MFORECASTS OF PHYSICIAN SUPPLY AND I REQUIREMENTS _ APRIL 1980 DOCUMEN Is DEPAR a“. mm; MAY 5 ,980 .i HER/3.5"! I, ‘ ”~5an 05 CA? WRRMA w CONGRESS 9F THE UNITED STATES Office 0! Technology Assessment .; Washington, D. c, 20510 ‘ ‘ ”RY Office of Technology Assessment Congressional Board Representative MORRIS K. UDALL, Arizona, Chairman Senator TED STEVENS, Alaska, Vice Chairman Senate House GEORGE E. BROWN, JR. California EDWARD M. KENNEDY Massachusetts ERNEST F. HOLLINGS JOHN D. DINGELL South Carolina Michigan ADLAI E. STEVENSON LARRY WINN, JR. Illinois Kansas ORRIN G. HATCH CLARENCE E. MILLER Utah Ohio CHARLES McC. MATHIAS, JR. JOHN W. WYDLER Maryland New York JOHN H. GIBBONS ex officio Director’s Office JOHN H. GIBBONS, Director DANIEL DE SIMONE, Deputy Director Advisory Council GILBERT GUDE CARL N. HODGES FREDERICK C. ROBBINS Chairman JEROME B. WIESNER Vice Chairman J. FRED BUCY CHARLES N. KIMBALL JOHN T. McALISTER, JR. CLAIRE T. DEDRICK RACHEL McCULLOCH JAMES C. FLETCHER ELMER B. STAATS LEWIS THOMAS The Technology Assessment Board approves the release of this report, which identifies a range of viewpoints on a significant issue facing the U.S. Congress. The views expressed in this report are not necessarily those of the Board, OTA Advisory Council, or of individual members thereof. @ORECASTS OF PHYSICIAN; SUPPLY AND REQUIREMENT; — APRIL 1980 OTA Reports are the principal documentation of formal assessment projects. These projects are approved in advance by the Technology Assessment Board. At the con- clusion of a project, the Board has the opportunity to review the report but its re- lease does not necessarily imply endorsement of the results by the Board or its indi- vidual members. ,“res r", \1’ “o éONGRESS OFT” NITED STATES ' \‘V‘ Office ot‘l’echnolo Assessme‘ry/ ‘ ' z /d"”‘<; 3“ my; as “’72 (P Kmfi L.» Library of Congress Catalog Card Number 80-600063 For sale by the Superintendent of Documents, U.S. Government Printing Office Washington, D.C. 20402 Stock No. 052-003-00746-1 ’Rflyflo ’7 Foreword é/ 5'9st / 9 go? 506%; Undertaken at the request of the Senate Committee on Labor and Human Re- sources, this report evaluates the assumptions, methods, and results of the two current models used to forecast the number and kinds of physicians the country is likely to need and have. Congress must rely heavily on such forecasts in shaping Federal policy and programs for aiding education in the health professions and for providing health resources and services. This report examines the two most important physician forecasting efforts— those of the Bureau of Health Manpower of the Department of Health and Human Services (DHHS) and those of the DHHS-chartered Graduate Medical Education Na- tional Advisory Committee. These two efforts together are generally representative of the kinds of techniques that are used to forecast physician and other health personnel supplies and requirements. The report points out that projections of physician supply and requirements de- pend on historical data to predict future events, but even recent historical data reflect past policies, not current ones. The limits of forecasts must be fully understood if they are to serve as effective tools in the shaping of Federal medical policy. Those limits could be made clearer by explicitly describing the assumptions behind any forecasts, by making alternative forecasts based on different sets of assumptions, and by expand- ing the forecasting process to include policymakers as well as technicians. This analysis was prepared by OTA staff. Drafts of the report were reviewed by an advisory panel convened for the study, by the Health Program Advisory Commit- tee, and by various individuals associated with the forecasting activities analyzed. m figt'w JOHN H. GIBBONS Director iii past OTA Health Program Advisory Committee Frederick C. Robbins, Chairman Dean, School of Medicine, Case Western Reserve University Stuart H. Altman Dean - Florence Heller School Brandeis University Robert M. Ball Senior Scholar Institute of Medicine National Academy of Sciences Lewis H. Butler Adjunct Professor of Health Policy Health Policy Program School of Medicine University of California, San Francisco Kurt Deuschle Professor of Community Medicine Mount Sinai School of Medicine Zita Fearon Research Associate Consumer Commission on the Accreditation of Health Services, Inc. Rashi Fein Professor of the Economics of Medicine Center for Community Health and Medical Care Harvard Medical School Melvin A. Glasser Director Social Security Department United Auto Workers Patricia King Professor Georgetown Law Center Sidney S. Lee Associate Dean Community Medicine McGill University Mark Lepper Vice President for Inter-Institutional Affairs Rush—Presbyterian—St. Luke's Medical Center C. Frederick Mosteller Professor and Chairman Department of Biostatistics Harvard University Beverlee Myers Director Department of Health Services State of California Mitchell Rabkin General Director Beth Israel Hospital Boston, Mass. Kerr L. White Deputy Director of Health Sciences Rockefeller Foundation Forecasts of Physician Supply and Requirements OTA Health Program Staff Joyce C. Lashof, Assistant Director, OTA Health and Life Sciences Division H. David Banta, Health Program Manager Lawrence Miike, Project Director Pamela Doty, Congressional Fellow Nancy Kenney, Administration OTA Publishing Staff John C. Holmes, Publishing Officer Kathie 8. Boss Debra M. Datcher Joanne Heming Advisory Panel Members E. Harvey Estes, Jr., Chairman Chairman, Department of Community and Family Medicine Duke University School of Medicine E. B. Campbell Ted Phillips Executive Vice President Associate Dean for Academic Affairs Lane College School of Medicine Jack Hadley University of Washington The Urban Institute Jane Record 10h“ Hatch Health Services Research Center School for Biomedical Education Kaiser Foundation City College of New York Lauren LEROY Alvin Tarlov Health POIICJ/ Program Department of Medicine SChOOl of Medicine School of Medicine University of California, San Francisco University of Chicago Charles Lewis Department of Medicine John Wennberg School of Medicine Department of Community Medicine University of California, Los Angeles Dartmouth College List of Acronyms AMA — American Medical Association AOA —- American Osteopathic Association BCHS — Bureau of Community Health Services BHM —— Bureau of Health Manpower BLS — Bureau of Labor Statistics CMG — Canadian medical graduate CPI — Consumer Price Index DHHS —— Department of Health and Human Services DO — doctor of osteopathy FMG — foreign medical graduate FTE — full—time equivalent GMENAC — Graduate Medical Education National Advisory Committee GNP — gross national product GP — general practitioner HIS — Health Interview Survey HMOs — health maintenance organizations HMSA — Health Manpower Shortage Area HSA — Health Service Area MD — doctor of medicine MUA — Medically Underserved Area NHI — national health insurance NHSC — National Health Service Corps vi Contents Chapter 1. SUMMARY AND CONCLUSIONS .............................. Introduction ................................................. Current Activities ............................................ Findings and Conclusions ....................................... Supply ................................................. Requirements ............................................ 2. SUPPLY .................................................... Aggregate Supply ............................................. Specialty Supply ............................................. Locational Distribution ........................................ Summary ................................................... 3. REQUIREMENTS ............................................ Introduction ................................................. Economic Models ............................................. The Bureau of Labor Statistics Model .......................... Bureau of Health Manpower Model ........................... The Framework ....................................... The Baseline Configuration .............................. Contingency Modeling ................................. Productivity ......................................... The Graduate Medical Education National Advisory Committee Model . . . Comparison of the BHM and GMENAC Models ..................... Productivity ................................................. Locational Requirements ....................................... BIBLIOGRAPHY ................................................. List of Tables Table No. 1. Derivation of Male and Female MD Retirement Rates and Death Rates by 5-Year Age Cohort ............................................................ 2. MD First-Year Enrollment Projections Using 1977 First-Year Enrollment as Base, to 1987 ............................................................ 3. DO First-Year Enrollment Projections Using 1976 First-Year Enrollment as Base, to 1987 ............................................................ 4. First-Year Enrollments in Medical and Osteopathic Schools Projected Under the Basic Assumption; 1978-79 Through 1987-88 .................................... . U.S.-Trained Physicians, Graduates; Projected 1978-79 Through 1989—90 .......... . U.S.-Trained Physicians, Graduates; Projected for 1980 and 1990 ................ . Supply of Active Foreign-Trained Physicians, Using Basic Methodology, Projected 1975-90 .................................................... 8. Basic, High, and Low Projections of the FMG Active Supply ................... “0‘01 15 15 23 30 38 45 45 47 47 49 50 53 59 61 62 72 77 80 85 Page 16 17 17 18 18 18 20 20 vii viii Contents—continued Table No. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. 44. 45. '46. 47. 48. Supply of Active Physicians by Country of Medical Education Using Basic Methodology: 1974 and Projected 1975-9O ................................. Supply of Active Physicians by Country of Medical Education Using Basic Methodology: Actual 1974, 1975; Projected 1980-90 .......................... Data Sources on Physician Specialty Supply ................................ Internship and Residency Data Sources .................................... First—Year Residency Distribution With Subspecialty Adjustment: Sept. 1, 1974 ..... Percent Distribution of U.S. /CMG First-Year Residency Projections Using Simple Linear Regressions ................................................... Percent Distribution of FMG First-Year Residency Projections Using Simple Linear Regressions ......................................................... Active Physicians, by Major Specialty Group: Actual 1974; Projected 1980-90 ...... Supply of Active Physicians, by Specialty: Actual 1974; Projected 1980-90 ......... Patient Care MDs by Selected Specialties and for High and Low States, 1976 ....... Non-Federal Physicians Providing Patient Care in Metropolitan and Nonmetropolitan Areas, 1963-75 ...................................................... New Physicians Entering Practice, 1973 to 1987 ............................. Percent of New Physicians Expected To Enter Primary Care .................... Projected County Classes of Newly Practicing Physicians ...................... Usual Source of Care for Urban Underserved Areas .......................... Estimated Principal-Provider Patient Loads of General Practitioners, Family Practitioners, and General Internists ...................................... Comparisons of Supply With Requirements Using Different Models .............. 1960’s Projections of Physician Requirements in 1975 ......................... Bureau of Labor Statistics Projections of Physician Supply and Requirements, 1985 . . Population Matrix Used in the BHM Model ................................ Health Care Categories Used in the BHM Model ............................. Projected Shifts in Age and Income Distribution, 1970-90 ...................... Projected Utilization Growth Factors, 1975-9O .............................. Prevalence of Selected Chronic Conditions Reported in Health Interviews, by Family Income .................................................... Number of Disability Days per Person per Year by Family Income, 1973 ........... Number of Physician Visits per Year by Poor and Not Poor Status and for Whites and Others, 1964 and 1973 ............................................. Number of Discharges From and Average Length of Stay in Short-Stay Hospitals, by Income Status and Color, 1964 and 1973 ................................ Estimated Growth in Per Capita Utilization, Four Forms of Health Care ........... Increase in Demand From Population and Per Capita Utilization Changes, 1975 to 1990, BHM Model ............................................. Dependence of Trend Projections on Alternative Starting Dates in the Baseline Data . Comparison of Linear Versus Logarithmic Extrapolation of Utilization Data ........ Allocation of Physicians by Type and Setting of Care for the 1975 Base Year, BHM Model ........................................................ Illustrative Computation of Manpower Requirements ......................... Specialty Areas and Subspecialties for Which Requirements Estimates Are Being Planned or Considered by GMENAC ..................................... The Average Practice Profile of General Surgeons ............................ Proportion of Persons Whose Experience With Physician Visits Is Beyond the Critical Threshold .................................................... Percent of Ethnic Groups Dissatisfied With Aspects of the Medical Care System ..... Regional Differences in Certain Health—Care Statistics, United States, 1969—70 ...... Shortage, Adequate, and Surplus Levels of Primary Care Physicians ............. Need for Primary Care Physicians and Psychiatrists in 1990 .................... Page 22 22 24 25 28 30 31 32 33 34 34 36 37 37 38 41 46 47 48 51 51 52 53 54 54 55 55 56 56 64 65 66 67 68 71 72 72 79 80 81 Contents—continued Table No. 49. 50. Criteria for Unmet Need Calculation by Area ............................... Physician Encounters per Physician and Physician Encounters per Physician Hour by Selected Cohorts, National Health Service Corps, 1976 ........................ List of Figures Figure No. 1. 2. HONOWVIO U1bJ>OJ Ht—l 12. 13. 14. 15. 16. 17. Diagram of Projection of Supply of Active Physicians Through 1990 ............. Trend Data on Number of First-Year Residents: Total, Primary, Nonprimary Care Specialties; 1960, 1968, 1974, and 1976 .................................... . Trend in Number of First-Year Allopathic Residents in Four Selected Specialties ..... “Projection of Active Physicians by Specialty ................................ . Frequency Distribution of Physician Availability Indexes—Primary Care Physicians and Surgeons for the 204 HSAs .......................................... . Per Capita Utilization of Physician Office Services, 1966-76 .................... . Per Capita Utilization of Short-Term Hospital Services, 1966-76 ................. . Per Capita Utilization of Dental Office Services, 1966-76 ...................... . Per Capita Utilization of Community Pharmacy Services, 1966-76 ............... . Non-Price-Related Per Capita Utilization Trends, Physician Office Services, 1966-76 . . Non-Price-Related Per Capita Utilization Trends, Short-Term Hospital Services, 1966-76 ............................................................ Non-Price-Related Per Capita Utilization Trends, Dental Office Services, 1968-76 . . . Non-Price-Related Per Capita Utilization Trends, Community Pharmacy Services, 1966-76 ............................................................ Illustrative Procedure for Arriving at Adjusted Needs Estimates ................. GMENAC Model for Estimating Physician Requirements ...................... Consumer Satisfaction With Physician Services ............................. Consumer Satisfaction With Physician Services, by Nature of the Experience ....... Page 82 84 Page 21 26 26 29 35 57 58 59 60 61 62 63 64 68 69 73 74 Reauthorization of the Health Professmns llEducational Assistance Act 0311th Law 94—484) is scheduled Tor 1980. ESsentiallw, theAQ-t rest fleets, Congress polieies ‘i‘iitowgrd Liiymedical and, 6ther‘ health professions educational support“: and toward identifying and addressing the prob- lems of medically underserved areas and pop- ulations man Resources, supported by the House Com— The senate Committee’ s letter pointed out that The request for this assessment originated With the Senate Committee on Labor and Hu- mittee on Interstate and Foreign Commerce. there have been wide variations in the numbers and types of phy51c1ans required and that as Congress begins to deal with the more d1fficult and the problem of medically underserved ~ Federal programs whose purposes are to build _ issues of spec1alty and geographic maldistri- -e-_-i7bution legislative policy will have to rely on V'Ttechnologies Tor estimating the adequacy of spe- -,c1alty and geographic distribution It would, therefore be helpful to. Congress that an analysxs‘ be undertaken oT the assumptions underlying-- the different forecasts, as well as the methods and conclusions of the forecasts themselves in ordeL to deter-mine which forecasting technolo-E gies are most reasonable Prejections of phySic1an supply and require- ments have influenced Federal policy toward and legislation on health profesmons education areas, and play an important role in existing serV1ces directly Until the 1976 Act, Federal policy was to in— ,crease the supply of physrc1ans and other health professionals, because the perception was that -,-»;of acute shortages AlthOugh the expiring legis—__ lation contains incentives to continue to accel- erate the supply of physmans, the general con-i- sensus now is that the aggregate supply of phy-g, ate supply Usually, this has meant that prima “care objectives have been phrased in terms of “Tithe percent of the aggregate physmian supply ,fthat should be in primary care. Such Ob]€Cth€s sicians‘f‘is at"“lea§t adequate and perhaps even in excess Hence, attention has turned toward the _,--_;-problems of speCialty and geographical or loca- yytional maldistribution Effort-s at correcting spec1alty maldistr1bution LL have concentrated on the primary care specul— __ ties, Wthl‘l are usually identiTed as general prac~ titioners,fam11y practitioners general inter- nists and general pediatric1ans All osteopathic physmans are also included although this pro—_ TessiOn is becoming more spectalized (about 40? percent are new spec1alists) Psychiatrists, ob-;_, stetr1c1an-gynecologists, and general surgeons " have sometimes been included , ., Definitional problems are obVious, and they :23 are important in determining the requirements for primary care physmans For example pri- mary eare phy51c1ans may include only those ' categories identified as primary care i. e diT- _-:1:i-Terent combinations of the cate ories identified above Ihe underlying rationale is that the way ”*‘in which medical care is prov1ded is cruC1al ___,Ihis approach sees primary care as requiring a change in attitude toward patient care a holistic approach to patients and their families and as prov1d1ng the appropriate entry pomt into the on officeabased ambulatory care regardless 0T, the specialty deSIgnation of the physmian pro-l Viding such serV1ces and estimate requirements ,, on that ba51s In addition to; definitional problems ap— proaches toward primary care have been remi- niscent of past approaches to aggregate physi- cian supply; the emphasis has been on simply 1ncreasmg the supply rather than simultane~ ously being concerned over what IS an appropri- ‘54: ' Eoreoigsts qfsPhysiciariSupply anti Reénireritentsit mgwould be inappropriate if aggregate supply were excesswe ‘ Geographical iior locational maldistribution 1s istgenerally a problem where health personnel and _ services are found inadequate by some defined standard to meet the health needs of the popu- lation of the identified communities areas, ,pr EV Locational maldistribu- tion is by definition a relative concept where institutional settings some of our people are determined to be at a dis— advantage relative to the rest of the United States. Once these are identified then ,tJ1e gapm between ealth personnel and services and thatEE‘" population 5 needs for them is quantified to1de- termine: 1) how man personnel are needed to bridge the gap, and 2) of the identified deflClen-1E Icy, how much of 1t Will be addressed through a ispec1f1c program Quantifying locational maldistribution serves wtwo purposes. First, it is uSed as part of the eligi~ EEEEEblllty criteria for; the Health ManpoWer Short- veinage Area (HMSA) deSignation for: 1) National CURRENT A £E::TIV|TIES Under the Health Professmns Educational the Department of Health and Human SerVices (DHHS) 18 required to proVide annual reports to the President and Congress on the status of health personnel 111 the-1.2:» " Asmstance Act of 1976, United States Estimatmg the present and future supply of and requirements for phy51c1ans and? other health professmns is the respons1bility of], Health Serv1ce Corps (NHS _, tion as serV1ce areas in which students who bor- row money under health professmns student reimbursement through Medicare and Medi-EE {Etmaldistribunon are used _,,_,11_,51ze of NHSC. That is, g _, ,_ "verse of ex1sting and tuture HMSAS pla 5-5;;11-1be made for determining how many of _ medical manpower shortage areas will be , ‘ii§é?"staffed by NHSC phySiCians Currently, the ma— students who Wlll be obligated to NHSC 11 ex- change for scholar hip support grams) s‘i'ould (could) be changed to meet these ) Sites, 2) des1gna- loan programs ca ractice in lieu of repaying , the loans' in money, 3) grants for various he th manpower training programs £4) eligibility or preference for grant funds for several Bureau of Community Health 8 Vices programs such as the urban and rural health in'tiatives and 5) certification of rural health clinics for nurse; practitioner s and phy51c1ans ass1stants sen/ices Second these methods Eto quantify locational planfor the future 'or source for those future HS pOSitions are ate theditéal eElucatio Its most immediate im- pact Will; come from ts recommend"""ons on??? how gra _,uate medical education (IESIdN cy‘ pro-1 _ goals GMENAC was given a the Health Resources Administration through“ _,1,its Manpower Analysis Branc of the Eureau at, “Heaith anpower (B 91- M) DHHS has produced sits first report (dated Anglist 1978 and reprinted in March 1979) and is in the::f1nal stages‘of re- view foi: its next re ‘E’ """ In addition DHHS chartered a Graduate Medical Education National Admsory Commit- distribution and methods for financmg 3r d prowded by V; economy, which proyects the future tional product (GNP) and its compo governmental expenditures :1 netexriiiortsiifiii dustrial Eoutpiit and productiwty, the l b force average weekly hours of work and E model fOr Its ph lain demand pizoje o E E pl6yn1ent forEE__deta11ed Industry groups‘f‘an‘ Thus there 3rE es two major effo s ‘ cupatlons ,, , - : The? Bureau of Labor §tatistics consxders the Y _ Q0 to 2 000 additlons per year from for gn ‘ edlcal schools 1n the 1980 s. sehols takeirr'first-Y‘e’ari‘ ollrhent pr01ect10ns, adjusted for attritiogs to arrive aW the number 6f graduates per year. Estlmates 6f flrstwyear en— rollments are based on trends in: _ capitation support 2)5 Federal grants a iVity, 3) new schoels already planned and 4) potential State and local support 0£ new schools ,_ __ E The net effect of overestimatmg domestlc sources an derestimatmg forelgn sources” could wash" each other out. , 6”, 000 physmrans in 19 Et the assumptions curt a fixed number y in use explxcrtly '51stance ’Act of 1976 whlch was éeSIgned to“ sharply curtail the 1mmlgration Eof ph ,- mto tJ1e United States “" ty of the prOJeCted supply and o£ the influence of Supply pr“ )ections leave: the impressmn that ize the influence of policy ntzésupply Estr- , mates based 6n different sets of assum ‘tions could provrde better indications of the vanabili- 6252;. Farecdsts Qf Physician Shpply and RedLuireLanenLth For foreign graduates tored into the model ty of ever gaining authorized capitation levels, although private medical schools continue to be 2:2;developed And the impaCt of Public LaW 94-484 on dampeni Lg foreign medical graduate sources maybe Circumvented by the increasing ,,,,, number of US. citizens studying Lmedicme abroad and eventually returning to the Llnited States to practice ; _ The specralty distribution of the proleCted supply 15 estimated by taking the number 0f ac- tive practitioners by (self-de51gnated) speCialty, Trends in first-year re51dency p051tions are used to predict future speCialty distribution because of lack; of data on final- -year residency ,p051tions However, first-year reSidency posi-: tions are often used for general clinical ex— LLL’perience prior to concentration in a particular itsubspeCialty or in another speCialty anH there- _ fore do nOt necessarily represent final specialty chmces ii;.e , ; first-year residency counts are duplicative for particular Specialties in that a _,E__proportion move on to subspeCialization or to LL another speCialty altogether. BHM’ 5 current proiections assume that the first- -yeLLar re31dency distribution trends for 19681197044, and 1976, L also apply through 1980~81 After 1980- 81, the re51dency distribution is held constant for the '9’? statistical reason that the base years chosen to establish the trend cover 6 year5 so BHM has chosen not to extend the extrapolation beyond 6 years. Downward adJustments are made to minimize double-"ounting, the greatest adjust- merits occur in general surgery (62 percent) and” internal medicme (32 percent) L ., As a percent of the total projected supply, physicians in general practice, family practice “internal medicme, and pediatrics (those usually -::counted as primary care speCialties) are pro- the presumed full 1111- pact of Public Law 94-484 is deliberately fac- 'eior domestic sources, full capitation and continued development of newLL’ medical schools in the 1980’ s are also asSumed The la er also reflects a presumed fuil impact of "’eXisting Federal law, but past experience and 272current consensus would deny the real p0551bili~ 2'9‘Tj'ect‘ed to comprise 39 percent in 1980, 41 percent Lin 1985, and 42 percent in 1990. The largest [spe— cialty among these, as well as among all the spe- c1alties, will be internal medicine, which will have more than twice as many phy51c1ans than any one Of the other speCialties 2 The locational distribution of the prolected supply, by specralty, is estimated by similar methad5 as for aggregate and specialty supply; _L i. e., current supply plus additions Thes5 loca- tiLonaLl proiections can be disaggregated in a vari- ety 0f ways; e. g; by geographic criteria such as by States, counties, Census-Defined State Eco- nomic Areas, 0r Health Service Areas, ., or by special populations such as institutional care ,(mental hospitals, prisons), the indigent, and, Native Americans ' Locational prOJections ar5 used to identify, Lthose lucations with the least number of physi-L 22;;rcians for programs which intend to place physi— Lycians (e. g., NHSC) or for which shortage des—y LLLignation is necessary to qualify for Government ~y:r~‘~funds Th5 process of de51gnating and staffing HMSAs presently includes estimating the future supply of physicians for-1: 1) rural counties; 2) ur- ban areas; 3) 3Federal, State, and lucalL prisons 4) 4State mental hospitals and community mental health centers, and 5) the Indian Health Service. Projections