fund Task Force on Academic Health Centers, 2001). During the same period, the number of uninsured patients cared for at AHC hospitals grew from 20 to 28 percent. Moreover, care for the uninsured appears to be growing as a proportion of all care provided at AHCs. In terms of hospital costs, uncom- pensated care was estimated at 7 percent of costs at AHCs in 2000, an increase of almost 2 percentage points since 1994 as compared with a 1 percent increase for other hospitals (Dobson et al., 2002). Public AHCs appear to play a larger safety-net role than private AHCs. The Commonwealth Task Force found that of the total charity care pro- vided in 1996, 31 percent was provided by public AHC hospitals and 13 percent by private AHC hospitals, a pattern similar to that exhibited by public and private hospitals generally (The Commonwealth Task Force on Academic Health Centers, 2001). In the previously noted study of 38 com- munities with AHCs, public AHC hospitals treated 17 percent of all unin- sured and 10 percent of all Medicaid patients in those markets, whereas private AHC hospitals treated 5 percent of all uninsured and just over 7 percent of all Medicaid patients (Reuter, 1999). A number of factors influence the safety-net role of AHCs. As noted above, public ownership is one factor. Geographic location is another, with many AHCs being located in central cities where large numbers of poor and uninsured people reside. It may be noted that the AHC safety net role has supported clinical education by providing students with a volume of pa- tients for their training experiences. A major source of support for hospitals that serve large numbers of poor people is disproportionate-share funds provided by Medicare and some state governments. According to figures presented to the committee, Medicare disproportionate-share funds are highly dispersed, going to ap- proximately 4,000 institutions, only some of which are AHCs (Anderson, 2002). One of the main concerns regarding disproportionate-share funding is that the formula does not adequately target hospitals that serve the greatest numbers of poor and uninsured (see Chapter 6). Furthermore, by paying for hospital care, the arrangement does not encourage the develop- ment of better models of care that are more responsive to the needs of these populations (such as the University of New Mexico example described earlier in this chapter). According to recent evidence, AHC hospitals that serve more poor and uninsured people have lower financial margins than other hospitals. In 2000, the aggregate total margin for public AHC hospitals declined to 3.7 percent (6.7 percent for aggregate operating margins), whereas total mar-
AHC AS A MODELER: THE PATIENT CARE ROLE 73 gins for private AHC hospitals increased to the 1998 level of 4.4 percent (1.1 percent for aggregate operating margins) (Dobson et al., 2002). The concern is that safety-net AHCs may have fewer resources and options available to them relative to other hospitals (Zuckerman et al., 2001). Although the implementation of information technology could pro- duce efficiencies that are needed by safety-net providers in particular, pur- suing such a strategy requires capital investment. Similarly, redesigning care will require working capital and could result in temporarily higher operat- ing costs as the organization transitions to new programs and operating designs. There is also concern about access by the poor and uninsured. Hospitals (including those at AHCs) that are facing reductions in revenues generally seek ways to become more efficient but may also limit access for medically indigent patients (The Kaiser Family Foundation, 2001). CHALLENGES TO AHCS IN DESIGNING BETTER MODELS OF CARE The pressures to redesign models of care can be expected to increase. As noted earlier, the shifts in the needs of the population and changing composition of the workforce will necessitate better approaches to care. Furthermore, as noted in Chapter 2, the increased demand for care brought about the aging population, combined with a slow growth in the size of the labor force, can be expected to result in increased labor costs, along with demands for productivity improvements. Labor shortages, such as in nurs- ing, will add to the pressures to redesign care. AHCs have an important role to play in redesigning models of care for at least four reasons. First, as part of the direct care delivery system, AHCs need to ensure that they are providing care designed to meet patients' needs and improve health. As noted earlier in this chapter, well-designed pro- cesses of care affect the health of patients. Second, through their research role, AHCs create the knowledge that drives the care received by patients. Part of translating that knowledge into practice is understanding and improving the organizational context in which the care is delivered. A research scientist may develop a procedure or other element of care that is technically sound, but if it is delivered through a poor design, its full benefits may not be realized. AHCs need to redesign care so that new knowledge discovered can also be delivered. Third, the care provided at AHCs needs to demonstrate evidence-based best practices for the students who are learning in these settings. Students should be taught to practice models of care that are designed to improve health. Therefore, it is important that AHCs view the clinical care setting as one component of their academic activities and use it to develop, test, refine, and improve processes of care.
