In California, these physician groups typically contract with HMO plans to deliver care in what has been described as a three-tier arrangement (the HMO as the first tier, the physician group as the second tier, and member physicians as the third tier) [3, 4]. The groups establish contracts with HMOs, usually for full or partial capitation, and therefore become responsible for utilization of some or all services by the HMO enrollees. The groups receive capitated payments from the HMOs but retain control over how they reimburse their own physicians and other providers (for example, through salary, capitation, or fee-for-service). The locus of control, therefore, lies with the physician group and its physicians, not with the HMOs. Physicians in these groups share the financial risk and will profit from patients enrolled in an HMO only if the cost to the group for use of services is lower than capitated payments. Further, although the HMOs require that the groups perform basic utilization management functions (such as maintenance of a utilization review committee), most decisions about the utilization management process are left to the groups themselves. Accordingly, the financial risk associated with capitation has challenged physicians to develop effective ways to manage their own utilization and costs while maintaining quality of care for capitated patients.

Although the literature has concentrated on the ethical dilemmas that arise when physicians are placed at financial risk [5-7] and when gatekeeping is used [1, 8], physicians practicing in capitated groups are implementing utilization management policies that go beyond gatekeeping. Utilization restrictions have previously been imposed on physicians by external parties, such as third-party payers and health plans, in the form of prospective, concurrent, or retrospective utilization management techniques [9, 10]. Capitated physician groups are adopting many of these same formal management techniques to control their own utilization. We refer to utilization management techniques that are initiated by physicians in response to capitation as "internally imposed utilization management."

Despite the growth of prepaid health care, we know little about these physician-initiated methods used to control utilization. We therefore examined the internally imposed utilization management techniques used by 94 physician groups in California caring for more than 2.9 million capitated patients.

Methods

Top Methods Results Discussion Author & Article Info References

We collected data on utilization management systems by administering two detailed self-report questionnaires to representatives (that is, medical directors or administrators) from a sample of California physician groups. Because the literature contains no descriptions of internally imposed utilization management systems, we based these instruments on a conceptual model of utilization management developed by reviewing literature on externally imposed utilization management; conducting exploratory, semistructured interviews with medical directors and utilization management directors in various managed care settings; and talking with the project's advisory panel, which is composed of medical directors and utilization management directors from eight physician groups (medical group practices and independent practice associations) in California.

Conceptual Model

Our model of internally imposed utilization management centers on five major components: gatekeeping, preauthorization, profiling of utilization patterns, guideline use, and education. The use of primary care providers as gatekeepers is designed to coordinate care and to control the use of specialty referrals and expensive tests by patients. Patients must first visit their primary care provider for evaluation of any medical condition. If further work-up is required, gatekeepers may be able to send patients directly to a specialist or to have a procedure, or they may have to submit preauthorization requests to the physician group. Preauthorization indicates that someone in the group, be it a nurse, doctor, or committee, must review the medical necessity of requests for specialty referrals, tests, and procedures before they are conducted. In general, groups choose the types of services that require preauthorization. For example, a group might decide that a referral to a cardiothoracic surgeon must always be preauthorized but that a gatekeeper can send a patient directly to an obstetrician without preauthorization. Retrospective profiling of physicians' utilization determines patterns of high and low utilization for various services. Such profiles may include the number of patients a physician referred for specialty care in a given time period, the mean number of specialty referrals for his or her colleagues, and possibly a recommendation for change if the particular physician's performance was far below or above the average. The dissemination of specific clinical practice guidelines to physicians represents another strategy for promoting appropriate utilization. Guidelines include recommendations and algorithms (developed by the group or by outside organizations) that specify appropriate use of preventive services such as mammography or specialized tests such as magnetic resonance imaging or that define appropriate management of clinical conditions such as diabetes. Finally, education programs, such as newsletters, conferences, and seminars, are designed to teach physicians to practice cost-effective managed care. In general, preauthorization and profiling involve more direct control and oversight by the group than do guideline use and education.

Questionnaire Design and Study Methods

We developed the questionnaire on the basis of the conceptual model detailed above. The first questionnaire examined group structural and organizational characteristics and was to be completed by the group's medical director or administrator. It contained 79 questions on physician characteristics, number and distribution of patients, reimbursement mechanisms, profitability, competition, and risk-sharing arrangements. The second questionnaire focused on the structure and intensity of utilization management and was to be completed by the physician most knowledgeable about the group's utilization management programs (generally the medical director or utilization management director). This questionnaire contained 80 questions on gatekeeping, preauthorization, profiling, education, guidelines, and quality assurance programs. Because capitated physician groups generally have several managed care contracts and establish utilization management policies that apply to all their capitated patients, questions focused on general utilization management policies rather than on those associated with one particular HMO.