of specialty and locational supply _ depend on the standard method of relying on historical data to predict future events, and in particular, On most recent experience to predict the most immediate future. This can be seen in the use of mid- to late 1960’s to mid-1970’s data-2‘ to predict 1980-90 patterns Aside from the in-LL Lev1table finding of’zi’ inadequate data" which, for zone of the most important marker specralties” y,_(in_t_ernal medicine), contains an errpr factor of, at least 32 and perhaps as high as 62 percent in the first-year residency count, the use of his- , torical data has two other limitations in these proiections of specialty and locational distribu- tion. The lat5 1960’s and 1970’s have Witnessed 1) Medicare and Medicaid and greater third- party priVate insurance coverage, 2) unprece- dented increases in medical school enrollments = and a large influx of foreign medical graduates, trend (see figure on p 9) Based on the BHM model an alternative apt” , proximation of the demand for physician serv__- ' changes, and assuming no long-term trend ‘ toward increases in per capita use, WOuld be 415, 00 phys1c1ans, an increase of 37, 000 from 378, 000' 1n 1975 But We could change, as could product1v1ty To sOme extent these are policy _ Choices to be made. If it is considered desirable for use to rise, for physicians to spend a Tew ex- ‘ tra minutes with each patient, Or for physicians .to have shorter workweeks, muCh of the pro- jected supply of 600, 000 phys1cians in 1990 could be appropnate As supply 15 estimated to be 680 000 in 1990, there is a dlfference of 185, 000 physicians be: _5_tween predicted Supply and estimated demand 5' 1n a static Situation. _, Some flex1b111ty 1n tHe model is necessary, Tor several reasons The enactment of national health msurance should lead to some increase in the demand far phys1c1an services. Second phy— ' physman pro ' 1990 which ' __ Increases in demand attributable to a histor ical trend toward increased per capita use are overestimated particularly for office services. The period 1968-76 is used to estabhsh the?" 55_trend,b,11t whereas a start date of 1968 yields ass; distinctly upward trend for physician office 5 services, a start date of197-1 yields a downward, grega " of the parent GMENAC panel’ the Work of the ind counseling and lead to'greater patient satlsfac- tion With the quallty 0E medical care of these change are desnable at55-the‘ cost that , 5 Will be borne by the soc1ety ices in 1990 adjusting only for demograph1c The GMENAC normative, medical Opinion model estimates all diSeases and conditions (on _, demograph1c bases such as age and sex) that shOuld be treated by physicians and the amount ”' ”of phys1c1an sen/ices, on a disease-by-disease or 'condltlomby-condltion basrs, that should be“ provided The theoretical level Of use is usually adjusted downwards to real-world estimates through consensus formatlon techmques quantifying use by health Care setting, these esti~ _ mates quantify use on a spec1alty-by-spec1alty _ _, ba51s , ,, _ Instead of Unlike the BHM model WhiCh Can project de— 5 mand year to year (prowctions n0w exist up to 2000), GMENAC's current future target is 1990 although its model ‘15 capable of prov1d1ng year- _, to—year projections GMENACS mo ling ef— fort, because its ultimate aim is to prov1de rrrecommendations on graduate medlcal educa~ gtion, professes t ss'concerned with ag- requirements. When addressed aggro, gate requirements will be more of a byproduct" idual spec1a SOURCE JWK International Jnc HRA 232-78 0111) 1579 On the other ha (1 he BHM model as pres—- ently“ constructed can 0 1y prowde aggregate and not speCIalty-sp H requ1re—» Rments because demand is g oLiped by health; iycare ettmg, met by spemalty Care. better capable of estlmatmg spec1alty-by-L; ‘spec1alty requirements bu cauld overestlmate difficulty bf reconc111ng overlappmg patlent care responmbtlltIes. This task 1 to be 11 d rtaken by the GMENAC pafi'el after he wark'of £115 spe- c1alty 1111:1515 completed “ i' A11 unresolved 1s§ue, 1,1 111 nts for the pnmary care spec1alhes There are basic d1fere11ces 01:1 what is primary care, d15- agrewmwnt over what spenatles constltute the prim ry care ones, and 1e pragmatlc ptobiem? 11131- other spec1allsts 1ml 1: nt _ue_ to prowde e1 are compkementary, sumatmg future ‘11 gr V, gate requirementsf m B‘ M 3 tie oft ‘ AGGREGATE SUPPLY The future aggregate supply of physicians is based on assumptions of the following factors (USDHEW, 1979a): 0 physicians currently active in practice, 0 new graduates of U.S. medical and osteo- pathic schools, and 0 immigration of physicians (including U.S. citizens studying abroad) educated in other countries. The estimates based on these production fac- tors assume that supply will not be affected by the demand for physicians; i.e., there is an in- elastic relationship between physician produc- tion and demand. Data on currently active physicians are ob- tained from the American Medical Association (AMA) and the American Osteopathic Associa- tion (AOA). Both AMA and ADA data rely on the physician's self-designation of specialty, so the published data are based on this self-iden- tification of primary specialty and activity and provide no information on activities in other specialty areas nor on the proportion of time spent in actual patient care. An additional factor is that the “not clas- sified" category in the AMA data, introduced in 1971, has grown from 300 in 1971 to over 30,000 in 1976, plus approximately 8,800 physi- cians whose addresses were unknown. Seventy percent of this ”not classified” category is below age 35 and most likely in active practice. In the trend analysis for estimating specialty distribu- tion, this “not classified" category is not in- cluded. However, ”not classified” is included in the aggregate projections, with the assumption that its specialty distribution is identical to physicians in residency programs. For physicians currently active in practice, the starting point (base year) is 1974. Data for MDs include age, specialty, and country of medical education. DO data for 1974 start with 1971 AOA data and add new DOS and subtract retirements and deaths between 1971 and 1974. Mortality and retirement rates for MDs are computed by age and sex as derived from studies on the physician population, not the general population. 1967 data on retirement rates are used, and mortality rates use an article published in 1975. These rates are also applied to osteopathic physicians. Both retirement and mortality data, therefore, do not reflect trends that might be occurring. Table 1 summarizes these estimates. Trends in new graduates of U.S. medical and osteopathic schools start with estimates of first- year enrollments to arrive at the number of graduates per year after adjustments for attri- tion. 1974 data were the original starting point, but data from the latest academic year, 1977-78, are now used. Estimates of first-year enrollments are based on trends in: 1) Federal capitation support, 2) Federal construction grants activity, 3) new schools already planned, and 4) potential State and local support of new schools. Separate com- putations are made for first-year enrollments in 15 16 0 Forecasts of Physician Supply and Requirements Table 1.—Derivation of Male and Female MD Retirement Rates and Death Rates by 5-Year Age Cohort Age Total MDs Number inactive Percent inactive Retirement rate Death rate Separation rate Male M Ds Less than 30 ........ 31,047 64 .0020 — .0007 .0007 31-34 .............. 39,470 64 .0016 .0000 .0007 .0007 35-39 .............. 38,562 88 .0023 .0001 .0014 .0015 40-44 .............. 37,501 107 .0029 .0001 .0022 .0023 45-49 .............. 32,989 156 .0047 .0004 .0043 .0047 50-54 .............. 27,319 188 .0069 .0004 .0066 .0070 55-59 .............. 25,100 370 .0147 .0016 .0111 .0127 60-64 .............. 19,452 708 .0410 .0053 .0188 .0241 65-69 .............. 13,368 1,483 .1 109 .0140 .0294 .0434 70-74 .............. 8,941 2,034 .2275 .0233 .0465 .0698 75 and over ......... 11,817 5,186 .4389 .0423 .1243 .1665 Female M Ds Less than 30 ........ 3,568 70 .0196 -- .0005 .0005 31-34 .............. 2,929 157 .0536 .0007 .0008 .0015 35-39 .............. 2,617 166 .0634 .0020 .0013 .0033 40-44 .............. 2,894 226 .0781 .0029 .0023 .0052 45-49 .............. 2,313 163 .0705 .0015 .0028 .0013 50-54 .............. 1,832 151 .0824 .0024 .0043 .0067 55-59 .............. 1,410 126 .0894 .0014 .0064 .0078 60-64 .............. 1,105 139 .1258 .0073 .0098 .0171 65-69 .............. 993 242 .2437 .0236 .0152 .0388 70-74 .............. 779 290 .3723 .0257 .0250 .0507 75 and over ......... 964 630 .6535 .0562 .0916 .1478 Based on: 1) American Medical Association, Department of Survey Research, Selected Characteristics of the Phys/cian Population, 1963 and 1967 (Chlcago, 1978), table 21, p. 162; and 2) R. Hendrickson, “Speclallsts Outllve Generallsts," Prism, December 1975. SOURCE: Interim Report at the Graduate Medical Education National Advisory Committee to the Secretary, Department of Health, Education, and Welfare, Washington, 0.0.: Health Resources Adminlstration, DHEW publicatlon No. (HRA) 79-633, p. 119. 3-year programs because of different attrition rates. Transferees into U.S. medical schools are also estimated. High, low, and basic projections are calcu- lated for these first-year enrollments. Basic pro- jections assume that full funding of capitation grants and moderate funding of construction grants will be achieved by 1981, that seven new medical and osteopathic schools will be estab- lished after the 1977-78 school year, and that there will be some limited further State, local, and private support for additional enrollment growth. The low-level projections assume full funding of capitation grants, but minimum funding of construction grants by 1981, the es— tablishment of four new schools after 1977-78, and no additional growth in enrollments arising from State, local, or other support beyond 1977-78. The high-level projections assume full funding of both capitation and construction grants by 1981, the establishment of 10 new schools after 1977—78, and additional growth in enrollments arising from State, local, or other support beyond 1977-78 at half the annual rate exhibited by the years 1953-54 through 1964-65 (before Federal programs had an impact). Tables 2, 3, and 4 summarize these estimates for MD and DO first-year students. Attrition rates are based on historical trends for 3- and 4-year MD programs, for osteopathic programs, and for foreign-trained U.S. medical students who transfer to U.S. medical schools. Table 5 summarizes actual and projected gradu- ates for 1978-79 to 1989-90, based on the fore- going assumptions. Table 6 summarizes similar projections made at about the same time for the Department of Health, Education, and Welfare’s (HEW) (now the Department of Health and Hu- man Services (DHHS)) annual report to the President and Congress (USDHEW, 1979b). The two tables show different projections for 1980 and 1990; for 1980, 16,375 v. 17,155; for 1990, 19,289 v. 19,987. The lower estimates are based on the foregoing assumptions. There are also discrepancies between the first— year enrollment assumptions and the projected Table 2.—MD First-Year Enrollment Projections Using 1977 First-Year Enrollment as Base, to 1987 1977-78 1978-79 1979-80 1980-81 1981 -82 1982-83 1983-84 1984-85 1985-86, 1986-87 1987-88 Low series ' . Total . , ............. 16,136 16,486 16,908 16,921 16,931 16,936 16,938 16,940 16,942 16,944 16,944 Base year ............... 16,136 16,136 16,136 16,136 16,136 16,136 16,136 16,136 16,136 16,136 16,136 Construction commitments — 300 700 700 700 700 700 700 700 700 700 New schools ............. — 50 72 85 95 100 102 104 106 108 108 Basic series Total ............... 16,136 16,725 17,350 17,525 17,612 17,690 17,765 17,838 17,909 17,980 18,047 Baseyear ............... 16,136 16,136 16,136 16,136 16,136 16,136 16,136 16,136 16,136 16,136 16,136 Construction commitments ,-— 450 950 1,025 1,025 1,025 , 1,025 1,025 , 1,025 1,025 1,025 New schools ............. — 74 134 169 191 204 214 222 228 234 236 Other ................... — 65 130 195 260 325 390 455 520 585 650 High series Total ............... 16,136 17,013 17,748 18,019 18,188 18,340 18,485 18,628 18,769 18,906 19,037 Baseyear ............... 16,136 16,136 16,136 16,136 16,136 16,136 16,136 16,136 16,136 16,136 16,136 Construction commitments — 650 1,150 1,225 1,225 1,225 1,225 1,225 1,225 1,225 1,225 New schools ............. — 98 204 271 311 334 350 364 376 384 386 Other ................... — 129 258 387 516 645 774 903 1,032 1,161 1,290 SOURCE: Interim Report at the Graduate Medical Education National Advisory Committee to the Secretary, Department of Health, Education, and Welfare, Washington, 0.0.: Health Resources Administration, DHEW publication No. (H RA) 19-633, p.135. Table 3.—First-Year Enrollment Projections Using 1976 First-Year Enrollment as Base, to 1987 1976-1977 1977-78 1978-79 1979-80 1980-81 1981-82 1982-83 1983-84 1984-85 1985-86 1986-87 1987-88 Low series Total ............... 1,068 1,218 1,309 1,354 1,411 1,429 1,447 1,464 1,481 1,498 1,515 1,532 Base year ............... 1,068 1,068 1,068 1,068 1,068 1,068 1,068 1,068 1,068 1,068 1,068 1,068 Construction commitments — 90 154 184 214 214 214 214 214 214 214 214 New schools ............. —- 60 87 102 129 147 165 182 199 216 233 250 Basic series Total ............... 1,068 1,258 1,364 1,437 1,522 1,562 1,603 1,643 1,682 1 ,722 1,762 1,801 Base Year ............... 1,068 1,068 1,068 1,068 1,068 1,068 1,068 1,068 1,068 1,068 1,068 1,068 Construction commitments — 95 164 199 234 234 234 234 234 234 234 234 New schools ............. — 84 111 138 177 207 237 266 295 324 353 382 Other ................... — 11 21 32 43 53 64 75 85 96 107 117 High series Total ............... 1,068 1,273 1,432 1,534 1,658 1,721 1,782 1,843 1,903 1,963 2,024 2,084 Base year ............... 1,068 1,068 1,068 1,068 1,068 1,068 1,068 1,068 1,068 1,068 1,068 1,068 Construction commitments — 100 214 264 264 264 264 264 264 264 264 264 New schools ............. — 84 188 241 282 322 361 361 400 439 478 517 Other ................... — 21 64 85 107 128 10 150 171 192 214 235 SOURCE: interim Report of the Graduate Medical Education National Advisory Committee to the Secretary, Department of Health, Education, and Welfare, Washington. D.C.: Health Resources Administration, DHEW publication No. (HRA) 19633, p.136 AI 0 filddns—Z 113 18 0 Forecasts of Physician Supply and Requirements Table 4.—First-Year Enrollments in Medical and Osteopathic Schools Projected Under the Basic Assumption; 1978-79 Through 1987-88 Total MD and DC MD first-year DO first-year Academic year first-year enrollments enrollments enrollments 1978-79 ............. 18,089 16,725 1,364 1979-80 ............. 18,787 17,350 1,437 1980-81 ............. 19,047 17,525 1,522 1981-82 ............. 19,174 17,612 1,562 1982-83 ............. 19,293 17,690 1,603 1983-84 ............. 19,408 17,765 1,643 1984-85 ............. 19,520 17,838 1,682 1985-86 ............. 19,631 17,909 1,722 1986-87 ............. 19,742 17,890 1,762 1987-88 ............. 19,848 18,047 1,801 SOURCE: Interim Report of the Graduate Medical Education National Advisory Committee to the Secretary, Department of Health, Education, and Welfare, Washington, D.C.: Health Resources Administration, DHEW publication No. (HRA) 19-633, p.146. Table 5.—U.S.-Trained Physicians, Graduates (MD and DO); Projected 1978-79 Through 1989-90 Academic year Total graduates MD graduates DO graduates 1978—79 ............. 16,044 15,048 996 1979-80 ............. 16,375 15,346 1,029 1980-81 ............. 16,997 15,789 1,208 1981 -82 ............. 17,662 16,354 1,308 1982-83 ............. 18,333 16,956 1,377 1983-84 ............. 18,699 17,241 1,458 1984-85 ............. 18,818 17,322 1,496 1985-86 ............. 18,928 17,394 1 ,534 1986-87 ............. 19,036 17,464 1,572 1987-88 ............. 19,142 17,532 1,610 1988-89 ............. 19,201 17,554 1 ,647 1989-90 ............. 19,289 17,604 1,685 SOURCE: Interim Report of the Graduate Medical Education National Advisory Committee to the Secretary, Department of Health, Education, and Welfare, Washington, D.C.: Health Resources Administration, DHEW publication No. (HRA) 19-633, p.147. Table 6.—U.S.-Trained Physicians, Graduates (MD and D0); Projected for 1980 and 1990 MD DO Year Schools Graduates Schools Graduates Total graduates 1960 .............. 86 7,081 6 427 1970 .............. 103 8,367 7 432 1975 .............. 114 12,714 9 698 13,412 1980 (projected) ..... 121 16,086 13 1,069 17,155 1990 (projected) ..... 121 18,318 13 1,069 19,987 SOURCEzA Report to the President and Congress on the Status of Health Professions Personnel in the United States, Washington, D.C.: Bureau of Health Manpower, Health Resources Administration, DHEW publication No. (HRA) 79-93, p.||-29. numbers of graduates. The AMA's annual re- port, Medical Education in the United States (AMA, 1978), lists 122 medical schools accred- ited or provisionally accredited and 16,134 first- year students in 1977-78, plus 2 schools accred- ited or provisionally accredited for the first 2 years of the MD program whose first-year en- rollments apparently were not included in the 1977-78 total of 16,134. And the Association of American Medical Colleges identified 2 addi- Ch. 2—Supply 0 19 tional medical schools in 1979, for a total of 126 (American Medical News, 1979). The projec- tions of first-year enrollments for 1977-78 match the AMA’s estimates of the number of first-year enrollees in medical school (16,136 v. 16,134). But the projections to 1980 and 1990 (table 6) state that there will be 121 medical schools and 13 osteopathic schools, compared to 114 medi- cal schools and 9 osteopathic schools in 1975. Thus, it is not clear whether the alternative estimates of 7, 4, or 10 new medical and osteo- pathic schools include some of the 122 medical schools already in existence, or whether they represent additional schools, as the explanation of the methodology seems to say. In addition, the assumption of full capitation funding by 1981 also is unrealistic, and the pro- jections also seem to indicate that full capitation is expected to be maintained after 1981. Cur- rently, the issue with capitation is whether it will continue at all, not whether fully author- ized levels will be appropriated. Immigration of graduates of foreign medical schools are calculated separately for Canadian medical graduates (CMGs) and other foreign medical graduates (FMGs). The Canadian addi- tion is currently estimated to equal losses from death, retirement, and emigration because the recent historical growth has leveled off. Additions from the rest of FMGs are particu- larly uncertain at this time because of the cur- tailing legislation in the Health Professions Edu— cational Assistance Act of 1976 (Public Law 94-484). Since historical trends will not be pre- dictive of future additions to supply by FMGs, the 1974-76 period has been used, with major adjustments that essentially try to guess at the impact of the legislative changes. Temporary- visa FMGs are assumed to equal the number of graduate medical positions available to them through regulations that implement Public Law 94-484, which require a stepwise reduction in positions until available positions to FMGs reach zero by 1990. The addition of permanent- visa FMGs to the supply is based on estimates of the number of FMGs passing the National Board of Medical Examiners' Visa Qualifying Exam- ination. This exam was begun in September 1977, so only 1 or 2 years of data are available. Permanent-visa FMGs and the proportion of temporary—visa FMGs estimated to establish permanent status through marriage (based on actual trends) are assumed to have the same death and retirement rates as U.S.-educated physicians. Of crucial importance is the apparent lack of analysis of the contribution from US. citizens studying medicine abroad, a situation currently under study by the General Accounting Office. The projections do account for students return- ing to the United States to complete their medical education in the United States, but they comprise only a small part of the pool of US citizens studying medicine abroad. Basic, high, and low projections are calcu- lated for the FMG addition to supply. The basic projection is summarized in table 7, with the results of the alternative estimates of the active FMG supply from the basic, high, and low esti- mates summarized in table 8. These supply projections are prepared in two matrices. The first matrix projects year-by-year future MD graduates and attrition from the ac- tive work force by country of medical educa- tion. The second matrix distributes these future graduates and attrition of active practitioners by specialty, each by country of medical educa- tion. The first matrix projects graduates and foreign additions utilizing estimates of first-year enrollments, student attrition, other medical- school-related trends, and the model of FMG (including Canadians) immigration. The second matrix distributes the graduates among medical specialties through projections of first—year residency trends, and computes deaths and re- tirements of active practitioners among the spe- cialties, using the mortality and retirement rates described earlier. Comparable disaggregation of the data on DOs has not been developed, although estimates of total DO supply have been made. The method is summarized in figure 1. Table 9 summarizes the projected supply of physicians through 1990. For comparative purposes, table 10 summarizes estimates made in early 1978 (USDHEW, 1979b). The estimates are ap- proximately equal. It should be noted that the 20 0 Forecasts of Physician Supply and Requirements Table 7.—Supply of Active Foreign-Trained Physicians, Using Basic Methodology, Projected 1975-90 New entry supply Losses Active supply Death and J-visa Year Total Permanent Temporary Total retirement emigrants FMG CMG 1974 ............ - — — — — — 70,940 5,510 1975 ............ 7,316 3,898 3,418 2,166 764 1,402 76,090 5,510 1976 ............ 6,609 3,399 3,210 2,569 815 1,754 80,130 5,510 1977 ............ 6,596 3,399 3,197 2,626 872 1,754 84,100 5,510 1978 ............ 4,150 1,152 2,042 2,680 917 1,763 85,570 5,510 1979 ............ 4,857 2,521 2,336 2,737 983 1,754 87,690 5,510 1980 ............ 3,847 2,521 1,326 2,107 1,047 1,060 89,430 5,510 1981 ............ 4,591 2,521 2,070 2,371 1,109 1,262 91,650 5,510 1982 ............ 3,581 2,521 1,060 1,751 1,184 567 93,480 5,510 1983 ............ 4,325 2,521 1,804 2,355 1,276 1,076 95,450 5,510 1984 ............ 3,315 2,521 794 1,735 1,351 384 97,030 5,510 1985 ............ 4,059 2,521 1,538 2,349 1,453 896 98,740 5,510 1986 ............ 3,049 2,521 528 1,739 1,538 201 100,500 5,510 1987 ............ 3,793 2,251 1,272 2,353 1,640 713 101,490 5,510 1988 ............ 3,023 2,521 502 1,923 1,740 183 102,590 5,510 1989 ............ 3,287 2,521 766 2,227 1,862 365 103,650 5,510 1990 ............ 3,023 2,521 502 2,153 1,971 182 104,520 5,510 SOURCE: interim Report 0/ the Graduate Medical Education National Advisory Committee to the Secretary, Department of Health, Education, and Welfare, Washington, D.C.: Health Resources Administration, DHEW publication No. (HRA) 19-633, p.140. Table 8.—Basic, High, and Low Projections of the FMG Active Supply Basic High Low Year FMG Canadian FMG Canadian FMG Canadian 1975 ............ 76,090 5,510 76,090 5,510 76,090 5,510 1980 ............ 89,430 5,510 92,340 5,510 86,270 5,510 1985 ............ 98,740 5,510 104,340 5,510 92,910 5,510 1990 ............ 104,520 5,510 112,580 5,510 96,320 5,510 SOURCE: Interim Report of the Graduate Medical Education National Advisory Committee to the Secretary, Department a! Health, Education, and Welfare, Washington, D.C.: Health Resources Administration, DHEW publication No. (HRA) 19-633, pp. 140-142. estimates in table 9 have lower projections of the graduate supply and higher projections of the FMG supply than the estimates in table 10. This is despite the optimistic projections of capitation funding and even further curtailment of the FMG supply that underlie the table 9 pro- jections. Interestingly enough, projections from the Bureau of Health Manpower (BHM) made in 1974 (USDHEW, 1974) were similar to those made in its report to the President and Congress (USDHEW, 1979b) but the contribution from the graduate supply was lower and that from FMGs higher. In other words, the Bureau's previous estimates, made before the 1976 law curtailing FMG immigration, are more internal- ly consistent with the estimates taking into con- sideration the effect of the 1976 law. So even though the aggregate projections of supply are similar for these different sets of assumptions, the contribution of the components of the ag- gregate estimates has differed significantly. Ch. 2—Supply 0 21 Figure 1.—-Diagram of Projection of Supply of Active Physicians Through 1990 aFYE = first-year enrollment. SOURCE: Interim Report of the Graduate Medical Education National Advisory Committee to the Secretary, Department of Health, Education. and Welfare, Washington, 0.6.: Health Resources Administration, DHEW publication No. (HRA) 79-633, p. 112. 22 0 Forecasts of Physician Supply and Requirements Table 9.—Supply of Active Physicians (MD and D0) by Country of Medical Education Using Basic Methodology: 1974 and Projected 1975-90 Category 1974 1975a 1980 1985 1990 Number of active physicians All active physiciansb . . . . 362,500 377,400 447,800c 523,600 596,800 U.S.-trained ................ 286,000 295,800 352,800 419,300 486,900 MD ...................... 272,400 281,700 335,100 396,100 457,000 DO ...................... 13,600 14,100 17,700 23,200 29,900 Canadian-trained MDs ....... 5,600 5,500 5,600 5,600 5,600 Foreign-trained MDs ......... 70,900 76,100 89,400 98,700 104,500 Rate per 100,000 population All active physicians ..... 171.1 176.8 201.5 224.8 245.1 U.S.-trained ................ 135.0 138.5 158.8 180.0 200.0 MD ...................... 128.6 131.9 150.8 170.1 187.7 DO ...................... 6.4 6.6 8.0 10.0 12.3 Canadian-trained MDs ....... 2.6 2.6 2.5 2.4 2.3 Foreign-trained MDs ......... 33.5 35.6 40.2 42.4 42.9 Percent distribution All active physicians ..... 100.0 100.0 100.0 100.0 100.0 U.S.-trained ................ 78.9 78.4 78.8 80.1 81.6 MD ...................... 75.1 74.6 74.8 75.6 76.6 DO ...................... 3.8 3.7 4.0 4.4 5.0 Canadian-trained MDs ....... 1.5 1.5 1.3 1.2 0.9 Foreign-trained MDs ......... 19.6 20.2 20.0 18.9 17.5 aAvailable estimates for 1975 and 1976 for active U.S.-trained MDs are 282,800 and 290,900 respectively; active FMGs are estimated at 76,200 and 79,700 respectively. Ac- tive Canadian-trained MDs are estimated at 5,500 for both years. Assumes that the percent active of the AMA “not classified" MDs is the same as the percent “professionally active" of the classified MDs including those with ad- dress unknown. cOriginal table added this column incorrectly to total 477,800. Population figures used (in millions): 1960: 185.4; 1970: 206.1; 1974: 211.9; 1975: 213.5; 1980: 222.2; 1985: 232.9; 1990: 243.5. SOURCE: Interim Report of the Graduate Medical Education National Advisory Committee to the Secretary, Department at Health, Education, and Welfare, Washington, D.C.: Health Resources Administration, DHEW publication No. (HRA) 19-633, p.144. Table 10.—Supply of Active Physicians (MD and DC) by Country of Medical Education Using Basic Methodology: Actual 1974, 1975; Projected 1980-90 Category 1974 1975 1980 1985 1990 Number of active physicians AII active physicians ..... 