735 328 1995 PUBL i i 3% _ Q i v CONGRESS ()l“ 'l‘llli UNI'I‘I‘ZI) S'l‘A’l‘liS (I().'\'(}RIZSSI()N\I, BL'IXII‘I'I‘ OFFICE STUDY SEPTEMBER 1995 Medicare and Graduate Medical Education MEDICARE AND GRADUATE MEDICAL EDUCATION The Congress of the United States Congressional Budget Office For sale by the US Govemment Printing Office Superintendent of Documents, Mail Stop: SSOP, Washington. DC 20402-9328 ISBN 0-16-048286-0 NOTES Numbers in the text and tables may not add up to totals because of rounding. Cover photo ©The Stock Market/Jose Luis Pelaez, 1994. Preface education--the training of resident physicians--in amounts that approach $6 billion annually. Those payments are projected to grow to more than $7.5 billion per year in 2000. Given the current concerns about the federal budget in general and the solvency of Medicare in particular, this is an opportune time to reevaluate Medicare's policy regard— ing graduate medical education (GME). T hrough the Medicare program, the federal government subsidizes graduate medical In response to a request from the House Committee on Ways and Means, this study provides an overview of the changing physician workforce and the effects of Medicare's GME policy on residency training and the teaching hospitals at which such training is based. Since today's residents are tomorrow's fully trained doctors, subsidies for training residents may have important implications for future access to medical care and the cost- effectiveness of that care. The study presents various options for GME policy and ex- plores their implications. In accordance with the Congressional Budget Office's (CBO's) mandate to provide objective, impartial analysis, the study contains no recommendations. James Baumgardner of CBO's Health and Human Resources Division prepared the study under the direction of Joseph Antos and Linda Bilheimer. Susan Labovich provided computer support, and Julia Matson Jacobsen offered valuable research assistance throughout all stages of the study. Outside CBO, Ruth Hanft provided many useful com- ments on an early draft. Within CEO, Richard Fernandez, Nancy Gordon, Corinne Manfredi, Alison Martin, Daniel Mont, Constance Rhind, and Roberton Williams contrib- uted helpful advice and assistance. Leah Mazade edited the study, and Christian Spoor provided editorial and production assistance. Sharon Corbin-Jallow prepared drafts of the manuscript. Kathryn Quattrone prepared the study for publication. June E. O'Neill Director September 1995 [if \1" LL/ K) / “ J K L ONE TWO THREE FOUR APPENDIX Contents SUMMARY ix INTRODUCTION 1 Background 1 The Physician Workforce 2 The Federal Government’s Role in Shaping the Physician Workforce 8 ECONOMIC FACTORS THAT AFFECT THE CHARACTERISTICS OF THE PHYSICIAN WORKFORCE 15 Effects of Medicare GME Subsidies on the Number of Residents 16 The Linkage Between the Markets for Residents and for Fully Trained Physicians 18 Historical Evidence 26 POLICY ISSUES AND QUESTIONS 29 Access to Care 29 Cost—Effectiveness of Health Care Delivery 31 Special Activities of Teaching Hospitals 34 POLICY OPTIONS 37 Modify the Current System of GME Financing 37 Restructure the System of GME Financing 40 End Federal Financing of GME 42 Supplementary Tables 45 vi MEDICARE AND GRADUATE MEDICAL EDUCATION September 1995 TABLES 1. A—2. A—3. A-4. FIGURES 1. GME Payments as 3 Fraction of Total Revenues of Teaching Hospitals, 1993 First-Year and Total Residents by FMG Status, 1988, 1990, and 1993 Distribution of Fully Trained Physicians Engaged in Patient Care, by Specialty, Selected Years, 1965 Through 1993 Fully Trained Physicians Engaged in Patient Care per 1,000 People, by Specialty, Selected Years, 1965 Through 1993 Distribution of Residents by Specialty, Selected Years, 1965 Through 1993 Residents Engaged in Patient Care per 1,000 People, by Specialty, Selected Years, 1965 Through 1993 Distribution of Fully Trained Physicians Engaged in Patient Care, by Specialty, Selected Years, 1965 Through 1993 Fully Trained Physicians Engaged in Patient Care per 1,000 People, by Specialty, Selected Years, 1965 Through 1993 Typical Patterns of Residency Training, by Specialty Distribution of Residents by Specialty, Selected Years, 1965 Through 1993 Distribution Among Teaching Hospitals of Marginal Subsidies per Resident from Medicare, 1993 Residents Engaged in Patient Care per 1,000 People, by Specialty, Selected Years, 1965 Through 1993 Relationship of the Ratio of Physician Income to Opportunity Cost and the Number of Physicians per 1,000 People, 1946 Through 1982 12 17 46 47 48 49 18 19 26 CONTENTS vii BOXES 1. 2. Medicare’s Subsidies for Graduate Medical Education Factors That May Influence the Number and Types of Residents 10 24 Summary medical education (GME)—-the training of medical residents—-in amounts that are ex- pected to exceed $6 billion for fiscal year 1995. The bulk of those subsidies are payments that the Medi- care program makes to teaching hospitals (hospitals with residency training programs). Medicare’s GME payments are projected to grow to more than $7.5 billion in 2000. That annual figure is a substantial amount of money for a controversial program--or really, a pair of programs-—that pay more to a teach- ing hospital the more residents it has. Given current concerns about both the federal budget in general and the solvency of Medicare in particular, this is an op- portune time to reevaluate the arguments for and against those GME payments and to consider various policy options. T he federal government subsidizes graduate Types of Medicare GME Subsidies The Medicare program makes payments for graduate medical education to the approximately 1,200 US. teaching hospitals through two significant programs: direct graduate medical education (DME) payments and the indirect medical education (IME) adjustment, which is sometimes called the indirect teaching ad- justment. DME payments to a teaching hospital are based on Medicare's share of the hospital's inpatient days, the direct costs per resident that the hospital incurred during a year in the mid-19805 (those costs have, in general, been updated for inflation), and the number of residents training at the hospital. In contrast, lME payments to a teaching hospital are calculated as an adjustment to the payments that Medicare makes to a hospital for inpatient services provided to Medicare beneficiaries. In other words, if two otherwise identical hospitals each admit a Medicare beneficiary for the same diagnosis, the teaching hospital will receive a higher payment than the nonteaching hospital. And the percentage add-on increases as the teaching hospital's ratio of residents to beds grows. DME and lME payments result in a substantial marginal subsidy for each hospital resident. The Congressional Budget Office's analysis of data from the Health Care Financing Administration indicates that for hospitals in the middle of the distribution in terms of subsidies per resident, an additional resident means that the hospital receives between $58,000 and $102,000 (in 1993 dollars) more in annual payments. About one-fourth of the teaching hospitals receive more than $102,000 annually for adding another resi- dent, whereas about one-fourth of the hospitals re- ceive less than $58,000 for one more. (For almost all residents who are beyond five years of residency or beyond the minimum number of residency years to be eligible for specialty board certification, the analo- gous range is about $47,000 to $86,000.) Clearly, both the absolute size of the subsidy per resident and the variation in subsidies among hospitals are sub- stantial. Average stipends for residents range between about $30,000 and $38,000 per year depending on the resident’s years of experience. Fringe benefits add an average of 18 percent to the stipend, resulting in an average range of compensation of between about x MEDICARE AND GRADUATE MEDICAL EDUCATION September 1995 $35,000 and $45,000. Thus, the average stipend and fringe benefits of residents are less than the marginal DME and IME subsidies for more than three—fourths ofthe nation’s teaching hospitals. An additional ben- efit to the hospitals from hiring residents is the value of the patient care services that residents provide. Additional costs that hospitals incur can include the time of the teaching staff that is taken up in training residents, payments to the teaching staff, overhead costs related to residency training, and the costs of extra tests or slower services because of the use of residents. Trends in the Number of Residents The number of medical residents in the United States has generally been increasing. The number of resi- dents grew from 83,000 in 1988 to 102,000 in 1993. Because the number of graduates of US. medical schools has remained approximately constant, much of the growth in the number of residents comes from foreign medical school graduates (FMGs) who fill hospital residency positions in this country. Between I988 and 1993, the number of FMGs grew from 14 percent of the total number of residents to 23 percent. Over 60 percent ofthe growth in the number of resi— dents can be attributed to increases in the number of FMGs who are training in the United States. (Previ- ous experience suggests that most FMGs who train here will eventually practice here.) Should the Government Maintain the Current GME Subsidies? Whether the federal government should subsidize GME is a controversial issue. At a theoretical level, neoclassical economic reasoning would lead to ques- tions about the need for government intervention to promote residency training. That line of argument proceeds as follows. Residents are receiving training and experience that will enable them to become com- petent, well-paid physicians. For an economically efficient level of training to result, the individual who will benefit from the training in the form of higher future earnings should pay the costs of it. In the con- text of the residency, a trainee pays for the costs that the hospital incurs by accepting a stipend that is lower in value than the services the resident provides. Whether costs are "direct" or "indirect" is a superfi- cial distinction-—those combined costs can be implic- itly paid by the residents in the form of stipends that fall short of the value of their services. GME subsi- dies, by increasing the apparent value of residents to the hospital beyond their actual value in providing services, encourage more training of residents than would otherwise occur. By that line of argument, valued medical services provided by residents to Medicare beneficiaries should be paid for by Medi- care, but training—related costs should not. The subsidization of the supply of physicians through GME payments by the federal government seems to be poorly coordinated with emerging trends in the marketplace. Many experts believe that the growth of managed care in both the private and government-provided insurance markets will lead to (or perhaps has already led to) a situation in which the demand for physicians, especially non-primary care physicians, falls relative to the supply. Yet the presence of the subsidies encourages larger increases in the number of physicians than would otherwise occur. The arguments in favor of continuing GME pay- ments revolve around the consequences of reducing or eliminating the subsidies. GME subsidies consti- tute a significant portion ofthe revenues of teaching hospitals: for private major teaching hospitals, GME payments are more than 7 percent of total hospital revenues, and for smaller teaching hospitals, GME subsidies are about 2.4 percent of revenues. Signifi- cant reductions in those payments could bring about a drop in the activities and services that teaching hos- pitals provide. Both the number of residents and the years of training of the typical resident might fall compared with what they would otherwise have been. Services might be curtailed in other areas besides the SUMMARY xi training of residents. Research, investment in new technologies, access for Medicare enrollees, medical school resources, and care for indigent people might all be reduced as a consequence of cuts in GME pay- ments. Because resident physicians become the trained doctors of the future, GME policy plays a role in de- termining the future characteristics of the medical workforce and the health care industry in the United States. Whether and, if so, how GME policy should be used to manipulate the shape and size of the future stock of physicians is another contentious issue. Policy Options Options for changing the financing of GME reflect a variety of policy goals including changing the level of federal expenditures, altering the future number of primary care (as opposed to more specialized) doc- tors, encouraging efficient amounts of training for new physicians, providing sufficient access to care for Medicare beneficiaries, encouraging medical re- search and development, and providing for the care of indigent people. Possible options range from modifying payments within the current structure of Medicare GME subsidies to more radical changes. 0 Reduce the [ME subsidy. This option would more closely align payments to teaching hospi- tals for inpatient services provided to Medicare beneficiaries with the costs that are correlated with training residents. However, residency training would probably be curtailed, and other services of teaching hospitals might decline. 0 Alter GME subsidies further in the direction of paying teaching hospitals relatively more for pri- mary care residents. This change would encour- age more primary care training. Whether that shift turned out to be an appropriate direction would depend on the validity of forecasts of a greater demand or need for primary care services. Form an all-payer GME fund. This approach could maintain revenues for teaching hospitals while decreasing the amount of money coming from Medicare. However, it amounts to substi- tuting one form of government—imposed financ- ing for another. Replace the current structure of the subsidies with a voucher system. This option could lead to greater flexibility in the site of residency train- ing, which opponents of the present system have argued is biased toward hospital—based training. The impact of such a sweeping change is highly uncertain. Set quotas on the number and types of residency positions, and use the potential loss of GME funds as an enforcement mechanism. This ap- proach could increase the speed with which the medical workforce adjusts to reach the primary care proportions advocated by some workforce experts. Quotas could, however, lead to a mis- match between the actual workforce and future demands for physicians' services. Replace DME and [ME payments with subsidies targeted more specifically toward the public goods that the government wishes to sponsor. This change would reduce the incentive to in- crease the future supply of physicians but lessen the possibility that the weight of the declines in GME payments would fall on indigent care, re- search, or access to care for Medicare beneficia- ries. Chapter One Introduction physicians are expected to exceed $6 billion for fiscal year 1995.' The bulk of those grad- uate medical education (GME) payments flow di- rectly to teaching hospitals through the Medicare program.2 (Teaching hospitals are hospitals that have residency training programs.) F ederal payments related to training resident Many occupations involve a period of general training-—much of it on the job—-in which the young adult pays for the various costs of that training in the form of low rates of pay or outright tuition payments. What makes medical training unusual is the substan- tial amount of federal funds that subsidize the general post-medical school training of physicians. The vast majority of graduates of U.S. medical schools receive residency training subsidized by federal dollars. Fur- thermore, that subsidization extends to graduates of foreign medical schools-—many of whom are foreign citizens--who fill a significant proportion of resi- dency positions that are supported by federal subsi— dies.3 Questions about the logic behind federal subsidi- zation of residency training and the sheer size of the 1. In this study, the general term "resident" also includes "interns" and "fellows." In the past. "intern" was used to describe individuals who were in their first year oftraining after medical school. The use ofthe term "fellows" varies among specialties. In some fields trainees in subspeeialty programs are called "fellows“; in other fields they are called "residents." 2. Besides the Medicare subsidies for GME, the Department of Veter- ans Affairs and the Department of Defense fund residency pro- grams in federal hospitals. Outlays for those programs were less than $1 billion annually in the early 19905. At an even lower level are Public Health Service grants that assist some primary care train- ing programs. 3. Some publications have replaced the term "foreign medical gradu— ate" with the term "international medical graduate." subsidies have led observers to suggest that the gov- ernment review its current GME policy. Such a re— view appears to be especially relevant in an era of growing concern over both the fraction of the econ- omy's resources that is devoted to medical care and the fiscal strains on the federal budget generally and the Medicare program in particular. Changes in federal policy toward residency train- ing would affect the size and characteristics of the future physician workforce. Thus, GME policy will necessarily affect the future mix, cost, availability, and quality of medical care. A number of private- sector institutions and individuals including doctors, hospitals, insurers, managed care organizations, em- ployers, and consumers will also be affected by GME policy decisions. This study examines trends in the size of the physician workforce and the distribution of specialties within it and describes the economic forces, including federal GME policies, that help to determine the workforce’s shape. The study also presents options for addressing several health care issues through changes in GME policy. Background Some historical background helps explain why fed- eral subsidization of medical residency training was once so readily accepted but now is increasingly questioned. In the early 19605, the medical sector of the US. economy bore little resemblance to the med- ical sector of today. No large, federally sponsored medical insurance programs covered elderly or indi- gent people, and the health care economy as a whole was much smaller. 2 MEDICARE AND GRADUATE MEDICAL EDUCATION September 1995 The mid-19603 was a time of sweeping govern- ment intervention in the health care industry. The Congress enacted legislation creating Medicare and Medicaid, and those programs brought about a large increase in the demand (or willingness to pay) for medical services. Not only did the government take on a substantial role in subsidizing the demand for medical care, but it also provided support on the sup- ply side. That support included payments to medical schools for increasing the size of their classes and an acceptance by the Medicare program that it should pay some share of the costs for residency training. (At that time, many people held the view that an in- crease in the number of physicians was desirable.) As the end of the century approaches, the past 35 years can be characterized as a period of remarkable successes for which the nation has incurred high costs. Elderly people in the United States are cov- ered by Medicare. Particular groups of low-income people are covered by Medicaid. Today, the number of fully trained patient care physicians for every 1,000 people is about 1.8, compared with around 1.1 in the mid-1960s.4 Substantial advances in medical technology have occurred. But costs have also been noteworthy. National health expenditures are now about 14 percent ofthe nation's gross domestic prod- uct--more than double their share of 30 years ago, which stood at 5.9 percent. The fraction of the fed- eral budget that goes toward health care is now al- most 20 percent. A number of questions claim the attention of to- day's health policymakers. One concern is whether too many of the economy's resources are being used to provide medical care. Several issues relate to the supply of physicians' services. How many physicians should be in training? What kinds of physicians, in terms of specialty, should be trained? How can the distribution of physicians among urban, suburban, and rural areas be altered? 