ing the performance of the health system as a whole and setting national health priorities. For hospital care, it would be possible to build upon existing claims formats and fairly uniform hospital discharge data across purchasers. However, for other settings, there is very little uniformity across purchasers. An area of particular concern is the lack of experience with obtaining detailed information from ambulatory care settings and particularly from physicians' offices. For Medicare and some private insurers, diagnostic data are now included on claim forms used for ambulatory care. This will be useful but still quite minimal for quality assessment purposes. Significant attention will have to be devoted to defining an appropriate set of minimum elements for this type of health care encounter and to ensuring that the information provided is reliable and valid. The recent revision to the 1981 National Ambulatory Medical Care Minimum Data Set by the National Committee on Vital and Health Statistics provides a start toward specifying such a set of data elements. Even with agreement on the appropriate data elements, the implementation of an integrated data system will not be simple. The resources required for collecting, processing, and maintaining this data base will be substantial and include both human resources and computer hardware and software. The integration of data across settings, providers, and purchasers will require the use of unique, common identifiers for providers and purchasers as well as for patients. The data coming into the system must be checked regularly to ensure their accuracy. The data will have to be organized so that all encounters for an individual patient, as well as all services provided by a particular provider, can be easily collated and analyzed. The system must also be flexible enough to accommodate the inevitable changes and improvements in data and quality assessment methods that will come with time. Safeguards for privacy and confidentiality will also need to be addressed. An Improved System of Local Review-Our reviews of the literature as well as the results of some quality review programs leave little doubt that significant numbers of patients are currently receiving inappropriate or poor-quality care. For example, in past studies, we have cited estimates of rates of inappropriate use of surgical procedures ranging from 14 to 32 percent as well as rates of inappropriate hospital admissions ranging from 7 to 19 percent. In addition, our evaluations of current quality assurance programs suggest that those programs are not identifying signifi General Accounting Office, Medicare: Improvements Needed in the Identification of Inappropriate Hospital Care, GAO/PEMD-90-7 (Washington, D.C.: December 20, 1989), 3-4. cant proportions of cases with potential quality problems. For example, SuperPRO regularly reviews a random sample of Medicare cases previously reviewed by PROs and typically questions the appropriateness of hospital admission in almost six times as many cases as the PROs. Similarly, our review of the initial screening of cases in military hospitals for occurrences indicating potentially substandard care found that many such occurrences were missed in the initial screening process." Despite the importance of continuous quality improvement strategies in the long run, our past work has shown that improvements in external quality assurance mechanisms are needed in order to achieve the goal of appropriate, high-quality medical care for all Americans. We believe that there are a number of key components for improving the conduct of quality assurance within the framework of a comprehensive, national strategy. First, the quality assurance activities need to be conducted by local review entities that are held accountable for identifying instances of poor quality and improving overall patterns of care within their geographical area. Second, the local review entities should have available a uniform set of methods for reviewing care (including practice guidelines and standards), developing and implementing interventions and reporting information on the results of reviews and interventions. Finally, a national organization is needed to develop the national guidelines and review methods and to coordinate and oversee the activities of the local review entities. By local review entities we mean organizations that are close enough to the local health care community that appropriate recognition of the unique circumstances of the community can be made and that the type of balanced quality assurance system we advocated earlier can be fostered and maintained. The state-level PRO program is one organizational model that approximates this goal. The individual PROS are charged with ensuring that the care provided to Medicare beneficiaries is appropriate and of high quality and, at the same time, with maintaining a positive, cooperative relationship with the provider community. Greater uniformity and effectiveness in review methods, intervention approaches, and reporting of results will be necessary in order to ensure that all patients are receiving an equally high level of quality. However, moving toward greater uniformity is not meant to imply that all reviews must be identical. • General Accounting Office, Medicare, 3. 