Make up your mind how many doctors the community needs to keep it well. Do not register more or less than this number.

G. B. Shaw

Preface to The Doctor's Dilemma

Health-care delivery in the United States is changing from a fee-for-service system, which required large numbers of specialists, to a system acutely sensitive to competitive market forces and in which financial incentives are created to reduce the provision of specialty care and, therefore, the need for medical subspecialists. In addition, there is a societal perception that this country has far too many specialists and too few primary-care physicians. These forces are transforming the focus of resident education from one of hospital-based specialty care toward more ambulatory and primary-care education. [1] The pending legislative changes in Congress as part of restructuring Medicare will exacerbate this trend. This change in health-care delivery has major ramifications in all aspects of health care, none more than academic medical centers, [2] where much anesthesiology graduate medical education (GME) resides.

The United States, unlike any other country, has never had a national physician workforce plan nor a health-care reimbursement and medical education financing strategy to address physician workforce needs. [3,4] In the United States, the education of physicians is conducted in 141 schools of allopathic and osteopathic medicine and more than 1,500 teaching hospitals [4] overseen by a "loose confederation" [5] of voluntary agencies. The agencies include academic medical centers, the nation's teaching hospitals, the American Board of Medical Specialties, and the Accreditation Committee for Graduate Medical Education (ACGME). Because of the complexity and lack of clear responsibility and/or authority regarding GME, there is a problem of matching the annual 17,500 U.S. and 6,750 international medical school graduate physicians in training [6] with society's needs for doctors. [7] .

Whereas it has always been the choice of medical students to enter whatever specialty interested them, governmental and various medical groups have now proposed that the number of specialists in training be reduced and that approximately 50% of graduating U.S. medical students become primary-care physicians (Table 1). Specific congressional proposals have been offered to support this initiative, [3,8],* and proposals limit the number of international medical graduates (IMGs) to 5–15% more than the number of U.S. graduates (the present percentage is 25%.)[9] The Council on Graduate Medical Education (COGME), which advises Congress on matters of GME and thereby influences the government's Medicare and Title VII financing policies, has argued strongly for fewer specialists and more generalists. [10] As one of the few links between federal funding and specialty physician production, COGME is important. Governmental GME financing policy has strongly influenced GME, especially the increase in the number of specialists. Medicare and other federal agencies (e.g., Veterans Administration) have encouraged the growth of specialty residents by paying hospitals through direct and indirect patient care reimbursement for specialty services provided by hospital house staff. Medicare, for example, paid an estimated $5.1 billion in 1992 for direct and indirect medical education allowances to hospitals. [4] This amounts to payment currently estimated at $78,000 in direct costs per resident to the hospital. Because public funding has contributed to a relatively high proportion of specialists compared to generalist physicians, it is only a matter of time until the financial incentives for creation of specialists are changed--this means financial disincentives inevitably will be enacted to discourage the education of specialists.

Independent of the relatively slow public and legislative mandates regarding change in GME are the very rapid changes in the health-care marketplace. The formation of health maintenance organizations and other health-care systems to compete with each other for the care of the U.S. population has exposed a growing surplus of U.S. specialists. [5,9,11–13],** Weiner predicted that, by the year 2000, there will be a surplus of 165,000 physicians in general and an astonishing oversupply of 60%(approximately 100,000) of all medical specialists. [13] COGME predicts a surplus of 170,000 specialists in 2010. [10] These predictions are based on the assumption that, in the managed-care era, between 40% and 55% of Americans will be members of managed-care networks, and it has been estimated that health maintenance organizations use approximately one-third fewer specialty physicians to care for their members than the number in other practice systems.*** Managed-care organizations curtail specialized services and employ and contract with fewer specialists; thus, many U.S. medical specialty physicians will not be needed for employment. [10] We believe the corollary is that fewer should be trained. Because the market is a cruel master, the newly trained physician specialists face the prospect of not practicing their specialty or practicing in a manner that results in harsh exploitation, that is, with compensation far less than that of other physicians who joined practice at an earlier time. Either prospect argues for a reduction in the production of anesthesiologists.