4. In this study, "fully trained physician" refers to a physician who is no longer a medical resident, A "patient care" physician is one whose principal activity involves the diagnosis or treatment of pa- tients as opposed to activities like administration or research. The physician counts reponed here are based on American Medical Association counts of doctors of medicine (M.l).s). Schools of osteopathy, whose graduates receive a doctor ot‘osteopathy degree, produce about 5 percent of a more broadly defined physician workforce. Even critics who have different philosophical and political perspectives may conclude that the number of physicians being trained is excessive and that overtraining may be a problem. Arrangements for financing medical residency training contribute to those views. Critics with a free-market perspective argue that if general training is worthwhile, the bene- fits, which will flow to the doctor in the form of higher future income, will outweigh the costs of training, which are paid by the young doctor in the form of low income during residency and difficult working conditions.5 Such critics maintain that GME subsidization distorts market signals by making train- ing appear less costly than in fact it is. According to that line of argument, it follows that too many doc- tors receive too much training. Critics with a planning perspective support tar- gets for the medical workforce based on assumptions about the number of different kinds of specialists that are required to fill the medical needs of the popula- tion. Such critics argue that there are too many doc- tors and that they are too highly specialized. Al- though these critics would reject a free-market ap- proach in the medical care sector, they too would tend to question the merits of current GME policy. GME subsidization encourages a greater supply of physicians than would otherwise result. Further- more, because the Medicare subsidies go exclusively to teaching hospitals, decisions about the site of training may be distorted by that financial incentive. Critics ofcurrent policy contend that such distortions have kept training away from ambulatory, primary care settings and reinforced its presence in more spe- cialized and more technologically costly settings. As a result, current GME policy is not consistent with the goals of a smaller, less specialized physician workforce. The Physician Workforce The characteristics of today's workforce of fully trained physicians represent the cumulative decisions 5. The term "general training" means that the resulting knowledge and skills will continue to be useful to trainees alter they have lefi the residency program and the particular institution at which they were residents. CHAPTER ONE INTRODUCTION 3 of many previous cohorts of medical residents. The future composition of the workforce depends on the makeup of the pool of physicians who are already trained and on the decisions made by current medical residents and future cohorts of residents about the specialties they will pursue. The workforce is also shaped by the characteristics of immigrating physi- cians. The Mix of Physicians in Primary Care Versus Non-Primary Care Specialties The relative balance of physicians in primary care specialties versus non-primary care fields continues to be an area of concern among health policymakers. Although no clear-cut definition of "primary care" exists, characteristics that tend to be associated with the concept ofa primary care physician include being a point of first contact for a patient with the medical care system and providing routine examinations and vaccinations. Another distinction that people some- times make is that primary care physicians are less oriented toward procedures (such as surgical inter— ventions) than are non-primary care physicians. Yet another characteristic that is sometimes associated with primary care specialties is the breadth of patient problems that such physicians handle. Some people might argue that a primary care physician is one who engages in a broad range of activities, whereas a non- primary care doctor is one who treats only a narrow range of illnesses. The above set of plausible characteristics for de- fining primary care is somewhat unsatisfactory be- cause certain physicians will fit some of the attributes but not others. Nonetheless, a widely accepted View is that general practice, family practice, general inter- nal medicine, and general pediatrics are primary care specialties; obstetrics and gynecology might also be- long in the primary care category" Over the past three decades, the general trend has been a decline in the fraction of fully trained physi- 6. Other observers argue that nurse-practitioners and nurse-midwives play an important role in providing primary care services and could play a greater role in the future. See Linda H. Aiken and Marla E. Salmon, "Health Care Workforce Priorities: What Nursing Should Do Now," Inquiry, vol. 31, no. 3 (Fall 1994), pp, 318-329. cians who are in primary care specialties. in 1965, 51 percent of doctors involved in patient care were in the primary care fields of general practice, internal medicine, and pediatrics. Obstetrics and gynecology accounted for a further 6.3 percent of trained doctors. Surgical specialties (including general surgery) ac- counted for 21 percent of the workforce; other non- primary care specialties and subspecialties of internal medicine and pediatrics accounted for 21 percent as well.7 Today, most fully trained physicians do not con- sider a primary care specialty to be their principal activity. The primary care fields of general practice, family practice, general internal medicine, and gen- eral pediatrics account for 34 percent of all fully trained physicians who are involved in patient care. Obstetrics and gynecology accounts for a further 6.5 percent. The remainder of the trained physician workforce is divided as follows: surgical specialties including general surgery, 19 percent; other non- primary care specialties including anesthesiology and radiology among others, 29 percent; and the sub- specialties of internal medicine and pediatrics, 11 percent (see Figure l).8 The fraction of the stock of fully trained physi- cians who actually spend some oftheir practice doing primary care exceeds the 34 percent who categorize themselves as primary care physicians. Physicians are self-categorized based on the specialty in which they spend the greatest amount of time in their prac- tice; however, some physicians also report practice hours spent in specialties other than their principal one. In 1992, 47 percent of physicians reported some practice time in a primary care specialty.9 Physicians in some specialties could make a tran- sition into primary care more readily than physicians in others. Doctors who practice the subspecialties of 7. The specialty counts of fully trained physicians are self-reported and refer to the principal specialty practiced by the doctor. The specialty category was unknown for about 1 percent ofthe doctors. 8, The specialty category is unknown for the remaining | percent of trained physicians. 9. Phillip R. Kletke, "Primary Care Versus Nonprimary Care Physi- cians: A False Dichotomy?" Physician Marketplace Report (Chi- cago: American Medical Association, Center for Health Policy Research), April 1994. 4 MEDICARE AND GRADUATE MEDICAL EDUCATION September 1995 internal medicine or of pediatrics have done resi— dency training in general internal medicine or pediat- rics and then received further education in a sub- specialty area. In principle, those physicians could move into primary care practice because they have already received formal training in a primary care specialty. But whether subspecialists reconfigure their practices toward primary care depends on the future marketplace for health care services and whether the relative returns from primary care prac— tice rise sufficiently to encourage such changes. Many of these subspecialists already spend some of their time practicing primary care. Over 75 percent of subspecialists in internal medicine and about 65 percent of subspecialists in pediatrics report spending some hours of their practice in a primary care spe— cialty. The degree of concern arising over the relative number of primary care physicians may depend on how rigidly one defines a primary care practitioner. As Figure 1 indicates, the fraction of physicians with a primary care specialty as their principal specialty has declined since 1965. If, however, one also in- cludes subspecialists of internal medicine and of pe- diatrics on the grounds that they are potential pri— mary care practitioners, the picture looks somewhat different. The combined fraction of actual and poten- tial primary care practitioners fell from I965 to 1975 but has since remained approximately constant. Moreover, the fraction of doctors who spend some time doing primary care significantly exceeds the fraction who call primary care their principal practice activity. Figure 1. Distribution of Fully Trained Physicians Engaged in Patient Care, by Specialty, Selected Years, 1965 Through 1993 Percent 100- , 1 i l 80 4o , , O 1965 1970 1975 1981 Unknown Other Non-Primary Care Specialties Surgical Specialties Obstetrics and Gynecology Subspecialties of Internal Medicine and of Pediatrics Primary Care 1985 1990 1993 SOURCE: Congressional Budget Office calculations based on data from American Medical Association (AMA), Distribution of Physicians in the U. 8. (Chicago: AMA, 1967 and 1971), Physician Distribution and Medical Licensure in the U.S., 1975 (Chicago: AMA, 1976), and Physician Characteristics and Distribution in the US. (Chicago: AMA, 1982, 1986, 1992, and 1994). NOTES: Data are for fully trained physicians who are actively engaged in patient care; that is, the data exclude residents and those fully trained physicians who are employed in professional activities other than patient care (for example, administration or medical research). The specialties of physicians are self-reported and refer to the principal specialty practiced by the doctor. Osteopaths are not included. Data for 1990 and 1993 are as of January 1. Data for all other years are as of December 31. See Appendix Table A-1. CHAPTER ONE INTRODUCTION 5 Most of the growth in the number of fully trained physicians who are involved in patient care has been concentrated in the non-primary care specialties (see Figure 2). The number of primary care physicians—- strictly confined to the specialties of general or fam— ily practice, general internal medicine, and general pediatrics--per 1,000 people grew only slightly be- tween 1965 and 1993. The larger set of doctors that includes subspecialists who have trained in internal medicine or pediatrics grew somewhat more, from just under 0.6 per 1,000 people in 1965 to 0.8 in 1993. The Geographic Distribution of Physicians Although the number of doctors per 1,000 people has risen substantially over time, a number of geographic locales--in rural and some urban areas--are still char- acterized as having an inadequate supply of physi- cians. An extreme example of an underserved area is a county with no doctor. Since 1976, the fraction of people living in such counties has barely changed. In 1976, 4.3 percent of US. counties did not have an Figure 2. Fully Trained Physicians Engaged in Patient Care per 1,000 People, by Specialty, Selected Years, 1965 Through 1993 Physicians per 1,000 people 2.0 ‘ 1.8 1 1.5 l 1.3 l / 1.0 l 7 0.8 l 0.5 0.3 1965 1970 1975 Other Subspecialtles of Internal Medicine and of Pediatrics * Primary Care 1981 1985 1990 1993 SOURCE: Congressional Budget Office calculations based on data from American Medical Association (AMA), Distribution of Physicians in the U. 8. (Chicago: AMA, 1967 and 1971), Physician Distribution and Medical Licensure in the U. 8., 1975 (Chicago: AMA. 1976), and Physician Characteristics and Distribution in the U. S. (Chicago: AMA, 1982, 1986, 1992, and 1994). NOTES: Data are for fully trained physicians who are actively engaged in patient care; that is, the data exclude residents and those fully trained physicians who are employed in professional activities other than patient care (for example, administration or medical research). The specialties of physicians are self-reported and refer to the principal specialty practiced by the doctor. Osteopaths are not included. The "Other“ category comprises fully trained physicians practicing in the specialties of obstetrics and gynecology, surgery, and other non-primary care fields. Data for 1990 and 1993 are as of January 1. Data for all other years are as of December 31. Data are plotted at the intervals shown. See Appendix Table A-2. 6 MEDICARE AND GRADUATE MEDICAL EDUCATION September 1995 active physician providing patient care; the popula— tion in those counties was 0.23 percent of the total US. population. In 1993, 4.8 percent of US. coun- ties did not have an active physician, and the popula- tion in those counties was 0.22 percent of the 1993 US. population. The Path of New Physicians into the Trained Physician Workforce After graduating from medical school, a new physi- cian typically enters a residency training program centered at one of the approximately 1,200 US. teaching hospitals. The "major" teaching hospitals-- those with resident-to-bed ratios of at least 0.25—-are less than one-fourth of the teaching hospitals but train about two-thirds of the residents. Besides the approximately 15,500 annual graduates of US. medi— cal schools, the cohort of first-year hospital residents includes over 6,500 doctors who attended a non-US. medical school. Some of the latter doctors are US. citizens who received their medical education in an- other country, but about 80 percent of the foreign medical graduates (FMGs) who train as medical resi- dents in the United States are foreign citizens. The fraction of medical residents who are FMGs has climbed from 14 percent in the 1988-1989 training year to 23 percent in 1993-1994. Approximately 70 percent to 75 percent of those FMGs are expected to eventually enter the physician workforce in the United States.10 Residency training ranges from three to about seven years depending on the specialty or subspe- cialty. The primary care fields have shorter training periods; the surgical subspecialties tend to have lon- ger ones (see Figure 3). Because of the sequential nature of decisions to enter some fields, a census of first-year residents is not a clear indicator of that cohort's ultimate distribu- tion among the various categories of medical prac- tice. For example, based on the behavior of the co- hort entering training in 1987, only 36 percent ofthe 10. See Fitzhugh Mullan, Roben M. Politzer, and C. Howard Davis, "Medical Migration and the Physician Workforce: International Medical Graduates and American Medicine," Journal of the Ameri- can Medical Association, vol. 273, no. 19 (May 17, 1995), pp. 1521-1527. residents who spend their first year of residency training in general internal medicine end their train- ing in that specialty. The other 64 percent of such residents finish their post-medical school training in a subspecialty of internal medicine, such as cardiol- ogy, endocrinology, or gastroenterology, or in an- other specialty, such as neurology or ophthalmology. Similarly, only 58 percent of residents who start in general pediatrics end their training as general pedia- tricians. Much of the gap is accounted for by those who decide to enter a subspecialty of pediatrics. In contrast, some fields show a high degree of predict— ability based on the first-year cohort: for example, 97 percent of those who start in family practice resi- dencies finish their residency training as family prac- titioners. Although the fraction of residents training in a primary care specialty or a subspecialty of a primary care specialty appears to have grown since 1965, sev- eral qualifications apply to that apparent trend (see Figure 4). Before 1969, official certification in fam- ily practice did not exist. Today, three—year resi- dency programs in family practice are well estab- lished. Thus, the apparent growth in family or gen- eral practice residents between 1965 and 1980 may reflect the introduction of the three-year family prac- tice option rather than an increased propensity for young doctors to enter primary care. Another point is that because of data limitations, internal medicine and its subspecialties are lumped together. As a re- sult, it is impossible to isolate the fraction of resi- dents who are training in a strictly defined primary care specialty. Several noteworthy patterns in the shares of resi- dents in different fields have emerged over the past 30 years. Between 1981 and 1993, the fraction of residents in family or general practice has declined, but it was almost constant between 1990 and 1993. The proportion of residents in a much more inclusive group of potential primary care physicians (including internal medicine and pediatrics and their subspe- cialties) grew from 41 percent in 1981 to 48 percent in 1993. Recent Cohorts of Trainees The cohort of doctors who began their residency training in 1992 is projected to roughly match the CHAPTER ONE INTRODUCTION 7 Figure 3. Typical Patterns of Residency Training, by Specialty Year of Graduate Medical Education 1 2 3 4 5 6 to 7 Family Practice General Pediatrics General Pediatrics Subspecialties of Pediatrics General Internal Medicine (IM) General Internal Medicine Subspecialties of IM # 7 IM Dermatology IM Neurology Psychiatry Obstetrics and Gynecology General Surgery (GS) General Surgery A PIastic/Colon/Thoracic Surgery GS Neurosurgery A GS Orthopedic Surgery GS Otolaryngology General Surgery Urology Pathology —7 if 7 I Anesthesiology Transitional Ophthalmology Yeara Physical Medicine Radiology SOURCE: Congressional Budget Office calculations based on National Resident Matching Program (NRMP). NRMP Directory (Evanston, Ill: NRMP, various years); American Board of Medical Specialties (ABMS), ABMS Annual Report and Reference Handbook--1994 (Evanston, Ill.: ABMS, 1994); and American Medical Association (AMA), Directory of Graduate Medical Education Programs (Chicago: AMA, various years). NOTE: These paths are only representative of training patterns for the different specialties, and not all fields are shown. In addition, many of the specialties above have subspecialties that entail additional years of training but are not included in this illustrative figure. a. The transitional year is one during which the resident develops basic clinical skills. 