The economic forces brought to bear in the health care marketplace by the recent growth in managed care have led to intense pressure on provider organizations to shift away from the current, prevalent pattern of specialist-based care [1] to one in which generalists provide longitudinal primary care and act as "gatekeepers" for access to specialty services [2]. Such a strategy, based almost entirely on the poorly documented notion that generalist-directed care is less costly than specialist-based care, has profound implications for government policies aimed at affecting physician mix; for academic medical centers [3] where medical training occurs; for physicians facing an uncertain demand for their services; and, ultimately, for public health. There is now an almost universal call for more generalists and fewer specialists, and proposals for answering this call abound [4], even in the face of a recent consumer backlash against the gatekeeper model of health care [5]. Few of these proposals have considered the quality of care issues that such a shift may engender. In fact, evidence drawn from disparate fields suggests that for selected conditions, specialists provide better care than generalists do.

Borowsky and colleagues [6] found that noncardiologists were less likely than cardiologists to refer their patients for "clinically necessary" coronary angiography, according to criteria developed by a diverse expert panel. Ayanian and coworkers [7] reported that cardiologists were significantly more likely than internists and family practitioners to prescribe therapies of proven efficacy for patients with acute myocardial infarction. Two population-based studies, one based on data from the Pennsylvania Health Care Cost Containment Council [8] and the other based on data from the Health Care Financing Administration [9], reported lower risk-adjusted mortality rates for patients with myocardial infarction who were treated by cardiologists instead of general internists. Cardiologists also have more accurate perceptions about cardiovascular risk and the health benefits of preventive services than internists and family practitioners do [10].

Recent reports have suggested that patients with acute ischemic stroke fared better if they were cared for by a neurologist [11]. Patients seen by AIDS specialists generally received zidovudine earlier than those who did not have consultation with specialists [12]. In this issue, Solomon and colleagues [13] review the literature comparing outcomes of patients with various musculoskeletal and rheumatic conditions who are treated by different kinds of providers, including generalists, orthopedic surgeons, rheumatologists, and chiropractors. They report that patients who have what is arguably the most complex of the disorders studied-rheumatoid arthritis-had better functional outcomes if they received regular care from a rheumatologist. Given the broad range of conditions, populations, and outcomes reviewed, it is not surprising that Solomon and colleagues found no consistent pattern of superior outcomes by provider group across all conditions. Their contribution, however, is more fundamental. The criteria they present for evaluating the literature on provider-based differences in outcome can be applied to any set of conditions.

The study of different outcomes by provider is a young field. Although "a clear understanding of the role of specialists and PCPs [primary care providers] ... requires scientifically based information regarding the relationship between training and experience and quality of care ... there is astonishingly little information about ... the most effective" use of specialists and primary care providers [14]. In addition, the message that can be read from the available information is often clouded by methodologic shortcomings that may be unavoidable in nonrandomized, observational studies, especially those based primarily on administrative databases, which often lack sufficient clinical detail to allow appropriate risk adjustment across distinct cohorts [15].

The criteria outlined by Solomon and colleagues for evaluating studies of provider differences closely parallel the guidelines of Naylor and Guyatt [16]. Wider application of these criteria will help elevate and clarify the ongoing discourse on how best to use the talents of generalists and specialists. We also agree completely with the authors' conclusions that more resources should be devoted to this question, especially in light of the considerable methodologic challenges that such studies face. It is also important to note that almost all of the studies comparing the relative quality of care provided by generalists and specialists have, for compelling reasons, focused on relatively short-term outcomes associated with particular medical conditions, whereas the essence of effective primary care is the attainment of good long-term outcomes in an unselected population. There is, therefore, a subtle but important bias that tends to undervalue the contribution of primary care to good health outcomes; this must be accounted for in future research.

But even if we had better-or even complete-knowledge of the relative effectiveness of specialist and generalist care for a wide range of conditions, we still would not have a blueprint for applying that information to health care delivery systems. As Flood [17] details in her review of the effect of organizational and managerial factors on the quality of care, identifying or even assembling talent is necessary but hardly sufficient for delivering effective, high-quality care. The sports team alluded to by the Vice President for Managed Care couldn't win a game without a coaching staff, a game plan, and teamwork; specialists and generalists need more than a list of conditions that they treat "better" than the other. They need to be supported in their joint efforts to achieve good health outcomes for patients in a collaborative environment that facilitates the sharing of information, responsibility, and expertise. Evidence from intensive care units demonstrates the importance of teamwork in promoting good clinical outcomes [18]. The feared and resisted "managed care revolution" may, in fact, provide the critical stimulus to encourage primary care and specialist physicians to organize together into administratively nimble organizations with modern information systems to facilitate collaborative care, outcomes tracking, improvement of clinical quality, and removal of parochial financial incentives from the decisions surrounding "who does what."

Focusing only on the team salary ignores the fact that the point of fielding a team is to win games. The primary goal of health care systems should not be to reduce costs but to deliver high-quality care within economic constraints. The best way to do that is to identify the optimal mix of providers on the basis of health needs and measurable outcomes and create the structure necessary to facilitate high-quality care.

David B. Nash, MD, MBA

Thomas Jefferson University Hospital; Philadelphia, PA 19107

Ira S. Nash, MD

The Mount Sinai Medical Center; New York, NY 10029

Dr. I. Nash: The Cardiovascular Institute, The Mount Sinai Medical Center, Box 1030, One Gustave L. Levy Place, New York, NY 10029.