The Future of Generalism

Subspecialization's Broad Impact

In the 26 years since the Millis Commission first recommended a clearer focus on generalism as a crucial aspect of national health policy [16], the debate about health manpower has been made even more complex by an increasingly tight association of manpower questions with issues of access, cost, and quality of care. These issues are intertwined to a degree that has made it very difficult to manage them individually. Accordingly, as the federal debate on a national health care system develops in a manner tightly linked to cost control, we are likely to see continuing change in the health manpower mix determined to an important degree at the national level, fueled especially by fiscal and access issues.

In the expansion of subspecialization, internal medicine, traditionally the source of broad, sophisticated care, has had, paradoxically, the greatest number of subspecialties in which certification is offered. Since its organization in 1936, the American Board of Internal Medicine (ABIM), has certified 111 409 physicians in internal medicine and 56 561 in 10 subspecialties, with the greatest numbers in cardiovascular disease and gastroenterology (2; also American Board of Internal Medicine. Personal communication; 1993) Figure 2, and in six areas of special competence. In addition, the proportion of ABIM general certificates followed by subspecialty or special competence certificates is increasing, from 42% before 1981 to 67% in the decade since [2]. The current ABIM array is shown in Table 1.

To an increasing degree this trend is viewed as resulting in care that is fragmented by discipline and site of delivery, overly technology intensive, excessively costly, and inaccessible to large segments of the U.S. population [3, 8, 9]. The linkage among technologic intensivity, costs, and subspecialty practice is indicated, or at least suggested, by the following: While Part B Medicare expenditures increased between 1985 and 1988 at an annual rate of 12.3%, the number of gastrointestinal endoscopies performed increased at a rate of 17.4% annually, more rapidly than any other technology. In fact, these procedures accounted for 65.9% of the Medicare revenues of gastroenterologists in 1988 [12]. In most instances, this degree of technologic intensity has not been shown to enhance patient outcomes, although data are scanty [13, 17].

The cost implications of the expanding physician population are exacerbated by increasing subspecialization in the system. By one projection [18], between 1978 and 1998 the number of subspecialist internists will have increased by 206%, and the number of general internists will have increased by 77%. The derivative problems, as The Council on Graduate Medical Education (COGME) has noted (Council on Graduate Medical Education, 1991. Unpublished data), become especially pressing in a system that values fee-for-service reimbursement, has an oversupply of physicians, has a high percentage of subspecialists, values patients' ability to self-refer to subspecialists, depends on subspecialists to provide significant amounts of primary care, and values technologies that (sub)specialists are uniquely trained to use. Furthermore, COGME projections indicate that the demand for access to primary physician/patient contacts would increase sharply, perhaps by 13% to 15%, if a program of universal health insurance were implemented, and would be unlikely to be met adequately by our present manpower mix.

If current trends toward subspecialization continue, the COGME study offers the following conclusions:

1. Primary care services will be provided increasingly by subspecialists who have little or no education in primary care.

2. Primary care services provided by subspecialists can be expected to cost more. Subspecialists providing primary care are likely to seek consultation more often because of the narrower focus of their education, expertise, and interests. Services provided will increase and result in fragmentation and replication of care as patients shop or are referred from one subspecialist to another.

3. Oversupply of subspecialists would be more costly than oversupply of generalists because subspecialists use advanced technologic procedures and hospital resources more often than do generalists.

4. Subspecialists tend to aggregate in urban areas, a trend likely to aggravate the existing discrepancy between levels of urban and rural physician manpower.

5. Subspecialists providing more generalist services will devote less time to their subspecialty areas, potentially threatening, to some degree, subspecialty competence, and physician satisfaction.

Reorientation of Primary Care

In response to these problems and projections, a reorientation centered powerfully on primary care has been advanced and widely embraced as the most rational approach to the intimately linked problems of quality of care, access, and costs.

The term primary care was introduced by White in 1961 [19] and has since been defined and understood in a number of ways. Definitions have variously emphasized first contact care and integration of the psychosocial aspects of health [16], a patient orientation together with consideration of family and community issues in illness [20], attention to resource allocation and sensitivity to the epidemiologic implications of disease and illness [21], and health promotion, disease prevention, and dealing with the diagnosis of undifferentiated syndromes [15]. The Institute of Medicine report of 1978 [6] emphasized accessibility, comprehensiveness, continuity, and coordination as the key characteristics of primary care, and these parameters have been widely accepted as its central features.