8 MEDICARE AND GRADUATE MEDICAL EDUCATION September 1995 Figure 4. Distribution of Residents by Specialty, Selected Years, 1965 Through 1993 Percent 100 60 40* 20 I 1965 1970 1975 1981 1985 Unknown Other Non-Primary Care Specialties i - Surgical Specialties i I ‘ ‘ ' 1 . Obstetrics and Gynecology _ i ~ Pediatrics and Pediatric Subspecialties ..uu Internal Medicine and Internal Medicine Subspecialties General Practice and Family Practice 1988 1 990 1 993 SOURCE: Congressional Budget Office calculations based on data for 1965 through 1985 from American Medical Association (AMA), Distribution of Physicians in the US. (Chicago: AMA, 1967 and 1971), Physician Distribution and Medical Licensure in the U. 8., 1975 (Chicago: AMA, 1976), and Physician Characteristics and Distribution in the US. (Chicago: AMA, 1982 and 1986); and on data for 1988 through 1993 from the American Association of Medical Colleges NOTES: Comparisons between the earlier and later years shown above must be made cautiously because specialties are unknown for a large proportion of residents in the earlier years and the three-year family practice residency was a new option in the 1970s In addition, different data sources were used for the earlier and later years. The “Unknown" category includes residents in a transitional year (a year of basic clinical training). Osteopathic doctors in MD. residency programs are included in the data. Data appear in Appendix Table A-3. current population of fully trained doctors in terms of their distribution among specialties. Ifthe transition patterns of the 1987 cohort of first-year residents are repeated in the 1992 cohort (almost all of whose members will have completed their graduate medical education by 1999), 33 percent will enter the work- force of trained physicians in the primary care fields of either family practice, general internal medicine, or general pediatrics. (In comparison, in 1993, 34 percent of the workforce of fully trained patient care physicians had a principal primary care specialty.) Six percent of the 1992 cohort would be expected to enter obstetrics and gynecology, and the remaining 61 percent would enter non-primary care fields." The projected percentages given above may be faulty to the extent that economic incentives alter the l I. These calculations appear in David A. Kindig and Donald Libby, "How Will Graduate Medical Education Reform Affect Specialties and Geographic Areas?" Journal of the American Medical Associa— tion, vol. 272, no. 1 (July 6, 1994), pp. 37-42. likelihood that a resident who is training in a particu- lar field during a particular residency year will either continue in that field or move to training in another field for the next year. Such patterns are not fixed, and they are likely to respond to changes in the eco- nomic incentives of teaching hospitals or residents. Those alterations could result from modifications in government policy or changes in the private market- place for physicians' services. The Federal Government’s Role in Shaping the Physician Workforce Federal policies may affect the number of physicians and their distribution among specialties by altering the economic incentives faced by both the people and the institutions that are or may be involved in medi— cal training. The people include medical students CHAPTER ONE INTRODUCTION 9 (those trained in both US. and foreign medical schools), medical residents, and potential medical students. The institutions include medical schools and teaching hospitals as well as health maintenance organizations (which have the potential to become more involved in training activities). In the context of graduate medical education for physicians, poli- cies aimed at residency training and policies that af- fect the economic rewards from practice in the differ- ent specialties can influence the size and composition of the physician workforce. Because the magnitude of the effects of various policies on the physician workforce is uncertain, this study confines itself to assessing the direction of those effects. Medicare Payments to Teaching Hospitals The most apparent federal funding of GME comes through the Medicare program. Under current law, the federal government, through Medicare, provides about $6 billion per year in subsidies related to the graduate medical education of physicians. The Medicare subsidy program has two parts: the direct graduate medical education (DME) subsidy and the indirect medical education (IME) teaching adjust- ment (see Box 1). The DME payment is based on the number of residents training at the teaching hospital, the typical costs that hospital incurred for training a resident physician in a period roughly corresponding to fiscal year 1984 (adjusted for inflation), and the Medicare patient load of the hospital. DME pay- ments, which for 1994 amounted to almost $2 billion, are intended to cover Medicare's share of the direct costs--such as residents' stipends (or salaries) and fringe benefits, salaries for teaching personnel, and overhead--of residency training programs.12 The [ME teaching adjustment is a subsidy to teaching hospitals in the form of a percentage add-on to the reimbursement for each Medicare patient ad- mitted to the hospital. The percentage adjustment increases with the resident—to-bed ratio. The adjust- 12, In some contexts, "DME" includes payments to hospitals to support training of nurses and other paramedical personnel. In this study, the term refers only to payments for the training of physicians. The payments for nonphysician training are many times smaller than the DME subsidies for physicians. ment is approximately 7.7 percent for going from a resident-to-bed ratio of zero to 0.1; however, the marginal adjustment falls as that ratio continues to rise (see Box 1 for the exact formula). Thus, a teach- ing hospital that has a resident-to-bed ratio of 0.1 receives an additional payment equal to 7.4 percent of the Medicare diagnosis-related group (DRG) pay- ments to that hospital for inpatient hospital services provided to Medicare beneficiaries.” A hospital with a resident-to-bed ratio of 0.2 receives an additional payment equal to 14.5 percent of Medicare DRG payments to the hospital. These "indirect" medical education payments amount to about $4 billion annu- ally. Supporters of IME payments view them as an attempt to reimburse teaching hospitals more than nonteaching hospitals because the costs of teaching hospitals are generally greater. Advocates cite a number of reasons, not all of which are related to teaching, to justify the payments. The reasons given for teaching hospitals' higher costs include the larger number of tests ordered by residents and other ineffi- ciencies caused by residents' lack of experience, the greater severity of the cases such hospitals treat, the need to be at the forefront of technology and re- search, and the provision of more uncompensated care and care for indigent people. The Size of GME Payments in Total Hospital Revenues GME payments constitute a significant percentage of the revenues of teaching hospitals. For all teaching hospitals, total GME payments account for almost 4 percent of total revenues (see Table 1 on page 12). (Total hospital revenues include patient revenues from all sources as well as donations, income from investments, and governmental appropriations.) How significant GME subsidies are in relation to total hospital revenues varies among different catego- ries of teaching hospitals. Private major teaching 13. The DRG payments are the payments that Medicare makes for in- patient hospital services. Medicare pays a predetermined rate for each inpatient stay based on the patient’s admitting diagnosis. 10 MEDICARE AND GRADUATE MEDICAL EDUCATION September 1995 Box I. Medicare‘s Subsidies for Graduate Medical Education The Medicare program provides teaching hospitals with two kinds of subsidies that are based on the size of their graduate medical education programs. Those payments are called direct graduate medical education (DME) payments and the indirect medical education (IME) adjustment. Direct Graduate Medical Education Payments For its DME payment, a teaching hospital receives an amount equal to the product of three factors: its "Medicare patient load," its adjusted number of full- time-equivalent (FTE) residents, and its allowed amount per resident. o The hospital's Medicare patient load is the frac- tion of its total number of inpatient days that Medicare beneficiaries represent. 0 The adjusted number of FTE residents is cal- culated by considering each resident in an ap- proved training program based at the hospital, calculating the degree to which that resident is in the program full time, and then multiplying by an adjustment weight. The weight equals 1.0 for residents who are in their "initial residency pe- riod" (IRP) and who have not been in training for more than five years.‘ (Residents in geriatric fel- lowships may receive a weight of 1.0 for two ad- ditional years.) Other residents receive a weight of 0.5. Graduates of foreign medical schools must have passed a competency exam to be counted toward DME payments. 0 The allowed per-resident amount differs among hospitals. It is based on the direct graduate med- ical education costs per resident incurred by the hospital in a period roughly corresponding to fis- cal year 1984, increased by 1 percent and up- I. The initial residency period is the minimum number ofyears ol‘rcsidency training to be eligible for board certification in a particular specialty. dated for changes in the consumer price index for urban consumers (CPI-U). For fiscal years 1994 and 1995, only residency positions in primary care and in obstetrics and gynecology receive the CPI-U update. Indirect Medical Education Adjustment The additional amount Medicare pays to a teaching hospital equals the hospital's total Medicare diagnosis- related group (DRG) payments for inpatient services multiplied by a factor that is calculated according to a specific mathematical formula: [ME . DRG payments x 1.89 x [(1.25%)405 . 1]. Under the formula, the hospital’s payments increase with the resident-to-bed ratio (the ratio of the number of FTE residents in approved training programs who work in the hospital to the number of beds). A hospi- tal receives approximately 7.7 percent more in pay- ments for each 0.1 increase in the ratio of residents to beds. The various provisions that reduce the weight of many residents in the calculation of DME payments do not apply to the count of FTEs that is used to cal- culate the indirect adjustment. Other Payments Medicare's payments to hospitals for capital-related costs also include an adjustment that gives larger pay- ments to hospitals that have more residents. Those payments are quite small in comparison with DME and [ME payments and are not discussed in this study. Illustrative Examples The table at right shows the relevant information for calculating the marginal subsidy per resident for three different teaching hospitals. The marginal subsidies CHAPTER ONE INTRODUCTION ll from the DME and IME programs combined appear in columns 7 and 8. Column 7 shows the additional pay- ments for adding another resident who is in the initial residency period but who has not exceeded five years of residency training. Column 8 applies to a resident who either is beyond the initial residency period or has trained for more than five years. Several interesting points emerge from these ac- tual examples. 0 The marginal subsidy drops for residents who are beyond the initial residency period (column 8 versus column 7) because the DME regulations give such residents a weight of 0.5 instead of 1.0. The marginal subsidies fall by less than half be- cause IME payments make no distinction be- tween residents within or beyond the initial resi- dency period and those payments constitute a substantial portion of the marginal subsidies. 0 Hospital 1 has a large marginal subsidy because it has a very large DME per-resident amount, an average Medicare patient load (Medicare days divided by total days), and average diagnosis— related group payments for a teaching hospital. Hospital 2 has a marginal subsidy near the me- dian despite the fact that it has a small DME per— resident amount. The large Medicare patient load of hospital 2, along with its high level of total DRG payments, which are probably the result of its sizable fraction of Medicare patients and the fact that it is a large hospital, helps boost its mar- ginal subsidy. Hospital 3 has a relatively small marginal sub- sidy, even though it has a large DME per-resident amount. A small Medicare patient load and a low level of DRG payments hold down the marginal subsidy for hospital 3. The low level of DRG payments probably reflects both the small Medi- care load and the fact that the hospital is small for a teaching hospital. Additional Additional DME/IME DME/IME Payments Payments DME for Another for Another Per- DRG Resident Resident Medicare Total Resident Payments in the Beyond Resident lnpatient Inpatient Amount (Thousands lRP" the [RP3 Beds FTEs Days Days (Dollars) of dollars) (Dollars) (Dollars) Examples (1) (2) (3) (4) (5) (6) (7) (8) Hospital 1 310 23 35,098 81,214 106,273 23,761 102,092 79,128 Hospital 2 670 7 78,897 150,202 38,737 53,289 80,877 70,704 Hospital 3 145 26 9,288 32,803 95,912 6,550 58,448 44,870 SOURCE: Congressional Budget Office calculations based on data for 1993 from the Health Care Financing Administration. NOTES: Payments are figured on an annual basis and in 1993 dollars. Hospital 1 has a marginal subsidy value for residents in the initial residency period at approximately the 75th percentile of the distribution among teaching hospitals. Hospital 2 has a marginal subsidy value for [RF residents near the median. Hospital 3 has a marginal subsidy value for lRP residents at around the 25th percentile. For residents who have exceeded five years of training (with certain exceptions), the numbers in column 8 rather than column 7 would apply. a. This column represents the sum of additional DME and lME payments. 12 MEDICARE AND GRADUATE MEDICAL EDUCATION September I995 Table 1. GME Payments as a Fraction of Total Revenues of Teaching Hospitals, 1993 (In percent) Percentage DME of All DME IME Plus IME Teaching Category Payments Payments Payments Hospitals All Teaching Hospitals 1.3 2.6 3.9 100 Teaching Statusa Major teaching, public 1.2 3.0 4.2 7 Major teaching, private 2.4 4.8 7.2 15 Other teaching 0.8 1.6 2.4 78 Disproportionate Shareb Large urban 1.6 3.1 4.7 38 Other 0.9 2.2 3.1 26 Non-Disproportionate Share 1.0 2.3 3.3 36 Type of Control Voluntary 1.3 2.7 4.0 81 Proprietary 0.8 1.6 2.5 7 Government 1.1 2.5 3.5 12 Number of Beds 1-100 0.8 1.2 2.0 7 101-200 0.8 1.4 2.2 21 201-400 1.0 2.1 3.1 44 Over 400 1.5 3.2 4.7 29 Urbanization Large urban 1.5 3.0 4.5 60 Urban 0.9 1.9 2.8 35 Rural 0.8 1.8 2.6 5 Geographic Regionc New England 1.7 4.0 5.7 7 Mid-Atlantic 2.1 3.3 5.4 22 South Atlantic 1.1 2.4 3.4 13 East North Central 1.2 2.7 4.0 21 East South Central 0.8 1.9 2.7 5 West North Central 1.0 2.1 3.1 9 West South Central 0.6 1.6 2.2 8 Mountain 0.6 1.6 2.2 4 Pacific 0.7 1.9 2.6 11 SOURCE: Congressional Budget Office calculations based on data from the Health Care Financing Administration (HCFA) for a sample of almost 1,000 teaching hospitals. NOTES: Total hospital revenues include patient revenues from all sources as well as donations. income from investments, and governmental appropriations. GME = graduate medical education; DME = direct graduate medical education; IME = indirect medical education adjustment. a. A "major" teaching hospital is one with a resident-to-bed ratio of 0.25 or greater. b. A disproportionate share hospital is one that qualifies for a payment adjustment because it serves a relatively large volume of low-income patients. 0. The regions noted here are as defined by the Bureau of the Census. CHAPTER ONE INTRODUCTION l3 hospitals obtain more than 7 percent of their revenues from Medicare GME payments. In comparison, pub- lic major teaching hospitals get just over 4 percent of their revenues from such payments, and other teach- ing hospitals receive only about 2.4 percent. Perhaps the difference between major teaching hospitals and other teaching hospitals is not surprising since major teaching hospitals have more residents and therefore receive more in GME payments. Other categories of hospitals for which GME payments constitute at least 4.5 percent of total revenues include disproportionate share teaching hospitals in large urban areas, teach- ing hospitals with more than 400 beds, and teaching hospitals in the New England and mid-Atlantic regions. Policies That Affect the Demand for Medical Care In addition to Medicare payments that are linked to residency training, other federal government policies affect the market for medical residents. Medical resi— dents provide patient care in teaching hospitals; thus, government policies that encourage the purchase of hospitalization insurance boost the demand for hospi- tal services and hence the demand for medical resi- dents. Such policies operate both directly (providing hospital insurance through Medicare and Medicaid) and indirectly (subsidizing the purchase of private hospitalization insurance through the favorable tax treatment of employment—based health insurance pre- miums)” In addition, Medicare and Medicaid poli- cies on fee schedules and coverage affect the rewards to various specialties and, in turn, the number of resi- dents who seek to practice in those specialties. Loan Policies The payment provisions of student loans may also affect the specialty a resident chooses. The longer a resident who has obtained a student loan (for either undergraduate or medical school education) is per- mitted to defer repayment because of being in train- ing, the greater is the incentive to continue the resi- dency period and move into a non-primary care field. Evidence indicates that residents with large debts who were permitted to defer repayment throughout their training were more likely to enter fields with longer residency periods than residents who had smaller debts. In contrast, some residents with large debts who had to begin repaying loans during their residency tended to choose fields with shorter resi- dency training. That evidence suggests that repay- ment provisions may influence the proportion of resi- dents who choose training in the primary care spe- cialties.l5 l4. See Congressional Budget Office, The Tax Treatment of Employment-Based Health Insurance (March 1994). 15. See Gloria J. Bazzoli, "Medical Education Indebtedness: Does It Affect Physician Specialty Choice?" Health Affairs, vol. 4, no. 2 (Summer 1985), pp. 98—104. Chapter Two Economic Factors That Affect the Characteristics of the Physician Workforce variety of economic factors from a number A of sources within the health care sector in- fluence the composition of the physician workforce. The demand for residents' services by teaching hospitals has an effect at the level of the training site. Farther downstream one finds the de- mand for the services of fully trained physicians by consumers. These demands originate with consum- ers, but they also reflect the type of insurance ar- rangement that consumers and their employers have, as well as the coverage and reimbursement policies of government-provided insurance. Residency training is a time of investment for a young physician, who will be able to earn a higher future income as a result of those efforts. The resi- dency period adds to the proficiency of the young physician through hands-on patient care under the supervision of the teaching staff, and it is a require- ment for eligibility for board certification in the vari— ous specialties.