Of the 33,500 anesthesiologists in the United States, 21,600 are actively practicing members of the American Society of Anesthesiologists (ASA).**** There are 27,041 certified registered nurse anesthetists (CRNAs), of whom 22,465 are in active practice.***** Thus, approximately 44,000 physicians and physician extenders in the United States are available to render a broad spectrum of anesthesia services. There has been great growth during the past decade in physician-anesthesiologists when hospitals and anesthesia program directors understandably increased anesthesia resident positions during the 1980s, because (1) there were an abundance of eager, qualified applicants who perceived a good job market;(2) residents are less costly to the hospital than CRNAs or staff physicians; and (3) many program directors aspired to establish an all-physician anesthesia workforce in the United States (as it is in many other developed countries). Interestingly, the increase in anesthesiologists contrasts with the constrained growth in other surgical specialties and in the education of CRNAs (Figure 1). Despite this large increase in the number of anesthesiologists, the beneficiaries of this increased workforce were mainly the large hospitals in metropolitan areas. [14] As with other medical specialties, rural and small hospitals attracted virtually none of these highly trained and skilled specialists. Figure 2depicts by state the distribution of anesthesiologists to population and reflects the regional imbalance. Geographic maldistribution of physicians is cited often as an example of failed national physician GME policy. [5] .

Residents enter anesthesia training either through the national resident match program (NRMP) or outside the NRMP by direct contract with a program. As medical students have become knowledgeable about the economic trends, there has been a marked decline in matching residents since 1992 but only a slight reduction in the number of positions offered in the NRMP (Figure 3). This results in a mismatch of supply and demand for anesthesia residency positions and raises the question of relevancy of the NRMP to anesthesiology residency class size. To fill the open NRMP positions, IMGs have increasingly been recruited, and 21% of residents were IMGs in 1994, most of whom were not in the NRMP. The overall number of anesthesiology residents for 20 yr and their medical school origin is in Table 2. It is clear from this table that the number of residency positions plateaued and has begun only a slight reduction. Also obvious from Table 2is the decreasing number of postgraduate year (PGY) 1 participants and an increasing percentage of IMGs. This demonstrates an apparent lack of attractiveness of the specialty to U.S. medical graduates and raises the issue of the perception of the future quality of anesthesia GME programs.

Nothing is more important to a medical specialty than the maintenance of its quality. The challenges for academic anesthesia GME programs are to maintain or improve the quality of its educational programs and simultaneously solve the effects of a reduced residency workforce on the delivery of clinical services in academic medical centers. Although many factors influence the establishment and maintenance of quality in a specialty, it begins with the recruitment and education of the "best" students. It is the responsibility of the faculty and directors of anesthesia training programs to define quality, accept students who meet it, and have ongoing measures to enforce the minimal standards of quality of education.

Definition of quality of medical education is not easy, but quality standards for residencies may be adopted by consensus. As an example, the Physicians Payment Review Commission (PPRC), which advises the U.S. Congress, has lumped quality measures into the following five categories: faculty, facilities, didactic curriculum, clinical experience, and characteristics of residents and graduates.******

These categories lend themselves to the development of objective criteria. Specific quality criteria have been developed by the Anesthesiology Residency Review Committee (RRC), an arm of the ACGME (Table 3). These minimal experience requirements, which go into effect in July 1996, along with board score performance by residents in the programs, could have an effect on the number of RRC-accredited programs sometime after 1996. If programs are to be eliminated, it should be on the basis of these objective measures and not on subjective impressions, reputations, or other soft criteria. Development of externally imposed and voluntary quality criteria will be contentious but necessary to develop an impartial and quality-based approach. Without such standards, external enforcement will take years of litigation if challenged in the courts.