74 ACADEMIC HEALTH CENTERS Fourth, AHCs need to ensure a good working environment within their own organizations to attract and retain a high-quality workforce. For ex- ample, one factor contributing to the current nursing shortage is dissatisfac- tion with the work environment, including a lack of respect, a lack of recognition, a lack of participation in decision making, and an erosion of the nursepatient relationship (Association of Academic Health Centers, 2002b). Since labor represents about half of a hospital's operating expenses (Sochalski et al., 1997), redesigning care to improve the health of patients will also require examining work processes and the use of human resources. AHCs are recognized for technological innovation, but they are not automatically associated with organizational innovation. Many of the ef- forts undertaken to date to reorganize care and people have involved "reengineering," or the simultaneous restructuring of work processes and organizational design (Walston et al., 2000). Reengineering reallocates and readjusts work flows and job responsibilities, and determines where work is located, who does the work, and how the work will get done. Reengineering efforts in health care generally have not lived up to their promise, however. Efforts have typically been based in departments, and have thus failed to address overall issues of organizational design (Aiken, 2001). For example, it may not be possible to resolve issues related to nurses' dissatisfaction with the work environment at the departmental level. Hospitals found to be more successful in attracting and retaining nurses are characterized by a professional practice environment that fosters greater autonomy for nurses, their greater control over support services, and better communication be- tween physicians and nurses (Steinbrook, 2002; Aiken, 2002). As discussed earlier, another challenge facing AHCs is the need to make greater investments in information and communications technology for monitoring and evaluating care, and for understanding the relationship between processes of care and outcomes. Assessing patterns of care for groups of patients will demand better information technology that can aggregate data across the patient's experience, especially across settings and over time. Information and communications technology can also serve as glue that holds care teams together, getting information to people whenever and wherever it is needed. In one example, the University of California (UC) system is installing a Web-based medical-event reporting system to improve patient safety in its medical centers and provide a means for rapid identification of areas for improvement (University of California, 2003). This is the first effort of its type, linking five AHCs--UC Davis, UC Irvine, UC Los Angeles, UC San Francisco, and UC San Diego--on a systemwide basis through the Internet, permitting front-line clinical workers to report on adverse and near-miss events from most computers in each of the par- ticipating medical centers. The system includes the establishment of a sever-
AHC AS A MODELER: THE PATIENT CARE ROLE 75 ity ranking to permit comparisons within and across campuses. Monthly conferences will be held to address findings. IMPLICATIONS FOR THE FUTURE AHCs need to develop the structures, processes, and team approaches necessary to achieve improvements in health for the patients and popula- tions that rely on them. Asking AHCs to redesign care will mean requiring that they conceptualize new models of care. This is different from improv- ing or refining a particular technique or procedure, for example. Although those aspects of care need to be developed and do produce improvements in care, conceptualizing new models of care will require AHCs to describe, design, and shape new approaches to care that are patient-centered and aimed at improving health. Some envision a more proactive model of care that identifies people at risk of major disease and intervenes early to pro- spectively alter the progression of disease (Williams et al., 2003). Conceptu- ally, interventions could include customized care that relies on the latest biomedical advances, but also community interventions aimed at specific subpopulations. Given the patient populations served by most AHCs, redesign should focus in particular on people at high risk for serious illness and those who are financially vulnerable. Redesign should also emphasize methods for encouraging patient self-management and adoption of healthy behaviors. Achieving such redesign will require that AHCs work across all of their component organizations, including nursing schools and public schools and programs, as well as with their local communities. Implementing new models of care will also require delivery system changes that include greater reliance on information systems, patient self- management that necessitates expanded health education and support, a team orientation, and decision support (Berenson and Horvath, 2003). However, current payment methods create several obstacles to making these types of delivery changes. First, the types of services that are most focused on improving and maintaining health are not as well supported by payers as medical services. For example, patient education for self-management is supported by Medi- care in only limited circumstances, such as diabetes care (Berenson and Horvath, 2003). Second, current methods have weak or no incentives to improve care or health, and are generally not designed to support coordination of care, interdisciplinary team approaches to care, or improvements in health. Fee- for-service payment rewards the delivery of individual units of care, an arrangement that inhibits coordination and team approaches and rewards treatment of illness (Institute of Medicine, 2001b). Providers focus on their
76 ACADEMIC HEALTH CENTERS own activities and functions without attention to the effect on costs or care across settings or providers (Berenson and Horvath, 2003). Third, providers can lose money by improving care. For example, a pilot project at Duke University improved outcomes of care and reduced annual expenses by almost 40 percent for patients with congestive heart failure. However, Duke lost money because patients stayed out of the hos- pital, avoiding procedures that are relatively well reimbursed, while incur- ring greater expenses for ambulatory care and patient education, which are more poorly reimbursed (Williams et al., 2003). In another example, a physician group that was paid through fee-for-services methods improved care for diabetes patients and achieved cost savings through reduced visits and hospitalizations but lost money in two ways (Institute of Medicine, 2001b). First, they incurred the expenses for tighter clinical management that produced the improved outcomes. Second, the savings due to reduced hospitalizations and visits accrued to the insurer rather than the provider that had made the savings possible. Providers cannot be expected to sustain care improvements if they will predictably lose money for doing so. Capitation payment arrangements should provide greater flexibility to coordinate care and allocate resources according to the needs of patient groups but appear to be diminishing as a payment method (Hurley et al., 2002). Furthermore, capitation arrangements may be narrowly defined, covering only office visits or ambulatory care, for example, rather than a comprehensive continuum of care that would be required under a chronic care or other model (Dudley and Luft, 2001). Shared-risk arrangements may offer a stronger potential for both payers and providers to gain from care improvements and cost savings. However, such arrangements would likely require a partnership between the payer and provider, as well as longer-term contracts to permit the needed investments and make it pos- sible to obtain the rewards of the improvements, rather than the annual arrangements most typical today (Institute of Medicine, 2001b). Redesigning care to improve health is not the responsibility of AHCs or of payers and employers--it is the responsibility of all. AHCs should help guide payers and policy makers with regard to the characteristics of care models that improve health for patients and populations and the features that best demonstrate evidence-based, continuously improving, cost-effi- cient practice, recognizing that payers and employers have to balance the cost and quality and access needs of a population of enrollees and beneficia- ries. Payers and employees should support demonstration projects that aim to build better models of care, recognizing the priorities of the other. Payers need to recognize that redesigning care will require some experimentation, that not all plans will work as designed, and that there is a cost for testing new approaches while not abandoning the status quo (in essence, maintain- ing dual systems).