‘ Some analysts have argued that be— cause residency training is a form of general training that can be useful elsewhere, employers such as teaching hospitals do not have a financial incentive to absorb the costs of that training. In contrast, the young doctor does have the incentive to make such an investment. In return for training and a stipend (or salary), residents provide patient care services for I. Board certification establishes that an individual has met certain training requirements and passed examinations set out by the respective specialty boards. Board certification is not the same as a legal license to practice medicine. Hospitals may use board certification. among other criteria, in decisions about whether to extend staff privileges. the hospital. Any difference between the value ofa resident's services to the hospital and the stipend (and fringe benefits) that the resident receives can be thought of as a "tuition" payment for the other costs that the hospital incurs in providing training. Those costs include both direct and indirect costs.2 Standard economic analysis suggests that in some sense, residents implicitly pay for the training costs that they impose on the hospital, regardless of the size of graduate medical education subsidies to teaching hospitals. That implicit payment equals the gap between the value of the resident to the hospital, which includes the resident's value in patient care and in GME subsidies, and the stipend (and fringe bene— fits) paid to the resident. If GME subsidies were in- creased, teaching hospitals would seek to hire more residents because their value to the hospital would have risen. Because the increased demand for resi- dents would put pressure on the market for such trainees, residents’ stipends might increase. Never- theless, residents would still pay for their training in a certain sense. The training costs (other than the stipends and fringes) imposed on the hospital (in- cluding its direct costs, such as salaries for the teach- ing staff, and overhead costs, plus any increase in operating costs related to the use of a less experi- enced physician) would be "paid for" by the differ- ence between the resident's value to the hospital (in- 2. See William D. Marder and Douglas E. Hough, "Medical Residency as Investment in Human Capital," Journal ofHuman Resources. vol. 18, no. 1 (Winter 1983), pp. 49-64, 16 MEDICARE AND GRADUATE MEDICAL EDUCATION September 1995 cluding the value of the subsidy) and the stipend (and fringes).3 In addition, if GME subsidies were in- creased, more residents would be training than if the subsidies were smaller. The available evidence suggests that the level of Medicare GME subsidies to teaching hospitals is an important factor influencing the overall number of residents. No rigorous, quantitative analysis of the impact of Medicare GME subsidies per se has yet been conducted, but an econometric study of the de- mand for residents shows that teaching hospitals' be- havior is consistent with economic theory.4 That the- ory predicts that subsidy formulas like those used to calculate direct graduate medical education payments and indirect medical education teaching adjustments will result in greater demand for residents than would otherwise be the case because the GME payments increase the residents’ value to the hospitals. Despite the fact that the number of US. medical school grad- uates has been roughly constant for the past decade, a trend toward an increasing number of residents is apparent. Such a change can be explained in part by a rise in the number of residents in the United States who are graduates of foreign medical schools. An- other source of change may be an increase in the length of residency training for the typical medical school graduate. Effects of Medicare GME Subsidies on the Number of Residents The available evidence suggests that the incentives to teaching hospitals to hire residents are so strong that the total number of residents continues to grow (see Table 2). Teaching hospitals typically get much more in Medicare GME subsidies for hiring another 3. The value of residents to a hospital may go beyond the strictly pecuniary contribution of their services and the GME payments. For example, ifa hospital valued the provision of care to indigent people, then a resident's value to the hospital would take account of the indigent care he or she provided. 4. See Robert H. Lee and Jack Hadley, "The Demand for Residents," Journal ofHeaIth Economics, vol. 4, no. 4 (December 1985), pp. 357-371. resident than they pay in stipends and fringe benefits to the resident. Among teaching hospitals in 1993, the annual subsidies for an additional resident varied widely: from about $58,000 at the 25th percentile of the distribution to about $102,000 at the 75th percen- tile (see Figure 5).5 Residents’ compensation was much lower. In 1993, stipends for third-year resi- dents taken from a sample of teaching hospitals ranged from $30,360 at the 25th percentile to $34,760 at the 75th. Adding fringe benefits--typi- cally 18 percent of the stipend--brings the range of residents‘ total compensation to between $35,800 and $41,000" And those differences between marginal subsidies per resident and compensation per resident do not even account for the value to the hospital of the patient care services that residents provide. The presence of residency training programs also imposes costs on hospitals besides the stipends and fringe benefits of the residents. The teaching staff takes time to instruct and oversee residents, for which they may or may not receive a salary. The hospital may also incur other training-related costs in the form of overhead expenses, additional tests or- dered by residents, or slower provision of services when residents are involved. The net value of a resi- dent to a hospital includes all such benefits and costs: the value of the services residents provide plus the value of Medicare GME subsidies, minus the various costs of having residents (including the costs of the stipend and fringe benefits). Continued Growth of Residency Programs Over the past several years, residency programs have generally grown, and the level of stipends has risen 5. The marginal subsidy figures given in the text are for residents in the "initial residency period" and not beyond five years of residency training (with minor exceptions). For residents who have gone beyond five years of training or beyond the years required for eligibility for board certification in their specialty, the marginal subsidies are smaller. Among teaching hospitals in 1993, the marginal subsidies for such residents were $47,000 at the 25th percentile, $67,000 at the median, and $86,000 at the 75th percentile. 6. See Association of American Medical Colleges (AAMC), Council of Teaching Hospitals: Survey of Housestaff Stipends, Benefits, and Funding, 1993 (Washington, DC: AAMC, October 1993). CHAPTER TWO ECONOMIC FACTORS THAT AFFECT THE CHARACTERISTICS OF THE PHYSICIAN WORKFORCE l7 Table 2. First-Year and Total Residents by FMG Status, 1988, 1990, and 1993 Category 1988 1990 1993 Number of Residents First-Year Residents Non-FMG 17,348 17,634 17,118 FMG Citizen 1,289 1,427 1,152 Noncitizen 2,204 3,636 5,595 Subtotal 3,493 5,063 6,747 Total 20,841 22,697 23,865 All Residents Non-FMG 71,235 75,870 78,578 FMG Citizen 4,329 4,990 5,165 Noncitizen 7,227 10,999 18,592 Subtotal 11,556 15,989 23,757 Total 82,791 91,859 102,335 As a Percentage of All First-Year Residents Non-FMGs 83.2 77.7 71.7 FMGs Citizen 6.2 6.3 4.8 Noncitizen 10.6 16.0 23.4 Subtotal 16.8 22.3 28.3 Total 100.0 100.0 100.0 As a Percentage of All Residents Non-FMGs 86.0 82.6 76.8 FMGs Citizen 5.2 5.4 5.0 Noncitizen 8.7 12.0 18.2 Subtotal 14.0 17.4 23.2 Total 100.0 100.0 100.0 SOURCE: Congressional Budget Office using data from the Graduate Medical Education Tracking Census of the American Association of Medical Colleges. NOTE: FMGs are graduates of foreign medical schools. Osteopathic doctors in MD. residency programs are included in the data. 18 MEDICARE AND GRADUATE MEDICAL EDUCATION September 1995 after adjusting for inflation. A number of factors in- cluding GME subsidies may have played a role in that growth. The number of first-year residents grew by more than 14 percent between 1988 and 1993. The number of residents in all years oftraining grew by more than 23 percent over the same five years. The total number of residents for every 1,000 people in the country generally grew throughout the 1965-1993 period and especially after 1980 (see Fig— ure 6). The number of general or family practice res- idents per 1,000 people remained approximately con- stant after 1985. But the rather inclusive group of potential primary care practitioners continued to grow both in absolute terms and in the number of such physicians per 1,000 people. Foreign Medical Graduates as the Marginal Supply Source Foreign medical graduates constitute an increasing fraction of medical residents. Since the size of the classes graduating from US. medical schools has been essentially constant, ranging between 15,100 and 16,400 since 1980, the principal source of addi- tional residents is FMGs. In 1993, FMGs made up about 28 percent of first-year residents compared with about 17 percent in 1988. The Linkage Between the Markets for Residents and for Fully Trained Physicians Besides the fact that resident physicians become the trained physicians of the future, another important link exists between the market for hospital residents and the market for the services of fully trained doc- tors. Namely, the incomes of trained physicians pro- vide incentives to young doctors that affect the num- ber of new physicians who choose to enter the re- spective specialties. For example, an increase in de- mand for the services of primary care physicians will Figure 5. Distribution Among Teaching Hospitals of Marginal Subsidies per Resident from Medicare, 1993 Initial Residency Period, Not to Exceed Five Years Thousands of Dollars 120 l 100i 1 sol ml mi 0 l 25th Percentile Median 75th Percentile Beyond Initial Residency Period or Beyond Five Years Thousands of Dollars 120 ‘ ‘ 100 l 80 l «l 201 o l fig i' 25th Percentile Median 75th Percentile SOURCE: Congressional Budget Office calculations based on data from the Health Care Financing Administration. NOTES: Marginal subsidy refers to the additional annual direct and indirect graduate medical education payments by Medicare that are made to a teaching hospital for adding another resident. The initial residency period is the minimum number of years required for eligibility for board certification in a particular specialty. ECONOMIC FACTORS THAT AFFECT THE CHARACTERISTICS OF THE PHYSICIAN WORKFORCE 19 CHAPTER TWO Figure 6. Residents Engaged in Patient Care per 1,000 People, by Specialty, Selected Years, 1965 Through 1993 Residents per 1,000 people 0.5 l ' ‘ , I i 0'4 i “ """"" Unknown 0.3 I 1 Other 0.2 Internal Medicine, Pediatrics, and the Subspecialties of Internal Medicine or of Pediatrics 0.1 . 'Family Practice and General Practice 1965 1970 1975 1981 1985 1988 1990 1993 SOURCE: Congressional Budget Office calculations based on data for 1965 through 1985 from American Medical Association (AMA), Distribution of Physicians in the US. (Chicago: AMA, 1967 and 1971), Physician Distribution and Medical Licensure in the U. 8., 1975 (Chicago: AMA, 1976), and Physician Characteristics and Distribution in the U. 8. (Chicago: AMA, 1982 and 1986); and on data for 1988 through 1993 from the American Association of Medical Colleges (AAMC). The dotted line in the figure indicates the shift from the AMA to the AAMC data. NOTES: Comparisons between the earlier and later years shown above must be made cautiously because specialties are unknown for a large proportion of residents in the earlier years and the three-year family practice residency was a new option in the 19705. The “Other” category comprises residents in the specialties of obstetrics and gynecology, surgery, and other non—primary care fields. The “Unknown” category includes residents in a transitional year (a year of basic clinical training). Osteopathic doctors in MD. residency programs may be included in the data. Data are plotted at the intervals shown. See Appendix Table A-4. increase the income of such physicians and also in— crease the number of young doctors who choose primary care residencies. Likewise, the total supply of new physicians (regardless of specialty) depends on the return from the investment in medical educa- tion relative to other potential careers. Certainly, nonfinancial factors, including differences in individ- ual preferences and abilities, affect the number of people who choose to become doctors and who choose various specialties. But as long as some indi- viduals place some weight on financial consider- ations, the number of entrants into a particular field will increase if the expected income in that field rises. This linkage between the residency market and the market for the services of trained physicians im— plies the existence of multiple policy options for reaching the same workforce goals. It also increases the possibility of following contrary policies in the two markets. For example, a policy that reduced GME subsidization of residency positions in non-pri- mary care specialties would by itself encourage an increase in the proportion of primary care residents and ultimately an increase in the proportion of trained primary care practitioners. However, if the Medicare fee schedule was adjusted to increase the relative compensation for surgical services, that pol- icy by itself would encourage relatively greater entry into non-primary care fields. The overall effect of the two policies on the future physician workforce would depend on the relative strength of the counter- incentives. Because the private sector accounts for almost two-thirds of payments to physicians, changes in the 20 MEDICARE AND GRADUATE MEDICAL EDUCATION September I995 patterns of demand in that sector may outweigh pol— icy adjustments by the federal government. Evidence suggests that the increasing prevalence of managed care insurance plans, especially the tightly managed forms of health maintenance organizations (HMOs), may be increasing the relative demand for primary care versus non-primary care physicians. The Market for Resident Physicians In this market, teaching hospitals obtain the labor services of resident physicians, who take on greater responsibility for patient care over the course of their residency training. In return, resident physicians re- ceive training from the teaching staff of the hospital and a stipend (or salary). The stipends of resident physicians are substantially lower than the typical income ofa fully trained physician. In 1993, the av— erage stipend of family practice residents, who typi- cally undergo three years of residency training, was about $31,000; the average net income after practice expenses for fully trained general or family practi- tioners was $1 16,800. Likewise, investment in surgi- cal training opens the way for much higher future earnings. In 1993, the average stipend of general surgery residents, who typically train for five years, was about $33,000, whereas the average income, net of practice expenses, for fully trained general sur- geons was $232,700.7 Demand Factors. The demand for hospital residents in general and for hospital residents in different spe- cialties is shaped by the insurance and demographic characteristics of the patient population, epidemio- logical factors, the equipment and technologies avail- able at hospitals, the reimbursement policies of pri- vate and government payers, the goals of hospitals (which may include such objectives as providing care to indigent people or being at the forefront of new medical innovations), and the subsidies to hospitals for GME. Some of those influences represent private market forces, and others are governmental in nature. 7. The resident stipend figures are based on Association of American Medical Colleges, Council of Teaching Hospitals. The physician income data are from the American Medical Association (AMA), Physician Marketplace Statistics, 1994 (Chicago, [1].: AMA, 1994). The aging of the population is a demographic factor that has increased the demand for hospital ser- vices and thus the demand for hospital residents. The effect of the aging population on demand has been further augmented by the presence of the Medicare program, which provides hospital insurance to almost all of the nation's elderly people. The demand for hospital residents has probably been amplified still more by the formulas used to calculate federal pay— ments to hospitals for residency training (see Box I on page 10). Those formulas yield marginal subsi- dies per resident that exceed $100,000 for a number of hospitals. Supply Factors. The distribution of the supply of resident physicians among specialties responds to the incentives to enter the various fields. Both the salary obtained while a resident and the expected stream of income once one becomes a fully trained physician have an effect on the number of residents who enter training in a particular specialty.8 Other characteris- tics—-the expected hours of work, the number of years in training, and the availability and payment terms of loans--also have some impact. Estimates suggest that a 1 percent increase in the income of fully trained doctors in a particular specialty (after adjusting for hours worked) relative to average earnings in all spe- cialties will increase the fraction of residents training for that specialty by between 0.3 percent‘ and 0.6 per— cent.9 8. Although a given hospital typically pays the same stipend to all residents, regardless of specialty, who are in the same year of residency training, the average stipend will still differ among specialty categories for several reasons. One is that different specialties have different training periods and hospitals pay larger stipends to residents who have completed more years oftraining. Thus, the typical surgical resident receives a larger stipend than the typical family practice resident. In addition, a number of hospitals allow program directors to pay residents additional amounts out of departmental funds. Differences among hospitals in the level of stipends can also be a source of variation in stipends by specialty because hospitals differ in the mix oftheir residency positions. 9. See Niccie L. McKay, "The Economic Determinants of Specialty Choice by Medical Residents," Journal of Health Economics, vol. 9, no. 4 (November 1990), pp. 335—357. CHAPTER TWO ECONOMIC FACTORS THAT AFFECT THE CHARACTERISTICS OF THE PHYSICIAN WORKFORCE 21 The Market for the Services of Trained Physicians The forces that determine the demand for and supply of the services of fully trained physicians determine the incomes of trained physicians. In turn, those in- comes provide a signal to young physicians about the rewards that they can expect in various fields. Thus, the demand and supply forces in the market for trained physicians' services affect the distribution of new residents among the various specialties or among aggregate groups of specialties such as pri- mary care versus non-primary care. Grouping specialties as either primary care or non-primary care is convenient, but the demand for particular specialties within each subset may move in different ways. For example, technical innovations and increases in enrollment in managed care plans may reduce the demand for inpatient care. As a re- sult, physicians in specialties like anesthesiology may experience a reduction in the demand for their services and hence a reduction in their income, whereas specialties that have adopted new outpatient surgical techniques, like orthopedic surgery, may see increased demand for their services. Both of those specialties are within the non-primary care subset, yet the demands for their services may move in oppo- site directions. Demand Factors. The demand for the services of fully trained physicians is shaped by the amount and type of insurance coverage of the population and its demographic characteristics, epidemiological factors, available technology, and the reimbursement and coverage parameters of both private and government insurers. An example ofa demographic influence on the demand for certain types of physicians' services would be a decrease in the population of women of child-bearing age. That decrease would lessen the demand for the services of obstetricians and ulti- mately reduce the demand for primary care practi- tioners who dealt with children, such as pediatricians. Over time, one would see a relative decline in the number of such practitioners as fewer residents chose such specialties and older practitioners retired. An example of a governmental influence on the demand for certain types of physicians' services would be an expansion of Medicaid coverage for pregnant women and children. Unless reimburse- ment levels were set so low that no providers were willing to serve the newly covered beneficiaries, such an expansion of coverage would increase the demand for the services of obstetrician/gynecologists and of pediatricians. Although women and children who became insured as a result of the Medicaid expansion would obtain more of those kinds of services than they would otherwise, their competing demands for services would make access to such practitioners more difficult for other consumers. One would also see a strengthening of fees and income for those practitioners. Over time, those impacts would dwin- dle as relatively more residents received training in and ultimately entered the relevant specialties. Implications of the Growth of Managed Care. The market for private health insurance is experienc- ing a rapid change from traditional fee-for-service or indemnity insurance plans to various forms of man- aged care organizations. That shift toward managed care represents a demand-side influence that may affect the relative income of physicians in different specialties and, in turn, the distribution by specialty of future cohorts of trained physicians. In 1992, approximately 20 percent of privately insured people were in HMOs, which are among the most tightly controlled of the managed care arrange- ments. Managed care represents a smaller fraction of the Medicare sector than of the sector for private health insurance; however, managed care is becom— ing somewhat more popular among Medicare benefi- ciaries. Medicare's risk-based HMO enrollment has grown from less than 4 percent of beneficiaries in 1990 to about 7 percent.'0 Managed care organizations are generally viewed as less costly than traditional fee-for-service arrange- ments because they exercise greater control over the medical expenditures of their enrolled populations through a variety of techniques. Those mechanisms include selective contracting with providers who agree to accept discounted prices for their services, reimbursement methods that give doctors less incen— lO. Risk-based HMOs are paid a fixed amount per Medicare enrollee. Cost-based HMOs are reimbursed by Medicare on the basis ofthe cost report they submit. 