1 General Accounting Office, DOD Health Care: Occurrence Screen Program Undergoing Changes but Weaknesses Still Exist, GAO/HRD-89-36 (Washington, D.C.: January 5, 1989). Some flexibility is needed to tailor review methods and interventions to specific situations. For example, generally speaking, reviewing the appropriateness of a hospital length of stay would be reasonable. However, since the Medicare Prospective Payment System reimburses hospitals a set amount regardless of the length of stay, the incentive for hospitals is to release patients earlier rather than later. Therefore, the review of the appropriateness of a hospital discharge under Medicare generally focuses on the possibility that premature discharge has occurred rather than on inappropriate days at the end of the stay. A variety of existing methods of quality assessment could serve as the core of the common review approaches. Reviews could be done prior to care being received (prospective review) that typically focus on the need for particular procedures, the appropriateness of the proposed setting (often the hospital), and the proposed length of stay. The limited information available suggests that these reviews are costeffective. Reviews could be done while the care is being delivered (concurrent review) and would typically focus on the need for continued care but might also address a lack of expected progress or improvement. This type of review tends to be expensive and is often limited to potentially high-cost cases. Reviews could be done after the care is completed (retrospective reviews) that examine the process and outcomes of care based on information contained in the medical record or on the claims form. Reviews based on the medical records are relatively expensive but can address a wide range of appropriateness and quality concerns, including both overuse and underuse. Reviews could be done of aggregate data from either claims or medical records (profiling; small area variation analysis) that focus on identifying providers who differ in one way or another from their peers in their process or outcome of care. These could be used to target both prospective and retrospective reviews. Reviews could be done of prescription drug use, prospectively or retrospectively, that focus on ensuring appropriate use and limiting adverse reactions and also allow the targeting of educational and other interventions for both patients and providers. Similarly, a number of intervention approaches that have been tried could serve as the basis for developing a uniform set of interventions for use by local review entities. One approach is that of undertaking educational interventions aimed at providing the medical community with information on the appropriate uses and costs of various medical services. The evidence on the effectiveness of this approach in changing provider behavior is mixed. Another is to provide feedback of review results to providers, either on individual cases or on aggregate practice patterns. While generally viewed as more effective than simple educational interventions, its usefulness has been limited by the unavailability of comprehensive data across purchasers and settings. Yet another approach consists of restrictions on providers' use of particular services (such as the total number of laboratory tests) or on their practice (such as hospital or operating room privileges). Restrictions have sometimes been met with resistance and often change behavior only as long as they remain in place. One more approach is to offer incentives (such as increased reimbursement, more patients, reduced administrative requirements) for providers to conform to particular standards of medical practice. These are being increasingly used, particularly in managed care organizations such as preferred provider organizations and health maintenance organizations. Last, monetary sanctions can be imposed or providers can be excluded from the program (as in the Medicare program) if they provide poor-quality care and are unwilling or unable to change their practice patterns. Additional development, experimentation, and evaluation of both assessment and intervention techniques will be needed in order to create an effective, comprehensive, national strategy. Finally, even though some flexibility in the implementation of reviews and interventions is necessary, a common set of reporting requirements, and particularly reporting categories, will be needed in order to oversee and evaluate the quality assurance activities at a national level. One of the greatest weaknesses of the current system of quality assurance is that there is no simple way to compare information on quality of care from one program to another or to monitor changes in levels of quality over time. This is another area in which developmental work is needed. A National Organizational Focus-We believe a national organizational focus is required to accomplish the many developmental, implementation, and evaluation tasks needed to set up and operate a comprehen sive, national system of quality assurance. Some of the developmental tasks have been alluded to abovesupporting research on the effectiveness of medical care and developing improved quality assessment and assurance techniques. Others include developing practice guidelines and standards, uniform reporting requirements for both medical data and data on the results of quality reviews, and methods of changing provider behavior, including approaches for fostering internal quality assurance activities. Implementation will require the development and oversight of local review organizations that have the necessary tools and skills in data integration and analysis, quality assessment, and quality