Long-Term Methodological and Data Implicit criteria provide greater flexibility and allow reviewers to adapt their assessments to specific features of individual cases. The application of medical judgment by professionals, however, may be time-consuming and expensive. An additional disadvantage is that the results are valid and reliable only if the reviewers are both expert and consistent in their application of medical judgment. Explicit review protocols may entail relatively high developmental costs, but once proven valid and reliable, they can be administered far more quickly and economically, usually by nonphysicians. A basic problem with explicit criteria is their relative inflexibility. This can lead either to lack of specificity, making the findings equivocal, or to the inclusion of too many or redundant criteria, creating incentives to provide unnecessary or inappropriate care. The spectrum from implicit to explicit criteria can be illustrated with existing review approaches. At one end of a continuum is work being done by the Rand Corporation. This approach is designed to generate highly detailed explicit criteria to be used in assessing the treatment of specific medical conditions. Research activities include the development of lengthy standardized medical records abstracting forms that organize the information required to assess the appropriateness of medical decisionmaking for each condition under study. Similarly, the Clinical Efficacy Assessment Project of the American College of Physicians has developed explicit guidelines for the appropriate care of patients with selected conditions, including cholecystitis and diabetes. Other approaches also focus on specific types of care or care settings, but have somewhat less detailed explicit criteria. For example, the National Medicare Competition Evaluation (see appendix VI) used a panel of physicians to help select 51 explicit criteria for reviewing "basic care" in ambulatory settings and between 14 and 86 criteria for reviewing four medical conditions (diabetes, hypertension, colo-rectal cancer, and congestive heart failure). Some review approaches include more generic sets of explicit criteria, often automated, that lead reviewers to apply additional, more implicit criteria for evaluating particular contingencies. A project funded by the HHS Office of the Inspector General used a standardized utilization review instrument, the appropriateness evaluation protocol, to help 3 A discussion of the Rand methodology is presented in M. R. Chassin, et al., “Does Inappropriate Use Explain Geographic Variations in the Use of Health Care Services?" Journal of the American Medical Association (November 13, 1987) pp. 2533-37. Long-Term Methodological and Data organize medical record data. Experienced nurse and physician reviewers then applied their own judgment in answering two "yes" or "no" questions for each case: (1) whether the care provided was "in accordance with professionally recognized standards of care," and (2) whether the patient was discharged prematurely. Entirely implicit criteria have been used in studies designed to examine a wide array of cases. One, for example, allowed physician reviewers to use their clinical judgment and experience to evaluate hospital care. The physicians were provided copies of hospital medical records for a random sample of union members and their families and asked to evaluate whether the care provided in each case was excellent, good, fair, or poor. Ratings were compared across hospital services, for physicians with various levels of specialty training, for hospitals with and without medical school affiliations, and the like. Comparative analyses found differences in the ratings on the quality of medical care ratings among facilities and physicians.1 Evaluating the relative effectiveness of these various approaches to process review is a prerequisite of establishing how explicit and implicit criteria should be used in Medicare case reviews. Each PRO is required to adopt written utilization and quality criteria to be used by physician advisors in reviewing cases referred to them by the initial reviewers. HCFA is responsible for determining that PRO review procedures conform to the specifications set out in the contractors' scope of work, and utilization criteria are examined closely by the SuperPRO and HCFA. The selection and refining of quality review criteria, however, is considered to be a professional medical responsibility left to the individual PRO. This ensures that local medical standards and practice can be incorporated into peer review, as Medicare law and regulations require. Despite the need to accommodate local standards when appropriate, case review by physician advisors is a component of a highly structured national review system. The determinations of physician advisors are the last step in a complicated process for assessing whether Medicare services are appropriate, necessary, and of acceptable quality, and in consequence, whether Medicare will pay for them. This makes the consistency and fairness of the system fundamentally important. While establishing explicit criteria for use by PROS is not a function assigned to HCFA, protecting beneficiaries (and the program itself) by