362,500 378,600 444,000 519,000 594,000 U.S.-t rained ................ 286,000 296,700 353,600 424,400 495,700 MD ...................... 272,400 282,600 335,900 401,100 465,900 DO ...................... , 13,600 14,011 17,700 23,300 29,800 Canadian-trained MDs ....... 5,600 5,700 6,000 6,100 6,200 Foreign-trained MDs ......... 70,900 76,200 89,400 88,500 92,100 Rate per 100,000 population All active physicians ..... 171.1 177.3 199.3 221.7 242.4 U.S.-tralned ................ 135.0 138.9 158.7 181.3 202.3 MD ...................... 128.6 132.3 150.8 171.4 190.1 DO ...................... 6.4 6.6 7.9 10.0 12.2 Canadian-trained MDs ....... 2.6 2.7 2.7 2.6 2.5 Foreign-trained MDs ......... 33.5 35.7 37.9 37.8 , 37.6 SOURCE: A Report to the President and Congress on the Status of Health Professions Personnel in the United States, Washington, D.C.: Bureau of Health Manpower, Health Resources Administration, DHEW publication No. (HRA) 79-93, p.A-25. Ch. 2—Supply ‘ 23 SPECIALTY SUPPLY Recall that aggregate supply was prepared in two matrices. The first matrix projects grad- uates and foreign additions utilizing estimates of first-year enrollments, student attrition, other medical-school-related trends, and the model of FMG immigration. The second matrix distrib- utes the graduates among medical specialties through projections of first-year residency trends, and computes deaths and retirements of active practitioners among the specialties. Comparable disaggregation of the data on DOs has not been developed, although esti- mates of total DO supply have been made. The DO distribution between primary care special- ties is difficult to predict because of the lack of basic data on graduate training positions and because the graduate osteopathic training sys- tem is changing. In addition, MD residency pro- grams accept DOs, which could lead to increas- ing specialization by younger DOS. Presently, about 58 percent of DOs are in primary care. If DO graduates enter first-year residency programs in the same proportion as projected for MDs, by 1990 only 52 percent would be in primary care. If DOs continue cur— rent trends in graduate osteopathic training, 64 percent would be in primary care in 1990 (USDHEW, 1979b). Although these are signifi- cant percentage differences, the absolute dif- ferences are not large. Out of a total DO supply of 30,000 in 1990, the 52-percent figure cor- responds to about 15,500 primary care DOS, and the 64-percent figure corresponds to 19,000. This is in contrast to 1990 estimates of total MD and DO supply of 600,000 and a primary care MD supply of 240,000. The projections for MD specialty distribution of the aggregate supply depend principally on first-year residency trends and on the attrition rates of the various specialties and subspe- cialties. The data sources for current specialty supply are summarized in table 11. Specialty designa- tions are obtained from the AMA master file, Board certification data, and specialty society memberships. The AMA file contains self-des- 60-618 0 — 80 — 5 ignation of specialty, tending to overestimate specialty supply and underestimate general practice supply. And, as only the primary activ- ity/ specialty is identified, nothing is known about patient care time spent in the identified specialty or in activities usually associated with other specialties. About half of the physicians identified in the AMA files are not identified in the Board certification data. Also, Board certifi- cation data and especially society membership data result in duplicate counting, as physicians can belong to more than one specialty board or society. There are 22 medical and surgical boards and over 130 specialty societies. BHM uses the AMA master file as its basic source, with the 1974 active supply as the start- ing point. First-year residency trends, which are used to project additions of MD graduates (foreign and domestic) to the specialties, contain three assumptions: 1) that the first-year residency distributions for 1968, 1970-74, and 1976, can be used to predict future first-year residency trends; 2) that first-year residency counts for particular specialties are duplicative in the sense that a proportion of these residents do not go on to complete that specialty training, but move on to subspecialization or to another specialty altogether; and 3) that some residency positions are shared by different institutions, which also leads to duplicative counting. Considering the kinds of interpretation problems that accom- pany trying to project specialty distribution among active practitioners from first-year resi- dency positions, the better method would be to analyze final-year residency counts, but the AMA does not keep year-by-year accounts of medical graduates, and first-year residency data represent the best available data. Residency data sources and comments on their strengths and limitations are summarized in table 12. The principal data source is the Directory ofAccredited Residencies. The particular years chosen to establish trends, 1968, 1970-74, and 1976, are the most recent years on which to base such calculations, 24 ' Forecasts of Physician Supply and Requirements Tab|e11.—Data Sources on Physician Specialty Supply Data sources . The American Medical Association Master File. —Contains data on all known M05 in the United States, ob- tained by surveys performed every 3 to 4 years, and updated annually by selected mailings to specific physi- cians for whom a change in status has been indicated .3 .o Strengths Most complete source of data on allopathic physi- cians. Published and up- dated annually providing trend data. Physicians are listed by se|f~designa- tion as to their specialty, activity, and location according to how they spend the majority of their time. Six- ty-eight specialities are included within which eight activity cate- gories are included. 9’ The American Osteopathic Associa- Most complete sources of tion Master File—Contains intorma— data on osteopathic physi- tion on both member and nonmem- ber osteopathic physicians as to lo- cation and updated annually. Aug- mented by surveys performed in 1956, 1967, 1971, and 1976 which yielded additional data on specialty, age, and activity status.e 3. Licensure data—Provides data on numbers of physicians licensed by State. Disaggregated by whether or not physician attended a US. or for- eign medical school. 5 . Board certification data—Gives in- formation the numbers of MDs certi- fied by the 22 medical and surgical boards and the numbers of DOS cer- cians. , Updated annually, and thus, only sources of trend data on osteopathic physicians. In some cases the data are comparable to AMA data. Contains data on physicians who have received licenses; therefore, one can be sure all uncredentialled physi- cians are excluded.9 Published and updated annually, so trend data are available. Most objective criteria of physicians postgraduate training in specific specialty areas. tified by the 14 osteopathic specialty boards.'1 5. Specialty society memberships.— lncludes numbers and distributions of MDs in over 130 specialty societies.i Published and updated annually so trend data are available. Gives some indication of physician’s interests in specific areas of medicine not revealed in other AMA specialty classifications. Published and updated an- nually. Limitations Self-designation of specialty gives no indication of spe- cific training in the area and also tends to overestimate specialty manpower, and underestimate general prac- tice manpower. Published data provide no information on the time devoted to other specialty areas and activi- ties making it difficult to determine full-time equivalent manpower.b Accuracy of data on FMGs is debatable as is the accu- racy of specialty distributions because increasing num- bers of physicians are being relegated to the “non- classified" category.c d Can be difficult and/or expensive to obtain unpublished tabulations. Published data usually 2 years out of date. The problems associated with self-designation relating to AMA data also apply to AOA data. Specialty data only available for survey years, and when published contains information up to 3 years out of date. Accuracy of specialty data questionable because large numbers of physicians are relegated to the non- classified category.i Not always comparable to AMA data. Underestimates true physician supply since it excludes all physicians who are not licensed, such as some of those in teaching and administration and research, and some FMGs who are providing important service despite their unlicensed status. No information on specialty and practice activity of licensed physicians. Duplication often occurs between various State licen- sure boards. Excludes almost half of MD supply as reported by AMA and 4/5 of the DO supply as reported by AOA. Duplicate counting occurs due to certification by more than one specialty board. Does not necessarily represent physician‘s present specialty activities. Gives no indication of physician‘s training or back- ground in a specific specialty area represented by the society. Duplicate memberships often occur. Does not necessarily represent the present activities of the physician. aAmerican Medical Association Physician Master File, American Medical Asso- ciation, Chicago, Ill., 1977. bFor example, a physician may report his or her professional activities in a typi- cal workweek consisting ol 30 hours of patient care and 20 hours of teaching and research, and in addition specialty activity is reported as 25 hours of inter- nal medicine and 25 hours of dermatology. This precludes determination of number of FTE physicians in direct patient care. 6According to cohort study of physicians immigrating to the United States be- tween 1961 and 1971 an estimated 33 percent of 27,710 immigrants in the co- hort were not on the AMA master tile. J. C. Kleinman, Physician Manpower Data: “The Case of the Missing Foreign Medical Graduates,“ Medical Care, 12:906, 1974. Others believe that the AMA does account for all FMGs. I. Butler and M. Schaltner, “Foreign Medical Graduates and Equal Access to Medical Care," Medical Care, 9 (2): 136-43, Marcthprll 1974. dFrom 258 in 1970 to 30,129 in 1976 for MDs, Louis J. Goodman, Physician Dis. tribution and Medical Licensure in the US, 1976. Chicago, Aumlcuu ' ' Association, 1977. .31976 Directory, American Osteopathlc Association, Chicago, 1976. 1From 901 In 1971 to 653 In 1976. M. E. Altenderter, Osteopathic Physicians in the US. A Report eta 1971 Survey, BHRD, DHEW publication No. (HRA) 75-60, 1975 and American Osteopathic Association, 1974 Master File, Liaison Com- mittee on Osteopathic Information, Osteopathic Manpower lnlormation Pro/'- ect, final report, May 20, 1977. 9At present it is estimated that there are about 36,500 physicians in the country who do not hold a regular State license. Louis J. Goodman, Distribution of Physicians, 1976, p, 577. hFor MDs the American Medical Association, Profile of Medical Practice 1977, Chicago, 1977, p. 101. For 003 see, Liaison Committee on Osteopathic In- formation, Osteopathic Manpower lnlormation Project, Final Report, May 20, 1977. IIn 1974, over 130 such societies existed, In which there were 342,090 members representing 104 percent of all active physicians during that year. American ' ‘ A s- ' Profile 0/ Practice 1975-76. Chicago, 1976. SOURCE: Interim Report of the Graduate Medical Education National Advisory Committee to the Secretary, Department of Health, Education. and Welfare, Washington,D.C.: Health Resources Administration, DHEW publication No. (HRA) 79-633, pp. 101-103. Ch. 2—Supply ' 25 Table 12.—lnternship and Residency Data Sources Data sources Strengths Limitations _.| . Directory of Accredited Residencies and previous editions of the Directory of Ap- proved Internships and Residencies.— Contains data on distributions of first-year training. Most complete source of data available on M03 in Usually it is 2 years out of date. Does not provide distributions of residents by institutions. and total residents by specialty (30 listed), Published and updated country of education, and affiliation status annually. of hospital. Also lists numbers of positions offered and filled by specialty and affiliation and num- bers of positions offered for the forthcom- ing year..91 2. American Osteopathic Association Al- teopathic hospitals by specialty and insti- tution.b nually. Most complete source of manac.—Contains data on residents in 03- data on 003 in training. Published and updated an- Physicians listed as first-year residents in some specialties may in fact be the second or third year of training. Includes only first-year and total counts—inter- vening years not given. No data on fellowships. Does not provide disaggregated data on residents by years in training. No information provided on DOs training in non- AOA-approved programs such as AMA-approved hospitals. Provides distributions of residents by institution. 3. Council of Teaching Hospitals—Provides Provides distributions of in- Does not provide distributions of resident special- data on interns and residents by institu- tion.° stitution. terns and residents by in- ty or years in training. Published and updated an- nually. Timely, 1976 data available in 1976. 4. National Intern and Resident Matching Program—Provides information on specialty distributions of first-year and other residents in AMA-approved hospitals who participate in the program.d 1976. Provides indications of stu- Does not provide trend information on unmatched dent specialty and institu- tional preferences. graduates and foreign medical graduates various- ly estimated at 10 to 30 percent of the total first year resident supply.e ' 9 Timely, 1976 data available in aAmerican Medical Association, Directory of Approved Internships and Residents 19754976, Chicago, 1976. bAmerican Osteopathic Association, ”Almanac, Supplement to Volume 76,” 1975 Journal of the American Osteopathic Association, April 1977. °Council of Teaching Hospitals, Directory, 1976, Association of American Medical Colleges, January 1976. dAmerican Medical Association, Directory of Approved Internships and Residencies. 9J. S. Graettinger, “Graduate Medical Education Viewed From the National In- tern and Resident Matching Program,” J. Med. Educ, 51, September 1976. fB. Biles, communication to staff of the Senate Committee on Labor and Public Welfare, June 6, 1976. QNIRMP does provide data for 1977 and 1978 on unmatched U.S. graduates. The program plans to collect such information periodically on all US. graduates, both those who use as well as those not using NIRMP. SOURCE: Interim Report of the Graduate Medical Education National Advisory Committee to the Secretary, Department of Health, Education, and Welfare, Washington, DC: Health Resources Administration, DHEW publication No. (HRA) 79-633, pp. 105-106. but they also are unfortunate choices in the sense that major changes were occurring in ad- dition to the general drive to increase the ag- gregate supply of physicians and particularly those in primary care. In 1971, the AMA de- cided to terminate the free-standing internship after July 1, 1975, and instead to integrate the first year of graduate medical education into specific residency programs. During this time, the number of first-year residency positions in- creased dramatically. Most of this increase oc- curred in the primary care specialties, especially internal medicine (see figures 2 and 3). It would be reasonable to presume that much of this growth was not related to interest in primary care as a career. Instead, the first year of primary care residency training most likely sub- stituted for the internship of previous years. This overcounting of specialists through the use of first-year residency data is not a phe- nomenom solely related to the discontinuation 26 0 Forecasts of Physician Supply and Requirements Figure 2.—Trend Data on Number of First-Year Residents: Total, Primary, Nonprimary Care Specialties; 1960, 1968, 1974, and 1976 20,000 15,000 10,000 5,000 V 1960 1 ' " 1968 1974 76 Year aNonprimary specialties are total less primary care specialties. Primary care specialties include general and family practice, internal medi- cine, and pediatrics. SOURCE: Interim Report of the Graduate Medical Education National Advisory Committee to the Secretary, Department of Health, Educatlon, and Welfare, Washington, D.C.: Health Resources Administration, DHEW publication No. (HRA) 79-633, p. 51. of the free-standing internship. It has been known for some time, although hard to quanti- fy, that some graduate trainees take a second first-year residency in a more specialized area of the same specialty or move on to more ad- vanced training in another specialty altogether. For example, there is an observed 22-percent in— crease between the first and second year in the surgical specialties (USDHEW, 1979a). The way in which the overcounting is mini- mized is to adjust the first-year residency data in the Directory of Accredited Residencies by sub- tracting the appropriate subspecialties from the general residencies 1 year later. These adjust- ments are performed for internal medicine, pe- diatrics, general surgery, psychiatry, and pa- thology (USDHEW, 1979a). For internal medicine, 9 percent of first-year residents are subtracted first, this percent is assumed to take another first-year residency in a Figure 3.—Trend in Number of First-Year Allopathic Residents in Four Selected Specialties 5,000 4,000 3,000 = Number of residents 2,000 1 ,000 1955 1960 1965 1970 1975 1980 Year 0 .,~ 1950 SOURCE: Interim Report of the Graduate Medical Educatlon Natlonal Advlsory Committee to the Secretary, Department of Health, Education, and Welfare, Washington, D.C.: Health Resources Administration, DHEW publication No. (HRA) 79-633, p, 50. medical subspecialty or in another specialty. Of the remainder, 25 percent is assumed to go on to subspecialty training. Thus, a total of about 32 percent (25 percent of the remainder of first- year residents after subtracting 9 percent, plus the original 9 percent of the total) of all internal medicine first-year residents are subtracted and “lost” to medical subspecialties or other spe- cialties. ‘ More recent data estimates that only 38 per- cent of first-year internal medicine residents end up in general internal medicine, as compared to the 68 percent summarized above (USDHEW, 1979a). However, these percentages are not directly comparable because of the way in which internal medicine subspecialties are counted or not counted as primary care. The 68 percent remaining in primary care internal medicine excludes gastroenterology, pulmonary disease, cardiovascular disease, and allergy, but Ch. 2—Supply ' 27 includes allergy and immunology, diabetes, endocrinology, geriatrics, hematology, im- munology, infectious diseases, neoplastic dis- eases, nephrology, nutrition, oncology, and rheumatology (table 13). Whether a subspe- cialty of medicine is included in the internal- medicine primary care count is of crucial im- portance, as the first-year residency distribution of primary care specialties is heavily weighted toward internal medicine. Internal medicine comprises more than 50 percent of all first-year primary care residency positions (table 13). The adjustment to pediatric first-year resi- dency positions is 9 percent, representing losses to pediatric allergy and pediatric cardiology. The surgical figures are adjusted downwards by 62 percent, representing all surgical specialties except obstetrics-gynecology and ophthalmol- ogy. Pathology is adjusted downwards by 2.7 percent, representing forensic pathology. And psychiatry is adjusted downwards by 20 per- cent, representing child psychiatry. The ”miscel- laneous" category is assumed to remain propor- tionate to the overall numbers of MDs through- out the projection period. The adjusted and un- adjusted first-year residency distribution for 1974 is summarized in table 13. The same adjustments made for the 1974 year are made for all historical years, 1968, 1970-74, 1976, that are used to establish the trend. For 1968, the use of these specific adjustment percentages may not be very relevant, since it was several years before announcement by AMA of its intention to discontinue the intern- ship. A more technically oriented summary is pro- vided in the following excerpt (with table numbers changed to match the sequence of this report) from the Graduate Medical Education National Advisory Committee (GMENAC) in- terim report (USDHEW, 1979a). Note that the trend in distribution of residents among the various specialty training programs assumes a similar trend for 1974 to 1980-81 as the base years of 1968-74 (which were also modified to include 1976 data). After 1980-81, the residency distribution is held constant. BHM’s justifica- tion for the constant distribution after 1980-81 is primarily that the base years which have been chosen to establish the trend covered 6 years, so the trend extrapolation is limited to 6 years. The total projection method is summarized in figure 4. Tables 16 and 17 summarize specialty projections for MDs to 1980, 1985, and 1990. Projections of filled first-year residencies were made by extrapolating the results of simple linear regression applied to the trend in filled first-year residency percent distributions for the years 1968, 1970-74, and 1976. The procedure was applied for each specialty individually ex- cept for the category ”miscellaneous,” which was assumed to remain constant at 6.7 percent (see tables 14 and 15). Also, rates were devel- oped separately for U.S., Canadian, and other medical school graduates. In these regression analyses, the slope of the regression line was computed from historical trends, and the constant term (base year) was taken from the first-year residency distribution of 1974, adjusted for the duplication caused by physicians first taking a residency in a general area and then in a specialty (table 13). Where this adjusted value differed significantly from the original value, as in general surgery, the yearly rate of change (slope) was decreased by the ratio of the unadjusted to the adjusted value. The degree to which simple linear regression rep- resents historical trends in individual specialties is reflected in the F and R2 values displayed in tables 14 and 15. Most specialty trends are ade- quately ”explained" using this statistical meth- od. However, recent cultural, political, and fiscal interventions have affected certain special- ties so that they behave erratically, and there- fore have statistically nonsignificant F and R2 values for a linear fit. In two cases, U.S./CMG general practice and radiology, the linear trend produced actual negative percent values. These values were set and held at zero for these two specialties. This is a reasonable assumption since general practice is being replaced by family prac- tice, and general radiology is being replaced by the diagnostic and therapeutic training pro- grams. The effects of recent legislation (Public Law 94-484) have not been evaluated and incorpo- rated into the projections. However, the percent of U.S./CMG filled first-year residencies in pri- mary care for 1980 are projected to meet the leg- islative mandates. Table 1 3.—First-Year Residency Distribution With Subspecialty Adjustment: Sept. 1, 1974 AMAa Adjustments Adjusted AMA USICMGs FMGs US/CMGs FMGS US/CMGs FMGs Specialty Number Percent Number Percent Number Percent Number Percent Number Percent (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) Total active physicians ........... 13,519 100.0 5,216 100.0 12,626 100.0 4,755 100.0 Primary care ........................ 5,978 44.2 1,746 33.5 4,735 37.5 1,394 29.3 General practice .................. 23 0.2 139 2.7 23 0.2 139 2.9 Family practice ................... 1,131 8.4 68 1.3 1,131 9.0 68 1.4 Internal medicineb ................. 3,591 26.6 962 18.4 — 1,144 — 306° 2,447 19.4 656 13.8 Pediatrics ........................ 1,233 8.4 577 11.1 — 99 — 88d 1,134 9.0 531 11.2 Other medical specialties ............. 335 2.5 46 0.8 1,155 9.1 266 5.6 Dermatology ..................... 248 1.8 16 0.3 248 2.0 16 0.3 Pediatric allergy ................... 46 0.3 13 0.2 46 0.4 13 0.3 Pediatric cardiology ............... 41 0.3 17 0.3 41 0.3 17 0.4 Internal medicine subspecialtiese. . . . — — — — + 820 + 220f 820 6.5 220 4.6 Surgical specialties .................. 4,395 32.5 1,454 27.9 3,280 26.0 936 19.7 General surgery ................... 1,803 13.4 836 16.0 — 1,118 — 5189 685 5.4 318 6.7 Neurological surgery ............... 114 0.8 15 0.3 114 0.9 15 0.3 Obstetrics and gynecology ......... 742 5.5 288 5.5 742 5.9 288 6.1 Ophthalmology ................... 468 3.5 36 8.7 468 3.7 36 0.8 Orthopedic surgery ................ 547 4.0 62 1.2 547 4.3 62 1.3 Otolaryngology ................... 227 1.7 43 0.8 227 1.8 43 0.9 Plastic surgery .................... 148 1.1 36 0.7 148 1.2 36 0.8 Colon and rectal surgery ............ 20 0.1, 10 0.2 20 0.2 10 0.2 Thoracic surgery .................. 97 0.7 50 1.0 97 0.8 50 - 1.1 Urology .......................... 232 1.7 78 1.5 232 1.8 78 1.6 Other specialties .................... 2,808 20.8 1,970 37.8 3,456 27.4 2,159 45.4 Anesthesiology ................... 367 2.7 348 6.7 367 2.9 348 7.3 Neurology ........................ 252 1.9 109 2.1 252 2.0 109 2.3 Pathology ........................ 397 2.9 410 7.9 — 11 — 11h 386 3.1 399 8.4 Forensic pathology ................ 17 0.1 7 0.1 17 0 7 0.1 Psychiatry ....................... 952 7.0 612 11.7 — 180 — 116i 771 61 496 10.4 Child psychiatry ................... 189 1.4 98 1.9 189 1.5 98 2.1 Physical medicine and rehabilitation . 29 0.2 93 1.8 29 0.2 93 2.0 Radiology ........................ 88 0.7 137 2.6 88 0 7 137 2.9 Diagnostic radiology ............... 452 3.3 101 1 9 452 3.6 101 2.1 Therapeutic radiology .............. 65 0.5 55 1 1 65 0.5 55 1.2 Miscellaneousi .................... — — — — + 840 + 316k 840 6.7 316 6.7 aDlrectory oi Accredited Residencies, AMA, Chicago, 1977. vided by 1973 FYFis (2,698) in general surgery is 62 percent, the proportion subtracted out of the 1974 bIncludes allergy and Immunology, diabetes, endocrinology, geriatrics, hematology, immunology, infec- FYRs in general Surgery. tious diseases, neoplastic diseases, nephrology, nutrition, oncology, and rhematology. h1974 FYRs (24) in forensic pathology divided by 1973 FYFts (898) in pathology is 2.7 percent, the propor- cFor explanation of 9- and 25-percent adjustments, see text. , tion excluded from the 1974 FYFis in pathology. d1974 FYRs in pediatric allergy and pediatric cardiology(117) divided by 1973 FYRs in pediatrics (1,699) is '1974 FY85 (287) in Child psychiatry divided by the 1973 FYRs (1.472) in psychiatry is 20 perCent. the Pro- 6.9 percent, the proportion excluded from the 1974 FYFls in pediatrics. ,portion subtracted out of the 1974 FYFls in psychiatry. eIncludes gastroenterology, pulmonary disease, cardiovascular disease, and allergy. llncludes aerospace medicine, public health, general preventive medicine, occupational medicine, 1,040 figure from footnote 6. “other," and unspecified. 91974 FYRs (1,679) in surgical subspecialties (excluding obstetrics/gynecology, and ophthalmology) di- kFor explanation, see text. SOURCE: Interim Report of the Graduate Medical Education National Advisory Committee to the Secretary, Department of Health, Education, and Welfare, Washington, DC: Health Resources Administration DHEW publication No. (HRA) 79633, pp. 124-125. ' sguamaxlnbag pun lilddns ungais/l‘lld jo s;sn3axo_.