assurance. Finally, the national organization will also require considerable expertise in data analysis, evaluation, and management in order to integrate the information coming from the various local review entities into a national picture of health care quality, to evaluate the performance of the local review entities, and to identify areas in which greater attention to quality is needed. The Role of Provider Accreditation and Certification Most of the discussion of quality assurance to this point pertains to the review of care provided to individual patients. However, it is also important to review the credentials, facilities, staff, and administrative procedures of health care providers (so-called "structural" quality assurance) to determine a provider's capability or potential for providing high quality care. While such review cannot ensure that quality care is actually provided, it is important for ensuring that at least the necessary elements for providing quality care exist and that providers without those elements are not allowed to participate. Established accreditation or certification programs exist for hospitals, nursing homes, and many ambulatory care settings. However, one setting in which little review of this type occurs is the individual physician's office. We believe that such review may be particularly important for physicians who do not have hospital admitting privileges and who are not part of a larger medical network through which their care might be scrutinized. For selected physicians in this category, on-site visits might be warranted to ensure that medical records are legible, integrated, and filed; that X-ray and laboratory equipment is properly calibrated, maintained, and used; and that the process of care (as revealed through a review of patients' records) is appropriate and high in quality. The Importance of Consumer Education Expanding access to care may bring some patients into the traditional health care system for the first time. They will need assistance in learning to access the system appropriately, select primary care physicians, and understand the importance of an ongoing relationship with an "accountable" provider. Providers will need assistance in working with these new patients and helping them to use the system wisely. All consumers will need assistance in using the increasingly available information on the appropriateness and quality of care to make prudent choices among providers. Conclusion We believe that a comprehensive national quality assurance strategy is needed in order to ensure that all Americans receive high-quality medical care. A comprehensive national strategy is important for several reasons: (1) to ensure that the treatment of individuals does not depend on how the care is financed; (2) to be able to examine the contents, appropriateness, and outcomes of care, regardless of when and where the care was provided or who paid for it; and (3) to meet the legitimate needs for information on quality of the many different actors in the health care system. APPENDIX I—EXPERT PANEL ON QUALITY ASSURANCE ISSUES Robert Brook, M.D., Sc.D. Deputy Director, Health Program The Rand Corporation John W. Bussman, M.D. Medical Director Oregon Medical Professional Review Organization Robert Keller, M.D. Executive Director Maine Medical Assessment Foundation Kathleen N. Lohr, Ph.D. Senior Professional Associate Institute of Medicine National Academy of Sciences Barbara Matula Director North Carolina Division of Medical Assistance Michael R. McGarvey, M.D. Corporate Vice President, Health Affairs Empire Blue Cross and Blue Shield Leslie Michelson President and Chief Executive Officer R. Heather Palmer, M.B., B.Ch., S.M. Gerald Plotkin, M.D. Medical Director, Medical Groups Division Harvard Community Health Plan Cary Sennett, M.D., M. Phil. Medical Director and Director of Technology Assessment AETNA Life and Casualty Michael Stocker, M.D. Executive Vice President U.S. Health Care Leon Wyszewianski, Ph.D. Department of Health Services Management and Policy The University of Michigan School of Public Health Edward Zalta, M.D. Chairman of the Board and Chief Executive Officer Capp Care APPENDIX II—MAJOR CONTRIBUTORS TO THIS REPORT Program Evaluation and Methodology Division Linda Demlo, Assistant Director for Program Evaluation in Human Services Areas Roger Straw, Project Manager Related GAO Reports Medicare: Improvements Needed in the Identification of Inappropriate Hospital Care (GAO/PEMD-90–7, December 20, 1989). Medicare: Assuring the Quality of Home Health Services (GAO/HRD-90-7, October 10, 1989). VA Health Care: Improvements Needed in Procedures to Assure Physicians Are Qualified (GAO/HRD-89-77, August 22, 1989). Health Care: Initiatives in Hospital Risk Management (GAO/HRD-89-79, July 18, 1989). Prescription Drugs: Information on Selected Drug Utilization Review Systems (GAO/PEMD-89-18, May 24, 1989). DOD Health Care: Occurrence Screen Program Undergoing Changes but Weaknesses Still Exist (GAO/ HRD-89-36, January 5, 1989). Medicare: An Assessment of HCFA's 1988 Hospital Mortality Analyses (GAO/PEMD-89-11BR, December 13, 1988). Medicare PROS: Extreme Variation in Organizational Structure and Activities (GAO/PEMD-89-7FS, November 8, 1988). VA Hospital Care: A Comparison of VA and HCFA Methods for Analyzing Patient Outcomes (GAO/ PEMD-89-29, June 30, 1988). Medicare: Improved Patient Outcome Analyses Could Medicare: Improving Quality of Care Assessment and Medicare: Preliminary Strategies for Assessing Quality of 15, |