22 MEDICARE AND GRADUATE MEDICAL EDUCATION September 1995 tive to provide services than under fee—for-service plans, and other management techniques that reduce hospital stays and visits to specialists. Increased use of managed care, as well as techno- logical changes in the direction of less inpatient care, implies a shift in the relative demands for different types of specialists. Managed care organizations use relatively more primary care physicians than are rep- resented in the physician population as a whole. One study of staffing patterns found that the fraction of primary care physicians in samples of several HMOs varied between 41 percent and 48 percent of their total physician workforces. In contrast, the propor— tion of primary care physicians in the entire nonfed- eral population of patient care physicians is only about 36 percent.‘1 That study suggests that if 40 percent to 65 percent of Americans were receiving care from a managed care network in 2000 and ev- eryone had some type of insurance, there would be a surplus of over 150,000 physicians, almost all of whom would be in non-primary care specialties. For several reasons, the evidence cited on the implications of managed care for changes in the com- position of the workforce might overstate the magni- tude of future changes in demand away from non- primary care physicians. First, the conclusions of that study depend on assumptions about the extent of managed care in 2000. Other analysts may argue that the growth of HMOs will not be as great as the study's author has assumed. Second, physicians and consumers who are most comfortable with HMO staffing patterns may be the first to join HMOs. As HMOs and other managed care plans expand to cover more of the market, those plans might have to adapt their staffing patterns to become more attractive to consumers who have not yet joined such plans. In that way, managed care staffing patterns might drift toward the more traditional patterns seen under con- ventional insurance. Third, if the expansion of man- aged care increases the demand for primary care phy- sicians, those doctors will become more expensive (particularly in the short run), and managed care or- ganizations may desire relatively fewer primary care 1 1. See Jonathan P. Weiner, "Forecasting the Effects of Health Reform on US. Physician Workforce Requirement: Evidence from HMO Staffing Patterns," Journal of the American Medical Association, vol. 272, no. 3 (July 20, 1994), pp. 222-230. doctors than was the case when managed care was less prevalent. Despite these caveats, if the trend toward man- aged care continues in either the private or the govemment-provided insurance sector, one would expect a relative decline in the income of non—pri- mary care physicians, although the particular non- primary care specialties may have quite different ex- periences.‘2 Over time, as the supply of newly trained residents responded to the relatively greater demand for primary care services, the direction of change in relative incomes could eventually reverse itself as the stock of physicians became oriented more toward primary care. Working in a counterdirection to those trends is one that seeks to prevent managed care organizations from determining the shape of their workforce. Some state legislatures have passed laws that may hinder the ability of managed care organizations to cut costs and also limit the relative shift in demand toward primary care specialties. For example, "any willing provider" (AWP) laws require a managed care organization to accept into its network any phy- sicians who agree to the plan's terms for participa- tion. Mandatory point-of-service (MPOS) options require managed care organizations to permit their enrollees to receive medical care from providers who are outside the organization; some, albeit reduced, payments are made for such services from the insur- ance pool. These kinds of laws might reduce an or- ganization's control over the delivery of medical care to its enrollees and dilute control over the composi— tion of the pool of physicians within the managed care organization.” Supply Factors. The supply of trained physicians' services depends on the size and composition of the 12. The income decline refers to relative changes in income, not changes in the ordinal rankings of incomes among sets of specialties. That is, the absolute level of primary care incomes would probably stay below surgical incomes even with shifls in demand toward primary care. All that would be necessary for some increase in the relative supply of primary care residents would be an increase in the ratio of primary care incomes to incomes in other fields. 13. Between 1984 and I994, about a dozen states reportedly implemented some form of AWP or MPOS laws. The states differ as to whether the laws include HMOs or only apply to looser managed care arrangements. CHAPTER TWO ECONOMIC FACTORS THAT AFFECT THE CHARACTERISTICS OF THE PHYSICIAN WORKFORCE 23 existing pool of trained physicians, the inflows of newly trained physicians and immigrant physicians, and the outflows of retiring physicians. Besides out- right retirement, some doctors may move to part-time practice or to non-patient care activities such as ad- ministration or research, which would effectively change the available supply of trained physicians' services. In addition to residency stipends and future expected income, other factors that affect the flow of new physicians and the total supply of physicians' services include incomes in alternative professions, the size of the population of young adults who could become physicians, the incomes available from med- ical practice in other countries, immigration laws, and governmental decisions about the subsidization of undergraduate and graduate medical training and about loan assistance to medical school students. These supply-side forces along with the demand- side ones discussed above jointly determine the amount of the different types of services that are pro- vided as well as the incomes of the various types of physicians. In turn, as discussed earlier, the residents who choose from among the different specialties re- spond to those incomes. Hence, a number of differ- ent nongovernmental and governmental factors can influence the number and types of hospital residents and ultimately shape the future physician workforce. (See Box 2 for a summary of those factors.) Speed of Adjustment of the Physician Stock Changes in the relative mix of primary care and non- primary care physicians in the workforce of trained physicians necessarily take a long time. Each year, the number of newly entering physicians is only about 4 percent of the total stock of active doctors. As a result, it can take a number of years for changes in the makeup of cohorts of newly trained physicians to affect the makeup of the total physician workforce. For example, assume that the fraction of newly trained physicians who would practice in primary care suddenly increased to 66 percent, nearly double the current proportion. A simple calculation, based on the assumptions that the absolute size of the phy- sician workforce stays constant and the youngest physicians replace the oldest ones, who retire, shows that it would take more than 12 years for the fraction of trained physicians in primary care to rise from one—third to one-half. Some caveats are necessary in regard to that il- lustrative calculation, however. The calculation might overstate the length of time the workforce needs to adjust because the example assumes that newly trained residents are the only source of new primary care doctors. But before their subspecialty training, medical or pediatric subspecialists have typ- ically undergone the same training as general inter- nists or pediatricians, respectively. If those subspe— cialists moved into primary care activities, the frac- tion of primary care physicians would grow to almost 45 percent. Decisions by some trained subspecialists to practice primary care, decisions by some non- primary care physicians to retrain in primary care, and changes in the relative retirement rates of pri- mary care and non-primary care providers are all fac- tors that could speed up the adjustment of the workforce relative to the calculation discussed above. Yet despite those qualifications, the calculation holds several general lessons for policymaking in relation to the medical workforce. If one accepts the argument that the current fraction of primary care physicians is far too small, a large increase in the proportion of residents in primary care would be nec- essary to avoid a long period of adjustment to the new goal. Those observers who favor an activist stance toward the medical workforce might defend such a policy against those who take a skeptical view of policymakers’ abilities to forecast future work- force needs. The activists would argue that even if one aimed at an inappropriate target for the final composition of the workforce, a workforce imbalance would not appear quickly because it would take a number of years for any policy to have large effects on the composition of the total workforce. Of course, the other side of this coin is that if one reaches an inappropriate balance or size of the workforce, it takes a number of years to correct the error. Yet an- other tack might be to question the point of activist workforce planning, given both the potential for error in forecasts of future demand and the long time be- fore such alterations in the cohorts of residents would have an effect. 24 MEDICARE AND GRADUATE MEDICAL EDUCATION September 1995 supply of residents. Factors That Affect the Demand for Residents Factors That Affect the Supply of Residents and the Future Supply of Trained Physicians’ Services Box 2. Nongovernmental Size of the population Age distribution of the population Consumers’ incomes Consumers’ education Other demographic characteristics Epidemiological factors Consumers’ and employers’ decisions about insurance coverage Insurance innovations such as managed care Research activities and technological advances Incomes of fully trained physicians Incomes in other professions relative to those in medicine Educational attainment of young adults Factors That May Influence the Number and Types of Residents A number of factors may affect the demand for and supply of residents. Likewise, numerous factors may affect the market for trained physicians’ services. Because the supply of residents responds to the incomes they expect as fully trained doctors, all of the factors that affect the demand for and supply of trained physicians’ services will affect the Some of those factors are governmental in origin; others are not. In many cases, a factor may evade easy classification and could be considered both governmental and nongovernmental in nature. Governmental Medicare’s direct graduate medical education payments Medicare’s indirect teaching adjustments Medicare/Medicaidmanaged care policies Income supplements to subgroups of consumers Provision of insurance including decisions as to who is covered and what services and technologies are covered Medicare/Medicaid reimbursement schedules Tax treatment of employment-based health insurance Laws and regulations that affect the insurance industry in general and managed care in particular Subsidization of biomedical research Policies toward FMGs Programs aimed at encouraging primary care practice Subsidization of medical school training ) CHAPTER TWO ECONOMIC FACTORS THAT AFFECT THE CHARACTERISTICS OF THE PHYSICIAN WORKFORCE 25 Nongovernmental Governmental Factors That Affect the Supply of Size of the young adult Subsidization of premedical school Residents and the Future Supply population training of Trained Physicians’ Services (Continued) Number of potential FMGs Working conditions Factors That Affect the Demand Size of the population Medicare/Medicaid reimbursement for Trained Physicians’ Services schedules Age distribution of the population Medicare/Medicaidmanagedcare policies Income supplements to subgroups of consumers Consumers’ incomes Consumers’ education Provision of insurance including decisions as to who is covered and what services and technologies are covered Other demographic characteristics Tax treatment of employment-based health insurance Epidemiological factors Consumers’ and employers’ l decisions about insurance Laws and regulations that affect the coverage insurance industry in general and managed care in particular Insurance innovations such as managed care Subsidization of biomedical research Research activities and tech- nological advances Factors That Affect the Current Supply of Trained Physicians’ Services NOTE: FMG = foreign medical graduate. Current stock of trained physicians Number of potential FMGs Physicians’ incomes in other countries relative to their incomes in the United States Working conditions Policies toward FMGs Programs aimed at encouraging primary care practice 26 MEDICARE AND GRADUATE MEDICAL EDUCATION -- —fi September 1995 k Historical Evidence The pattern of behavior observed during the late 19605 and the 19705 after the introduction of Medi- care, Medicaid, and their precursor programs illus- trates the response of the supply of physicians to eco- nomic incentives.14 The historical evidence is consis- tent with the view that the introduction of those gov— ernment insurance programs brought about a surge in demand for physicians' services. That surge led to an increase in physicians' income and in the rate of re- turn on medical education from the mid-19605 to the early to mid—19705. To understand the incentives facing potential physicians, economists consider the relationship be— tween physicians' earnings and the "opportunity costs" of becoming a physician. The opportunity cost of being a physician is calculated by considering the income that a doctor could have earned had he or she pursued the career of a typical college graduate. In addition, the total opportunity cost includes the direct costs of tuition for medical school and takes into account the cost of beginning one's career sev- eral years later than a typical college graduate.l5 The income-to-opportunity-cost ratio for physi- cians began growing in the early to mid—19605 and peaked near 1.5 in 1971. That strong economic in- centive brought about a predictable supply response (see Figure 7). The number of physicians surged, a response that was amplified by paying subsidies to increase the size of medical school classes. An addi- tional component of the supply response was pres- sure for entry by foreign-trained physicians. During the 1971—1981 period, the returns on investing in 14. Amendments to the Social Security Act in I960, 1961, and 1962 provided for matching grants to the states for medical assistance to low-income elderly people. In 1965, Medicare and Medicaid were enacted. 15. Specifically, the total opportunity cost of being a doctor is defined as the level of annual physician income that would bring about equality between the present value of net earnings as a doctor and the present value of earnings ofa college graduate. Figure 7. Relationship of the Ratio of Physician Income to Opportunity Cost and the Number of Physicians per 1,000 People, 1946 Through 1982 Physician Income to Opportunity Costa Ratio 2.2 L ”1 1 1.8} 1.6 1.41 > V \ ‘\ i 1.2.\/ fl . _\ / ”sag ' isso ' 1554 ‘ 19'58 ' 1967 ' 59‘s.; I i956 ‘ i954 ’ 1936 ' 1982 Physicians per 1,000 Peopleb Number of Physicians 2.2 ,, 2.0 ,, / / , Active and , ’ 1-3 Inactive Physicians , 4, // 1.6 . // 1.4 L ’ 1 Active 1 I Physicians 1.2 i o ’ ................. 1946 1950 1954 1958 1962 1966 1970 1974 1978 1982 SOURCE: The data for the ratio of physician income to opportu- nity cost were adapted by the Congressional Budget Office (CBO) from Monica Noether, "The Growing Supply of Physicians: Has the Market Become More Competitive?" Journal of Labor Economics, vol. 4, no. 4 (October 1986), pp. 503-537. The data for the number of physicians per 1,000 people are 080 cal- culations based on data from American Medical As- sociation (AMA), Distribution of Physicians in the U. 8. (Chicago: AMA, 1967 and 1971), Physician Distribu- tion and Medical Licensure in the US, 1975 (Chi- cago: AMA, 1976), and Physician Characteristics and Distribution in the US. (Chicago: AMA, 1983); and from the Bureau of the Census. 3. The ratio is the mean U.S. physician income to the total oppor- tunity cost of becoming a physician. The total opportunity cost of being a doctor is defined as the level of annual physician income that would bring about equality between the present value of net earnings as a doctor and the present value of earnings of a college graduate. b. Resident physicians are included in the data series; osteo- paths are not. From 1950 through 1980, data are plotted at five-year inter- vals. v CHAPTER TWO ECONOMIC FACTORS THAT AFFECT THE CHARACTERISTICS OF THE PHYSICIAN WORKFORCE 27 medical education began to decline toward their ear- lier levels as the stock of physicians swelled.