{ . gz Ch. 2—Supply 0 29 Figure 4. —Projection of Active Physicians by Specialty Adjustment of ' Regression ' j first-year analysis of ., residencies percent .. , for duplication specialty . caused by general 1 distribution 7 and specialized . of adjusted ' residencies - first-year ,, residencies 7‘ to develop I I trend Extrapolation of regression results after specialty distribution is held constant Judgments and ' = Adjustment so . ' Application of analysis of that percent ‘ projected special circum- specialty adjusted first- stances g distribution year residency ' ' - sums to 100% }‘ distribution to . graduates over ‘ all specialties f to 1980-81. There- 1 Addition to active phy- sician supply by specialty Subtraction of separations due to retirement and death based _ ‘ on age distribu- Net active physician supply by specialty '» I. for projection tion for each ' year specialty SOURCE: Interim Report at the Graduate Medical Education National Advisory Committee to the Secretary, Department of Health, Education, 'and Welfare, Washington, 0.0.: Health Resources Administration, DHEW publication No. (HFiA) 79-633, p. 122. As mentioned, judgment was used in specific instances where straightforward extrapolation appeared to produce intuitively unreasonable re- sults. Such was the case with specialties showing a curvilinear trend. This trend was terminated in 1977 for these specialties, not only because the projected numbers appeared unreasonable, but also because the historical data in no instance showed a strong curvilinear trend in one direc— tion that lasted more than 3 years. The USMC radiology trend was allowed to fall to zero by 1980 because of the reported phasing out of general radiology as a prerequisite for entry into one of the radiology subspecialties. It is readily acknowledged that such use of re- gression in this analysis implied an assumption that the conditions underlying and responsible for past trends will also be in force in the future. Even though the situation is rapidly changing in the GME environment, it is nonetheless believed that such projections, when interpreted in a cau- tious manner, can be of value as ”baseline” esti- mates, indicating what the specialty configura- tion might be if residency developments con- tinue as they have in the past. Using this approach, extrapolations of each distinctive residency trend were developed. Be- cause each residency category was projected separately, however, a few minor changes had to be made to adjust the overall distribution to the control total for all residencies. In other words, when the projected percentage distribu- tion of residencies did not add to 100 percent because of unusually strong trends in one spe- cialty, the specialties which remained constant (10 out of 29 in the case of USMGs) in the his- torical trend period were adjusted slightly to make up the difference. For several reasons, this methodology was employed for the period of 1980-81. Thereafter, the 1980-81 residency distribution was held con- stant. One reason for this is that extrapolation is not statistically justified for longer periods in the future than are represented by the historical data on which it is based, in this case 6 years. Anoth- er reason is that most trends of the type ob- served have a tendency to level off after their ini- tial spurt. (Additional research and trend anal- ysis is continuing on this aspect of the projec- tions.) These assumptions are the working assump- tions of the Division of Manpower Analysis, BHM, for specific purposes. They have not been endorsed by GMENAC, which will develop its own assumptions concerning requirements rates, foreign medical graduates projections, specialization rates, capitation grants to medical schools and other issues. It is GMENAC's intent to investigate and, as needed, modify these as- sumptions (USDHEW, 1979a). 30 ' Forecasts of Physician Supply and Requirements Table 14.—Percent Distribution of U.S.ICMG First-Year Residency Projections Using Simple Linear Regressions (1976 actual, 1977-81 projections) Specialty 1976a 1977 1978 1979 1980 1981 Fb R2b Totalc ......................... 100.0 100.0 100.0 100.0 100.0 100.0 Primary care ..................... 43.2 43.8 46.0 48.1 50.2 52.2 General practice ................ 0.1 0.0 0.0 0.0 0.0 0.0 18.2 .78 Family practice ................. 12.0 13.5 15.1 16.6 18.1 19.5 385.9 .99 Internal medicine ............... 22.6 21.2 21.7 22.2 22.7 23.2 10.1 .67 Pediatrics ..................... 8.5 9.1 9.2 9.3 9.4 9.5 9.6 .66 Other medical specialties .......... 9.8 9.6 9.5 9.7 9.8 9.9 — — Dermatology ................... 1.6 1.7 1.6 1.6 1.5 1.5 2.7 .36 Pediatric allergy ................ 0.4 0.4 0.4 0.4 0.4 0.4 0.1 .02 Pediatric cardiology ............. 0.2 0.3 0.2 0.2 0.2 0.2 2.6 .34 Internal medicine subspecialtiesd . 7.6 7.2 7.3 7.5 7.7 7.8 16.1 .76 Surgical specialties ............... 22.5 21.8 20.7 19.4 18.0 16.9 — — General surgery ................ 4.8 4.2 3.8 3.4 3.0 2.6 60.3 .92 Neurological surgery ............ 0.6 0.6 0.5 0.5 0.4 0.3 24.9 .83 Obstetrics and gynecology ....... 5.9 5.8 5.8 5.7 5.6 5.6 0.4 .07 Ophthalmology ................. 3.1 2.8 2.6 2.3 2.0 1.8 500.0 .99 Orthopedic surgery ............. 3.4 3.6 3.4 3.2 3.0 2.8 7.9 .61 Otolaryngology ................. 1.4 1.4 1.3 1.1 1.0 0.9 36.6 .88 Plastic surgery ................. 1.0 1.1 1.1 1.1 1.1 1.1 8.8 .40 Colon and rectal surgery ......... 0.1 0.2 0.2 0.2 0.2 0.2 22.7 .82 Thoracic surgery ............... 0.7 0.6 0.6 0.6 0.5 0.5 5.4 .52 Urology ....................... 1.4 1.5 1.4 1.3 1.2 1.1 5.3 .51 Other specialties ................. 24.6 24.8 23.8 22.9 22.1 21.2 — — Anesthesiology ................ 2.2 2.0 1.7 1.4 1.1 0.9 113.9 .96 Neurology ..................... 1.8 1.8 1.6 1.5 1.4 1.3 6.1 .55 Pathology ..................... 3.2 2.9 2.8 2.7 2.6 2.5 11.7 .70 Forensic pathology ............. 0.1 0.1 0.1 0.2 0.2 0.2 0.5 .11 Psychiatry ..................... 4.8 4.5 3.9 3.3 2.7 2.2 26.0 .84 Child psychiatry ................ 1.3 1.5 1.5 1.5 1.5 1.4 0.01 .002 Physical medicine and rehabilitation ................ 0.4 0.2 0.2 0.1 0 1 0.0 5.1 .50 Radiology ..................... 0.2 0.0 0.0 0.0 0.0 0.0 57.1 .92 Diagnostic radiology ............ 3.5 4.6 4.8 5.0 5.3 5.5 1.5 .28 Therapeutic radiology ........... 0.4 0.5 0.5 0.5 0.5 0.5 0.07 .02 Miscellaneouse ................. 6.7 6.7 6.7 6.7 6.7 6.7 — — aActual figures. he degree to which simple linear regression represents actual historical trends in the individual specialties is reflected In the F and Ft2 values. 6Figures may not total due to Independent rounding. dlncludes gastroenterology, pulmonary disease, cardiovascular disease. and allergy. elncludes aerospace medicine, public health, general preventive medicine, oc- cupational medicine, "other,” and unspecified. The “F Test," as applied to the regression on historical residency data. meas- ures the statistical significance of the linear trend as an estimate of the past changes in the number of first-year residents by specialty 1968-76. Values of F greater than 6.6 are statistically significant at the 95-percent confidence level. R‘, the square of the Pearson product-moment correlation coefficient, is fre quently referred to as "The Correlation Index.” On a scale from zero to one, it measures the degree to which the linear trend estimates the actual changes in the number of first-year residents, by specialty, 1968-76. SOURCE: Interim Report of the Graduate Medical Education National Advisory Committee to the Secretary, Department of Health, Education, and Welfare, Washington, D.C.: Health Resources Administration, DHEW publication No. (HRA) 79-633, pp. 127-128. LOCATIONAL DISTRIBUTION Where physicians reside and practice medi- cine is generally obtained from the same data sources as for aggregate and specialty supply. Distribution is usually described at the State, county, or Health Service Area (HSA), and, for comparative purposes, quantified as a physi- cian-to-population ratio. The limitations of the data sources in simply describing where physicians live and practice are similar to the data limitations in describing aggregate and specialty supply. In addition, de- scribing locational distribution by States, by counties, by metropolitan versus nonmetropoli- tan areas, and even by HSAs may be most con- Ch. 2—Supply ' 31 Table 15.—Percent Distribution of FMG First-Year Residency Projections Using Simple Linear Regressions (1976 actual, 1977-81 projections) Specialty 1976a 1977 1978 1979 1980 1981 Fb R1b Total.c ....................... 100.0 100.0 100.0 100.0 100.0 100.0 Primary care ..................... 33.2 31.6 32.1 32.6 33.1 33.6 General practice ................ 4.4 3.5 3.6 3.7 3.8 3.9 0.7 .12 Family practice ................. 3.1 3.1 3.5 3.9 4.3 4.7 27.2 .87 Internal medicine ............... 13.2 13.3 13.1 12.9 12.7 12.5 2.0 .28 Pediatrics ..................... 12.5 11.7 11.9 12.1 12.3 12.5 1.3 .21 Other medical specialties .......... 5.4 5.2 5.2 5.1 4.9 4.7 — — Dermatology ................... 0.3 0.2 0.2 0.2 0.2 0.1 6.7 .57 Pediatric allergy ................ 0.3 0.3 0.3 0.3 0.3 0.3 0.1 .02 Pediatric cardiology ............. 0.4 0.3 0.3 0.3 0.3 0.2 10.4 .68 Internal medicine subspecialtiesq . 4.4 4.4 4.4 4.3 4.2 4.1 2.2 .31 Surgical specialties ............... 20.5 19.0 18.2 17.6 17.0 16.2 — — General surgery ................ 7.0 6.6 6.4 6.1 5.9 5.7 16.4 .77 Neurological surgery ............ 0.8 0.5 0.5 0.5 0.5 0.5 0.1 .02 Obstectrics and gynecology ...... 5.4 5.0 4.6 4.1 3.7 3.3 138.7 .97 Ophthalmology ................. 0.5 0.6 0.5 0.5 0.5 0.4 3.6 .42 Orthopedic surgery ............. 2.0 1.7 1.7 1.8 1.9 1.9 2.3 .31 Otolaryngoiogy ................. 0.9 0.9 0.9 0.9 0.9 0.9 0.7 .13 Plastic surgery ................. 0.8 0.9 0.9 1.0 1.0 1.0 41.3 .89 Colon and rectal surgery ......... 0.3 0.3 0.3 0.4 0.4 0.4 9.5 .65 Thoracic surgery ............... 0.9 0.8 0.7 0.6 0.5 0.4 14.9 .75 Urology ....................... 1.8 1.7 1.7 1.7 1.7 1.7 .00005 .00001 Other specialties ................. 40.9 44.3 44.5 44.8 45.1 45.5 — — Anesthesiology ................ 6.2 6.6 6.2 5.9 5.5 5.2 6.5 .56 Neurology ..................... 1.9 2.3 2.3 2.4 2.4 2.5 2.7 .35 Pathology ..................... 7.1 7.5 7.2 6.8 6.5 6.2 6.7 .57 Forensic pathology ............. 0.2 0.2 0.2 0.2 0.2 0.2 0.4 .10 Psychiatry ..................... 8.5 9.2 9.4 9.6 9.9 10.1 1.5 .23 Child psychiatry ................ 2.0 2.3 2.5 2.7 2.9 3.1 55.5 .92 Physical medicine and rehabilitation ................ 2.5 2.5 2.7 2.9 3.0 3.2 38.5 .89 Radiology ..................... 1.3 1.6 1.2 0.8 0.5 0.1 29.5 .86 Diagnostic radiology ............ 3.0 3.6 4.1 4.6 5.1 5.6 239.7 .98 Therapeutic radiology ........... 1.5 1.8 2.0 2.2 2.4 2.6 154.7 .97 Miscellaneouse ................. 6.7 6.7 6.7 6.7 6.7 6.7 — — aActual figures. he degree to which simple linear regression represents actual historical trends in the individual specialties is reflected in the F and Pi2 values. cFigures may not add to total due to independent rounding. dincludes gastroenterology, pulmonary disease, cardiovascular disease, and allergy. eIncludes aerospace medicine, public health, general preventive medicine, oc- cupational medicine, “other," and unspecified. The ”F Test," as applied to the regression on historical residency data, measures the statistical significance of the linear trend as an estimate of past changes in the number of first-year residents by specialty 1968-76. Values of F greater than 6.6 are statistically significant at the 95-percent confidence level. R’, the square of the Pearson product-moment correlation coefficient, is fre- quently referred to as “The Correlation Index." On a scale from zero to one, it measures the degree to which the linear trend estimates the actual changes in the number of first-year residents, by specialty, 1968-76. SOURCE: Interim Report of the Graduate Medical Education National Advisory Committee to the Secretary, Department of Health, Education, and Welfare, Washington, 0.0.: Health Resources Administration, DHEW publication No. (HRA) 79-633, p. 129-130. venient from a data availability point of view, but it does not necessarily follow that physi- cians are available to the populations they are matched against. Nor are populations identified on these bases comparable, and one area (e.g., county) may have people with significantly dif- ferent health problems than people in other areas. So in addition to the basic problem of being able to count the numbers of practicing physicians and their clinical specialties in an identified area, there is the additional problem of whether these physicians actually provide medical services to the designated population (including whether they may be providing serv- ices to people in adjacent areas). This qual- ification becomes important when such com— parative data are used to implement programs that single out ”health manpower shortage areas” for support. In these aid programs, a spe- cific physician-to-population ratio is chosen as the cutoff point and used in conjunction with other indices of medical need to determine 32 0 Forecasts of Physician Supply and Requirements Table 16.—Active Physicians (MD), by Major Specialty Group: Actual 1974; Projected (under the basic assumption) 1980-90 Projected Specialty group Base year 1974 1980 1985 1990 Number of active physicians Total .................. 348,960 430,150 500,340 566,940 Primary care ................ 133,240 166,790 203,370 239,830 Other medical specialties ..... 18,930 26,580 33,800 41,080 Surgical specialties .......... 97,720 113,200 122,160 129,610 Other specialties ............ 99,070 123,580 141,050 156,410 \ Percent distribution Total .................. 100.0 100.0 100.0 100.0 Primary care ................ 38.2 38.8 40.6 42.3 Other medical specialties ..... 5.4 6.2 6.8 7.2 Surgical specialties .......... 28.0 26.3 24.4 22.9 Other specialties ............ 28.4 28.7 28.2 27.6 Rate per 100,000 population Total .................. 165.5 193.1 213.8 231.3 Primary care ................ 63.2 74.9 86.9 97.9 Other medical specialties ..... 9.0 11.9 14.4 16.8 Surgical specialties .......... 46.3 50.8 52.2 52.9 Other specialties ............ 47.0 55.5 60.3 63.8 SOURCE: Interim Report of the Graduate Medical Education National Advisory Committee to the Secretary, Department of Health, Education, and Wellare, Washington, D.C.: Health Resources Administration, DHEW publication No. (HRA) 79-633, p. 155. whether an area is eligible or not for aid. This concept of “shortage" is a question more basic to comparing supply with requirements and is discussed in the chapter on requirements. The distribution of physicians by the most common methods of description and quantifica- tion is summarized in tables 18 and 19 and fig- ure 5. Table 18 provides physician-to-popula- tion ratios for selected specialties in the aggre- gate and for the States with the highest and low- est ratios. Table 19 contrasts non-Federal MDs in metropolitan with nonmetropolitan counties. And figure 5 contrasts the supply of surgeons and primary care physicians as grouped by HSAs. DHHS also compiles these statistics through a “GINI index," a statistical tool that expresses unevenness as a single number. To compute the GINI index, the percentage of the total population is graphically accumulated on one axis and the percentage of practitioners similarly accumulated on the other axis, starting with the area with the lowest physician-to- population ratio and going to the area with the highest ratio. If the distribution were perfect, the result would be a 45-degree “line of equal- ity.” The GINI index is the ratio of the area be- tween the actual curve and the line of equality to the total area under the line of equality (USDHEW, 1979b). The GINI index value varies between zero, in- dicating no maldistribution, and 1.0, indicating the greatest possible maldistribution. In general, smaller index values (indicating less unevenness) can be expected when making comparisons among larger geographical units. This can be seen in the following GINI index for active non- Federal MDs in 1973: By State (50 States) ...................... 0.161 By Census-Defined State Economic Area (173 areas) ......................... 0.292 By county (3,071 counties) ................. 0.361 (Source: USDHEW, 1979b) Osteopathic physicians (estimated at 17,700 in 1980 out of a total supply of active physicians of 447,800) are unevenly distributed among the States because they were not allowed to practice in some States until recently and because of the limited number of schools. In 1977, Michigan had 2,760 osteopaths, Alaska, only 7. More than 20 States had less than 100 osteopathic physicians and students. Ch.2——Supply ' 33 Table 17.—Supp|y 01 Active Physicians (MD), by Specialty: Actual 1974; Projected 1980-90 Number of physicians Percent distribution Specialty 1974 1980 1985 1990 1974 1980 1985 1990 Total active physicians ........ 348,960 430,150 500,340 566,940 100.0 100.0 100.0 100.0 Primary care ..................... 133,240 166,790 203,370 239,830 38,2 38.8 40.6 42.3 General practice ............... 46,530 39,290 32,870 26,350 13.3 9.1 6.6 4.6 Family practice ................ 9,480 22,380 39,190 56,480 2.7 5.2 7.8 10.0 Internal medicine .............. 54,780 73,280 91,020 108,530 15.7 17.0 18.2 19.1 Pediatrics ..................... 22,460 31,830 40,290 48,470 6.4 7.4 8.1 8.5 Other medical specialties .......... 18,930 26,580 33,800 41,080 5.4 6.2 6.8 7.2 Dermatology .................. 4,470 5,830 6,720 7,610 1.4 1.4 1.3 1.3 Pediatric allergy ............... 480 870 1,210 1,500 0.1 0.2 0.2 0.3 Pediatric cardiology ............ 590 850 1,030 1,200 0.2 0.2 0.2 0.2 Internal medicine subspecialtiesa. 13,120 19,030 24,850 30,730 3.8 4.4 5.0 5.4 Surgical specialties .............. 97,720 113,200 122,120 129,610 28.0 26.3 24.4 22.9 General surgery ................ 32,100 34,700 35,210 35,140 9.2 8.1 7.0 6.2 Neurological surgery ........... 2,990 3,470 3,360 3,710 0.9 0.8 0.7 0.7 Obstetrics and gynecology ...... 22,080 26,620 30,040 33,230 6.3 6.2 6.0 5.9 Opthamology .................. 11,220 12,630 13,210 13,730 3.2 2.9 2.6 2.4 Orthopedic surgery ............. 11,550 14,280 16,170 17,890 3.3 3.3 3.2 3.2 Otolaryngology ................ 5,870 6,640 6,980 7,310 1.7 1.5 1.4 1.3 Plastic surgery ................. 2,330 3,370 4,280 5,150 0.7 0.8 0.9 0.9 Colon and rectal surgery ........ 660 800 890 980 0.2 0.2 0.2 0.2 Thoracic surgery ............... 2,100 2,750 3,080 3,350 0.6 0.6 0.6 0.6 Urology ....................... 6,790 7,960 6,620 9,150 1.9 1.9 1.7 1.6 Other specialties ................. 99,070 123,580 141,050 156,410 28.4 28.7 28.2 27.6 Anesthesiology ................ 13,330 15,600 16,210 16,830 3.8 3.6 3.2 2.9 Neurology .................... 4,200 6,070 7,360 8,520 1.2 1.4 1.5 1.5 Pathology ..................... 12,310 15,860 18,120 20,020 3.5 3.7 3.6 3.5 Forensic pathology ............. 220 360 540 700 0.1 0.1 0.1 0.1 Psychiatry .................... 24,740 28,560 29,900 30,690 7.1 6.6 6.0 5.4 Child psychiatry ............... 2,730 4,460 5,970 7,730 0.8 1.0 1.2 1.3 Physical medicine and rehabilitation ................ 1,780 2,450 2,780 2,990 0.5 0.6 0.6 0.5 Radiology ..................... 11,900 11,710 10,950 9,970 3.4 2.7 2.2 1.8 Diagnostic radiology ............ 3,650 8,180 13,440 18,660 1.0 1.9 2.7 3.3 Therapeutic radiology ........... 1,200 2,000 2,760 3,420 0.3 0.5 0.6 0.6 Miscellaneousb ................ 23,010 28,320 33,030 37,670 6.6 6.6 6.6 6.6 aIncludes allergy, cardiovascular disease, gastroenterOIOQy. and pulmonary disease. Includes aerospace medicine, general preventive medicine, occupational medicine, public health, unspecified, and “other specialties." NOTE: Figures may not add to subtotals and totals due to independent rounding. SOURCE: interim Report of the Graduate Medical Education National Advisory Committee to the Secretary, Department of Health, Education, and Welfare, Washington, DC: Health Resources Administration, DHEW publication No. (HRA) 79-633, p. 153. These estimates of MD and DO‘locational distribution do not address the question of fu- ture distributional patterns. Such predictive ef- forts are used for programs which intend to place physicians in identified areas of shortage and for which such shortage designations also make the identified areas eligible for govern- mental (Federal and State) funds. Prior to the Health Professions Educational Assistance Act of 1976, different criteria had been developed for designation as a Health Manpower Shortage Area (HMSA) for BHM programs and as a Medically Underserved Area (MUA) for the Bureau of Community Health Services' (BCHS) programs. Following the pas- sage of the 1976 Act, these definitions have been consolidated under the HMSA designation. Thus, once designated a HMSA, such areas: 1) would be eligible for National Health Service Corps (NHSC) staffing of Corps practice site, 2) would be areas in which students who borrowed money under health professions student loan programs could practice in lieu of repaying the loans in money, 3) would be eligible for grants in various health manpower training programs, 34 ' Forecasts of Physician Supply and Requirements Table 18.—Patient Care MDs (non-Federal) by Selected Specialties and for High and Low States (1976) MDs per 100,000 Specially population, all States Ratio for highest State Ratio for lowest State All specialties ................. 137.0 New York ............. 198 South Dakota ................ 78 Primary carea .................. 58.4 New York ............. 84 Alaska ..................... 35 General and family practice ...... 23.9 California ............ 30 Alabama .................... 19 Internal medicine .............. 22.2 Massachusetts ........ 43 South Dakota ................ 7 Pediatrics .................... 8.9 Maryland ............. 17 South and North Dakota ....... 3 Obstetrics and gynecology ...... 9.3 Maryland ............. 16 South Dakota ................ 3 General surgery ............... 13.5 New York ............. 20 South Dakota and Alabama. . . . 2 Psychiatry .................... 9.0 New York ............. 20 South Dakota and Alabama. . . . 2 Opthalmology ................. 4.9 New York ............. 7 South Dakota ................ 2 Orthopedic surgery ............. 4.9 Massachusetts, Connect- South Dakota, Alabama, and icut, and California. .. 7 Mississippi ............... 3 Anesthesiology ................ 5.3 Massachusetts ...... ‘. . 9 South Dakota ................ 1 aDefined as general and family practitioners, internists, and pediatricians. SOURCE: Interim Report of the Graduate Medical Education NatlonalAdvisory Committee to the Secretary, Department of Health, Education, and Welfare, Washington, D.C.: Health Resources Administration, DHEW publication No, (HRA) 79-633, pp. 85-87. Table 19.—Non-Federal Physicians (MD) Providing Patient Care in Metropolitan and Nonmetropolitan Areas 1963-76 Number of physicians MDs per 100,000 population Percent of MDs All Metro- Non- All Metro- Non- All Metro- Non- Year counties politan metropolitana counties politan metropolitan counties politan metropolitan 1963 ...... 225,427 178,403 47,024 120.3 144.2 73.8 100.0 79.1 20.9 1964 ...... 232,067 184,298 47,769 122.0 146.8 73.9 100.0 79.4 20.6 1965 ...... 239,482 189,211 48,271 123.2 148.7 73.6 100.0 79.7 20.3 1966 ...... 241,473 192,871 48,602 123.7 148.9 74.1 100.0 79.9 20.1 1967 ...... 247,256 200,880 46,376 125.4 150.0 73.3 100.0 81.2 18.8 1968b ..... 236,458 192,242 44,216 118.7 141.6 69.7 100.0 81.3 18.7 1969 ...... 245,368 200,247 45,121 121.8 145.7 70.5 100.0 81.6 18.4 1970 ...... 252,778 206,676 46,102 124.2 148.7 71.5 100.0 81.8 18.2 1971 ...... 261,335 217,187 44,148 127.5 152.9 70.1 100.0 83.1 16.9 1972 ...... 266,587 225,424 41,163 128.5 152.9 68.6 100.0 84.6 15.4 1973 ...... 270,412 231,529 38,883 129.1 150.9 69.4 100.0 85.6 14.4 1974 ...... 276,070 235,994 40,076 130.9 153.3 70.4 100.0 85.5 14.5 1975 ...... 287,837 249,218 38,619 134.8 156.9 74.1 100.0 86.6 13.4 1976 ...... 294,730 255,102 39,628 137.4 158.2 74.4 100.0 86.6 13.4 aFor 1963-66, metropolitan counties include those in SMSAs on basis of 1962 population and nonmetropolitan counties include those adjacent to metropolitan counties and isolated rural and semirural counties. For 1967-76, metropolitan counties include those in SMSAs on basis of 1967 population and nonmetropolitan counties include potential metropolitan counties and all others outside SMSAs. Beginning in 1968, the AMA changed its methods of classifying physicians to reflect the number of hours spent in various activities and specialties. This resulted in a loss in physicians in patient care with corresponding increases in physicians in "other activities" and inactive. Based on annual reports on the distribution physician in the United States by the American Medical Association, 1963-67. SOURCE: Interim Report of the Graduate Medical Education National Advisory Committee to the Secretary, Department of Health, Education, and Welfare, Washington, D.C.: Health Resources Administration, DHEW publication No. (HRA) 79-633, pp. 91-92. 4) would be eligible or given preference for grant funds for several BCHS programs such as the urban and rural health initiatives, and 5) would be the only areas in which rural health clinics could be certified for reimbursement of nurse practitioner and physicians' assistant serv- ices under Medicare and Medicaid. Through the 1976 Act shortage designation for eligibility for NHSC physician services is now available not only to geographic areas (the old emphasis on alleviating rural shortages), but also to population groups and institutional set- tings of care. The former include Native Amer- icans, migrants, and the aged. The latter include hospitals, state mental health facilities, rehabil- itation facilities, long-term care facilities, com- munity health centers, community mental health centers, migrant health centers, and Fed- eral and State prisons. Ch. 2—Supply ' 35 Figure 5.—Frequency Distribution of Physician Availibility indexes—Primary Care Physicians and Surgeons for the 204 HSAs 69 (.39 .4 .59 .6 .79 .8 0.99 Below equivalent share for target population I Surgeons Primary care physicians 1.0 1.29 1.3 1.59 1.6 1.99 >2.0 Above equivalent share for target population The availability index is a weighted average of the ratio between the portion of the Nation’s physicians in each of the HSA‘s counties and the portion of the Nation's population living in each of those coun- ties. If the HSA has attracted a portion of the Nation’s physicians equivalent to its portion of the U.S.’s population, its physician availability index would be 1.0. SOURCE: Interim Report of the Graduate Medical Education National Advisory Committee to the Secretary, Department of Health, Educa- tion, and Welfare, Washington 0.0.: Health Resources Administration, DHEW publication No. (HRA) 79-633, p. 95. From this wide array of potential shortage areas, the NHSC program had to develop sub- sets that would actually receive Corps attention. NHSC includes obstetrician-gynecologists as primary care physicians, in addition to family practitioners, general practitioners, pediatri- cians, and internists. Psychiatrists are included for mental health facilities and osteopaths also are included, although not specifically in their projections, which concern allopathic physician supply. For nonmetropolitan areas there is a primary care physician-to-population designation ratio of 1:3,500, which means that only those areas .36 ' Forecasts of Physician Supply and Requirements with that ratio or higher (fewer physicians) would be eligible for Corps staffing. The target ratio is 122,000. Once designated and selected for Corps staffing, Corps physicians would be provided until a ratio of 1:2,000 was achieved (USDHEW, 1978a). As part of the process of planning for how much effort is needed in specific practice set- tings, estimates must be made of the numbers and types of physicians who will settle in these areas voluntarily. DHHS, through the joint ef- forts of its BHM of the Health Resources Ad- ministration and the Office of Planning, Evalua- . tion, and Legislation of the Health Services Ad- ministration, has developed a computerized model to project the number and distribution of active, non-Federal, office-based patient care physicians in the 50 States and all counties from 1972 through 1990. As the Health Professions Educational Assist- ance Act of 1976 enlarged the definition of HMSAs, separate estimates are made for: 1) rural counties, 2) urban areas, 3) Federal and State prisons, 4) State mental hospitals and community mental health centers, and 5) the In— dian Health Service. The computerized model is used to project future supply in rural counties, but it cannot be used to project supply in the other categories. Hence, other methods must be used for these other categories. For prisons, mental health facilities, and the Indian Health Service, the sponsoring agencies have provided estimates of both supply and requirements. Es- sentially, these agencies forecast a steady state supply (USDHEW, 1978a). The NHSC task force's estimates of the sup- ply of primary care physicians in rural and ur- ban areas are summarized below. Rural areas.