In addition to establishing and maintaining quality, another central principle of anesthesia GME should be to educate the number of anesthesiologists needed by society. To attain this goal requires forecasting societal needs. Physician workforce forecasting is fraught with difficulty, [6,12],******* but failing to do so results in either over- or undersupply of one of this country's most precious resources, its medical-care workforce. We present three approaches of varying sophistication to predict the number of anesthesiologists that will meet society's future needs.

A. The 50% proposal has been widely advanced (Table 1). In this model, 50% of U.S. medical student graduates would be encouraged or mandated to become primary-care physicians. To achieve an approximate 50:50 generalist:specialty physician ratio and have anesthesiology contribute its mathematical share of specialty education reduction, the current number of anesthesia residents of about 5,400 (in the 3-yr continuum) would decline to 2,900 (46%, or 966 per class). [15] The percentage of IMGs in anesthesia training programs would decrease to less or equal to 10% only if mandated by some governing agency--otherwise, the percentage of IMGs will continue to grow if U.S. students are not attracted into anesthesiology. Anesthesiology would not be alone in suffering these drastic reductions, because it is estimated that all specialty programs would be curtailed by 40–50%. [16,17] The 50% proposal is purely hypothetical, and the continued production of more than 900 anesthesiologists per year would create, by the year 2010, a ratio of approximately 11.8 anesthesiologists per 100,000 U.S. population, compared with 9.2 per 100,000 reported in 1994 (Figure 2). [13] .

B. Some managed-care predictions have been published [13,18] that are based on health maintenance organization staffing models primarily in the U.S. western states. In these instances, the number of anesthesiologists required to serve a 100,000 population in market areas with a high percentage of managed-care patients ranges from 3.6 to 9.0. [13] Using the 9.0 per 100,000 ratio, which is a maximum managed-care estimate, only 16 states need anesthesiologists now (Figure 2). To maintain the current national ratio of about 9.2 per 100,000, we should accept (depending on age of retirement) 634–872 residents per year into anesthesia GME. The criticism of the managed-care model is that not all U.S. practice will be managed care in the foreseeable future, and managed care tends to cover segments of the population that do not require as much specialty care as, for example, the older Medicare group. Nevertheless, because we have a ratio of an estimated 9.2 anesthesiologists per 100,000 U.S. citizens in our practice system, still largely fee-for-service, the future needs are likely to be less and, as a consequence, fewer than 600–800 residents per class will be required as the United States moves to greater managed care.

C. A needs-based, practice-sensitive model was sponsored by the ASA and conducted by Abt Corporation.******** Their projections are based on anticipated surgical, obstetric, educational, administrative, pain-management, and intensive-care physician duties in the year 2010. Time estimates for anesthesiologists' work were used and actuarial data computed regarding the U.S. population and anesthesiologists' death, retirement predictions, and number of hours worked per week. All data used to model the physician workforce needs were put within four major practice options,******** because the practice setting (i.e., how physicians work with physician-extenders, such as CRNAs) is the single most important determinant of physician workforce needs:

1. Physician-intensive model, in which 75% of anesthesia procedures involve only a physician, 20% involve a team of anesthesiologists and CRNAs, and 5% involve a CRNA alone;

2. First team model, in which 45% of anesthesia procedures involve only a physician, 45% involve a team of anesthesiologists and CRNAs, and 10% involve only a CRNA;

3. Second team model (which approximates current U.S. practice patterns), in which 25% of anesthesia procedures involve only a physician, 65% involve a team (typically with teams consisting of two CRNAs per physician), and 10% occur in settings with only CRNAs; and

4. CRNA-intensive model, in which 10% of anesthesia procedures involve only a physician, 80% involve a team (typically with teams involving more than two CRNAs per physician), and 10% rely exclusively on CRNAs.