ensuring that *This study is discussed in A. Donabedian, Quality Assessment, vol. 3, pp. 186-93. Long-Term Methodological and Data Longitudinal Research peer reviewers use those criteria, protocols, or review methodologies Currently, there are basic limitations in the information that can be gleaned from Medicare administrative data about the quality of care received by beneficiaries. Medicare pays for only about half of the elderly's medical care.5 Thus, even complete and accurate records of Medicare-covered services cannot provide a full picture of beneficiaries' health care experiences or problems. Some services, such as most longterm care, and many routine medical costs, such as physical exams, drugs, eyeglasses and hearing aids, are not covered at all. Furthermore, existing national population surveys such as those conducted by the National Center for Health Statistics, or national medical expenditure surveys conducted by the National Center for Health Services Research and Health Care Technology Assessment typically do not include data from medical records, and may not include sufficient numbers of Medicare beneficiaries to allow analysts to determine the extent or distribution of problems of quality or lack of access to the full range of health services over time. Longitudinal epidemiological data on patients with chronic diseases or impairments could help identify problems related to subacute care. For example, a review of patients' health status and functional abilities in addition to their use of medical services might indicate whether longer hospitalizations, or specific patterns of inpatient or outpatient rehabilitative services were associated with better rates of recovery following specific types of surgery (for example, hip surgery). Gathering information on subacute and institutional care as well as basic inpatient and physician services could also be useful for assessing the interrelationships among Medicare and Medicaid-covered services, including effects of coverage policies on the use, costs, and outcomes of care. Longitudinal studies designed to identify variations in medical procedures and treatment plans are often conducted for targeted samples of 5D. Waldo and H. Lazenby, “Demographic Characteristics and Health Care Use and Expenditures by the Aged in the U.S.: 1977-1984," Health Care Financing Review (Fall 1984), p. 1. Concluding Long-Term Methodological and Data cases. But nationally representative longitudinal data could identify trends in treatment patterns, use of services, and outcomes over time, highlighting problems that could be related to changes in the use of services resulting from modifications in reimbursement methods or the organization of health services. Rather than focusing on specific known problems or the identification of "outlier" providers, studies of nationally representative populations can answer broader questions about trends in health care problems and the quality of medical care provided to the Medicare population as a whole. Comprehensive patient-level data of this type also would aid in evaluating possible differences in quality associated with fee-for-service, versus prepaid, medical care and in assessing the implications for patient outcomes of variations in the use of or availability of medical services in different regions or localities. The costs of developing epidemiological data on the full range of health care used by the Medicare population would vary according to the size of the sample(s), the nature of the data collection (for example, medical records abstraction, patient interviews), whether additional administrative or validation data were obtained, the frequency of data collection, and so forth. Based on the costs of current studies collecting extensive patient-level data, it is likely that a set of epidemiological studies addressing a range of issues such as those discussed here would cost several million dollars per year. The usefulness of these data, both in terms of increased public accountability and for targeting future program changes, could make such expenditures worthwhile. We believe, however, that designing such studies should be carefully linked to a wider plan or program for developing needed information on quality of care in the Medicare program. Addressing the measurement and methods issues discussed here— Long-Term Methodological and Data In our preliminary report, we characterized two separate aspects of this problem. First, while the coordination of Medicare quality review activities is assigned to HCFA, the responsibility for assessing quality of care across settings and for evaluating changes in levels of quality over time has not been clearly assigned to any unit within HCFA. Second, there is no clearly identified “organizational structure" within HHS for developing, coordinating, or disseminating information about either the methods and procedures for quality assessment or their findings. Our subsequent work has increased our concern about where and how this type of work can be done most productively. These issues are discussed in the next chapter. 6U.S. General Accounting Office, Medicare: Preliminary Strategies for Assessing Quality of Care, GAO/PEMD-87-15BR (Washington, D.C.: July 1987). |