—The projections are based on: 1) existing supply (DOS and MDs, domestic and foreign graduates); 2) anticipated deaths, retire- ments, and relocation of existing physicians; 3) anticipated graduations, specialty choices, and practice locations; and 4) anticipated popula- tion growth. Beginning with 1972 as the base year, total physician and primary care and non- primary care physician-to-population ratios are projected for each county (urban as well as rural) for each year. The method is thus similar to agggregate and specialty supply, with the added factor of accounting for where physicians locate. Counties are assumed to maintain population as fixed proportions (derived from 1972 data) of their respective State populations. Starting with the 1972 active, non-Federal, office-based, patient-care physicians in each county, year-by-year reductions from deaths and retirement are calculated. The number of new physicians expected to enter practice is estimated by the method de- scribed earlier for aggregate supply. These num- bers are summarized in table 20. Total new physicians entering practice each year are reduced for: 1) Federal employment, 2) nonpatient care activities, and 3) practices in areas other than the 50 States. The physician distribution among the States for 1973-87 (since updated to the 1990's) is based on the historical pattern of distribution for grad- uates from 1965 to 1969, modified for percent changes in State populations projected for 1972 to 1987. The percent of physicians entering practice from 1973 to 1987 as primary care specialists is Table 20.—New Physicians Entering Practice, 1 973-87 Year USMG FMG DO 1973 .............. 8,367 3,665 472 1974 .............. 8,974 5,081 485 1975 .............. 9,551 5,202 649 1976 .............. 10,391 2,709 587 1977 .............. 11,613 3,799 702 1978 .............. 12,714 5,265 809 1979 .............. 13,561 3,517 908 1980 .............. 13,607 3,895 964 1981 .............. 14,598 1,903 996 1982 .............. 15,048 2,187 1,029 1983 .............. 15,346 1,849 1,208 1984 .............. 15,789 2,372 1,308 1985 .............. 16,354 2,034 1,377 1986 .............. 16,956 2,246 1,458 1987 .............. 17,241 1,908 1,496 SOURCE: Memorandum from the Chairman, NHSC Needs Task Force A, to the Director, Bureau of Community Health Services, Health Services Ad- ministration; the Deputy Director, Bureau of Health Manpower, Health Resources Administration; and the Chairman, NHSC Needs Task Force, Washington, 0.0., May 26,1978. Ch. 2—Supply ' 37 projected to increase to the proportions listed in table 21. This should be distinguished from the .percent of the total physician supply that is pro- jected to be in primary care (tables 16 and 17). Physicians are projected to enter practice in county classes in the percentages summarized in table 22. Newly entering physicians are allocated to in- dividual counties by specialty according to the 1974 observed pattern of 30- to 44-year-old phy- sicians of the same specialty. This new supply is added to the existing supply, modified yearly for attrition from deaths and retirements. Table 21.—Percent of New Physicians Expected To Enter Primary Care Percent in Year primary care 1973 ................................. 25.2 1974 ................................. 26.4 1975 ................................. 28.1 1976 ................................. 33.5 1977 ................................. 37.5 1978 ................................. 43.2 1979 ................................. 43.8 1980 ................................. 46.0 1981 ................................. 48.1 1982 ................................. 50.2 1983 ................................. 52.2 1984-87 ............................... 52.2 SOURCE: Memorandum from the Chairman, NHSC Needs Task Force A, to the Director, Bureau of Community Health Services, Health Services Ad- ministration; the Deputy Director, Bureau of Health Manpower, Health Resources Administration; and the Chairman, NHSC Needs Task Force, Washington, 0.0., May 26,1978. Urban area.—Predicting the future supply of physicians for urban areas in order to assess the need for additional physicians is not computed on a county basis. If measured by county, the number of primary care physicians is usually adequate, so the needs in urban areas are meas- ured by the needs of certain population groups which have financial and sociocultural barriers to access instead of the geographic barriers of rural areas, for which physician-to-population ratios serve as substantial proxy measures. Thus, an estimate of the number of physicians required to meet the needs of metropolitan low- income areas as defined by the Bureau of the Census is used. Such identified low-income area populations declined from 17,936,000 in 1974 to 16,554,000 in 1976, or a decline of 3.8 percent per year. But the task force concluded that the decrease is not expected to continue indefinitely and that there is a current trend for physicians in central cities to move to the suburbs. It therefore assumed that the decrease in low-income population will be offset or more than offset by the emigration of physicians from the inner city. In other words, present supply as expressed in physician- to-population ratios also predicts what the fu- ture supply will be. This average is 13.3 full- time-equivalent primary care physicians per 100,000 population. Parenthetically, it was determined that 42.3 primary care physicians per 100,000 population Table 22.—Projected County Classes of Newly Practicing Physicians MD County classa Family practice Primary careb Nonprimary care DO 1 ............... 2.7 0.2 0.1 3.0 2 ............... 7.5 2.0 0.5 6.7 3 ............... , 10.6 5.1 1.6 6.4 4 ............... 9.4 6.9 3.7 5.2 5 ............... 2.3 2.2 1.2 2.2 6 ............... 26.8 19.5 24.6 17.6 7 ............... 13.2 12.8 13.1 13.9 8 ............... 21.9 35.5 37.0 40.1 9 ............... 5.6 15.0 18.2 4.9 100.0 100.0 100.0 100.0 aAMA demographic county classification (1-4 = rural; 5-9 = urban). bExcluding family practice. This definition of ”primary care“ includes obstetrics-gynecology, in addition to general practice, family practice, Internal medicine, and pediatrics. SOURCE: Memorandum from the Chairman, NHSC Needs Task Force A, to the Director, Bureau of Community Health Serv- ices, Health Services Administration; the Deputy Director, Bureau of Health Manpower, Health Resources Admin- istration; and the Chairman, NHSC Needs Task Force, Washington, 0.0., May 26,1978. 38 0 Forecasts of Physician Supply and Requirements (a staffing ratio of 1:2,000) would be needed in these low-income areas and that the 13.3 num- ber meant that 31.4 percent of need was already met. The analysis then goes on to say that sepa- rate analyses of the underserved population’s “usual source of care" resulted in the figures in table 23 and confirmed the 31 .4-percent figure. The analysis then goes on to equate the sum of care from “hospital,” “neighborhood health center," and “none” with unmet need of 62.6 percent of the population and goes on to esti- mate additional physicians needed on this basis. Yet only 8.6 percent of the 62.6 percent received no care. Hospital care does not represent “un- met need" but involves the question of what is appropriate care. And since this analysis was made to estimate the number of NHSC physi- cians that might be placed in these areas, identi- SUMMARY Comparing the methods for estimating supply with those used for estimating requirements, we would expect more certainty in the supply pro- jections. Yet the foregoing description of supply projections shows that there are many weak- nesses in the data bases, some questionable as- sumptions underlying the projections, different interpretations given to some commonly used terminologies such as “primary care" and ”full- time-equivalent, ” etc. The description of how supply projections are made can quickly become quite detailed. So let us summarize: 1) the components and primary assumptions of the supply estimates, and 2) some definitional problems that are linked to substantive issues and which are compounded by weaknesses in the data. Supply is the sum of practicing physicians and additions of foreign and domestic medical and osteopathic school graduates (there are no foreign additions to the osteopathic supply). At- trition from deaths and retirements for the prac- ticing physician component is estimated by age- specific rates. For specialty supply, the same death and retirement rates are applied to each specialty. Table 23.—Usual Source of Care for Urban Underserved Areas Percent Private physicians ...................... 31.4 Hospital (emergency room and outpatient treatment) ................. 31.3 Neighborhood health center ............. 22.7 None ................................. 8.6 Other ................................ 6.0 Total ............................... 100.0 SOURCE: Memorandum from the Chairman, NHSC Needs Task Force A, to the Director, Bureau of Community Health Services, Health Services Ad- ministration; the Deputy Director, Bureau of Health Manpower, Health Resources Administration; and the Chairman, NHSC Needs Task Force, Washington, DC, May 26,1978. fying “neighborhood health center" care as “no care" must mean that the presumption is that such centers will be staffed only by NHSC physicians in the future. Additions to supply are the sum of foreign and domestic graduates. Estimates of first-year enrollments and attrition prior to graduation are made to arrive at the number of domestic graduates. The high, low, and basic first-year- enrollment estimates all assume full capitation funding by 1981, and 4, 7, or 10 new medical schools after 1977-78. For FMG additions, the Canadian addition is currently estimated to equal losses from death, retirement, and emigra- tion. Estimates of the addition to supply from other FMGs rely heavily on the presumed im- pact of the 1976 Act, which contained major restrictions on FMG immigration. It should be noted that the resulting total projections of sup- ply made before and after the 1976 Act have not varied greatly (estimated at approximately 600,000 in 1990). The component projections, however, have varied widely. In essence, pres- ent projections of domestic graduates are larger than previous estimates, and present projections of FMG additions are less than prior estimates. In the current projections, the assumption of full capitation funding is not very realistic and tends to increase the supply projections. On the other hand, the additions from the FMG supply may be too optimistic in terms of legislative impact i—- ax Ch.2——Supply ° 39 on decreasing this source of supply, particularly with the large pool of US. citizens studying medicine abroad, for whom immigration re- strictions are not applicable, although they have to pass a competency exam in order to practice medicine in the United States. Additions to specialty supply use projections of first-year residency trends to allocate foreign and domestic graduates among the specialties. The predictive power of first-year residency choices is a problem because they may not re- flect ultimate specialty practices. This is particu- larly true for internal medicine, where at least one-third of the first-year residency positions is used as a general medicine traineeship for physi- cians ultimately subspecializing or choosing another specialty altogether. The trend in specialty choice is determined by the trend reflected in the years 1968, 1970-74, and 1976, years in which major changes were occurring in the structure of postgraduate medical training programs. Statistical techniques also limit the applicability of these trend years up to 1981, at which point the distribution is held constant for future years. Methods for predicting the locational distri- bution of physicians are generally similar to those for aggregate and specialty supply. For rural areas, the active, non-Federal, office-based physician supply in each county is reduced for deaths and retirements. The supply of new phy- sicians is allocated to individual counties by spe- cialty according to the 1974 observed patterns of 30- to 44-year-old physicians of the same specialty (counties are allocated to nine classifi- cations from rural to urban). For urban (inner city) areas, physician supply is assumed to de- crease (no numbers given), reflecting continued emigration to the suburbs. For prisons, mental health ,services, and the Indian Health Service, future supply is generally assumed to hold cons- tant at its present rate. In addition to absolute numbers, a relative standard is used, the physician-to-population ratio, which is also commonly expressed as the number of physicians per 100,000 population; e.g., 1:1,000 or ICC/100,000. This ratio is used to provide a more complete quantification of supply; i.e., we need to know not only the 50—618 0 — 80 - 6 numbers of physicians in practice, but also the populations which they serve. For aggregate and specialty supply, the Cen- sus Bureau's Series II (or Series III, which pro-V jects slower growth) estimates of population are used. As these reflect the 1970 Census, more ac— curate information will be available from the upcoming 1980 Census. For locational distribution, the population estimates try to be more specific, as supply estimates are part of programatic efforts to iden- tify HMSAs. For rural areas 1972 State popula- tion estimates are used, and counties are as- sumed to maintain fixed proportions of their respective State populations. For urban areas, the population is that identified by the Bureau of the Census as metropolitan low-income areas. Although these low-income populations have declined (17,936,000 in 1974 to 16,554,000 in 1976, or a decline of 3.8 percent per year), the physician-to-population ratio is assumed to hold constant in the future because of the previously mentioned expectation of continued emigration of physicians out of the inner cities. It should be obvious that supply, as refer- enced to population, depends not only on the physician supply projections, but also on the population projections. An example is the dis- tinction between projections of physician supply which include or exclude Federal physicians. In projecting supply for rural areas, table 19 esti- mated that there were 287,800 active non-Fed— eral MDs in 1975. Table 9 estimates that there were 363,400 active MDs (including Federal) in 1975 (377,500 minus 14,100 DOS). The 287,800 figure was used to allocate physicians among all counties. However, in subtracting the Federal physician supply, no effort was made to de- crease the population by a proportional amount (these Federal physicians were active and pre- sumably providing patient care). In addition, 1967 population estimates were used in table 19, whereas table 9 used population estimates that included projections of population growth. On the physician side of the ratio, table 9 assumed that a portion of “not classified" MDs were active; whereas table 19 excluded this cate- gory from its count (approximately 30,000 phy- 40 0 Forecasts of Physician Supply and Requirements sicians, or about 10 percent). In addition, table 9 included DOs, table 19 did not. So even though at first glance the differences in the number of active physicians represented in tables 9 and 19 (377,500 v. 287,800) might seem accounted for from the exclusion of Feder- al MDs, ”not classified” physicians, and DOS in table 19, the method of quantifying the popula- tion also contributed to the different physician- to-population ratios (176.8/100,000 v. 134.8/ 100,000). We have already seen that there are defini- tional problems associated with quantifying physician supply. These definitional problems will take on even more significance once we begin to quantify requirements and try to match that with supply. Two basic problems are in- volved: 1) the amount of patient care that is at- tributed to the average physician, and 2) the type of patient care provided. The first problem is usually couched in terms such as “productivity” or ”full~time-equiv- alent,” which attempt to provide a common ref- erence by which the number of physicians can be equated to a certain volume of patient care services. For example, physicians’ assistants in prisons were assumed to be equal to 0.5 physi- cians (USDHEW, 1978a). In this case, a physi- cian was equal to a full-time-equivalent (FTE), whatever the particular hours or number of pa— tients seen by prison physicians. Implicit in this definition are assumptions on physician produc- tivity. Other uses of FTE are more explicit. In- diana uses a definition of a FTE primary care physician as a general or family practitioner in the age group 35 to 39, which has the highest output in terms of visits per year for that specialty (Hindle et al., 1978). A more common method is to use average productivity figures by specialty, either as measured by the average pa- tient care hours worked per week (hospital and ambulatory care), the number of patients seen per week (usually expressed as the number of ambulatory visits), or both. And still another method is to estimate what percent of time is spent on nonpatient care activities and subtract that percentage from the total (aggregate or by specialty) supply. These productivity or FTE estimates are crucial when comparing supply with requirements, because they are the basic methods underlying the comparison. Given the same basic numbers of physician supply as pro- vided through “head counts" by the methods summarized earlier, whether supply equals, falls short, or exceeds demand obviously depends on the productivity assumptions applied to the physician. The definition of ”primary care" involves more than the simple identification of which specialties “qualify" for that designation. Yet we have already seen that what specialties count as primary care is quite confusing. Even if we limit the specialties to general practice, family prac- tice, general internal medicine, and general pediatrics, there can be great variations in the quantification of primary care physicians, be- cause over 50 percent are in the internal med- icine category. Yet, at various times, some sub- specialties of internal medicine are included and some are not. For example, table 13 excludes dermatology, gastroenterology, pulmonary dis- ease, cardiovascular disease, and allergy from the primary care internal medicine subspecial- ties, but includes allergy and immunology, dia- betes, endocrinology, geriatrics, hematology, immunology, infectious diseases, neoplastic dis- eases, nephrology, nutrition, oncology, and rheumatology. The Institute of Medicine (1978) reviewed 38 definitions of primary health care, and con- cluded that primary care cannot sufficiently be defined by the location of care, by the provider's disciplinary training, or by the provision of a particular set of services. It then goes on to elaborate on what it considers primary health care's five essential attributes: 1) accessability, 2) comprehensiveness, 3) coordination, 4) con- tinuity, and 5) accountability. In a study examining the general care content of different specialty practices, the data was dis- aggregated into five components: 1) first en- counter, 2) episodic care, 3) principal care, 4) consultative care, and 5) specialized care (Aiken et al., 1979). Principal care was defined as: There is evidence of continuity; the physician reports having seen the patient before and con- siders him or her to be a regular patient. Com- Ch. 2—Supply 0 41 prehensiveness is suggested, since the physician indicates that he or she provides most of the pa- tient’s care. Principal care thus falls short of the Institute of Medicine's definition of primary health care. Obviously, quantifying the supply andrequire- ments for specific physician specialties will dif- fer between these definitions, and they will sub- stantially affect the quantification of specialty distribution. This difference also points out the use of specific assumptions on FTEs and productivity. If many specialty types are providing principal care, the use of FTEs will serve as proxy meas- ures for some part of the total demand for the specific specialties. But the different specialties may also have different productivity rates. For example, internists generally see sicker and older patients than seen by general and family practitioners, and their average patient loads may be considerably less than the latters’ (table 24). Different results can be easily obtained on: 1) what exactly is primary health care, 2) which Table 24.—Estimated Principal-Provider Patient Loads of General Practitioners, Family Practitioners, and General Internists Average number of persons covered Specialty perphysician General practitioner ................ 870- 965 Family practitioner ................. 1,004 -1,127 General internal medicine ........... 468- 523 SOURCE: L. H. Aiken et al., ”The Contribution of Specialists to the Delivery of Primary Care,” 1979, table 4. specialties qualify, 3) the proportions within each specialty which provide principal health care, and 4) the use of different productivity values for each specialty. And different re- quirements projections also easily result when these factors, in addition to the specialty care responsibilities of each specialty, are used to translate these FTE/productivity values into head counts for each specialty. And similar es- timates must be made for the supply head counts in order to ultimately compare the sup- ply with the requirements projections. 3. Requirements INTRODUCTION Faced with what seemed to be a straightfor- ward question on what Our country's physician supply is and will be, we have found that supply projections are not definitive and need to be revised periodically. And in somewhat circular fashion, supply projections can be substantially dependent on deliberate Federal actions at the same time that Federal actions are somewhat de- pendent on supply projections. Even if there were agreement on the assump- tions underlying the supply projections, what the projections mean in terms of meeting the re- quirements for physicians would still be un- clear, as the numbers represent self-designated specialties, general estimates of physicians in ac- tual patient care activities, and general estimates of the locational distribution of the supply of physicians. In order to compare supply with re- quirements, common standards of quantifica- tion (e.g., full-time-equivalents (FTEs), what counts as primary care and what types of physi- cians provide it, etc.) must be applied to both estimates. The limitations imposed by definitional prob- lems compound the problems that arise from the specific assumptions underlying the supply pro- jections. Reaching agreement on the assump- tions underlying the projections (e.g., the future prospects for capitation and its effect on the number of domestic graduates) is a separate question from agreeing on questions such as what counts as primary care and how the vol- ume of primary care is to be translated into specific numbers of physicians in the various specialties. In other words, the separate projec- tions for supply and requirements and compari- sons between the two have both best-guess and value-laden assumptions undergirding the meth- odologies. The first task in estimating requirements for physicians is to decide what method should be adopted. A 1977 review of the literature (USDHEW, 1977a) concluded that the ap- proaches could be categorized into one of four groupings and pointed out that the first relates to treatment of all medical needs, as determined by disease prevalence and morbidity data, and that the other three categories deal with the de- mand for care, as derived from the opinions of health professionals, calculations of service re- quirements and manpower productivity, and observed staffing patterns in prepaid group practice settings (health maintenance organiza- tions, or HMOs). The four categories were de- fined as: 0 Medical need based ratio—A ratio which describes the number of health profession- als required to care for a given population if all disease conditions that require treat- ment (existing utilization plus unmet needs) are actually treated. Data on morbidity or disease incidence and prevalence must be used. 0 Professional judgment based ratio. —A ratio which reflects the opinion of health professionals or health manpower experts regarding the number of health profession- als needed to meet the expressed demands for care of a given population. This cluster encompasses ”ideal," ”adequate,” and ”minimally acceptable" ratios. The stand- ard is derived from an aggregate assess- ment of the manpower situation, without detailed consideration of utilization or pro- ductivity factors. 0 Demand/productivity based ratio.——A ra- tio which describes the number of health professionals needed to care for a given population, as derived from specified as- sumptions about services demanded and/ or manpower productivity. Calcula- tions may account for changes in technol- ogy, health insurance coverage, composi- tion of the population, utilization of allied health personnel, and similar factors. 