Using this relatively sophisticated methodology, workforce needs were predicted, and the number of anesthesiologists and the number per 100,000 population are presented in Table 4according to practice mode, retirement age of 65 yr, and work week (either 50 or 62 h per week). The number of anesthesiologists required in 2010 ranges from 14,351 to 34,093. Using these predictions, the number of trainees needed per year ranges from 0 to 1,242. The number of residents is presented as a range depending on the amount of residents spending time in nonclinical training, such as research. Needs-based estimates like this suffer from the fact that needs are based on current practice, and current practice will change in the future.

(Table 5) summarizes the three predictive models and contains the authors' recommendation. From our analysis of the data, we believe a reasonable model to adopt is the first team staffing model of Abt, with physicians working 62 h per week and retirement at age 65 yr. We think that, with the influx of physicians into the workforce, it is likely that the first team model will prevail, albeit many anesthesiologists will have lower financial compensation levels than now. Using these assumptions and the Abt methodology, the number of trainees required per year is between 506 and 599. Thus, we believe that 550 residents is an appropriate class size target for the next 10 yr. After 10 yr, another assessment should be made. Coincidentally, this is approximately the number of residents who matched in the 1995 NRMP (Figure 3). This target would require about a 70% reduction (from 1,800 to 550) of the current residency size. The effect of this production of anesthesiologists over time is shown in Figure 4along with other class size forecasts.

To reduce the size of the residency workforce to 550 will require closing programs and reducing class size in programs that remain open. This will demand voluntary and involuntary solutions.

A. Voluntary means of reaching the goals for each year include the following strategies:(1) Allow attrition of residents to go unanswered. Attrition occurs between the PGY2 and PGY3 years and between the PGY3 and PGY4 years. (2) Another method and the most common will involve the voluntary reduction of class size each year of 10–20% for 3 yr. The only way to enforce this method is to publish the number of residents in each program each year and identify non-compliant programs. It has been shown before that professional peer pressure influences program directors and hospitals. This was seen when it was decided by the Society of Academic Anesthesia Chairmen that all anesthesia programs should participate in the NRMP. (3) To attain the desired number of positions, program directors should submit to the NRMP only slightly more positions in anesthesia than 550 per year. (4) A third method would be to agree to accept into residency no more than 10% IMGs, letting U.S. medical students set the size of the class in response to the perceived need of U.S. graduates. This would eliminate IMGs in the NRMP and rely on after-match placement of IMGs but may not be feasible for ethical and legal reasons. One argument prominently used against reducing the number of IMGs is that some hospitals who care for the poor would be compromised in their clinical-care mission if IMGs were not available for care. This has been found to be valid: 77 of 106 hospitals that depend on IMGs for much of the workload serve the uninsured and poor. [19] (4) The voluntary closing of programs would be a welcome contribution to achieving the goal of reduced resident training. If program directors who have very small and/or marginally acceptable quality residency programs elected to close their programs, the national goal could be met without subjecting all programs to across-the-board decreases of greater or equal to 50%. Voluntary program closures would allow concentration of residency education in facilities of academic excellence and in areas of societal need. The largest concentrations of anesthesia residency positions are in New York and Massachusetts, yet this region has the highest number of anesthesiologists per population (Figure 2). It has recently been shown that 51% of all GME graduates practice their specialty in the state in which they trained. [7] If this is true for anesthesiology and if regional imbalances are influenced negatively by the number of specialty providers in that region, then it might be argued that GME programs in New York and Massachusetts should reduce the number of GME positions more than those in other regions of this country.