During the winter of 1993, we mailed the two self-administered questionnaires to the medical director of each of the 133 groups that have capitated contracts with one of the largest network-model HMOs in California. Groups not responding after 4 weeks were mailed a second copy of the questionnaires. Members of the advisory panel also called the medical directors of all nonresponding groups to encourage participation.

The study was endorsed by members of the project's advisory panel and by the Unified Medical Group Association, a national association of more than 60 medical group practices that provide managed care in more than 350 separate sites. The Unified Medical Group Association did not provide any funding for the study. Our sources of grant support provided funding for the investigators' salaries and for data collection but did not participate in acquiring, analyzing, or interpreting the data.

Results

Top Methods Results Discussion Author & Article Info References

Seventy-one percent (n = 94) of the groups responded to both surveys. Most responses came from the groups' medical directors (73%), utilization management directors (13%), and administrators (9%). Independent practice associations and medical group practices were evenly represented, and groups with a widely ranging number of capitated patients and practice sites were included in the sample (Table 1). Groups cared for capitated patients an average of 8 years. Forty-six respondents represented independent practice associations, and 48 were from medical group practices. Twenty-one nonrespondents represented independent practice associations, and 18 were from medical group practices. Respondents and nonrespondents did not differ in the median number of primary care physicians (40 compared with 37, respectively).

Overall Utilization Management Methods

All the groups used primary care gatekeeping and preauthorization; 79% retrospectively profiled physicians' utilization patterns; 70% used guidelines; and 69% had instituted some form of managed care education. After the groups rated eight possible factors for their influence on the structure of the group's utilization management strategy, 62% ranked "financial control" as having the greatest influence, whereas 23% believed that "quality of care provided" was most important.

Gatekeeping and Preauthorization

Although all groups used primary care gatekeepers, most groups also required that gatekeepers submit preauthorization requests for most types of referral (Table 2). For example, 66% and 60% of the 94 groups required preauthorization for a referral to a mental health specialist and to an internal medicine subspecialist, respectively. Only a small percentage of the groups (1% to 7%) allowed patients to refer themselves to most types of specialists. The exception was for referrals to optometry, obstetrics, and gynecology. For these specialties, as many as 27% of groups allowed patient self-referral, and primary care gatekeepers had greater liberty to refer without preauthorization. In addition, more groups required preauthorization for expensive tests than for less expensive tests (Table 3). Specialists were also subject to preauthorization requirements Table 4, even for tests and procedures in their own fields.

In all groups, at least one physician performed utilization management (mean, 5.0 physicians), and 93% had at least one nonphysician reviewer (mean, 2.7 nonphysicians). Thirty-six percent of the groups had at least one physician who spent at least 50% of his or her time on utilization management functions. In addition, 89% of the groups indicated that they had a utilization review committee that met, on average, once a week, with a mean of 11 physicians attending. An average of 41% of approvals for preauthorization requests were granted by nonphysician reviewers; 38%, by individual physician reviewers; and 20%, by a utilization management committee. For denied requests, an average of 9% of decisions had been made by nonphysician reviewers; 54%, by individual physician reviewers; and 37%, by a utilization management committee. When the group considered denying a request, 28% of the groups consistently (that is, at least 50% of the time) asked a specialist in the relevant field to formally review the request before rendering a decision.

Groups rarely denied requests for referrals and tests. Seventy-seven percent of groups indicated that they infrequently (< 10% of the time) denied high-cost procedures and tests (cost greater than $500), and 86% infrequently denied low-cost procedures and tests ($200 or less). Patients or providers appealed an average of 17% of denied requests, and the groups reversed an average of 35% of these decisions.

Profiling

Seventy-one percent of the groups profiled utilization rates for ambulatory specialty referrals at least once a year, 69% profiled rates for ambulatory tests and procedures, 66% profiled hospitalization rates, and 44% profiled pharmaceutical use. Among the 74 profiling groups, 72% never financially penalized physicians and 54% never terminated physicians' positions or contracts with the group if utilization patterns were greater than average. Fifty-eight percent of all groups reported utilization profiles to individual physicians. Seventy-five percent of these groups recommended changes for practice based on these profiles. However, 61% of the groups never did basic case-mix adjustment (for age and sex) of patients in the physicians' panels.

Guideline Use and Education Programs

Forty-nine percent of groups had developed their own written practice guidelines for use by their physicians or for the preauthorization process, 21% used guidelines developed by others, and 30% did not use any written guidelines. Fewer than 35% of groups used written guidelines to assist in making preauthorization decisions on expensive medical procedures and tests such as lumbosacral magnetic resonance imaging, angiography, and angioplasty.