45 46 ' Forecasts of Physician Supply and Requirements 0 HMO based ratio—A ratio which directly reflects or is derived from observed staffing patterns of prepaid group practice. This review concluded that three of the four approaches required special “leaps of faith:" Need-based standards are appropriate for planning only if consumers are both able and willing to express all medical problems or de- mands for services. Acceptance of professional judgment based standards requires blind faith in the knowledge and foresight of those con— sulted—especially ironic in that ”medical exper- tise” (knowledge of diagnosis and treatment re- quirements) is only one of many types of infor- mation needed to estimate manpower require- ments. Adoption of HMO based standards (ad- justed HMO staffing patterns) assumes that most care will be provided through this delivery mode which integrates both financial and deliv— ery system variables. The fourth approach, the calculation of re- quirements estimates based on explicit utiliza- tion and productivity assumptions, requires no leaps of faith, but neither does it provide unam- biguous guidance. It is viewed, nevertheless, as the most valid approach available for estimating health manpower requirements. Its value to the health planner is that it focuses attention on the parameters which determine manpower require- ments. In doing so, it lays the groundwork for integrating manpower with health system plan- ning. In this broader context, policies can be for- mulated to influence these parameters and mod— ify health manpower requirements in a socially desirable manner. Another criticism frequently leveled at pro- jections of the future requirements for physi- cians is that the modeling effort may be a static, snapshot picture of the health care system and does not sufficiently account for change. This criticism is most frequently raised when a single physician-to-population ratio, as derived from a particular modeling effort, is used to project future requirements. The use of a fixed ratio represents the conclusion that physician require- ments should (or would) change in direct pro— portion to population growth. A fixed ratio does not take into account any trends toward in- creasing per capita use of physician services. Thus, the use of a fixed physician-to-population ratio versus the use of changing ratios will lead to quite different estimates of the future re- quirements for physician services. If changing ratios are used: 1) the rate of change adopted will have a significant effect on the calculation of future requirements, and 2) at some point, a leveling off of the rate of change has to occur. Table 25 compares supply with requirements using the studies in the literature review just cited (USDHEW, 1977a) that led to the highest estimates of requirements. Each of these studies estimated requirements for a specific year: for the HMO method, estimates were made for 1972; for the need-based method, estimates were for 1974; for professional judgment, 1975; and for demand/ productivity, the year chosen was 1980. As these projections excluded osteo- pathic physicians, the supply column in table 25 is limited to allopathic physicians (MDS). Taken together, table 25 shows that there was an ap- proximate balance between supply and re- quirements in the 1970’s. Table 26 is modified from a 1970 analysis (Hansen, 1970) of physician requirements pro- jections for 1975, the year which is the basic starting point for current projections. Again, we see a fairly wide range of requirement estimates, with the midrange approximately in balance with the actual supply. Table 25.—Comparisons of Aggregate Physician (MD) Supply With Requirements Using Different Models Rate/100,000 Target year Total Supply HMO ................. 153.6 1972 321,000 333,000 Need-based ........... 167.8 1974 355,600 351,000 Professional judgment. . 187.3 1975 400,000 366,000 Demand/productivity . . . 182.8 1980 407,000 430,000 (projected) SOURCE: See text. Ch. 3—Requirements 0 47 Table 26.—1960’s Projections of Physician Requirements in 1975 (actual supply in 1975 equals 378,000) Deficit (— ) or Projection study Requirements Supply surplus ( +) Bane Committee Report (1959)a ..... 330,000 (minimum) 312,800 — 17,200 318,400 —- 11,600 Bureau of Labor Statistics (1966)b . . . 305,000 — — Fein (1967)c ...................... 340,000 to 350,000 361,700 + 21,700 to + 11,700 372,000 to 385,000 — 10,300 to - 23,300 National Advisory Commission on Health Manpower(1967)d ........ 346,000 (minimum) 360,000 + 14,000 Bureau of Labor Statistics (1967)3 . . . 390,000 360,000 — 30,000 Public Health Service (1967)‘ ........ 400,000 360,000 — 40,000 425,000 — 65,000 aSurgeon General’s Consultant Group on Medical Education, Frank Bane, Chairman, Physicians for a Growing America (Washington: Government Printing Office, 1959). bU.S. Bureau of Labor Statistics, “America's Industrial and Occupational Manpower Requirements, 1964-1975,” In: The Out- look for Technological Change and Employment, Appendix Volume l to Technology and the American Economy, report of the National Commission on Technology, Automation, and Economic Progress (February 1966). °Rashi Fein, The Doctor Shortage: An Economic Analysis (Washington: The Brookings Institution, 1967). dNational Advisory Commission on Health Manpower, Report, vol. II. (Washington: Government Printing Office, 1967). eBureau of Labor Statistics, Health Manpower 1966-1975, A Study of Requirements and Supply (Washington: Government Printing Office, 1967). fPublic Health Service, Health Manpower, Perspectives 1967, (Washington: GPO, 1967). SOURCE: W. L. Hansen, “An Appraisal of Physician Manpower Projections," Inquiry 7: 102-113,1970. Two other observations are of interest. Fein’s estimates are provided in two sets of numbers. The first set, 340,000 to 350,000 is based on population growth alone. This would be ana- logous to the use of a fixed physician-to-popula- tion ratio to predict, in 1967, the demand in 1975. The second set of numbers, 372,000 to 385,000, is based on an increase due to all fac- tors. It would be analogous to including further increased requirements due to increasing per capita utilization of physician services. Finally, it should be noted that the estimates of “require- ments" of the Bureau of Labor Statistics (BLS) were revised markedly between 1966 and 1967, increasing from 305,000 to 390,000. In essence, it was a judgment that, given rising demand for physician services, 85,000 more physicians could be employed. ECONOMIC MODELS The Bureau of Labor Statistics Model The US. Department of Labor’s BLS provides projections of demand and training needs for the major occupations every 2 years (BLS, 1979a). Thirteen occupational groupings are analyzed, including the health occupations. The In pursuit of quantifying future ”appropriate demand" or “requirements," little attention has been paid on past and current balances or im- balances between the supply of and require- ments for physician services. Policy has been less concerned with whether the projections were correct than with the effect of such projec- tions on current decisionmaking. Thus, this as- sessment concentrates on the two modeling ef- forts that will have the most impact on Federal health manpower policy, the sustained model- ing and projection activities of the Bureau of Health Manpower's (BHM) Division of Man- power Analysis, and the yet-to-be completed deliberations of the Graduate Medical Education National Advisory Committee (GMENAC). health occupations are medicine; dentistry; nursing; medical technologists, technicians, and assistants; therapy and rehabilitation; and other health occupations. These projections relate manpower to projected economic demand (ex— 48 0 Forecasts of Physician Supply and Requirements penditures) as provided by the Bureau’s model of the future economy, which projects the future gross national product (GNP) and its compo- nents—consumer expenditures, business invest- ment, governmental expenditures, and net ex- ports; industrial output and productivity; the labor force; average weekly hours of work; and employment for detailed industry groups and occupations. Current projections are based on the follow- ing assumptions: 0 the institutional framework of the US. economy will not change radically; 0 current social, technological, and scientific trends will continue, including values placed on work, education, income, and leisure; 0 the economy will gradually recover from the higher unemployment levels of the mid- 1970’s and reach full employment (defined as an unemployment rate of 4.7 percent in 1985 and 4.5 percent in 1990); and 0 trends in the occupational structure of in- dustries will not be altered radically by changes in relative wages, technological changes, or other factors. Beginning with population projections by age and sex from the Bureau of the Census, projec- tions of the total labor force are derived by using expected labor force participation rates for each group. The labor force projection is then translated into the level of GNP that would be produced by a fully employed labor force. The GNP projection is then divided among its four major components—consumer expendi— tures, business investment, governmental ex- penditures, and net exports. Each component is broken down by producing industry. Medical care falls under the consumer expenditure com- ponent. Estimates of future output per hour are derived from productivity and technological trends in each industry, and industry employ- ment projections are derived from the output es- timates. These projections are then compared with employment projections through the use of regression analysis. Comparison of the two methods is used to identify inconsistencies of one method with past trends or with the Bu- reau's economic model, and the projection is ad— justed as needed. Projections of industry em- ployment are translated into occupational em- ployment projections for each industry (201 in- dustry sectors and 421 occupation sectors). The growth rate of an occupation is determined by: 1) changes in the proportion of workers in the occupation to the total work force in each indus- try, and 2) the growth rate of industries in which an occupation is concentrated. In addition to occupational employment pro- jections, attrition from the existing work force is calculated to estimate average annual replace- ment needs for each occupation over the projec- tion period. Supply estimates assume that past trends of entry into the particular occupation will con- tinue. These estimates are developed independ- ently of the demand estimates; i.e., wages do not play a major role in equating supply and de- mand. Table 27 summarizes these projections. 1985 projected employment (demand) of 520,000 physicians, up from 375,000 in 1976, equals the projected supply (see tables 9 and 10). Table 27.—Bureau of Labor Statistics Projections of Physician Supply and Requirements, 1985 Employment, 1976 ................... 375,000 Projected employment, 1985 .......... 520,000 Percent growth, 1976-85 .............. 37.8 Average annual openings, 1976-85 ...... 21,800 Growth .......................... (16,000) Replacement ..................... (5,800) Available training data: Projected 1976-85 1975-76 (annual average) MD degree ....... 14,163 15,997 DO degree ....... 806 1,128 SOURCE: Occupational Projections and Training Date, Bureau of Labor Statis- tics, Department of Labor, Washington, D.C., Government Printing Office, Bulletin 2020, 1979, pp. 64-65. BLS, in its forthcoming revisions of its esti- mates to include 1990 projections, will adopt the midpoint of the range of projections from the BHM model for physician requirements. BLS considers the BHM model as more sophisticated for two reasons. First, while the BLS model in- corporates population as a variable, no con- sideration is given to the varying utilization rates of the demographic components. Women Ch.3——Requirements ' 49 use more physician and hospital services than men, and older people require more care than the young. The BHM model does capture these aspects of demand. Second, the BHM model estimates the effects of possible price elasticity for physician services and provides a narrow range of the resultant expected demand for physician services. BLS also points out that both its and BHM's models cannot measure and project the capacity of physicians to stimulate demand. A down- ward bias in the projections would result if physician-stimulated demand reflects real need for medical care. To the extent that physician- stimulated demand reflects unnecessary care, the demand estimates would introduce an up- wards bias to the projections. BLS also points out two other factors which would bias the BHM projections in an upward direction. The inclusion of data points from 1968 and 1969 for projecting utilization rates reflect the sharp growth in utilization attrib- utable to Medicare and Medicaid startup. (As we will explain later, we agree that 1968 and 1969 data points should be excluded in deter- mining utilization trends, but not for the reason given by BLS). Second, BLS points out that as the coinsurance (that part of fees paid by the pa- tient) has been and is projected to continue dropping, the effective price of physician serv- ices has dropped as well. BHM'S assumption of a constant price elasticity means that a percent- age drop in coinsurance from 20 to 10 percent would result in the same percent increase in the demand for physician services as would a drop in coinsurance from 50 to 40 percent. A more likely effect would be that consumer demand would gradually be saturated as the percentage coinsurance decreases (BLS, 1979b). The Bureau of Health Manpower Model BHM uses a demographic projection method which makes certain assumptions about medical care utilization and physician productivity to arrive at its estimates for physician require- ments. The general method is not specific to physicians and is applied to 29 general types of health manpower, ranging from selected MD specialties to broad allied health groups. Al- though specialty-oriented estimates for physi- cians can be calculated from the total popula- tion’s utilization of specific types of care by cross-multiplying the projected size of each pop- ulation subgroup by its associated utilization rate, the method is presently not considered reli— able enough to use in projecting specialty-by- specialty requirements. The basic assumptions are: 0 No major unforeseen events will occur in the projection period, including the enact- ment of a comprehensive program of na- tional health insurance (NI-II). 0 Supply and demand were in balance in 1975. As we have seen from tables 25 and 26, past estimates are in general agreement with this assumption. 0 Physician productivity does not change substantially between 1975 and 1990. 0 Price elasticity, the sensitivity of demand to net price, remains constant between 1975 and 1990. (Different choices of price elas- ticity of demand coefficients, representing alternative rates of consumer response to price changes for a given personal health service, are used to present the range of estimates.) 0 Nondollar costs of obtaining care do not change substantially during the projection period. 0 No major health care or manpower substi- tution occurs between service categories. The model is developed in three stages which, in practice, may be separated or combined as desired. The most rudimentary stage is termed the ”framework." It estimates future demand from the anticipated impact of population growth and demographic shifts (age, sex, in- come distributions) on the economic demand for medical services, with physician productivity assumed to remain the same. In this first or framework stage: 0 The U.S. population is projected to 1980, 1985, and 1990 by age, sex, and income subgroups in 40 components (5 age by 4 in- come by 2 sex groupings). 50 0 Forecasts of Physician Supply and Requirements 0 For each of the 40 subgroups, a utilization rate for each of 20 types of health service settings (e.g., general medical office visit, inpatient hospital admission, nursing home stay, vision care, laboratory services, etc.) was estimated from recent National Health Interview Survey data and other sources for the 1975 base year and the future years. 0 The percentage increase in utilization for each of the 20 types of care is obtained by summing over the population and dividing the future year utilization by the 1975 base year utilization for that type of care. The second stage, called the ”baseline" config- uration, attempts to factor the effects of histori- cal trends in per capita utilization of medical services into the general model. In this second stage: 0 Utilization, by each of the 20 care types, is adjusted to account for future increases that represent continuation of recent trends. Economic and noneconomic trends are analyzed separately. Economic trends interpret the effects of price and insurance copayment charges on changes in utiliza- tion. As noted earlier, the price elasticity is assumed to be constant over the period 1975-90. The utilization projections in each of the 20 care types are then individually adjusted for the future years, based on these projections of price and coinsurance payment. After removing price effects from the past utilization trends, the remainder (representing education, consumer prefer- ence, patient-care physician supply, etc.) is the noneconomic trend. This noneconomic effect is assumed to continue in a linear fashion into the future and per capita uti- lization adjustments for each care type are made. 0 The adjusted increases in future care uti- lization are then applied to the 1975 esti- mates of the distribution of the 29 man- power categories in each of the 20 care types to give a preliminary estimate of the future manpower required to provide the projected care utilization. 0 Adjustments are then introduced to ac- count for trends in the manpower required to provide a given amount of care utiliza- tion. For physicians, it has been previously noted that the assumption is that produc- tivity remains constant from 1975-90. 0 Demand is then summed over the 20 types of care to arrive at the baseline forecast of future demand. These estimates assume no major departure from historical experience. The third stage, the ”contingency” modeling capability, is a separate component that has been used to explore the possible impact on the economic demand for physician services of the following contingen- c1es: 0 The impact of NH1 can be estimated by ad- justments to the utilization increases through new economic trend adjustments which provide for different NHI copayment possibilities. 0 Expanded-function task delegation can be factored in by new adjustments to the man- power staffing trends based on alternative assumptions of midlevel practitioner em- ployment and their productivity rates. ' Expanded use of HMOs is treated by alter- native estimates of the population enrolled in HMOs in future years and separate uti- lization rates, trends, and staffing assump- tions for this part of the population. The Framework To predict the impact of population growth and demographic shifts, the model starts with the basic assumption that supply and demand were in balance in 1975 and calculates per capita utilization rates for each of 40 population segments (age by sex by income categories) with respect to 20 forms of health care. Table 28 displays the population matrix utilized; table 29 indicates the various health care categories. Note that the BHM health care categories in- clude care setting (office-based, short- and long- term hospital care, nursing homes) as well as types of care (pediatric, surgical, psychiatric). Thus, the capability of the BHM model to make distinctions among physician specialties is not very fine-grained. The basic distinctions made are between general medical, pediatric, obstet- rics-gynecology, psychiatric, vision, surgical, Ch.3—Requirements ' 51 Table 28.—Population Matrix Used in the BHM Model Family income Under $5,000 Males Under14 14-24 25-44 45-64 65 + Females Under 14 14-24 25-44 45-64 65 + $5,000-$9,999 $10,000-$14,999 $15,000 and over SOURCE: JWK International Incorporated, Evaluation of Project SOAR (Supp/y, Output, and Requirements), draft report, DHEW contract No. HRA 23278-0140, 1979. Table 29.—Health Care Categories Used in the BHM Model Form of care General care Pediatric care Obstetric-gynecological care Psychiatric care Vision care Other medical office care Short-term hospital Outpatient care Surgical care Medical care Long-term hospital Psychiatric care Other long-term hospital care Other care settings Nursing home care Dental care Veterinarian services Optometric care Podiatric care Other patient care, not elsewhere specified Pharmacy services Laboratory services Noncare activities, not elsewhere specified Setting Medical office Noncare settings SOURCE: JWK International Incorporated, Evaluation of Project SOAR (Supply, Output, and Requirements), draft report, DHEW contract No. HRA 232-78-0140, 1979. podiatric care, and “all other.” Finally, it should be noted that the overall model applies to 29 categories of health manpower (including vet- erinary medicine). Accordingly, all 20 health care categories do not come into play in estimat- ing physician demand. Utilization rates are'calculated for each ap- plicable category of health care for the 40 population subgroups (based on recent National Health Interview Survey data, etc.). Changes in population size and in demographic distribution (age by sex by incomes) are calculated for future years. 1975 utilization rates for each category are then projected onto the future population estimates and the results are summed to obtain expected utilization of services. Based on population and demographic shifts, physician demand is expected to increase by 9.5 percent between 1975 and 1990 (Cultice, 1979) yielding an estimated demand of 414,336 in 1990. This corresponds to a projected increase in the population of 10.5 percent in the same period. Of the 414,336 physicians, 229,276 are expected to be in demand in office—based settings (versus 205,196 in 1975), and 126,008 in short- term hospital care (versus 117,573 in 1975). The remaining 59,052 physicians are expected to be in demand in other care settings. The anticipated population changes between now and 1990 that are most pronounced and ap- pear most likely to impact on utilization pat- terns are: 1) an aging population, and 2) strong signs of upward income mobility, measured in constant (1970) dollars. Although the U.S. population is aging, that does not mean the 1990 population will be an elderly one. Growth in the over-65 category will, in fact, be small (from 9.5 percent of the total population to 12.3 percent). Rather, the bulk of the population will be in the prime of adulthood, 25 to 44. The percentage of Amer- icans in this age category is projected to grow 52 0 Forecasts of Physician Supply and Requirements from 23.6 to 32.9 percent. Correspondingly, the percentage of persons under 25 is projected to drop sharply (from 46.3 to 35.4 percent). The projected shift in income distribution is more dramatic than the age shift. The percent- age of individuals whose family income, meas- ured in 1970 dollars, lies below $10,000 is pro- jected to shrink from 50.6 to 20.2 percent, while the percentage of those with family incomes of $15,000 or greater is projected to grow from 23 to 58.1 percent—a complete reversal of the per- centages in the below $10,000 and above $15,000 groups respectively. These population changes are summarized in table 30. Changes in the age and income composition of the population can be expected to bring about changes in patterns of utilization of medical services. Specifically, we can anticipate that those forms of care more heavily used by the elderly and by those with higher incomes will grow more rapidly than those forms used by the young and by the poor, and that physician serv- ices most closely associated with the former will also experience a higher rate of growth. Thus, the greatest growth rates are projected for nurs- ing home and podiatric care (both utilized more heavily by the elderly) and for psychiatric, vi- sion, and dental care (services used more by those with higher incomes). Conversely, the lowest growth projections are for pediatric care and hospital outpatient care (more frequently used by the poor). The population shift into the 25 to 44 age bracket is also expected to increase demand for ob-gyn care somewhat, because these are the childbearing years. It is worth ob- serving that though the U.S. population is ag- ing, we will not be faced with providing for the medical needs of a heavily geriatric citizenry as of 1990. The major growth is going to be in the age group 25 to 44, which is traditionally a com- paratively healthier age group relative to both children and the elderly. It is also noteworthy that most of the health care areas projected to grow as a result of demo- graphic shifts are areas in which services are not provided by physicians (dental care, podiatry) or only partially so (vision care, nursing home services). The projected growth rates for the 20 types of health services is summarized in table 31. Note that the growth factor attributable solely to overall population growth is 110.5 percent; i.e., the population will have grown 10.5 percent be- tween 1975 and 1990. Thus, surgical care is pro- jected to grow at barely over the overall popula- tion growth rate (0.4 percent higher), and both general medical office-based care (108.9 per- cent) and medical care in short-term hospitals (103.3 percent) are projected to grow at less Table 30.—Projected Shifts in Age and Income Distribution, 1970-90 Percent distribution, by yeara 1970 1975 1980 1985 1990 Age Under 14 ................... 26.8% 23.2% 20.5% 19.5% 19.4% 14 - 24 ..................... 19.5 20.8 20.5 18.5 16.0 25 - 44 ..................... 23.6 25.1 28.1 31.1 32.9 45-64 ..................... 20.6 20.4 19.8 19.2 19.4 65 + ...................... 9.5 10.5 11.1 11.7 12.3 Total .................. 100.0% 100.0% 100.0% 100.0% 100.0% Incomeh Under $5,000 ............... 19.9% 16.6% 11.6% 9.9% 8.6% $5,000 - $9,999 .............. 30.7 23.4 20.2 16.9 11.6 $10,000 - $14,999 ............ 26.4 24.1 24.3 21.2 18.7 $15,000 - and over ........... 23.0 35.8 43.7 51.9 58.1 Total .................. 100.0% 100.0% 100.0% 100.0% 100.0% aColumns may not add due to rounding. bIn 1970 dollars. SOURCE: JWK international Incorporated, Evaluation of Project SOA Ft (Supp/y, Output, and Requirements), draft report, DHEW contract No. HRA 232-78-0140, 1979. Ch. 3—Requirements 0 53 Table 31.-—Projected Utilization Growth Factors, 1975-90 Projected growth 1975-90 Medical office General care .......................... 108.9 Pediatric care ......................... 101.1 Ob-gyn care ........................... 120.5 Psychiatric care ....................... 124.7 Vision care ............................ 123.8 Other care ............................ 111.0 Short-term hospital Outpatient care ........................ 95.6 Surgical care .......................... 110.9 Medical care .......................... 103.3 Long-term hospital Psychiatric care ....................... 118.9 Other care ............................ 110.53 Other care settings Nursing home care ..................... 127.3 Dental care ............................. 121.6 Veterinarian services ................... 110.5a Optometric care ....................... 116.4 Podiatric care ......................... 127.2 Other care ............................ 110.6a Noncare settings Pharmacy services ..................... 111.7 Laboratory services .................... 110.5a Noncare activities ...................... 