‘6 None- theless, the income-to-opportunity-cost ratio re- mained high in comparison with its levels prior to the mid-1960s, which indicates that medicine continued to be a financially rewarding field. That observation is consistent with the continued entry of new, young physicians into the workforce and the continued growth in the number of physicians per 1,000 people. The growth of the health sector since the mid- 1960s appears to have been spurred by the introduc- tion and expansion of the Medicaid and Medicare programs (and precursors of those programs), which l6. See Monica Noether, "The Growing Supply of Physicians: Has the Market Become More Competitive?" Journal of Labor Economics, vol. 4, no. 4 (October 1986), pp. SOS-537‘ subsidized the demand for medical care. Those sig- nificant demand-side subsidies have been combined with such supply-side subsidies as payments to medi- cal schools for increasing class sizes and GME pay- ments to teaching hospitals. The number of physi- cians is probably greater than would have been the case in the absence of those government policies. Foreign medical schools provide a pool of gradu- ates who can augment the stock of practicing U.S. physicians beyond the supply capacities of US. med- ical schools. In 1970, about 14 percent of fully trained physicians in patient care were FMGs. By 1980, the fraction of such doctors had grown to al- most 20 percent, and in 1993 that fraction was almost 23 percent. Foreign medical graduates represent an increasing fraction of the stock of trained physicians as well as an increasing proportion of medical resi- dents. Chapter Three Policy Issues and Questions hanges in federal policy regarding graduate ‘ medical education may have both budgetary impacts and effects relating to broad goals of public policy in the health sector. In terms of bud— getary impact, direct graduate medical education payments are almost $2 billion per year and are pro- jected to grow to over $2.5 billion annually by the end of the decade. Federal payments for the indirect medical education adjustment are about $4 billion per year and are projected to grow to almost $5 bil- lion annually by the end of the decade. The broader goals that GME policy may affect include such issues as sufficient access to care, the cost-effectiveness of health care delivery, and the support of special activi- ties undertaken by teaching hospitals. GME subsidies, rather than directly affecting those broader goals, probably have their impact through several intermediate outcomes: 0 The number and specialties of medical residents 0 The number, specialties, and training background of future fully trained physicians o The incentives for teaching hospitals to treat Medicare beneficiaries o The overall financial resources of teaching hospi- tals Opinions may differ concerning the best approach for affecting those outcomes. In addition, critics may disagree about targets for the outcomes in a world of limited resources. Differences in opinion on GME policy will de- pend in part on different views about market incen- tives. Disagreements may arise over the degree to which students and young doctors are aware of or will respond to economic incentives. Policymakers may also disagree about whether the economic incen— tives transmitted through the marketplace are the proper signals to guide the allocation of resources. Access to Care Concerns about access to care are often raised in re- gard to particular subgroups of the population. Those subgroups include Medicare beneficiaries, indigent people, and people who live in geographic regions, such as rural areas or inner cities, that have few, if any, physicians. GME policy may affect access to care through several mechanisms. The current formulas for GME subsidies are written in such a way that they offer incentives for teaching hospitals to both provide more services to Medicare beneficiaries and hire more residents. Changes in the formulas might alter teaching hospitals' incentives to treat Medicare bene- ficiaries. In the longer run, GME policy might affect access to care through its impact on the future supply of physicians. In the shorter run, policies that affect the number of residents could affect access to care in communities in which medical residents provide a significant amount of hospital—based patient care. Moreover, because changes in GME policy might affect the revenues of teaching hospitals, many ser- 30 MEDICARE AND GRADUATE MEDICAL EDUCATION September 1995 vices that teaching hospitals provide, including care for people who are uninsured and indigent, could re- spond to changes in the level ofGME subsidies. Medicare Beneficiaries The current formulas for GME subsidies encourage teaching hospitals not only to employ more residents than they otherwise would but also to be more will- ing to provide services to Medicare beneficiaries. A hospital's DME payment is based on the product of its Medicare patient load, the allowed amount per resident, and the number of residents. As a result, a hospital's DME payments increase with its Medicare caseload. The IME formula bases payments on the product of the diagnosis-related group payments owed to a hospital for Medicare inpatient services and another factor that depends on the resident—t0- bed ratio. Hence, a hospital's IME payments rise with its Medicare admissions (adjusted for diagno- ses). It follows that changes in the DME or [ME for— mula would alter the incentive to admit and treat Medicare patients. Access to care for Medicare beneficiaries could be at risk if the government insurance program reim- burses providers for services at lower rates than do private insurers. The fees for physicians’ services that Medicare pays have been approximately two- thirds of the level paid by private insurers.l For 1993, the payment-to-cost ratio for hospital services for Medicare patients is estimated to have been about 69 percent of the ratio for private payers.2 Although in general, access does not appear to be a significant problem for Medicare beneficiaries, difficulties could develop if payment rates became sufficiently low.3 GME payments may affect access because they pro- vide an incentive for teaching hospitals to treat Medi- care patients. 1. Sec Physician Payment Review Commission, Annual Report to Congress, 1995, pp. 75-80. 2. See Prospective Payment Assessment Commission, Medicare and the American Health Care System: Report to Congress (June 1995). p. 21. 3. See David W. Lee and Kurt D. Gillis, "Physician Responses to Medicare Payment Reform: An Update on Access to Care," Inquiry (Fall I994), pp. 346-353; and Physician Payment Review Commis- sion, Annual Report to Congress, 1994, pp. 325-366. People Who Are Uninsured and Indigent Providing medical care to poor people who are unin- sured and who do not qualify for Medicaid coverage is one of the activities of teaching hospitals that may be viewed as a public good (see the discussion later in this chapter). Access to care for uninsured, indi— gent people may be affected by the level of GME subsidies for two reasons. First, since GME subsi- dies are a significant source of revenue for a number ofteaching hospitals, changing the level of the subsi- dies would probably affect the amount of various services that teaching hospitals provide, including care for uninsured people. Second, to the extent that care for indigent people is provided by hospital resi— dents, changes in subsidies could affect access for that group by influencing the number of residents available to provide such care. Geographic Areas That Are Underserved Standard economic theory predicts that a physician will locate in an area only ifthe benefits from living and working there are at least as great as those to be obtained in a different location. Abstracting from different amenities (for example, high-quality schools and social and cultural opportunities) avail— able among locales and specific preferences for prac— ticing in particular areas, one would expect that all communities that have doctors would have sufficient demand (or willingness to pay) so that each doctor in those communities earned at least a competitive re- turn on his or her investment in a medical education. Communities will not have doctors if their level of demand is too low for a single doctor to earn an in- come equivalent to the income that could be made in locations that have physicians. Communities that have doctors will have enough of them to bid in- comes (or, in economic terms, utilities) down to the level available in other locales that are being served. This standard model of physician location makes clear why both general subsidies to physician supply (including GME subsidies) and more targeted at- tempts to encourage doctors to locate in underserved geographic areas (such as the National Health Ser- CHAPTER THREE POLICY ISSUES AND QUESTIONS 31 vice Corps) appear to have limited success. Accord— ing to the model, if the supply of physicians in- creased, the number of physicians would increase in all the communities that already had physicians. The increased competition in those communities would bid down doctor's incomes (or utilities). Some for- merly unserved communities would add physicians only if the incomes (or utilities) available in served communities became sufficiently low that settling in an underserved community appeared attractive. Therefore, it would not be surprising to find that the counties with the lowest levels of demand would continue to go without a physician even ifthe num- ber of physicians in the country as a whole grew. Indeed, the empirical observation that the number of counties with few or no physicians has been little changed by the appreciable increase in the physician- to-population ratio in the country as a whole is con- sistent with the theory. The model also implies that a policy that directly places a new physician in an area with a low physician—to-population ratio is likely to be offset eventually by the departure of another phy- sician in that area for another market or by fewer physicians settling in that area in the future.4 An example of a factor other than government policy toward GME or toward physician supply that might affect the geographic location of doctors is the growth of managed care. Growth in that sector of the health care market might cause a shift in the distribu- tion of some types of physicians toward less densely populated areas. For that shift to occur, two condi- tions must hold: the infiltration of managed care into a locale must squeeze down physicians’ incomes in that area, and the entry of managed care organiza- tions must be greater in more densely populated areas than in less densely populated ones--which is consis- tent with the view that scale economies are relevant in the formation of managed care networks. Under one plausible scenario, the entry of managed care into an area would result in higher incomes for pri— mary care physicians and lower incomes for non-pri- mary care physicians. Then, if managed care entry occurred to a greater extent in more densely popu- lated areas, one would expect a tendency for primary 4. See Joseph P. Newhouse and others, "Does the Geographical Dis- tribution of Physicians Reflect Market Failure?" Bell Journal of Economics. vol. 13, no. 2 (Autumn 1982), pp. 493-505. care physicians to relocate toward more densely pop- ulated locales whereas non-primary care specialists would diffuse into less densely populated regions. An approach for reducing the number of medi- cally underserved areas is to raise the demand for medical services high enough in those locales for physicians to find them economically attractive. To the extent that amenities are fewer in underserved areas, the financial incentives must be even greater if physicians are to be attracted to practice in them. In areas with a significant population of uninsured or underinsured people, the government could provide or subsidize health insurance to increase demand, thereby making the area more attractive to physi— Clans. One way to subsidize the provision of medical care in such locales would be to offer an additional payment to physicians for each medical service they provide. The Medicare program already furnishes that kind of bonus payment--IO percent for services provided in what it terms “health professional short- age areas.” The bonus program is relatively new; therefore, assessing whether the program has had a measurable impact on physicians’ decisions about where to practice is premature. It might be the case that some regions are so sparsely populated, or have such low demand for medical care for reasons such as low incomes or lack ofinsurance, or are so lacking in amenities, that extremely large bonus payments would be needed to draw a physician to those areas. Cost-Effectiveness of Health Care Delivery The cost-effectiveness of health care delivery in the future will be influenced by the size, specialty mix, and training of the future physician workforce. To- day's residents are tomorrow's fully trained physi- cians; consequently, GME policy--through its impact on the characteristics of the residency pool--plays a role in cost-effectiveness. At a more subtle level, the style of practice of future physicians might be altered if GME policy changed in ways that affected the ba- sic model and sites of postgraduate medical training. In a broader context, not only the physician work- 32 MEDICARE AND GRADUATE MEDICAL EDUCATION September 1995 force but also the size and characteristics of the nurs— ing and allied health workforces will have an impact on cost-effectiveness. One particular concern is that the current physi- cian workforce is too oriented toward specialty prac- tice to be compatible with emerging arrangements for managed care insurance. The question then becomes whether an activist policy toward the physician workforce can forecast future demands well enough to forsake relying on the market to determine the number and characteristics of physicians. Different philosophical and theoretical perspectives lead to different opinions about the appropriate policy to follow. The Appropriate Number of Physicians Some health policy experts have concluded that the United States has too many physicians.5 Those claims are based on opinions about the appropriate number and types of physicians required to fill the basic medical needs of the population. Some people have also suggested that the growing number of phy- sicians is a cause of rising medical expenditures and that reducing the number of physicians per 1,000 people would help to contain costs. Some of the arguments underlying that view ofa surplus of physicians are not consistent with a con- ventional economic approach to labor supply. The latter approach places more emphasis on the discre— tion of consumers regarding the desirable amount of medical care than does the more rigid approach of determining physician requirements. Some econo- mists are likely to question whether the number of physicians is the fundamental cause of high levels of medical expenditures. Instead, a reasonable interpre- tation is that other underlying factors cause both large numbers of doctors and high levels of expendi- tures. Those factors include the incentive to use more services when those services are insured, the favorable tax treatment of employment-based health 5. See the summary in Steven A, Schroeder, "Managing the US. Health Care Workforce: Creating Policy Amidst Uncertainty," In— quiry. vol. 31. no. 3 (Fall I994), pp. 266-275. The groups that have concluded that there are too many physicians include the Council on Graduate Medical Education, the Bureau of Health Professions. and the Physician Payment Review Commission. insurance, the fee-for—service nature of most government-provided insurance, and federal subsi- dies directed toward the supply of physicians. Con- ventional economic analysis would stress that in the absence of major noncompetitive factors or distor- tions, the market has the inherent ability to efficiently allocate people tojobs. Thus, ifthe number of physi- cians was excessive, the relative fees and incomes of doctors would fall, the medical profession would be- come less desirable to potential entrants, and the ex- cess would be eliminated over time.6 Analysts with a medical requirements perspective and analysts with a market-oriented economic per- spective work from widely different underlying as- sumptions, but both parties would generally conclude that the nation has more than the optimal number of physicians. The requirements, or planning, approach would base that conclusion on calculations of the number and types of doctors needed to provide what is viewed as appropriate medical care for the popula- tion. The market economies approach would base its conclusion on the existence of significant demand- and supply-side subsidies that have caused the medi- cal industry to become larger than it otherwise would be.7 (That is, the distortions that are present appear to work toward making the medical industry larger than optimal.) The two types of analysts would gen- erally disagree on the proper policy response. For example, planners might argue for continuing the subsidies and imposing quotas on the number and 6. One theoretical viewpoint that places relatively more emphasis on the notion that the number ofphysicians is an underlying cause of higher levels of medical expenditures is the "demand inducement" theory. The basic idea is that doctors take advantage of consumer ignorance and respond to potential reductions in their own income by providing more services to their patients than they would other- wise. The empirical importance of demand inducement is contro- versial among economists who study the health industry. 7. At one time, it was thought that the number of physicians was be- ing held at an artificially low level to boost physicians‘ incomes. Proponents ofthat view argued that licensing laws and limitations on the number ofplaces in medical schools were the tools by which those artificial restrictions on supply were enforced. Under such a scenario, a market economist might have favored subsidizing the supply of physicians to offset the anticompetitive restrictions on that supply. Beginning in the 19705, the market for the supply of physicians appears to have become more competitive (see Monica Noether, "The Growing Supply of Physicians: Has the Market Be- come More Competitive?" Journal ofLabor Economics, vol. 4, no. 4 (October 1986), pp. 503-537). If immigration polices were al- tered to restrict further the entry of foreign medical graduates, argu- ments for offsetting anticompetitive restrictions on supply might become more relevant. CHAPTER THREE POLICY ISSUES AND QUESTIONS 33 types of residency positions. Market economists would argue for eliminating the subsidies and not imposing any legal restrictions on the number or types of positions.8 The Appropriate Mix of Physicians by Specialty A number of experts have argued that there are too many non—primary care physicians relative to pri- mary care physicians in the current workforce and that a large decline in the fraction of non-primary care residents is necessary to attain a more appropri- ate balance. Two principal arguments appear to un- derlie the desire to increase the relative size of the primary care medical workforce. One argument is that more primary care doctors are needed to fill a crucial “gatekeeper” role in managed care organiza- tions and that such organizations represent the future of health care delivery. A second argument is that primary care physicians offer a cheaper, less techno— logical, more preventive and holistic approach to the delivery of health services. As noted earlier, some experts prefer an activist workforce policy and are skeptical of relying on mar- ket forces to determine the size and distribution of the nation’s cadre of physicians. In their view, sev- eral good reasons support intervention by the govern— ment. Proponents of an activist policy argue that the inherently slow adjustment of the physician work- force makes it too costly to rely on market signals alone to encourage greater entry into primary care specialties. Another argument against using market signals is that young physicians are poorly informed about the marketplace for trained physicians. There- fore, only a minimal change in supply would occur in response to shifts in the demand for physicians' ser- vices. Market skeptics might also question the valid- ity of demands by consumers for physicians' services. If consumers cannot judge what is best for them, re- 8, Market-oriented economists who are concerned with allocating resources efficiently might go even further. They might argue that ifthe distortions toward too large a medical industry, which arise from the tax treatment of employment-based insurance and subsi- dized govemment-provided insurance, cannot be eliminated, the distortions could be offset by taxing the “inputs" used to produce medical care. But it might be difficult and impractical to impose taxes in a neutral fashion across those inputs. lying on market signals based