B. Involuntary measures may be imposed by those outside of anesthesia and outside of organized medicine. The federal government, state authorities, deans, hospital directors, and accrediting bodies (ACGME and/or RRC) are all possible agents of reduction of residency size. Because federal laws have been proposed, it is not difficult, despite the Congressional election of 1994, to believe that policy-makers, with the help of organized medicine, are intent on forcibly reducing the number of specialists. The simplest method will be through withdrawal of funding to hospitals for Medicare payment to support anesthesia residents. COGME also reported that federal dollars should be used to "slow down the growth in the supply of specialty physicians." [10] Congresswoman Nancy Johnson made such a proposal in the current congress.********* In such a system, a hospital director who was previously financially encouraged to hire anesthesia residents would instead oppose it. Table 6lists other recommendations from the PPRC to Congress regarding GME. [20] Another federal mechanism would involve limiting the number of IMGs through issuance of training cards and J-1 visas to a specific number of IMGs. The Educational Commission on Foreign Medical Graduates could be used to decide who qualifies for the training cards. [6] State licensure could be used as a tool to direct physicians into practices for which the state determines a need. It is also likely that the RRC will want to improve the quality of anesthesia programs by requiring that residents be in programs primarily for educational purposes and not service obligations. Numbers of procedures per resident will be scrutinized as well as pass/fail rates on examinations. Some marginal programs will be put on probation, and after all the appeals are finished (years after starting), there will be closure of programs. In addition, the RRC will require prospective approval of increasing the number of residency positions, something heretofore unheard of in anesthesiology but common in other specialties. The PPRC has stated that the RRCs should see to the reduction of specialty resident positions.****** Another governmental solution to achieving the correct apportionment of trainees is a voucher system [21] similar to the "training card" used in Ontario, Canada. In this system, a certain number of specialists are deemed appropriate by the government, and graduating medical students are given vouchers to receive residency education at any approved residency location. Only residents with vouchers are permitted educational opportunities. Finally, Board Certification could be used to ensure that residents come from an approved residency with an approved number of resident positions.

Powerful involuntary market pressures will force program directors to reduce the size of residency classes or, more probably, make U.S. medical students less likely to choose anesthesia as a specialty because an oversupply will diminish the remuneration to those completing residency.********** This happened in the 1995 match, which saw only 44% of CA-1 (PGY2) positions matched, 32% fewer than 1994 (Figure 3). Because there is a perceived relative shortage of primary-care physicians, market forces will make nonspecialty primary-care training and practice more financially attractive. Ironically, it already is argued that too great a switch in the ratio of generalists to specialists will ultimately create a surplus of generalists [1,22]; nevertheless, the short-term result for anesthesia is that fewer medical students are attracted into anesthesiology. These market forces are dramatic and may create an imbalance that many years from now requires another major correction. Market forces led to our oversupply and could produce an undersupply of anesthesiologists if we do not plan appropriately.

Those who do not wish to initiate changes in anesthesia resident workforce numbers argue that involuntary methods (including market forces) are preferable to voluntary reduction in training positions. Failure to voluntarily reduce the residency class size could prove more painful to those involved than aggressively and voluntarily taking this action. If an overcorrection is made on a voluntary scheme, it can voluntarily and quickly be changed; this is not true with involuntary approaches. Neurosurgery [11] and, more recently, cardiology [23] have set an example for all specialties to follow with regard to voluntary reduction of trainees.

A major obstacle to reduction of the resident workforce is the clinical service they provide in academic medical centers. The large residency workforce has helped academic medical centers care for the sick in its hospitals, and to fulfill service needs with fewer residents involves several strategies. The first and most essential is that operating room scheduling be made more efficient. This is advantageous to the hospital, which is financially vulnerable when large numbers of operating rooms are used less than 50–60% of the staffed time. We believe a reasonable goal for operating room use is 80%. If this involves the closing of operating rooms, it should be done. Academic medical centers must become more cost-efficient. [2,10,24] The Conference of Teaching Hospitals understands this strategy and will work with program and operating-room directors toward efficient scheduling and use of operating rooms. Likewise, nonoperating room scheduling of anesthesia services will require modifications so that the length of a normal schedule is evenly distributed throughout the day. The number of resident positions that could be easily reduced by appropriate scheduling is unknown but considerable.