110.5a aAge-, sex-, and income-specific utilization rates are either inappropriate or un- available tor these categories. The growth shown (110.5) is that attributable solely to the overall growth of the population. SOURCE: JWK international Incorporated, Evaluation of Project SOAR (Supp/y, Output, and Requirements), draft report, DHEW contract No. HRA 23278-0140, 1979. than the overall population growth rate of 110.5 percent. These projections deal with utilization of medical services as an expression of economic demand, not medical need. We do know, how- ever, that higher income has traditionally been associated with improvements in health status; i.e., less need for medical care. Tables 32 to 35 give some indication of the differential health status and utilization patterns of high- and low- income groups. Clearly, low-income groups are sicker and use more medical services, particular- ly hospital care, than higher income groups. An evaluation completed under contract to BHM of its general demand model (JWK Inter- national, 1979) raises the question whether “new arrivals" in the upper income brackets would exhibit the same patterns of utilization or the same underlying patterns of medical need as long-time members of upper income categories. Conceivably, the “new arrivals" in the upper in- come categories might lag in exhibiting the lower utilization rates of the higher income brackets. We might also question whether the relation- ship between utilization of medical services and income is indeed constant over time, which is what is implied by projecting 1975 income uti- lization rate differentials to the 1990 population. If we had done such an exercise 15 years ago (that is, attempted to project 1975 utilization from a 1960 data base) we would have underes- timated the actual utilization rates of low- income groups. Over this time period the utili- zation rates of the poor have risen more rapidly, and, in the area of physician visits, the poor now use more services than the nonpoor, whereas, previously they used less (see table 34). These questions are raised not so much to cri- tique the assumptions that went into the BHM framework model—which are quite plausible assumptions—but merely to point out that even the most plausible of assumptions may finesse a great deal of uncertainty. The Baseline Configuration The ”baseline" configuration factors in the ef- fects of historical trends in per capita utilization of medical services. Two major component trends are distinguished in the analysis. The price-related component is that portion of the Change in utilization that is expected to result when changes in the price of a particular form of care or medical service affect consumer deci- sions to seek that care or service. The non-price- related component is a residual that measures the effects of all other possible influences com- bined, including such factors as changes in the accessibility of care, changes in population, changes in consumer taste and preference, changes in medical technology, environmental changes, and changes in disease prevalence and incidence. The non-price-related component is the greater of the two. The language of the BHM model is the lan- guage of economics, which tends to treat medi- cal care like any other ”consumer product,” 54 0 Forecasts of Physician Supply and Requirements Table 32.—Prevalence of Selected Chronic Conditions Reported in Health Interviews, by Family Income (United States) Chronic Heart Hyper- Impairment,D Hearing Vision Arthritis Asthma bronchitis Diabetes conditions tensiona back or spine impairment impairment Family income (1969) (1970) (1970) ' (1973) (1972) (1972) (1971) (1971) (1971) Number per 1,000 persons 17-44 years Under $5,000. . . 46.9 34.1 28.4 11.4 32.5 48.9 59.4 55.4 43.2 $5,000-$9,999 . . 40.5 23.6 22.3 8.7 23.3 40.8 50.5 44.0 31.7 $10,000-$14,999 38.7 24.4 21.8 8.4 22.5 35.9 47.4 39.3 28.7 $15,000 and over 35.9 26.8 23.7 8.0 24.3 29.8 42.4 35.8 30.9 Number per 1,000 persons 45-64 years Under $5,000. . . 297.8 53.5 44.2 74.1 139.3 172.7 102.8 158.9 114.1 $5,000-$9,999 . . 200.3 33.5 38.7 43.8 92.5 125.4 67.2 118.1 57.4 $10,000-$14,999 163.7 23.7 29.0 37.8 74.3 121.3 62.3 107.3 45.9 $15,000 and over 159.8 22.7 30.3 30.5 66.6 105.3 52.2 85.9 48.9 aWithout heart involvement. bExcept paralysis. SOURCE: National Center for Health Statistics, “Selected Reports from the Health Interview Survey," Vital and Health Statistics, Series 10. Table 33.—Number of Disability Days per Person per Year by Family Income (United States, 1973) Restricted Family income activity days Bed disability days Work-loss days Days per person ages 17-44 years Under $5,000 ............. 21.1 8.3 6.5 $5,000 - $9,999 ............ 14.6 5.7 5.9 $10,000 - $14,999 .......... 11.9 4.8 4.8 $15,000 and over .......... 11.4 4.4 4.6 Days per person ages 45-64 years Under $5,000 ............. 45.7 15.5 7.5 $5,000 - $9,999 ............ 25.1 8.7 7.3 $10,000 - $14,999 .......... 16.9 5.9 5.5 $15,000 and over .......... 14.0 4.5 5.3 SOURCE: National Center for Health Statistics, ”Current Estimates from the Health Interview Survey, 1973,” Vita/and Health Statistics, Series 10, No. 95; and unpublished data. with the assumption that utilization is generated by consumer demand. It has frequently been argued, however, that much utilization of medi- cal care is physician generated. The model does take such factors into account. What it does not do is differentiate between consumer- versus provider-generated changes in per capita utiliza- tion of medical care. The price-related utilization trend is cal- culated through out-of-pocket costs to the con- sumer. Over the years, the price of health care relative to the Consumer Price Index (CPI) will have risen for some forms of care and declined for others. In all (or almost all) instances, however, the actual out-of—pocket expenses to the consumer will have declined due to the in- creasing use of private and public health insur- ance. For those forms of care for which the net price to the consumer has declined, per capita utilization would be expected to increase. We will not describe in any detail the tech- nical aspects of utilization trend analysis. Con- ceptually, however, the process involves the fol- lowing steps: 0 The observed utilization trend over the baseline time period (1968-76) is factored into its price- and non-price-related compo- nents. 0 A linear least squares regression equation is fitted to the non-price-related utilization trend and the resulting straight line is ex- trapolated forward in time to 1990. Ch. 3—Requirements ' 55 Table 34.—Number of Physician Visits per Year by Poor and Not Poor Status and for Whites and Others (United States, 1964 and 1973) Total Whites All others Age and year Poor Not poor Poor Not poor Poor Not poor Number of physician visits per person per year 17 - 44 years 1964 ......... 4.1 4.7 4.5 4.8 3.3 4.2 1973 ......... 5.7 5.0 5.8 5.0 5.6 4.8 45 - 64 years 1964 ......... 5.1 5.1 5.2 5.1 4.9 4.6 1973 ......... 6.3 5.4 6.1 5.4 7.1 5.3 Percent with no physican visits in past 2 years 17 - 44 years 1964 ......... 24.2 18.1 23.2 17.7 26.6 22.9 1973 ......... 13.4 12.8 13.1 12.7 14.5 13.5 45 - 64 years 1964 ......... 29.2 21.7 28.0 21.3 33.1 29.0 1973 ......... 20.6 16.9 21.4 16.9 17.0 16.9 Definition of poor is based on family income: Under $3,000 in 1964. Under $6,000 in 1973. In each case, this included about one-fifth of the population. SOURCE: National Center for Health Statistics, unpublished data from the Health Interview Survey. Table 35.—Number of Discharges From and Average Length of Stay in Short-Stay Hospitals, by Income and Color(United States, 1964 and 1973) Total Whites All others Age and year Poor Not poor Poor Not poor Poor Not poor Number of discharges per 1,000 population 17 - 44 years 1964 ......... 181 161 188 164 163 132 1973 ......... 198 148 190 148 223 149 45 - 64 years 1964 ......... 146 148 159 151 102 111 1973 ......... 225 152 238 153 174 133 Average length of stay in days 17 - 44 years 1964 ......... 6.9 6.3 6.8 6.2 7.1 8.0 1973 ......... 6.4 6.0 6.0 5.9 7.2 7.0 45 - 64 years 1964 ......... 14.4 9.7 12.8 9.5 22.6 13.5 1973 ......... 12.8 9.3 12.3 9.0 15.3 13.0 Definition of poor is based on family income: Under $3,000 in 1964. Under $6,000 in 1973. In each case, this included about one-fifth of the population. SOURCE: National Center for Health Statistics, unpublished data from the Health Interview Survey. 0 In a separate process, the historical values of coinsurance and price, defined as the ratio of CPI for health care to CPI for all items combined, are calculated. These val- ues are then multiplied to yield the pro- jected net price (out-of-pocket costs) to the consumer for the year or years in question. 60—618 0 - 80 — LI 0 The projected net price to the consumer is then applied to a demand curve with an as- sumed elasticity. This demand curve relates the price (to the consumer) of a given form of care to the demand for that care. Based on that relationship, the extrapolated non- price-related utilization is adjusted upward 56 0 Forecasts of Physician Supply and Requirements or downward to reflect the estimated im- pact of price changes during the years ahead. Current trend analysis employs ”high" and ”low" assumptions of price elasticity. Thus, two alternative price-adjusted utilization growth trends are projected. Table 36 shows alternative utilization growth rates based on high- and low- elasticity assumptions for four types of medical care. The utilization trend analysis is an extremely important element in the BHM general demand model. 1990 demand is projected to be 596,217 as compared to a 1975 demand of 378,376. Of the 217,841 additional physicians, 181,881, or 83 percent of the difference, is attributable to the assumptions made about rising per capita uti- lization. Table 37 summarizes the projected rise in physician demand between 1975 and 1990, separated into population/ demographic and per capita utilization trend components. The projection of the per capita utilization trend is perhaps the most problematic aspect of Table 36.—Estimated Growth in Per Capita Utilization, Four Forms of Health Care Projected per capita utilization in 1990 relative to baseline utilization in 1975 High Low elasticity elasticity Medical office services ...... 1.45 1.37 Short-term hospital services. . 1.38 1.29 Dental office services ....... 1.28 1.29 Community pharmacy services 1.58 1.50 SOURCE: JWK International Incorporated, Evaluation of Proiect SOAR (Supply, Output, and Requirements), draft report. DHEW contract No. HRA 232-780140, 1979. Table 37.—Increase in Demand From Population and Per Capita Utilization Changes, 1975 to 1990, BHM Model 1975 1990 Demand .................. 378,376 596,217 (assumed equal (projected) to supply) Increase .................. —— 217,841 (projected) Population effect ........ (35,960) Per capita utilization effect (181,881) SOURCE: See text. the BHM general modeling effort because of the difficult philosophical and methodological is- sues involved in identifying and projecting trends. Methodologically, one problem is that the non-price-related utilization trend includes changes in per capita utilization that occur as a result of demographic changes, including age and income shifts. Thus, the BHM model double counts the effects of age and income, and the estimates are inflated accordingly. While this problem is relatively easy to cor- rect (and the model is in the process of being ad- justed so that age- and income-related trends will be counted only once), the sensitivity of the estimates to alternative starting dates for trend projection is more difficult to remedy. The first task in trend analysis is to identify a time period for which data will be collected and analyzed. Until recently, the time period cov- ered began in 1966 and extended through the latest year for which suitable data were avail- able (currently 1976). The starting date was moved forward to 1968 because of concern that the increases in utilization of medical services that occurred during the startup period of Medi- care-Medicaid tend to misleadingly inflate the trend data when 1966 is used as a starting point. If the desire is to remove the unique historical impact of Medicare-Medicaid startup from the projection of a longer term societal trend in utilization, it is questionable whether moving the starting date of the trend analysis from 1966 to 1968 is sufficient. With respect to Medicaid, many States did not yet have their Medicaid programs fully in operation in 1968, and impor- tant amendments to the original legislation were also passed during this period, most notably, expanded nursing home coverage. The choice of starting date is extremely im- portant because, for most kinds of medical care (short-term hospital care excepted), the slope and even the direction of the trend line changes rather dramatically if the starting date is moved from the late 1960's to the early 1970’s. Figures 6 to 9 summarize unadjusted per capita utilization and high and low elasticity, non-price-related utilization for the years 1966 Ch. 3—Requirements 0 57 Figure 6.—Per Capita Utilization of Physician Office Services, 1966-76 Per capita utilization (visits per year) 500 4.75 ‘4.50 4.25 4.00 3.75 3.50 3.25 I V 3.00 1 966 1967 1968 1 969 1970 1971 1972 1973 1974 1975 1976 Year SOURCE: JWK International, Inc.. Evaluation of Project SOAR (Supply, Output, and Requirements), draft report, DHEW contract No. H RA-232-78-0140, 1979. to 1976. Only one set of data points (utilization of short-term hospital services) displays a con- sistent, natural linear trend. In the case of the other services (physician office visits, dental care, community pharmacy), the utilization data being extrapolated do not consistently dis- play a pattern for which linear extrapolation can be justified. The mathematics of trend projection (linear regression analysis) involves fitting a straight line to the data points. Clearly, such a concep- tual simplification of reality is less distorting of the world's true complexity when the data points themselves tend naturally to conform to a straight line pattern when graphically por- trayed. Philosophically, the issue is one of con- sistency. Trend projection in general and the methodology of linear regression in particular assume that reality—in this instance per capita utilization of medical care—exhibits a reason- ably consistent pattern; whether it be one of in- creasing, decreasing, or remaining steady. Figures 10 to 13 take the high-elasticity data points of figures 6 to 9 and fit straight lines to the data points for the years 1966-76, 1968-76, and 1971-76. Except for short-term hospital services, distinctly different trends can be pro- jected, depending on the starting date. For phy- sician office services, a start date of either 1966 or 1968 is seen to yield a distinctly upward trend, whereas a start date of 1971 shows a downward slope. In the case of dental office visits, starting the trend analysis in 1968 pro— duces a flat projection; starting in 1971 produces a decline. For community pharmacy services, the trend lines started in 1966 and 1968 show a sharp upward rise even though per capita uti- lization has been declining since 1973. The magnitude of the disparities in per capita utilization growth produced by different start dates is quantified in table 38. 58 0 Forecasts of Physician Supply and Requirements Figure 7.—-Per Capita Utilization of Short-Term Hospital Services, 1966-76 Percapna ufiflzafion (admissions per year) 0.150 0.125 0.100 0.075 . . .. . , 1966 1967 1968 1969 1970 1971 1972 1973 Year 1 974 1975 1976 SOURCE: JWK International, Inc., Evaluation of Project SOAR (Supply, Output, and Requirements), draft report, DHEW contract No. H RA-232-76-O140, 1979. Clearly, the considerable variability of these results (except in the case of short-term hospital services) points to the need for caution in ex- trapolating utilization trends to project future physician demand. A slight change of 1 or 2 years in the time period covered in the trend analysis can produce enormous differences in the projections. It is evident, however, that the choice of a pre-1970 start date followed by linear trend extrapolation is bound to result in drastically different estimates from estimates which use a post-1970 start date. BHM's review of its general demand model (JWK International, 1979) suggested that in- stead of linear extrapolation, a logarithmic fit would produce a somewhat better trend extrap- olation (for all but short—term hospital services), less sensitive to the Medicare-Medicaid startup years. Table 39 shows the difference in the non- price-related per capita utilization trend be- tween 1975-90 produced by employing logarith- mic versus linear extrapolation. The review concluded, however, that, while it is possible to reduce the disparities produced by selecting different start dates for trend analysis by using an alternative form of extrapolation, it was difficult to rationalize the use of a particular functional form of trend fitting, and that there was simply no reason a priori to expect the data to follow a logarithmic as opposed to a linear (or any other) pattern. In addition to math- ematical retooling, the evaluation suggested more use of human judgment rather than mech- anistic methods, perhaps through projections of what would be “most likely," based on classic Delphi techniques or simply averaging the re- sponse of a suitably selected group of knowl- edgeable individuals. Ch. 3—Requirements 0 59 Figure 8.—Per Capita Utilization of Dental Office Services, 1968-76 Per capita utilization (visits per year) 2.00 . 1.75 1.50 L25 . w 1‘: 1 .00 ‘ ~ : V ' ' 1 966 1 967 1 968 1969 1970 1971 1972 1973 1974 Year 1975 1976 SOURCE: JWK International, lnc.. Evaluation of Project SOAFi (Supp/y, Output, and Requirements), draft report, DHEW contract No. HRA-232-78-O140, 1979. BHM states that the reasons it continues to use the 1968 start date are: 1) the historical ob- servation of rising per capita use of physician services, 2) the more base years used to establish the trend, the sounder the methodology of trend extrapolation, and 3) its conclusion that the 1968 and 1969 data points do not reflect Medi- care and Medicaid startup activities (Cole, 1980). The problem, however, is not whether these statements are correct or not. Instead, the prob- lem stems from the specific rate of increase in per capita utilization of physician services that were used to calculate physician demand in ad- dition to that derived from demographic changes. We have simply pointed out that BHM's use of 1968 as the starting point results in a trend line radically different from actual uti- lization data from 1971 to 1976 (see figure 10). It is necessary to point this out, because the pro- jections of physician demand issued by BHM combine the estimates of demand due to demo- graphic changes with that due to increasing per capita use of physician services without an in- dication of the very different results that occur if the starting date for trend projections is changed by just 2 or 3 years. Even BHM’s internal eval- uation has suggested that more human judg- ment, rather than mechanistic methods, be used. Contingency Modeling The contingency modeling capability has been used to explore the possible impact on eco- nomic demand for physician services from: 1) alternative forms of NHL 2) various rates of growth of the HMO movement, and 3) in- creased use of midlevel practitioners or ”task delegation.” The effects of NH1 on utilization are assumed to be mediated through a lowering of the out-of- 60 0 Forecasts of Physician Supply and Requirements Figure 9.—Per Capita Utilization of Community Pharmacy Services, 1966-76 Percapfia ufifizafion (prescriptions per year) 7.5 7.0 6.5 6.0 5.5 5.0 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 Year SOURCE: JWK International, Inc., Evaluation of Project SOAR (Supply, Output, and Requirements), draft report, DHEW contract No. HRA-232-78-O140. 1979. pocket costs to consumers for medical services of various types. It is assumed that coinsurance under NHI would be lower than it would be without NHI. Clearly, the greater the gap be- tween the average coinsurance rates projected for future years in the absence of NHI and seems likely under NHI, the more utilization would be expected to rise under NHI. Three NHI plans have been modeled, assum- ing rates of coinsurance of 5, 10, and 15 percent. A recent rough estimate of the impact of NHI in 1990 assuming a 10-percent coinsurance rate with and a 25-percent coinsurance rate without NHI, employing a log linear fit, results in a 13- percent upward demand shift. Trend estimates suggest, however, that over- all, as coinsurance falls, the gap between out-of- pocket costs to the consumer with or without NHI is narrowing. This means that while those portions of the population who currently have little or no medical insurance would surely ex- perience a great drop in out-of-pocket expenses and would increase their utilization of medical services accordingly, in the aggregate, most Americans would not experience such a major drop in out-of—pocket costs. Thus, if the coin- surance rate continues to decline without NHI, the expectation would be for the eventual im- pact of NHI on utilization and physician de- mand to be less, the longer the delay in enacting an NHI plan, particularly if the plan enacted had a comparatively high coinsurance rate (e.g., 15 percent rather than 5 percent). With respect to the impact of HMO growth on utilization, currently available data on HMO growth suggest that only 6 percent of the pop- ulation can be expected to belong to I-IMOs in 1990. This is not enough to appreciably lower utilization or physician demand. Ch. 3—Requirements 0 61 Figure 10.—Non-Price-Related per Capita Utilization Trends, Physician Office Services, 1966-76 Non-price-related per capita utilization 4.00 3.75 3.50 3.25 195 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 Year SOURCE: J WK International, Inc., Eva/uailon of Project SOAR(Supp/y, Output, and Requirements), draft report, DHEW contract No. HRA-232- 78-0140, 1979. In the case of task delegation to midlevel practitioners, the productivity enhancement fac- tor is estimated at 30 percent. Here again, how- ever, the effect of expanded task delegation is projected to be negligible in 1990—on the order of 2 to 3 percent. This minor effect is due to the projected limited supply of nurse practitioners and physicians’ assistants, based on current training levels. Productivity Finally, productivity of physicians in 1990 is assumed to remain the same as productivity of physicians in the base year 1975. Productivity in the base year is addressed indirectly in the form of a staffing matrix which shows the number of units of manpower engaged in each separate form of health care activity during that year. Because productivity is assumed to be constant, the ratio of services to manpower will be the same in 1990 as in 1975, and it is on this basis that manpower estimates are generated. The total number of estimated active physicians as of 1975 was obtained from the American Medi- cal Association (AMA), with the various spe- cialties regrouped to provide a more compact typology. Physicians were then allocated to par- ticular specialties and care settings based on care profiles. As a result, the numbers used are not head counts of physicians claiming particular specialties, but are estimated FTE physicians providing a particular type of care. For exam- ple, the percentage of time the average general practitioner (GP) devotes to providing ob-gyn care is assigned to the ob-gyn category, while the estimated percentage of GP practice devoted to pediatric care is assigned to pediatric care. Thus, physician demand is calculated in terms of the nature of the services provided, as well as 62 . Forecasts of Physician Supply and Requirements Figure 11.—-Non-Price-Related per Capita Utilization Trends, Short-Term Hospital Services, 1966-76 Non-price—related per capita utilization 0.110 .. 0.105 0.100 0.095 0.090 0.085 ..; -. . .1 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 Year SOURCE:JWK International, Inc., Evaluation of Project SOAR (Supply, Output, and Requirements), draft report, DHEW contract No. H RA-232-78~0140, 1979‘ in terms of physician specialty categories. Table 40 shows the allocation of physicians by type and setting of care in the 1975 base year. Table 41 is an illustrative computation of the 1990 manpower demand for pediatricians. These figures are derived from the “framework" model; i.e., the projected growth in utilization in table 41 is based only on anticipated popula— tion growth and demographic shifts. THE GRADUATE MEDICAL EDUCATION NATIONAL ADVISORY COMMITTEE MODEL As we have seen, the BHM estimates are products of a market-oriented approach that tries to predict the future economic demand for medical services if current trends in utilization continue without major disruption. In contrast, GMENAC seeks to define physician require- ments in terms of the type and amount of care that medical professionals believe should be utilized in 1990, in light of available data and medical judgment as to the prevalence of bio- logic conditions and the ability of the medical profession to provide useful therapeutic and preventive care. The main aim of the GMENAC modeling effort is to generate estimates of physi- cians trained in particular specialties so that graduate medical education programs can be re- vamped accordingly. Table 42 summarizes the specialties and subspecialties for which esti- mates are being planned or considered by GMENAC. Of these categories 14 to 26 are ex- Ch. 3—Requirements 0 63 Figure 12.—Non-Price Related per Capita Utilization Trends, Dental Office Services, 1968-76 Non-price-related per capita utilization 1.75 p p . 1.50 1 .25 1.00 1966 1967 1968 1969 1970 1971 1972 Year 1973 1974 1975 1976 SOURCEzJWK InternationaI, Inc.. Evaluation of Project SOAH (Supply, Output, and Requirements), draft report, DHEW contract No. HRA-232—78-0140, 1979. pected to be completed (McNutt, 1979). While an aggregate estimate of physicians required in 1990 is not the principal objective of GMENAC, such a number can readily be generated simply by adding the estimates for each specialty, once those numbers are available and if all specialties are covered. As of this time, no estimates for any specialty group are available from GMENAC. The esti- mates should be released in 1980, provided GMENAC undergoes no further delays and meets its scheduled date for publication of the final report. The formal definition of “need for care” em- ployed by GMENAC is as follows: An individual is said to need medical care if a pathologic finding exists or if the individual will benefit from such care. Need for care thus refers to: 1) persons with a given morbidity for whom intervention by a physician is appropriate for di- agnosis and treatment, and 2) persons without morbidity for whom preventive services are ap- propriate. Thus, in the GMENAC model, population- based estimates of morbidity (biological need) are adjusted to determine the proportion of per- sons with a given morbidity who are in need of physician intervention. In addition, the quantity and type of preventive services appropriate for certain population subgroups are normatively estimated. Further, the model takes into account other uses such as insurance physical examina- tions and visits by the ”worried well." The result is what GMENAC terms an “adjusted-needs” model which is used in conjunction with U.S. Census population projections to estimate the need for physician care in 1990. Figure 14 illus- trates the procedure of arriving at an adjusted needs estimate for one particular type of bio- logical condition, varicose veins. 64 0 Forecasts of Physician Supply and Requirements Figure 13.—Non-Price-Related per Capita Utilization Trends, Community Pharmacy Services, 1966-76 Non-price—related per capita utilization 7.0 6.5 6.0 5.0 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 Year SOURCE: JWK International. lnc.. Evaluation of Proiect SOAR (Supply, Output, and Requirements), draft report. DHEW contract No. H RA-232-78-0140, 1979. Table 38.