on consumers' de- mands for medical services would not tend to yield the appropriate number or types of physicians. However, the experts who argue for a dramatic increase in the relative number of primary care doc— tors may err for one of two reasons (or for both): they may be using incorrect forecasts, or they may be making inappropriatejudgments about what consum- ers want. Some expert opinions depend on forecasts of future increases in the prevalence of managed care, including health maintenance organizations, in health care delivery.9 Experts on the other side of the debate point out that previous forecasts of HMO growth have overestimated the actual rates. Another point of contention concerns the appropriateness of expert opinions in determining the correct balance of medical care needs. Consumers may prefer more highly technological, more interventionist, and more specialized medical care than the experts deem suit- able. Market-oriented economists would point out that the distribution of the supply of physicians among the different specialties will respond to economic incentives, which in turn will reflect the attitudes of consumers about the types of health care delivery systems that they prefer, given the costs of alterna- tive systems. If managed care organizations that make less use of the services of specialists carry the day in the consumer marketplace, the incomes of non-primary care physicians will fall relative to those of primary care physicians, and relatively fewer young doctors will seek non-primary care training. However, free-market economists would warn that, because experts might either incorrectly forecast market trends or incorrectly determine what is best for consumers, it would be preferable to follow a neutral government policy regarding the entry of physicians into various specialties. Many advocates of a market approach would agree that the government can assist the marketplace by providing information (to help overcome failures in the market for information). But they would also tend to argue that relying on the demands of imper- 9. For example, see Jonathan P. Weiner, "The Demand for Physician Services in a Changing Health Care System: A Synthesis," Medical Care Review, vol. 50, no. 4 (Winter 1993), pp. 4] 1-449. L_—— 34 MEDICARE AND GRADUATE MEDICAL EDUCATION September I995 fectly informed consumers is preferable to relying on the political process or a commission of experts to determine the number and types of physicians the country should have. The government could provide information to consumers, medical students, and young physicians and then allow them to make deci- sions in a free market. Empirical evidence suggests that both the total supply of physicians and the distribution among spe- cialties respond to changes in relative incomes. However, an interventionist policy may still have merit despite the existence of such evidence. For example, one may accept the view that students, resi- dents, physicians, and hospitals respond to economic incentives; however, one may also view the response time as too slow or feel that consumers are in- adequately informed and therefore that signals from the marketplace convey inappropriate incentives. Policymakers who subscribe to those views may ar— gue for intervention by the government to alter the economic incentives. Such alterations might include paying relatively larger subsidies for primary care residencies than for residencies in other specialties or changing the fee schedule for Medicare to reward primary care activities relatively more than non—pri- mary care activities. The Appropriate Site for GME Training The current system of GME subsidies to teaching hospitals may reinforce the traditional approach to GME centered at the hospital site--an approach that some experts have questioned. Many of the patient care activities of fully trained primary care practi- tioners occur in a nonhospital setting. But because non-primary care specialties have relatively more of their services associated with a hospital setting, the current mechanisms for GME financing may encour- age more new doctors to focus on non-primary care specialties than if GME subsidies were also available to group practices outside of hospitals. Some indus- try experts have argued that HMOs find that newly trained primary care physicians have not developed adequate skills to work in an HMO environment be- cause their training has used the hospital rather than the medical group practice as its principal focus.lo Hospital-based training has many advantages, however, and hospitals might remain the principal site of post-medical school training even if relative subsidies to the residency programs at teaching hos- pitals were reduced. The hospital, for instance, al- lows a resident to see more difficult and more severe cases and to see a variety of cases and procedures in one venue. Special Activities of Teaching Hospitals Because changes in the level of subsidies would af- fect teaching hospitals' revenues, other activities and services that they offer besides GME might be al- tered in response to changes in the generosity of GME payments. Indeed, the special activities and special services of teaching hospitals are cited as ad- ditional reasons for providing GME subsidies. Economists classify some of those special activi- ties as public goods--goods that, because their bene— fits cannot be restricted to those who purchase them, tend to be undersupplied in the absence of subsidies. The special activities of teaching hospitals in that regard may include engaging in research and devel- opment, treating and educating the public about com- municable diseases, and providing care to indigent people. Other special activities may include funding some medical school education and dealing with more difficult cases than are seen in other hospitals. A substantial amount of the revenues of medical schools is derived from patient care income gener- ated at affiliated teaching hospitals; in addition, the facilities of some teaching hospitals are used for edu— cation at the medical school level. Since GME pay- ments affect the level of revenues of teaching hospi— tals and subsidize residency programs whose teach- ing faculty may also be affiliated with a medical 10. For example, see Robert Gumbiner, "Perspectives of an HMO Leader," Inquiry, vol. 3l, no. 3 (Fall 1994), pp. 330-333. CHAPTER THREE POLICY ISSUES AND QUESTIONS 35 school, the resources of medical schools are probably affected by GME subsidies. To the extent that some of the activities of teaching hospitals are seen as pro- moting the public welfare, policymakers could pro- vide subsidies for those special activities specifically rather than indirectly supporting them through GME payments. Chapter Four Policy Options could pursue with regard to Medicare gradu- ate medical education policy. The range of options is fairly broad: from modifying the formulas for the subsidies but maintaining the current basic structure of direct graduate medical education and indirect medical education payments, to more radical restructuring of government payments for GME, to ending federal GME subsidies. P olicymakers have a number of options they All of the options have potentially significant consequences for access, cost-effectiveness, and the special activities of teaching hospitals, issues that were raised in Chapter 3. In addition, almost all of the options could be tailored to generate a number of budgetary outcomes, ranging from savings to in- creased spending. For example, modifying the GME formulas to encourage relatively more primary care residencies could be done in a budget—neutral fashion by increasing the DME or IME subsidies for primary care (or both) while decreasing the subsidies for other specialties. Alternatively, policymakers could produce budgetary savings by maintaining the cur— rent subsidies for primary care while reducing them for other specialties, or by reducing subsidies for all residencies but making the greatest reductions in sub- sidies for non-primary care. As another example, the option to end federal financing of GME could yield budgetary savings, but those savings would be smaller ifthe Congress enacted additional programs of direct subsidies for public goods produced by teaching hospitals (such as research and development or care for indigent people). Modify the Current System of GME Financing Several options are available for changing the mecha- nisms for federal financing of GME within the cur- rent structure ofthe DME and [ME payment formu- las. Those options all have potential effects on the size and distribution among specialties of the current pool of residents and on the future workforce of trained physicians. They could have impacts as well on the incentive for teaching hospitals to treat Medi- care beneficiaries and the financial position of such hospitals. Reduce the IME Teaching Adjustment The government’s Prospective Payment Assessment Commission (ProPAC) has suggested that the current IME subsidy be reduced to reflect more accurately those increases in teaching hospitals' costs that are associated with larger resident-to-bed ratios. Based on an analysis of recent data, ProPAC has recom- mended that the lME adjustment be reduced in phases from its current rate of about 7.7 percent to 4.5 percent, for a 0.1 increase in the resident-to-bed ratio of the hospital.1 A decrease in [ME payments would be expected to lead to fewer residents than I. Prospective Payment Assessment Commission, Report and Recom- mendations to the Congress (March 1. 1995), pp. 41-42. 38 MEDICARE AND GRADUATE MEDICAL EDUCATION September I995 teaching hospitals would otherwise have trained, al- though the effect of such a decrease on the mix of residents among specialties is not clear. Ultimately, these changes would have some im- pact on the characteristics of the trained physician workforce. Teaching hospitals would probably also scale back some activities besides residency training, but predicting the magnitude and nature of other re- ductions in services is difficult. Access to care at teaching hospitals might be diminished, especially for hospitals at which residents provide a significant amount of care. Moreover, because the lower IME payments would have the effect of reducing what the hospital was paid for providing inpatient services to Medicare beneficiaries, teaching hospitals would have weaker incentives to serve Medicare patients than under the current formulas. Change the Relative Subsidization of Particular Specialties An option that would encourage relatively more resi- dents than otherwise to enter primary care would be to alter the formulas for the IME and DME subsidies to pay relatively more to hospitals for primary care residents than for non—primary care residents. Be- cause the subsidies affect the value of residents to teaching hospitals, altering the formulas to increase the relative reward for filling primary care residen- cies versus non-primary care training slots would raise the demand by teaching hospitals for primary care versus non-primary care residents. The option could be implemented in a budget-neutral manner or as part of a package that would change the total amount of federal GME payments. Over the long run, if no other changes occurred in the market for physicians' services, tipping subsi- dies more strongly toward primary care training would probably have several effects. In terms of ac- cess to care, consumers would have more access to primary care services and less access to non-primary care services. In terms of fees and physicians' in— come, primary care physicians would experience downward pressures on their fees and income relative to those of non-primary care physicians. Thus, a pol- icy of relative subsidization at the residency level would create some windfall gains for those who were already fully trained in non—primary care specialties and some windfall losses for those who were already fully trained in primary care. All of those effects refer to comparisons with how the market would otherwise be (that is, in the absence of specific policy action). For example, if the trend toward managed care continued and brought with it an increase in demand for primary care services, the downward pressure on primary care fees from the subsidization of primary care residen- cies would be offset by the increased demand for pri- mary care services. The net change in the income of primary care practitioners would depend on which influence turned out to be stronger. The greater subsidization of primary care resi- dencies would bring about a faster transition of the trained medical workforce to its increased primary care proportions than would otherwise have occurred. Also in evidence would be a lower path of primary care fees and income relative to non-primary care fees and income. However, during the transition pe— riod, primary care fees and income would grow rela— tive to non—primary care fees and income if the in- creased demand for primary care services outstripped the effects on supply of greater subsidization of pri- mary care residencies. Reactions by the government to changes in de— mand for various services would also affect the in— comes of different types of physicians and patients' access to different types of doctors. Suppose, for example, that the net effect of changes in the medical insurance market and favorable subsidization of pri- mary care residencies turned out to put upward pres— sure on non-primary care fees as a result of the rela— tively smaller number of non—primary care practi— tioners. If the fee schedule for Medicare failed to incorporate the higher fees, increases in the income of non-primary care physicians would be held down-- at the cost of potential problems in access for Medi- care beneficiaries, who could experience a shortage of non-primary care providers. In contrast, if the net effect of changes in the insurance market and govern- ment GME policy put upward pressure on primary care fees, Medicare might have to raise its primary care payments to prevent a perceived shortage of pri- mary care providers for Medicare beneficiaries. CHAPTER FOUR POLICY OPTIONS 39 Other options that would increase the relative number of residents in primary care while reducing federal outlays for GME include eliminating subsi— dies for non-primary care positions or ending such payments for residents beyond three years of train- ing. The primary care specialties have the shortest residencies. Thus, if the [ME and DME formulas were altered to pay hospitals only for residents who were in their first three years of residency training, that change would reduce what hospitals perceive as the net value of non-primary care residents. Such options, which imply an overall reduction in the ex- tent of the Medicare subsidies for residency training, would lead to smaller hospital residency programs that were more disposed to primary care training than would otherwise be the case. The disadvantages ofthis approach would be the decline in the number of residents at teaching hospi- tals, which might lead to a decline in the availability of medical services at such hospitals or increased costs for those services. It could also lead to a reduc- tion in other outputs of teaching hospitals including care for indigent people. Some of the impact might be passed on to residents in the form of reduced sti- pends. Reductions in either type of GME subsidy would reduce the hospitals' incentive to treat Medi— care patients. Reduce or Eliminate GME Subsidies for Noncitizens Yet another strategy to decrease GME payments within the current framework of subsidies would be to reduce or eliminate payments for residents who are not US. citizens. About one-fourth of all current residents graduated from foreign medical schools; the majority of those foreign medical graduates are not US. citizens, although most of them are expected to enter the trained U.S. physician workforce at some pomt. Reducing or eliminating subsidies for noncitizens would probably result in fewer residents than other- wise, with the decrease primarily the result of fewer noncitizen residents than if the current system re- mained unchanged. A disadvantage ofthe policy is that it might lead to a two-tier residency system with a defined wedge between the stipends of citizens and noncitizens or to increased segregation of citizens into more desirable programs and noncitizens into less desirable programs. Again, potential problems with access and reductions in the services provided by teaching hospitals become an issue as they do un- der other policies that reduce the overall level of GME subsidies (unless, of course, other grants or payment programs replace the subsidies). Also, the future supply of physicians in this country would probably be smaller than otherwise, and F MG physi- cians might average less residency training than they would ifthe current policy was maintained. Establish an All-Payer System to Finance GME A policy that would continue to subsidize GME yet reduce the funding required for the Medicare pro— gram would be the formation of a GME fund sup- ported by mandated payments from all insurers, both public and private. That approach would increase the cost of both employment-based and directly pur- chased health insurance and would probably encour- age people, in a modest way, to buy less coverage. By raising the price of insurance, the policy would reduce slightly the take-home pay of workers with employment-based insurance. However, it would also allow Medicare to reduce its funding of GME with no loss in total GME payments to teaching hos- pitals. The policy would simply shift some of the government-imposed financing that is required to fund GME subsidies from the Medicare program to purchasers of private insurance and to other public insurers. One effect of an all-payer fund could be to weaken a teaching hospital’s incentive to treat Medi- care patients. That outcome could occur if the cur- rent subsidy formulas were maintained except for replacing Medicare's inpatient load and diagnosis- related group payments with the insured inpatient load and insured payments for inpatient hospital charges. In that case, dollars from the GME fund could be earned on all insured patients rather than on Medicare patients alone, which would dilute the in- centive, inherent in the current system, to treat Medi- care beneficiaries. 