A second major strategy for providing clinical service is to substitute others to perform clinical work currently provided by the anesthesia resident. Some clinical activities provided by anesthesia residents might be performed by registered nurses or technicians. This is done in many nonoperating room settings, e.g., endoscopy, when conscious sedation is monitored by nonanesthesia personnel. Where appropriate, consulting faculty anesthesiologists could supervise nonphysician and non-CRNA personnel in a variety of locations now staffed by residents. Where anesthesia clinical personnel are required, there are several substitution models:(1) all CRNAs or other nonphysician (physician assistants) providers of clinical care;(2) all staff (faculty) physicians;(3) a group of nonfaculty physicians, "clinical associates," who donate their time;(4) fellows (nonresident learners); and (5) a combination of any or all of the foregoing, which seems the most practical. Many other specialties have successfully employed physician assistants and nurse practitioners to substitute for residency labor (Table 7) just as anesthesia programs have employed the CRNA and, to a lesser degree, anesthesia physician assistants. With regard to academic faculty working alone in operating rooms, it should be apparent that funding from clinical dollars for nonoperating room functions, such as research, will be reduced, and faculty, to stay gainfully employed, may need to spend an excessive percentage of their time practicing in operating rooms. Of course, this portends strong implications for the academic future of anesthesia unless mechanisms are developed to see that nonclinical time is appropriately allocated to the faculty.

We recognized the need to reduce the number of residents in our residency in 1992 and charged a faculty task force with computing the size of the residency that would constitute the best educational experience. Included in the analysis were factors such as numbers of relatively scarce cases per resident (e.g., pediatric cases younger than 1 yr and uncomplicated obstetric care) and call as well as clinical obligations in intensive care units and rotations, such as pain clinic. An optimal educational experience was designed in our hospital for ten residents per class (Table 8). This recommendation was proposed to the anesthesia subspecialty division chiefs, approved, and then ratified by the entire faculty. A formal proposal was sent to the hospital chief operating officer in the spring of 1994 to begin an aggressive reduction in the number of residents, implementation of which was contingent on several elements. First and foremost was establishing a rational surgical scheduling system of operating room time, which effected the closing of some operating rooms and lengthened the use of others to 7 PM. Additionally, nonoperating room anesthesia service schedules are strictly controlled, meaning that anesthesia is scheduled for both morning and afternoon sessions rather than the "favored morning for all." An anesthesiologist was appointed to fill the newly created position of Director of Perioperative Services. The major functions of this position are to improve the overall use of operating rooms, recovery areas, and preoperative screening clinics and to align goals and rewards between nursing, surgery, and anesthesia for optimal patient care and use of perioperative personnel and facilities. A hospital commitment was obtained to hire 5 CRNAs (coincident with 25 fewer residents), and a portion of faculty have agreed to perform cases. In reducing the size of the resident workforce, all members of the department are affected, and each has special concerns. The residents have and must insist on clear educational goals and guarantees not to exceed maximum length of days, number of night calls, and duration of experiences in such settings as intensive care units. One inevitable outcome is that certain intensive care units will be staffed with nurses and physician assistants in addition to residents. To successfully reduce the number of residents requires cooperation of all members of the anesthesia department, hospital management, and our colleagues in surgery, medicine, psychiatry, and radiology. The Duke University Hospital clinical chairs have agreed to reduce by 30% all resident positions (to anticipate general financial realities of future GME payments), and thus the relatively large contribution by the anesthesia department makes a welcome and meaningful contribution toward this institutional goal. [25] .

A. Concern regarding violation of antitrust regulations as an argument not to reduce the size of resident classes often is used against change. [5,20] The PPRC position is that, because the process of limiting residency size is "federally sanctioned," the RRC and others who orchestrate the reduction of residency size would not be subject to antitrust action. [20] Many other specialty groups have published workforce surveys and the need for downsizing, none more prominently than cardiology. [23] Nevertheless, organizations, such as the ASA and Society for Academic Anesthesia Chairmen, probably will not mandate resident quotas for fear of legal action under the restraint of trade laws.