Fifty-one percent of groups reported that primary care physicians spend at least 1 day per year on managed care education. The most common types of educational tools used were periodic seminars or lectures on cost-effective practice (56% of groups), written procedures on the care of capitated patients (52%), and managed care orientation for new physicians (45%).

Differences between Independent Practice Associations and Medical Group Practices

Medical group practices had about two thirds as many primary care physicians as independent practice associations (an average of 41 physicians per group compared with 65 physicians per group), but more than twice the mean number of capitated patients (43 130 patients and 18 891 patients, respectively) and twice the mean percentage of capitated patients in each provider's panel (66% and 32%, respectively). In addition, compared with independent practice associations, medical group practices had fewer practice sites (11 sites compared with 35 sites, on average) and had existed an average of three times longer (25 years compared with 7 years).

Medical group practices and independent practice associations also differed in the way they performed a few aspects of utilization management. In particular, independent practice associations were more likely than medical group practices to require that gatekeepers submit preauthorization for specialty referrals to optometrists (53% compared with 31%), gynecologists (65% compared with 31%), and obstetricians (35% compared with 25%). Further, independent practice associations were more likely to require preauthorization for medium-priced tests, such as echocardiograms (87% compared with 58%), and were less likely to provide periodic seminars or lectures focusing on cost-effective practice (44% compared with 69%).

Discussion

Top Methods Results Discussion Author & Article Info References

In an effort to contain use, all groups in our study instituted internally imposed utilization management techniques for capitated patients. This finding suggests that capitation at the group level has influenced physicians to devise their own management systems to contain costs. Groups have implemented utilization management methods such as gatekeeping, preauthorization, profiling, education, and guidelines. Education programs and use of guidelines, methods that entail less direct oversight by practicing physicians, were used less frequently than preauthorization and profiling. Independent practice associations may more strongly emphasize preauthorization than do medical group practices.

Most groups asked both primary care physicians and specialists to obtain preauthorization for many specialty referrals, procedures, and tests; patients were limited in referring themselves to specialists. Groups devoted a substantial amount of physician and personnel time to performing utilization management functions. On average, physicians were involved in more than 50% of approved decisions and more than 90% of denied requests. Despite these preauthorization mechanisms, most groups denied requests less than 10% of the time. This low rate of denial may reflect a sentinel effect in that physicians know which services will be approved and thus primarily submit appropriate requests.

Some utilization management methods appeared to be in an early stage of development. For example, fewer than 35% of the groups used specific written practice guidelines to help them make utilization management decisions, even for expensive tests likely to require preauthorization. These groups relied instead on implicit, case-by-case, clinical decision making. Although this practice may allow more flexibility, preauthorization decisions made without reliance on written guidelines may result in different approval criteria being used for patients with similar clinical conditions. Similarly, the finding that 61% of groups that provided physicians with utilization profiles did not adjust physicians' profiles for case-mix calls into question the ability of such profiles to equitably reflect an individual physician's utilization. Nonetheless, groups sometimes made recommendations on the basis of these profiles.

Our study has two main limitations. First, these results are based on self-reports from medical directors and administrators. Second, the results are not necessarily generalizable to other groups or to practice settings outside of California. However, the groups included in our study had a broad range of practice arrangements and together cared for more than 2.9 million capitated patients.

Internally imposed utilization management is fundamentally changing the way physicians practice medicine. As managed care expands, physicians in other states will probably face the challenge of capitation and will respond with utilization management strategies. The growth of capitation may give physicians greater control over utilization management policies and result in less reliance on externally imposed management. Such a transition may have positive consequences on practice if physicians sense that they have greater control over decision making and have confidence in the validity of the utilization management methods. In a recent position paper, the American College of Physicians called for reform of medical oversight (utilization review and profiling), which they viewed as currently being conducted by external entities [11]. Internally imposed utilization management represents a physician-directed approach for practicing medicine when physicians, in addition to being responsible for caring for an enrolled population, also have fiscal control. Future studies should examine the influence of this growing trend in medical practice on quality of care, costs, health outcomes, and physician and patient satisfaction.

Drs. Hays and Brook: RAND, 1700 Main Street, Santa Monica, CA 90407-2138.

Dr. Mittman: Center for the Study of Healthcare Provider Behavior, Veterans Affairs Medical Center (152), 16111 Plummer Street, Sepulveda, CA 91343.

Dr. Siu: Mount Sinai Medical Center, 1 Gustave Levy Place, Box 1077, New York, NY 10029.

Dr. Leake: Department of Medicine, University of California, Los Angeles, School of Medicine, B-261 Louis Factor Building, 10833 Le Conte Avenue, Los Angeles, CA 90095-1736.