40 MEDICARE AND GRADUATE MEDICAL EDUCATION September I995 Payments to teaching hospitals from an all—payer fund would not necessarily have to retain the current or even modified financing formulas. For example, policymakers could use the all-payer fund to finance the more radical options discussed below. Restructure the System of GME Financing Other options represent more fundamental changes in the mechanisms by which the federal government provides assistance for GME. Two possibilities that come from different ends of the spectrum in terms of using market incentives to shape residency training are shifting to a GME voucher system or establishing quotas for residency positions by specialty. Open Up the Market for GME by Offering Vouchers One method by which the government could subsi- dize graduate medical training would be to offer vouchers to medical school graduates. The young doctor could transfer the voucher to a medical group or hospital as part ofa contract in which the organi— zation would provide training and a stipend to the resident in exchange for his or her services and pay- ments from the federal government. An advantage of that approach is that it would subsidize post-medical school education at sites that appeared to offer the mode of training most relevant to the new profes- sional's career. Young physicians would use market signals and expectations about future changes in the marketplace for physicians to make decisions about the specialty they wished to pursue and the kind of facility at which they wanted to receive training. That approach would uncouple GME subsidiza— tion from teaching hospitals, which would have to compete with other entities such as group practices or health maintenance organizations that were willing to start their own training programs. Over the long run, teaching hospitals might remain the predominant pro— viders of GME. However, ifthey remained so under a voucher system, it would probably be because of true advantages in the use of teaching hospitals for GME rather than because of a federal system that subsidized training only through those hospitals. It is difficult to predict how specialty boards would react to such changes. The current require- ments for board certification were shaped in a tradi- tion of training centered at teaching hospitals that was reinforced by Medicare GME payments to such facilities. If the boards were reluctant to consider training at nonhospital sites as equivalent prepara- tion, the GME system might remain essentially un— changed. The disadvantages of a voucher approach would include potentially adverse impacts on teaching hos- pitals from reductions in their Medicare GME reve- nues. Another disadvantage would be the loss of the inherent incentives for teaching hospitals to treat Medicare patients that are built into the current for- mulas for the GME subsidies. In addition, because the voucher system is based on responses to market incentives, it depends on reasonably informed deci- sionmaking. Without adequate information, chang- ing to a voucher system might result in responses by the workforce that were slower than some policy- makers would find desirable. Consequently, the argument that the government should provide or assist in providing information may be even more relevant to the voucher option than to other options. The federal government could inform medical or premedical students about trends and innovations in the market for physicians' ser- vices. Publications could update students on trends in physicians' incomes, employment opportunities for new physicians by specialty, and other changes in the nature of medical practice. Increasing the informa- tion available to students and residents could yield a faster supply response to changing conditions in the market than would otherwise occur. That quicker response would address concerns that the market ad- justs too slowly or that participants in the market lack the information needed for relying on the market- place to allocate the number and types of physicians. Policymakers could vary several parameters within a voucher framework. They could set up the program to approximate current federal expenditures per resident or base the value of the vouchers on CHAPTER FOUR POLICY OPTIONS 41 some other level of total expenditures. If the govern- ment wanted to influence the balance of the medical workforce among specialties, it could provide vouch- ers of different values or vouchers only for particular specialties. Citizenship could be considered a poten- tial requirement for receiving a GME voucher (but not a requirement for receiving training). Establish Quotas for Residency Positions Quotas that designate the number and types of resi- dency positions have been suggested as a means of increasing the relative numbers of primary care doc— tors and reducing the overall supply of physicians. One proposal that has been offered is to limit the to- tal number of first-year residency positions to 110 percent of the number of annual graduates from US. medical schools and to set specialty quotas allocating half of the residency positions to primary care and half to non-primary care. Proponents of such mea- sures point out that it takes many years for the com— position of the workforce to adjust and that responses to financial incentives may not be sufficient for meet- ing their workforce goals. Proponents of letting mar- ket forces allocate the workforce would object to that notion, maintaining that quotas could lead to a work- force that was not configured according to the under- lying desires of consumers or future market realities, which can be forecast only in a limited fashion. A difficulty with using quotas is assigning the slots. If the number of new medical graduates who wish to enter a particular field exceeds the quota, the decision as to which graduates obtain the available positions might appear to be arbitrary. Under quotas, some committee or group would have to allocate the different kinds of specialty slots among the many teaching hospitals. Those decisions might also ap- pear to be arbitrary, given that some hospitals would not be permitted to offer positions that otherwise they would offer. Moreover, unless the quotas were coordinated with changes in the demand for physicians' services, they would lead to an increase in the relative incomes of fully trained physicians in the rationed fields-- which could encourage people to try to get around the quotas. For example, if policymakers established a quota that effectively restricted the number of non- primary care residents but the projected shift in de— mand toward primary care services did not continue as expected, the result would be an increase in the relative incomes of non-primary care physicians. That situation would lead even more residents to want to enter non—primary care fields. Another source of pressure might come from fully trained for— eign specialists. Such specialists might immigrate, which would not violate the quotas per se (since the quotas apply to residency positions) but would partly undo the goals 0fthe quotas. Another problem with using quotas is that the definitions of various specialties may change or be changed, especially in the face of economic incen— tives for doing so. Programs may label themselves in accordance with the quotas, but the actual content of training may instead resemble the way the market would have looked without quotas. For example, some "internal medicine" or "family practice" pro- grams might begin to incorporate more subspecialty training if the demand for and returns from being a subspecialist remained high. In that fashion, the ulti- mate goals of advocates of quotas might be undone. A policy restricting the total number of residency positions to limit the total number of physicians could face similar problems. Rationing of the limited slots could appear to be capricious with apparently qualified young physicians being turned away from residency training. If no reduction in demand for physicians' services occurred, physicians' incomes would increase more than they otherwise would have, and that would increase the number of young people competing for the limited number of slots. In addi- tion, more fully trained foreign physicians would want to immigrate. A positive or negative view 0fthe immigration of already trained doctors that could result from a quota policy depends on one's perspective. Potential immi- gration (or the lack thereof) would be seen in a posi- tive light if one believed that the quotas that had been set truly reflected the direction of the market in terms of future demand for physicians' services, and that quotas were needed to speed the adjustment of the workforce to that demand (since adjustment of the workforce would be faster with immigration than 42 MEDICARE AND GRADUATE MEDICAL EDUCATION September 1995 without it). Immigration would be viewed negatively if one supported quotas from the belief that market forces would not yield the appropriate number or types of physicians in the workforce. (Immigration would tend to undo movement toward the workforce goals that were the original basis for the quota pol- icy.) To enforce quotas, policymakers could make the receipt of funds (which could be called GME funds) contingent on a hospital's remaining within the resi- dency quotas that it received. The financial positions of teaching hospitals could be maintained by keeping their levels of funding near the current real (adjusted for inflation) dollar values of GME payments. Alter- natively, the policy could alter those funding levels-- although in that case, possible effects on the services provided by teaching hospitals could not be ruled out. For example, the incentives of teaching hospitals to treat Medicare patients could be influenced by the funding formulas. Because the current GME formu- las build in some incentive to treat Medicare benefi- ciaries, a change to, say, a lump-sum payment-—with the condition that the hospital must remain within the quotas--would eliminate that particular incentive. The incentive might be maintained by keeping the formulas in current law but making them conditional on observance of the quotas. End Federal Financing of GME Eliminating federal financing of GME is yet another policy option. Estimates indicate that eliminating both DME and IME payments would reduce federal outlays below what they would otherwise have been by over $7.5 billion for fiscal year 2000 and over $8.5 billion for fiscal year 2002. One argument in support of this option is that the hospitals that obtain the services of residents and the residents who receive training would be faced with the true costs and benefits of the services and train- ing. That newfound clarity would lead to decisions about residency training that more accurately re- flected the costs and benefits of providing and receiv- ing it. In that spirit, Medicare’s payment rules for physicians' services could be loosened to allow resi— dents to bill for beneficial medical services that they provide to Medicare patients. That policy, however, might introduce new questions about payment levels and qualification requirements for residents. Eliminating the subsidies would also address the concern that there will be more physicians than is appropriate if the supply of physicians continues to be subsidized. If GME subsidies were reduced or eliminated, hospitals would be expected to demand fewer residents. Many FMGs who would have done residencies in this country and who would probably have become part of the future U.S. workforce of fully trained physicians might choose not to do so. Also, graduates of US. medical schools might re- ceive less residency training than otherwise. The effects on access in the future to fully trained physi- cians in underserved locales would probably be very small for the same reasons that large increases in the total supply of physicians have caused minimal in- creases in such areas. However, the number of phy- sicians per 1,000 people in locales that are already being served would be fewer than they would other- wise have been. Eliminating GME subsidies (without other, off- setting changes) would significantly lessen the reve- nues of teaching hospitals and would bring about a reduction in some of their services. It is difficult to predict the particular impact on different services. Various public goods-—such as providing care for in- digent people, conducting research and development, and undertaking efforts to control infectious diseases --might be cut back in response to reducing or elimi- nating GME payments. Because teaching hospitals would have significantly weakened incentives to hire residents, they would attempt to fill fewer residency positions. Some of the impact would probably in- clude lower residency stipends. Medical schools would also be affected because of their ties to teach- ing hospitals. In addition, teaching hospitals would have less incentive to serve Medicare patients, al- though easing restrictions on Medicare billings for physicians' services provided by residents might indi- rectly diminish that impact. Another alternative that would address reduced incentives to serve Medicare patients and lower revenues for hospitals would be to allow larger inpatient payments to hospitals that cur- rently receive GME payments but to dispense with CHAPTER FOUR POLICY OPTIONS 43 formulas like the present ones that relate additional payments to the number of residents. To minimize adverse consequences, policy- makers could gradually reduce GME payments over time according to a stated schedule. A gradual drop in the subsidies would allow teaching hospitals to plan and adjust their facilities and services to lower levels of revenue. Potential medical students would also be able to plan better: they would have the in- formation that the subsidies were being reduced and the understanding that such reductions would proba- bly result in lower residency stipends and fewer resi- dency positions, as well as higher future incomes for trained doctors in the respective specialties, than would otherwise have been the case.2 2. The Prospective Payment Assessment Commission has recom- mended that the reduction it proposes in [ME payments, which does not call for ultimate elimination of the [ME adjustment, be phased in and reevaluated over several years. Policymakers could also use more specifically targeted subsidies to encourage teaching hospitals to provide public goods and to deal with specific issues that now may be indirectly paid for through GME funding. Payments to providers for indigent care, research activities, and prevention of infectious dis- eases could be directed specifically toward those ac- tivities. If teaching hospitals indeed receive more difficult cases than other kinds of hospitals, further research could provide a basis for modifying the DRG system of reimbursing hospital services under Medicare. In addition, the concern that medical stu— dents and residents are uninformed about trends in the medical market could be addressed by increased efforts to provide such information. ' Appendix Supplementary Tables ’ I |his appendix contains tables that present the data used to construct Figures 1, 2, 4, and 6 in the text. 46 MEDICARE AND GRADUATE MEDICAL EDUCATION September 1995 - Table A-1. Distribution of Fully Trained Physicians Engaged in Patient Care, by Specialty, I Selected Years, 1965 Through 1993 (In percent) Specialty 1965 1970 1975 1981 1985 1990 1993 Primary Care 50.5 42.2 39.3 35.1 34.6 34.4 33.8 i Subspecialties of Internal I Medicine and of Pediatrics 1.7 4.2 4.8 8.2 9.3 10.2 10.9 Obstetrics and Gynecology 6.3 6.8 6.8 6.8 6.4 6.6 6.5 Surgical Specialties 21.3 22.5 22.6 21.3 20.5 20.0 18.9 i Other Non-Primary Care I Specialties 19.6 23.0 25.6 27.6 28.1 28.4 29.1 Unknown 0.6 1.3 0.9 0.9 1.1 g __Q._8_ Total 100.0 100.0 100.0 100.0 100.0 100.0 100.0 SOURCE: Congressional Budget Office calculations based on data from American Medical Association (AMA), Distribution of Physicians in the US. (Chicago: AMA, 1967 and 1971), Physician Distribution and Medical Licensure in the U. 8., 1975 (Chicago: AMA, 1976), and Physician Characteristics and Distribution in the U. 8. (Chicago: AMA, 1982, 1986, 1992, and 1994). NOTES: Data are for fully trained physicians who are actively engaged in patient care; that is, the data exclude residents and those fully trained physicians who are employed in professional activities other than patient care (for example, administration or medical research). The specialties of physicians are self-reported and refer to the principal specialty practiced by the doctor. Osteopaths are not included. - i Data for 1990 and 1993 are as of January 1. Data for all other years are as of December 31. APPENDIX SUPPLEMENTARY TABLES 47 Table A-2. Fully Trained Physicians Engaged in Patient Care per 1,000 People, by Specialty, Selected Years, 1965 Through 1993 Specialty 1965 1970 1975 1981 1985 1990 1993 Primary Care 0.55 0.46 0.46 0.49 0.53 0.56 0.59 Subspecialties of Internal Medicine and of Pediatrics 0.02 0.05 0.06 0.11 0.14 0.17 0.19 Other M M Q 0.79 0.87 0.90 0.96 Total 1.10 1.09 1.16 1.40 1.54 1.63 1.74 SOURCE: Congressional Budget Office calculations based on data from American Medical Association (AMA), Distribution of Physicians in NOTES: the U. 8. (Chicago: AMA, 1967 and 1971), Physician Distribution and Medical Licensure in the U. 8., 1975 (Chicago: AMA, 1976), and Physician Characteristics and Distribution in the U. 8. (Chicago: AMA, 1982, 1986, 1992, and 1994). Data are for fully trained physicians who are actively engaged in patient care; that is, the data exclude residents and those fully trained physicians who are employed in professional activities other than patient care (for example, administration or medical research). The specialties of physicians are self-reported and refer to the principal specialty practiced by the doctor. Osteopaths are not included. The "Other" category comprises fully trained physicians practicing in the specialties of obstetrics and gynecology, surgery, and other non-primary care fields. Data for 1990 and 1993 are as of January 1. Data for all other years are as of December 31. 48 MEDICARE AND GRADUATE MEDICAL EDUCATION September 1995 Table A-3. Distribution of Residents by Specialty, Selected Years, 1965 Through 1993 (In percent) Specialty 1965 1970 1975 1981 1985 1988 1990 1993 General Practice and Family Practice 1.6 2.6 5.6 8.4 9.2 8.6 7.8 7.8 Internal Medicine and Internal Medicine Subspecialties 18.5 19.6 25.2 24.6 26.0 30.8 31.1 31.0 Pediatrics and Pediatric Subspecialties 6.4 7.0 8.3 8.4 8.3 9.0 9.0 9.3 Obstetrics and Gynecology 6.0 4.9 6.3 6.3 6.1 5.6 5.5 5.3 Surgical Specialties 26.1 25.3 24.9 21.6 21.9 19.5 18.7 18.0 Other Non-Primary Care Specialties 24.0 24.1 21.5 18.4 23.4 24.7 26.3 27.2 Unknown 17.4 16.5 8.2 12.4 5.0 1.7 1.6 1.5 Total 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 SOURCE: Congressional Budget Office calculations based on data for 1965 through 1985 from American Medical Association (AMA), Distribution of Physicians in the U. 8. (Chicago: AMA, 1967 and 1971), Physician Distribution and Medical Licensure in the US, 1975 (Chicago: AMA, 1976), and Physician Characteristics and Distribution in the U. 8. (Chicago: AMA, 1982 and 1986); and on data for 1988 through 1993 from the American Association of Medical Colleges. NOTES: Comparisons between the earlier and later years shown above must be made cautiously because specialties are unknown for a large proportion of residents in the earlier years and the three—year family practice residency was a new option in the 19705. In addition, different data sources were used for the earlier and later years. The "Unknown" category includes residents in a transitional year (a year of basic clinical training). Osteopathic doctors in MD. residency programs are included in the data. r—————~—————— . APPENDIX SUPPLEMENTARY TABLES 49 Table A-4. Residents Engaged in Patient Care per 1,000 People, by Specialty, Selected Years, 1965 Through 1993 Specialty 1965 1970 1975 1981 1985 1988 1990 1993 General Practice and Family Practice a 0.01 0.01 0.02 0.03 0.03 0.03 0.03 Internal Medicine, Pediatrics, and the Subspecialties of Internal Medicine and of Pediatrics 0.05 0.07 0.09 0.09 0.11 0.13 0.15 0.16 Other 0.12 0.13 0.14 0.13 0.16 0.17 0.18 0.20 Unknown M M 0._2 0.03 0.02 0.01 0.01 0.01 Total 0.22 0.25 0.26 0.27 0.31 0.33 0.36 0.39 SOURCE: Congressional Budget Office calculations based on data for 1965 through 1985 from American Medical Association (AMA), Distribution of Physicians in the US. (Chicago: AMA, 1967 and 1971), Physician Distribution and Medical Licensure in the U.S., 1975 (Chicago: AMA, 1976), and Physician Characteristics and Distribution in the U. 8. (Chicago: AMA, 1982 and 1986); and on data for 1988 through 1993 from the American Association of Medical Colleges. NOTES: Comparisons between the earlier and later years shown above must be made cautiously because specialties are unknown for a large proportion of residents in the earlier years and the three-year family practice residency was a new option in the 19705. In addition, different data sources were used for the earlier and later years. The "Other" category comprises residents in the specialties of obstetrics and gynecology, surgery, and other non-primary care fields. The ”Unknown" category includes residents in a transitional year (a year of basic clinical training). Osteopathic doctors in MD. residency programs are included in the data. a. Less than 0005 residents per 1,000 people. ISBN 0-16-048286-0 H “ 90000 80160 482861 U. C. BERKELEY LIBRARIES CDL‘W70553