The ambiguity resulting from this array of definitions has considerably muddied the debate about the nature of primary care and what it should or should not offer. Importantly, in most of the published considerations, primary care is equated with generalism and is usually considered to subsume family practice, general internal medicine, and general pediatrics. Efforts to define the levels of clinical content in primary care practice, however, are largely lacking from current descriptions, except for emphasis on the care of ambulatory patients; and the linkages of such a system to the subspecialist in internal medicine and to hospital practice thus remain unclear.

In most constructs, then, primary care is concerned at the clinical level with challenges and responsibilities of undefined degrees of complexitya crucial set of considerations. In relation to general internal medicine, data from The National Ambulatory Medical Care Survey (NAMCS) [22] for 1989 sampled office visits to internists. The NAMCS is a year-long probability sample survey of office-based non-federally employed physicians practicing in the United States. The sample is taken from the records of The American Medical Association and The American Osteopathic Association and consisted in this instance of physicians who classified themselves as internists and whose specialty was confirmed at the time of the NAMCS interview. Most are ABIM certified. Limitations of the study include a small physician base (148 internists with a response rate of 64%) and a sample size of only 2774 patient records. The data, subject to these limitations, indicate a patient population that reflects increased use of physician services by older segments of the population Figure 3 and a wide array of symptoms prompting the visits (Table 2); most of these may reflect either major or minor disease and therefore require considerable expertise if they are to be evaluated efficiently and with precision. Furthermore, neoplasms, endocrine and metabolic disorders, especially diabetes mellitus, and diseases of the cardiovascular or respiratory systems are prominent as causes of the presenting symptoms (Table 3); the specific disorders treated included hypertension, ischemic heart disease, cardiac dysrhythmias, heart failure, and obstructive pulmonary disease, as well as other conditions requiring secondary or tertiary care as they progress (Table 4). An indirect estimate of the disease intensity addressed may be gleaned from tabulation of the drugs most often prescribed at these visits, more than half of which are used to treat major cardiovascular disease (Table 5). Furthermore, data such as these reflect diagnoses already established but fail to capture the often complex diagnostic process involved in generalist functions, as well as the further clinical demands imposed by the presence of multiple simultaneous disorders, a particular characteristic of the elderly patients being seen in increasing numbers.

These characteristics of ambulatory practice require further definition based especially on additional studies of the array and intensity of the syndromes encountered as well as the specific diseases and combinations of diseases that the clinician must manage. In addition, studies of the patterns of use of various clinical procedures would be useful, including diagnosis-specific hospitalization rates. Understanding the extent to which general internists deliver secondary and tertiary care would be enhanced by studies of the frequency and circumstances under which patients hospitalized by general internists for disorders that fall within the province of internal medicine are referred to internal medicine subspecialists for overall in-hospital management, as opposed to referral for technologic procedures or consultative opinions.

Although there is a clear need for further information, the data from the NAMCS as well as previous studies [23] indicate that general internists care for a patient population with serious clinical problems that have considerable potential for major morbidity or death. The medical outcomes study of Greenfield and colleagues [24] suggests that, in contrast, the patients of family practitioners are younger (40.0 versus 46.9 years) and are less sick, as measured by the percentage of patients hospitalized (4.77 versus 5.43), the percentage having tests and procedures performed (40.0 versus 44.2), the number of prescription drugs required per patient (1.40 versus 1.46), and the number of chronic diseases per patient (0.70 versus 1.02). In addition to the NAMCS, in a study of one urban general internist's practice, about 10% of the ambulatory patients seen in a 1-month sample had unusual or particularly complex disorders [23]. Thus, substantial demands for diagnostic expertise and for secondary and tertiary levels of care are integral elements of the practice content of general internal medicine. In this connection it is worth noting that primary care is not the only ambiguously framed level of clinical intervention; secondary care and tertiary care, terms appearing increasingly in the literature, are also ambiguous and should be defined more clearly. Tertiary is meant here to indicate, for example, such disorders as gross gastrointestinal bleeding, fever of unknown origin, congestive heart failure, most pneumonias, and ventilatory failure, but not, in general, intensive care unit management. The internist, like every physician, should be aware of the limitations of his or her capacities so that appropriate use may be made of subspecialty consultation for clinical management and for technologic procedures.