B. Replacement costs of the anesthesia resident workforce is real but unmeasured. If funds are not available to support requisite substitute clinical services, particularly in underserved areas, such as inner cities and rural areas, it is certain that clinical services will suffer to the detriment of the health and welfare of society. [19] If GME funds are not part of managed care, the admonition that "the real threat that managed care poses to the academic medical center is financial insolvency" [25] will become reality. All payer funding to support GME is required, including the costs of replacement personnel. We strongly agree with DeBakey that any health-care "reform" that does not recognize the necessity to fund education and research is "self-defeating." [26] .

C. Cultural change is perhaps the most difficult of all barriers to transcend. During the past 20 yr, anesthesia has had unprecedented growth in quality and number of trainees. All who have witnessed this spectacular success have admired the accomplishments of anesthesia GME during this time. It is difficult for many to accept that, with our enormous success in solving the early workforce deficit of anesthesiologists in this country, we now have a new problem: a workforce surplus of anesthesiologists. Nevertheless, to preserve the quality of our GME programs and produce the projected future need, it is essential that we reduce the number of residents being trained.

It is necessary for the continued improvement of anesthesiology that we concentrate on the educational quality of our trainees and training programs and reduce the quantity of residents and programs. Forces are at work that, if ignored, will end the attraction of the brightest and best U.S. medical students into our ranks. Additionally, society and our fellow physicians recognize the need to educate more primary-care physicians and fewer specialists. There is abundant evidence that academic anesthesia can change when challenged to do so. The most prudent course is for voluntary consensus agreement on what constitutes quality programs. Those programs that do not meet minimal quality standards should close. The recruitment of the best medical school graduates into smaller but better anesthesia GME programs must be the single highest priority of our training hospitals. Improvements in the efficient use of our clinical personnel and replacing resident service obligations with other clinicians is required. Emphasis on anesthesia GME must turn from providing service to obtaining education. With these changes, the future of anesthesiology as a vibrant specialty is assured as we enter the next century.

The authors thank the members of the American Society of Anesthesiologists/Society for Academic Anesthesia Chairmen task force who have read, contributed, and assisted in the formulation of this paper but who do not endorse its content. Members of the task force were Ronald F. Albrecht, MD, James E. Cottrell, MD, Roy F. Cucchiara, MD, Simon Gelman, MD, Charles P. Gibbs, MD, Alan W. Grogono, MD, Harvey Shapiro, MD, and Robert Vaughan, MD They also thank Rhonda Comfort and Laraine Tuck, for clerical assistance in the preparation of this manuscript.

*Association of American Medical Colleges: COGME discusses workforce issues and financing policy. Washington Highlights 1995; 6:3–4.

**Greater NY Hospital Association 1994: Graduate medical education reform and teaching hospitals in the New York City area. New York, Greater NY Hospital Association, 1994.

***Whitcomb ME: Physician workforce policy. Chicago, Council on Graduate Medical Education, 1993.

****Johnson G: Personal communication.

*****Zambricki C: Personal communication. 1995.

******Ginsburg PB: Allocating residency positions under graduate medical education reform: The potential of using quality in making decisions. Washington, DC, Physician Payment Review Commission, 1994.

*******Weiner JP: Forecasting physician supply: Recent developments. Health Affairs 1989;Winter:173–9.

********Abt Associates Inc: Forecasting anesthesia manpower needs for the year 2010. Park Ridge, American Society of Anesthesiologists, 1994. Available from the ASA: 520 North Northwest Highway, Park Ridge, Illinois 60068.

*********Johnson NL: Proposal for graduate medical education. Washington, DC, Congress of United States, Committee on Ways and Means. 1995.

**********Anders G: Once hot specialty, anesthesiology cools as insurers scale back. Washington, DC, Wall Street Journal, March 17, 1995, p 1.