These considerations suggest the following: 1) Primary care does not embrace a homogeneous patient population in terms of the intensity or complexity of the disorders to be managed; in particular, primary care general internal medicine does not appear to be congruent with primary care family practice [24, 25] as it relates to adults. 2) Although primary care is an important element of general medical care, it is an incomplete expression of general internal medicine because it does not encompass the secondary and tertiary care needs of the patient population subsumed. 3) A special attribute of general internal medicine, given the patient population treated, should be a capacity to offer major diagnostic expertise and to render high-quality secondary care and important elements of tertiary care, within the limits of the physician's capacity.

In addition, there is a further consideration: the new concerns about accessible, comprehensive, longitudinally stable health care emerge from a practice climate in which we have increasingly stratified clinical problems in internal medicine along organ- or system-specific (subspecialty) lines. In our search for a more rational design we should be careful not to stratify clinical care just as powerfully, and as counterproductively, along lines dictated by intensity of illness and tied to a structure in which the limits of responsibility of the general internist are defined by his or her primary care activities. In that scenario, we risk expanding the problems we are trying to address by increasing demand for a subspecialist referral network that must then expand in response.

Implementing Reform

Given these complex, intertwined issues, what steps should be taken? Physicians in internal medicine have a particular responsibility to be innovative because this is the specialty with the largest generalist component and the most highly developed subspecialty thrust.

General internal medicine should be redefined with attention to its inclusion of, but noncongruence with, primary care. In the process, accepting again our traditional responsibilities for first contact and ongoing care across the array of syndromes and acute and chronic disorders germane to our field of competence, the redefined general internist should be characterized by a capacity to deliver care in more complex situations; this should be an expectation held by the general internist and reflected in the structure of our educational and training programs.

How and to what extent would this more sophisticated generalist be responsive to problems in the present structure of our system or to the implications of general internal medicine if confined to its primary care responsibilities?

First, the sophisticated generalist would provide care that is more integrated and coherent across multiple levels of intensity. Over time the patient would have, as the chief caregiver through a variety of health vicissitudes, the physician most knowledgeable about his or her clinical and psychosocial health history. Fragmentation of care and of loci of care would thereby be ameliorated.

Second, the costs of care would be reduced by virtue of a reduced reliance by the generalist on subspecialty consultations [26] and associated use of subspecialty-related technologic procedures. In addition, referrals by subspecialists providing general care to other subspecialists, for problems outside their particular areas of competence, would be reduced and so, therefore, would associated costs and fragmentation of care.

Third, to the extent that a revivified general internal medicine assumed responsibility for general care previously provided by subspecialists, quality would probably improve; general internal medicine is a specific area of expertise, requiring ongoing exposure to the full array of problems germane to it. The subspecialist practicing outside his or her area of expertise and interest is at a disadvantage in this regard.

Fourth, as the practice content of general internal medicine is extended to include clinical problems of greater range and complexity, the attractiveness of internal medicine as a whole, and of general internal medicine in particular, to students and trainees is likely to be enhanced, and to some degree the current level of dissatisfaction with internal medicine practice may be mitigated.

Fifth, teaching is likely to be enhanced, both in in-patient and ambulatory care sites, as breadth, coordination, and cohesion in diagnosis and management demonstrate a more rational, patient-oriented, and intellectually attractive model to students and residents, as well as to faculty.

Increasing expertise in primary care and in the care of the sicker patient requires expansion of the internist's capacities at both ends of the scale of disease and illness severity. Simply redesigning the current 3 years of residency training to invest more time in hospital ambulatory care sites, practitioners' offices, community clinics, and chronic care facilities will not suffice. More training time should be devoted to developing the enhanced clinical skills required to diagnose and manage patients with more severe illnesses, in or out of the hospital. This necessity, together with the expanding body of knowledge relevant to internal medicine, requires extension of basic residency training to 4 years. Stein [25] and others [27, 28] have made this suggestion before, with varying recommendations concerning the content of the additional year. In Stein's formulation, the trainee would spend 4 months in selected areas of nonhospital care and 8 months in concentrated training in one or two subspecialty areas, allowing the development of additional expertise; two or even three such rotations would probably be better, providing further training characterized by breadth as well as focused experience. Other variations in training can be envisioned. A major goal of the added year should be to enhance the physician's capacity in secondary and tertiary care across a broad array of clinical problems. The additional year would also permit significant extension of the trainee's ambulatory care experience and, if linked to a sophisticated instructional program, would effectively address many of the deficits in current training programs relating to the primary care responsibilities of the internist, especially meeting the diagnostic challenges presented by new, unfiltered patients and the critical triage functions intrinsic to ambulatory practice. Other needs, such as experience in chronic care and community care facilities and exposure to dermatology, office gynecology, orthopedics, and other fields dealing with common office problems, could be scheduled in the third and fourth years.

A redesign along these lines would be likely to make general internal medicine more attractive as a career choice and would encourage re-emergence and recognition of the broad, specific expertise it subsumes. This schema would also permit us to move toward a reduction in subspecialty fellowships through substitution of fourth year medicine residents in subspecialty services. In this fashion, the service needs of the subspecialties would continue to be met. The costs of this additional year of training should be largely offset by reduction in subspecialty training positions. The stipend and benefit costs of clinical fellows are 63.7% funded from patient revenues; the comparable figure for residents is 78.7% [29], but is likely to be reduced as health care cost containment initiatives lower Medicare payments for graduate medical education.

Redesign of residency training in internal medicine should include examination of objectives and of curricula. We should be careful not to focus too closely on attempts to replicate the content of medical practice. Although it is essential that the physician trained in internal medicine programs possess the basic capacities to recognize and manage the common disorders relevant to the field, it is less clear that he or she must have direct exposure to every potential exigency of practice, that residency training should amount, in effect, to practicing for practice. It is likely to be more productive, and in fact more realistic, if we view the objective of residency training as the acquisition of essential competencies. These include, most importantly, the capacity to obtain a sophisticated, reliable (reproducible) patient health history, to perform a sophisticated, reliable (reproducible) physical examination, to assign appropriate weights to clinical data, and to evaluate and manage clinical information in a logical, sequential, clinically focused manner. Additional important competencies include the capacity to evaluate the literature critically and to integrate information gathered in this way with the trainee's clinical experience. Thoroughness, an important attribute of the good physician, may also be conceived of as a clinical competency; it is certainly one that can and should be learned during residency training, as should development of a capacity to express the human support functions of the physician. Other desirable competencies can, of course, be identified. In this connection it would be wise to determine, to the extent possible, the particular contributions to be sought in each of the training sites to which the resident is exposed. Some things are, in fact, better learned in working with ambulatory patients, some in working with in-patients, and some in working with intensive care unit patients. None of these is restricted in its applicability to the site where it is learned: The objective should be to link training in various sites to the competencies best taught or learned in each of them.

What is required to begin redefining the role of the general internist? Is such an adjustment likely to be generated by the private sector mechanisms, especially academic leaders, in whom the design of residency training programs is vested? The chief determinants of the quality, content, and duration of residency training are the criteria of the specialty boards and of the Accreditation Council for Graduate Medical Education (ACGME) and its residency review committees. Ebert and Ginzberg [30] have noted that this autonomous governance system results in an uncoordinated national medical manpower distribution. Vanselow [31] views these autonomous external bodies as largely incestuous in their composition, in that their membership is restricted almost exclusively to the specialty in question and overlap in their membership is considerable. As a result, decisions on number of trainees, length of training, and program content and specialty/subspecialty distribution are made within the context of what is perceived as best for the specialty, rather than with regard for broader health manpower needs.

Vanselow [31], Petersdorf [2], Ebert and Ginzberg [30], and Scherr and colleagues [32] have concluded that a national commission for graduate medical education will be required to adjust the physician array to meet the national need to produce a less autonomous, more demand-driven system. It seems increasingly probable that some move toward a national manpower policy for medicine is coming; internal medicine is uniquely positioned, by virtue of its size and its influence in the medical governance system, to lead the way. The ABIM, which spearheaded subspecialty development, has not found a way to help adjust the generalist-subspecialist imbalance. In addition, the recently expressed view of The Association of Professors of Medicine [33], that the paramount objective of the medical schools is the discovery and dissemination of new knowledge, with the implication that education at both undergraduate and graduate levels is an additional but secondary responsibility, is not reassuring despite Association of Professors of Medicine support of a goal of having 50% of internal medicine residency graduates enter the practice of general internal medicine. Both groups should adopt a more innovative approach to balance the physician mix.

The ABIM should increase the length of training required to sit for the certifying examination from 3 years to 4. The membership of the ABIM should include more generalists, probably at least half the total number. Individuals intending to pursue subspecialty training might, as Stein [25] has suggested, sit for the certifying examination after 3 years of residency training or, perhaps better, might enter fellowship after 3 years and take the certifying examination after 4 (at the end of the first fellowship year).

The ACGME and the Residency Review Committees should tie new program approvals to the national need by controlling and, at least in the case of internal medicine, reducing the number of subspecialty training slots available, and by establishing guidelines concerning the competencies to be acquired in a 4-year residency program. The recent moratorium on recognition of new subspecialties, declared by the ACGME (ACGME News release, June 1992), is an important step in the right direction.

The Association of Professors of Medicine and the Association of American Medical Colleges should embrace the 4-year training paradigm and commit to a reduction in subspecialty training slots. They should develop competency-oriented training curricula articulated to the training sites required. Expansion of generalist faculty should be sought, and the relevant scholarship and tenure issues addressed with particular attention to what Boyer [34] has called the scholarship of integration of information, the scholarship of application, and the scholarship of teaching, in addition to the scholarship of research. And finally, the academic establishment, at both institutional and national levels, should seek avenues to increased recognition and application of the particular talents and capacities of generalists in internal medicine, through appointment to appropriate administrative and other academic responsibilities.

Obstacles to Reform

What are factors that weigh against redefinition of the general internist and reduction in production of subspecialists? One, clearly, is the powerful subspecialty structure and orientation of our departments of medicine and the dependence of our academic centers, at both the medical school and hospital levels, on clinical practice income and income from the use of hospital beds and facilities. Similar difficulties are likely to be encountered in community hospitals as well, because in many instances they compete with academic centers for patients requiring subspecialty care and procedures. Potential countervailing forces include the availability of fourth-year residents to substitute for subspecialty fellows in relation to clinical care and procedures; the heightened capacity of a cadre of sophisticated general internists to manage the care of the seriously ill; adjustments in graduate medical education cost reimbursement through the Medicare system; and, importantly, progressive readjustment of reimbursement for services of the general internist compared to those provided by subspecialists performing technologic procedures. Such readjustment will come primarily through federal mechanisms and represents a powerful, and probably essential, element in the recruitment of young physicians to general internal medicine careers. Although some movement in this direction has started, considerably more is needed.

Antitrust issues must be addressed by the internal governance structure of medicine and may prove to be thorny. A national commission for graduate medical education might offer a mechanism that would shield private sector elements, such as the specialty boards and the ACGME, from antitrust interventions, especially if such a commission had a mix of public and private sector elements in its composition. The powerful case for change to benefit the public interest must be made. As a new national system for health care financing emerges, specialty distribution may be affected significantly by federal policies. How these will play out in relation to the functioning of the Federal Trade Commission remains to be seen.

It is unlikely that training cost considerations will be prime determinants, at least in relation to financing the residency training slots; loss of clinical practice income in subspecialty divisions is likely to be modest because, in most instances, the clinical services of subspecialty residents do not make a significant financial contribution to the division or department.

Any change will probably be linked to alterations in our national health care system: that major changes are coming no one can doubt. It will be impossible to control health care costs unless the manpower mix in American medicine is changed. The private sector must be involved prominently in each step of the redesign. Recent important movement to encourage the emergence of generalist physicians has come from the ACGME as well as from the Association of Professors of Medicine. Also, the Josiah Macy, Jr. Foundation conference on graduate medical education [35] has recommended a new federal commission with public and private sector membership to address specialty distribution and the allocation of funds to encourage appropriate change. These efforts, combined with a new view of the responsibilities of the general internist, would move the system powerfully in the right direction.

Changes in medical school admissions policies and in curricula in undergraduate and graduate medical education will produce only limited shifts in the direction of generalism in internal medicine, as Levinsky [36] has noted. Changes in the reimbursement pattern, together with re-emergence of the field so that the intellectual content and clinical challenges of general internal medicine are enhanced are more likely to redress the generalist-subspecialist imbalance. To do this, private sector institutions and leaders must join those designing new payment arrangements in the public sector. A revivified general internal medicine offers a strong rational basis for such a joint effort.

Based on The Jeremiah Metzger Lecture, American Clinical and Climatological Association, October 21, 1992.