—Dependence of Trend Projections on Alternative Starting Dates in the Baseline Data Projected growth in non-price-related per capita utilization, 1975-90(1975 = 100) Start date Elasticity 1966 1968 1971 Physician office services ..... High 116.3 123.0 89.9 Low 123.1 127.3 95.4 Short-term hospital services . . High 123.8 131.4 123.4 Low 123.9 129.9 129.9 Dental office services ........ High NA 97.5 56.3 Low NA 123.3 106.9 Community pharmacy services High 135.6 129.4 100.1 Low 140.1 134.1 105.6 SOURCE: JWK International Incorporated, Evaluation of Proiect SOAH (Supp/y, Output, and Requlrements), draft report, DHEW contract No. HRA 232-78-0140, 1979. Ch.3—Requirements 0 65 Table 39.—Comparison of Linear Versus Logarithmic Extrapolation of Utilization Data Ft2 (percentage of variance Projected growth in non-price-related per explained by regression) capita utilization, 1975-90 (1975 = 100) Linearextrapolation Logarithmic extrapolation Linearextrapolation Logarithmic extrapolation Form of services and price elasticity Physician office High ................. 43.2% 43.5% 116.3 103.1 Low ................. - 65.7 66.2 123.1 104.4 Short-term hospital High ................. 75.7 73.5 123.8 104.4 Low ................. 90.1 88.5 123.9 104.3 Dental ollice High ................. 00.7 00.4 97.5 101.7 Low ................. 59.1 60.3 123.3 106.5 Community pharmacy High ................. 76.4 77.6 135.6 110.6 Low ................. 81.1 82.3 140.1 111.3 SOURCE: JWK International Incorporated, Evaluation of Proiect SOAFt (Supp/y, Output, and Requirements), draft report, DHEW contract No. HRA 232-78-0140, 1979. Use of such adjusted needs estimates has im- portant implications. If, for example, we com- pare estimates of physician requirements based on biological need and appropriateness of med- ical intervention with estimates based on pro- jecting current patterns of utilization of physi- cian services into the future, we can anticipate some differences in the types of physician serv- ices on which the estimates of overall physician requirements would be based. The GMENAC model, for example, implies that patients will not receive physician services merely because they want such services and can pay for them; i.e., factors that translate into ”effective eco- nomic demand" in a market-oriented model. Using the GMENAC model, a proportion of the current and future economic demand for care might be discounted, because the persons seek— ing physician services might lack sufficient biological need for services or might be seeking services inappropriate to their biological condi- tion, or might be seeking care for biological con- ditions for which no useful physician interven- tions are presently available. From what is known about current patterns of utilization, such a downward adjustment of “economic de- mand” to meet standards of true biological need and appropriate, useful physician intervention would have the greatest impact on primary care (defined as “first contact” physicians). The rea- son is that a high volume of complaints seen by primary care practitioners are nonserious, self- limiting conditions for which no effective med- ical treatment currently exists (e.g., "colds,” nonbacterial sore throats, and similar condi- tions). However, the adjusted needs approach does take into account some proportion of the demand for care that is generated by the so- called ”worried well" and persons with vague symptoms, probably psychological in origin, for which the patient seeks a physical cause and medical cure. Conversely, there are also medical conditions for which beneficial interventions are available which do not, however, generate demand. The would-be patient may be unaware of the condi- tion or of the availability of effective treatment or preventive cure, or, for whatever reason, has chosen not to seek it. On this dimension, a need- based model would tend to overestimate physi— cian requirements. The GMENAC model accordingly provides for some downward adjustment of medical need estimates to conform to patterns of future uti- lization that can realistically be anticipated, even though such adjustments imply an accept- ance that the true medical need for physician services will never be wholly met. Finally, much has been written in recent years concerning over and unnecessary utilization of medical services that is physician rather than consumer gener- ated. To the degree that over and inappropriate utilization are factors in current patterns of utilization and ongoing trends in utilization, economic demand models include such unneces- sary services in projecting physician require- ments. In contrast, an adjusted need-based mod- 66 0 Forecasts of Physician Supply and Requirements Table 40.—Allocation of Physicians by Type and Setting of Care tor the 1975 Base Year, BHM Model Medical office General Pediatric Total care care Ob-gyn care Psch. care Vision care Other care Physicians (MD) ......... 340,280 46,493 21,453 16,255 15,080 8,820 76,406 Generala ............. 116,430 36,476 9,932 2,895 1,081 —— 26,995 Pediatric ............. 21,746 568 12,061 —— —— —— 271 Obstetrics-gynecology. 21,731 2,634 ——- 12,964 79 — — —— Opthalmolgy ......... 11,129 —— —— —— —— 8,820 —— Psychiatry ........... 26,502 —— —— —— 13,837 —— —— Surgeryb ............. 76,017 3,516 —— 396 63 —— 24,003 Secondary specialistC. . 48,322 3,299 —— —— —— —— 25,137 Noncare specialist ... . 18,403 -—— —— —— —— —— —— Physicians (DO) ......... 14,532 11,072 47 47 35 24 464 Short-term hospital Long-term hospital Out pt. care Surgical care Medical care Psychiatric care Other care Physicians (MD) ......... 8,481 63,701 35,680 9,334 3,314 Generala ............. 5,660 1,351 21,292 3,103 1,476 Pediatric ............. 638 49 6,179 —— 131 Obstetrics-gynecology. —— 5,156 —— —— —— Opthalmology ........ —— 1,991 -—-— —— —— Psychiatry ........... 1,746 —— 1,392 6,231 —— Surgeryb ............. —— 45,289 —— —— —— Secondary specialistc. . 437 4,374 6,124 —— 1,707 Noncare specialistd . . . —— 5,491 693 —— —— Physicians (DO) ......... 312 254 1,210 —— 166 Other care settings Noncare settings Nursing Dental Vet. Opt. Pod. Other Pharm. Noncare home care care care care care Lab service service activities Physicians (MD) ......... 594 —— —— —— —— 2,611 4,309 —— 27,749 Generala ............. 594 —— ——-— ——- —— —— —— —— 6,115 Pediatric ............. —— —— —— —— —— —— —— —— 1,849 Obstetrics~gynecology. —— —— ——-— —— —— -—— —— —— 898 Ophthalmology ....... —— —— —— —— —— —— —— —— 318 Psychiatry ........... —— —— —— —— —— —— —— —— 3,276 Surgeryb ............. —— --— —— ——-— —— —— —— —— 2,750 Secondary specialistc. . —— —— —— —— —— -— —— ——-— 7,244 Noncare specialistd . . . -—— —— —— —— —— —— 4,309 —— 5,299 Physicians (DO) ......... —— —— —— —- —— —— —— —— 901 aIncludes general and family practice, internal medicine, and ”specialty un- specified" (presumed to be predominantly in primary care). Includes general surgery, neurological surgery, orthopedic surgery, otolaryn- gology, plastic surgery, colon and rectal surgery, thoracic surgery, urology, and anesthesiology. cIncludes allergy, cardiovascular diseases, dermatology, gastroenterology, pe- diatric allergy, pediatric cardiology, pulmonary diseases, radiology, diagnostic radiology, therapeutic radiology, neurology, physical medicine and rehabilita- tion, and “other specialties." dIncludes occupational medicine, general preventive medicine, public health, aerospace medicine, pathology, and forensic pathology. SOURCE: JWK International Incorporated, Evaluation of Project SOAH (Supp/y, Output, and Requirements), draft report, DHEW contract No. HRA 232-78-0140, 1979, eling effort such as GMENAC's tries to factor out some unnecessary services from its esti- mates. There is controversy over the definition of “unnecessary" services. The GMENAC model would presumably reflect expert opinion in re- spect to whether particular conditions require a physician visit, and whether these conditions could benefit or not from further treatment. Fig- ure 15 summarizes the component processes that are involved in translating adjusted medical need estimates into projections of physician re- quirements by specialty. Epidemiological data on the frequency of specific biological condi- tions in the population are used as the starting point. Data on conditions that are known to be treated by physicians in a given specialty or spe- cialty groups are selected based on analyses of current practice content by self-designated spe- cialists and estimates of the training content in each specialty. Ch. 3—Requirements ' 67 Table 41 .—ll|ustrative Computation of Manpower Requirements . 1. According to table 40, the Nation’s pediatricians were involved in 1975 in the following forms of health care activity, in the numbers shown: Number of pediatricians engaged in this activity (1975) ~ Medical office General care ........................................... 568 Pediatric care .......................................... 12,061 Other care ............................................ 271 Short-term hospital - Outpatient care ........................................ 638 Surgical care .......................................... 49 Medical care ........................................... 6,179 Long-term hospital Other care ............................................ 131 Other settings Noncare activities, not elsewhere specified ................ 1,849 Total ............................................... 21,746 2. Runs conducted by the Division of Manpower Analysis indicate that between 1975 and 1990, utilization of each of the foregoing forms of care will have undergone the following growth (or reduction): Projected utilization growth factor (1975-90) Medical office General care ........................................... 108.9 Pediatric care .......................................... 101.1 Other care ............................................ 111.0 Short-term hospital Outpatient care ........................................ 95.6 Surgical care .......................................... 110.9 Medical care ........................................... 103.3 Long-term hospital Other care ............................................ 110.5 Other settings Noncare activities, not elsewhere specified ................ 110.5 3. Applying these projected growth factors to the corresponding 1975 supply of pediatricians, the following table of projected 1990 manpower requirements is produced: Number of Projected 1990 pediatricians Projected manpower engaged in this utilization requirements (column activity (1975) growth, 1975-90 1 times column 2) Medical office General care ..................... 568 108.9 619 Pediatric care .................... 12,061 101.1 12,194 Other care ....................... 271 111.0 301 Short-term hospital Outpatient care .................. 638 95.6 610 Surgical care ..................... 49 110.9 54 Medical care ..................... 6,179 103.3 6,383 Long-term hospital . Other care ....................... 131 110.5 145 Other settings Noncare activities, not elsewhere specified ...................... 1,849 110.5 2,043 Total ........................ 22,349 SOURCE: JWK international Incorporated, Evaluation of Project SOAFi (Supply, Output, and Requirements), draft report, DHEW contract No. HRA 232-755-0140, 1979. 68 0 Forecasts of Physician Supply and Requirements Table 42.—Specialty Areas and Subspecialties for Which Requirements Estimates Are Being Planned or Considered by GMENAC Figure 14.—|llustrative Procedure tor Arriving at Adjusted Needs Estimates Obstetrics-gynecology Dermatology Adult medical care Family practice General internal medicine Allergy and immunology Hematology Cardiovascular disease Infectious disease Endocrinology and metabolism Nephrology Pulmonary disease Gastroenterology Rheumatology Medical oncology Pediatric medical care Family practice General pediatrics Allergy and immunology Pediatric hematology-oncology Pediatric nephrology Pediatric endocrinology Pediatric cardiology Neonatal-perinatal medicine Otolaryngology General surgery Colon and rectal surgery Orthopedic surgery Thoracic surgery Ophthalmology Urology Neurosurgery Plastic surgery Pathology Radiology Psychiatry and neurology Anesthesiology Preventive medicine Nuclear medicine Total US. population A. Persons with he or more episodes of varicose v -‘ SOURCE: Interim Report of the Graduate Medical Education National Advisory Committee to the Secretary, Department of Health, Education, and Welfare, Washington, DC: Health Resources Administration, DHEW publication No. (HRA) 79-633, p. 206. These data on current incidence and preva- lence of conditions, it is important to note, are subject to various limitations in terms of validi- ty and reliability. The following quote, taken from the workbook prepared for the general surgery advisory panel, is illustrative: SOURCE: GMENAC Workbook for General Surgery Panel, 1979. contains the most extensive coverage of condi- tions is the Health Interview Survey (HIS). However, conditions in HIS are self-reported and thus represent an individual’s knowledge and perception of a condition, and not necessari- ly an accurate measure of disease or injury. In addition, in HIS, reporting of morbidity is con— tingent upon a person's taking one or more of various actions, such as restriction of usual ac- tivities, bed disability, work loss, or seeking medical advice. A further problem exists in ob- taining reliable estimates for low-prevalence conditions (less than 1 to 2 percent of the pop— ulation). The sample of persons with rare condi- tions is usually small, and thus these estimates tend to be unreliable. Since many of the general surgery conditions occur with low frequency in the total population, they are difficult to esti- imate accurately. In general, the available mor- bidity estimates presented for the general sur- gery conditions are thought to be underestimates of actual morbidities in the US. population. Given the limitations of the national morbidity There is a general problem in the national data sets with coverage of conditions treated by gen- eral surgeons. Population based clinical exam- ination data provide the best source of data for estimates of incidence and prevalence of disease and injury. Such data for general surgery condi- tions are limited, however. The data set that data, the estimates of the proportions of persons with general surgery conditions is presented to the panel for review and revision. Accordingly, the GMENAC advisory panels of experts for each specialty area use their pro- fessional judgment to take account of possible Ch. 3—Requirements 0 69 Data sources True Figure 15.—GMENAC Model for Estimating Physician Requirements need (biologic . ,. incidence of ‘ ' HIS conditions) Somers AHA . - Westoffl VA Zelnick/NCHS, NIMH etc. Population-based surgical rates, etc. Ambulatory Institutionalized 2 Hospital utilization 3 , .21, Total Adjusted service needs over, under, or misreporting of conditions. For example, estimates of the frequency of venereal disease would be adjusted upwards, since the frequency of these diseases is known to be un- der-reported. In the next phase the advisory panels of ex- perts are asked to estimate the probable fre- quency of these same conditions in 1990, based on the data on current frequency adjusted by their judgments concerning changing disease patterns, host responses and technology, effi- cacy of preventive strategies, and any other fac- tors they might believe to have an important im- pact. The specialty panel of experts also esti- mates what proportion of episodes of a given requirements Physician manpower requirements: FTE aggregates 7 Nonphysician '> service requirements Physician manpower requirements: head counts by specialty condition should receive a physician’s care in 1990 and what proportion of these should be seen by the panel’s medical specialty (e.g., gen- eral surgery, general pediatrics, psychiatry) ver- sus some other specialty. Here the panel mem- bers again employ their own intuitive judgments concerning the frequency of self-limited condi- tions, the availability of effective therapeutic or preventive care, and any other such factors that might be expected to influence the degree of ben- efit persons with particular conditions might be expected to derive from receiving care from a physician and from a particular type of special- ist. In making these and other similar kinds of judgments, specialty panels are instructed to think in terms of the average patient. A modi- 70 O Forecasts of Physician Supply and Requirements fied Delphi process is used to achieve consensus. The final product emerging from the delibera- tions of the expert panel at this phase is a list of diseases, diagnoses, preventive activities, opera— tions, and counseling requirements expressed in terms of population rates and disease or diag- nostic categories. GMENAC staff then apply these estimates of medical need to census projections of the size, age, and sex distribution of the U.S. population in 1990. The GMENAC model apparently does not consider future changes in income distribu- tion and the impact these changes might be ex- pected to have on population health needs. Ad- justments for the unusual needs of some groups of people as well as those previously excluded from the health care system are introduced at this phase of the model. In the next phase, the panels of experts deter- mine norms of care for each disease or diagnos- tic category. Here again, the panels will have available to them data on actual utilization rates from a variety of sources such as HMOs and the National Ambulatory Medical Care Survey's published research studies. Each panel may rec- ommend increases or decreases in the prevailing rates of utilization, based on its perceptions of what constitutes good medical care and what technology is likely to be available in the future. The norms of care may be expressed as visits per episode of illness or annual encounter rate per chronic condition or some other unit of service. During this phase of the study the panels also consider which conditions can be treated in the office, which require hospitalization, and what is the appropriate length of hospital stay. Again, each panel examines existing data on uti- lization (e.g., Hospital Discharge Survey) and adjusts it up or down for its estimates of appro- priate care for the average case. Finally, the panels estimate the proportion of inpatients and office visits that can be delegated to physicians’ assistants or nurse practitioners. Although ac- tual figures are not available, GMENAC staff report that the panels have been willing to dele- gate significant amounts to paraprofessionals compared to what is currently delegated. GMENAC staff predict that the specialty panels will recommend increased task delegation and will specify where increased task delegation to paraprofessionals is most appropriate. Perhaps the single most problematic aspect of the GMENAC modeling effort occurs in the next phase. This is the reconciliation of conflicting estimates by the various specialties as to what proportion of a given disease or diagnostic cate- gory “belongs” to each specialty. The extent of the problem is likely to be mitigated somewhat because each specialty panel contains a few rep- resentatives from other specialties. In particu- lar, generalists (general practitioners, family practitioners, and general pediatricians) are rep- resented on the specialty panels (surgery, der- matology, etc.) and vice versa. This is impor- tant because it is essential that a specialty's estimate of the conditions it should handle match those of the generalists who make the re- ferrals. At this point, it is difficult to determine how many problems there will be in mediating disputes between specialties. An indication of the complexity of the task facing these panels is that only 14 to 26 of the 37 specialties listed in table 42 are expected to be completed by the end of 1980. Among the difficult questions that must be mediated during this phase is the issue of how much primary care should be provided by sec— ondary and tertiary care specialists. The issue is a knotty one that cannot be easily settled. One reason is that wider geographic distribution of subspecialists, outside major cities, virtually re- quires a part—time practice of the subspecialty for some percentage of these physicians, be- cause the conditions treated by subspecialists are comparatively rare. The final task in the modeling process is to translate estimates of the volume of physician services into FTE physicians and then into ac- tual head counts. FTEs are arrived at by divid- ing service estimates allotted to each specialty by the expected productivity for each physician in that specialty. Productivity may be expressed in terms of encounters, operations, or some other unit, depending on which is most ap- propriate. As in the BHM model, it is generally assumed that physician productivity will be the same in > C11.3—Requirements 0 71 1990 as it is now, although this depends on each specialty panel. GMENAC staff review the available data on the typical physician practice profile and arrive at estimates of productivity for the various kinds of services the specialty provides. Table 43 displays the average practice profile of general surgeons and the preliminary productivity estimates to be used to calculate FTE general surgeons. It should be noted that, in addition to the estimate of 43 office visits per week, alternative numbers of 77.2, 58, and 51 were also cited from the data. Finally, some esti- mates are also made of the productivity en- hancement for physicians of employing physi— cians’ assistants and nurse practitioners and the productivity gains from organizational arrange- ments such as group practice and the interaction between these two factors. Overall, the productivity estimates used in the GMENAC model are somewhat problematic for two reasons: 1) the data sources on which productivity estimates are based often exhibit considerable disagreement (in part because the definitions of service units vary; for example, some measures of time allotted to surgical operations may count operating room time only, while other measures may include all the care associated with a procedure including pre- operative and postoperative office and hospital visits) and 2) little information is available about trends in productivity over time, particu- larly by specialty. Table 43.—The Average Practice Profile of General Surgeons Hours Hours worked Average number of weeks worked per year . 47.0 Average number of hours worked per week. 52.0 Time allocation within week Hours in hospital ....................... 31.3 Hours in operating room .............. 11.5 Hours in inpatient visits ............... 19.8 Hours in office ......................... 13.4 Other professional time ................. 7.3 Total professional time ............... 52.0 Weekly productivity Office visits per week ................... 43 Inpatient visits per week ................ 45 Operations per week .................... 3.4 Operations per week (CRV units) .......... 34.4 CRV SOURCE: GMENAC General Surgery Workbook. 60—618 0 - 80 - z The final step in the GMENAC model is the conversion of FTE physicians into actual head counts. In essence, this involves making some additional allocation to cover those physicians who do not practice full time but are instead in— volved in full- or part-time research, teaching, or administration or have taken some time out from practice for continuing education or other activities. As a complement to the modeling effort, GMENAC commissioned a study of selected ele— ments of consumer dissatisfaction with health care. The study, based on a scientifically de- signed opinion survey, was carried out by re- searchers at the Center for Health Administra- tion Studies of the University of Chicago (USDHEW, 1979c). It is not known how GMENAC plans to incorporate the report's findings into its final estimates, if indeed it plans to do so at all. The report does, however, con- tain some interesting findings with potential im- plications for manpower policy. Broadly speaking, the report suggests that there is a tradeoff relationship between physi- cian productivity and consumer satisfaction, and that a decrease in current productivity lev- els might result in greater patient satisfaction. The assumption here is that if physicians saw fewer patients, they would be able to spend more time with patients. Presumably, with more time, they would be better able to express concern, courtesy, and consideration; improve- ments in the quality of the doctor/ patient rela- tionship that the data indicate some patients be- lieve are needed. . Tables 44 and 45 and figures 16 and 17 sum- marize the findings of the consumer study in respect to levels of consumer satisfaction/ dissatisfaction with various aspects of care. Note that the single major source of consumer dissatisfaction, high out-of—pocket costs (figure 16), is not particularly amenable to solution via manpower policy. Table 45 indicates that ethnic minorities are more dissatisfied with all aspects of care than the majority white population. Finally, the study found that consumer per- ceptions of the availability of care correlated highly with actual data on physician availabili- 72 0 Forecasts of Physician Supply and Requirements ty. In other words, people were more likely to report that their area lacked sufficient doctors in areas where there actually were lower physi- cian—to-population ratios. Consumer percep- tions of physician shortages were particularly sensitive in the case of medical specialists. Table 44.—Proportion of Persons Whose Experience With Physician Visits Is Beyond the Critical Threshold Number of patients beyond critical threshold (in millions) Percent of persons beyond Critical threshold critical threshold Aspect ofvisit Travel time ...................... 30 minutes or more 11% 15 Appointment time ................ Over2 weeks 10 14 Waiting time ..................... 30 minutes ormore 27 37 Time with doctor ................. Less than 10 minutes 28 39 Information from doctor ........... A little or nothing 27 37 Out-of-pocket costs ............... $10 or more 38 52 SOURCE: The Consumer Viewpoint: “What is Health Care and Whatdo We Want?" & A Response, prepared for the Graduate Medical Education National Advisory Com- mittee, Washington, 0.0: Health Resources Administration, DHEW publication No. (HRA) 79-632, p. 13. Table 45.—Percent of Ethnic Groups Dissatisfied With Aspects of the Medical Care System Aspects of the medical care system Office waiting time Out-of—pocket cost to seethe doctor ofthe medicalvisit Humaneness and quality ofthe visit Waiting time for Ethnic group an appointment Majority white .............. 15% 27% 36% 10% Urban black ................ 27 38 43 15 Rural southern black ......... 26 39 45 12 Spanish heritage, Southwest. . 33 32 39 16 SOURCE: The Consumer Viewpoint' ”What is Health Care and What do We Want?” & A Response, prepared for the Graduate Medical Education National Advisory Com- mittee, Washington, DC: Health Resources Administration, DHEW publication No. (HRA) 79-632, p. 19, COMPARISON OF THE BHM AND GMENAC MODELS The two major modeling efforts currently un- derway, those of BHM and GMENAC, exempli- fy two quite different philosophical approaches to the task of estimating future physician re- quirements. The GMENAC approach is to esti- mate how many physicians would be needed to provide appropriate services to meet the pop- ulation’s medical need for care. Estimates of medical need and appropriate services are in turn based on a combination of the frequency of particular illnesses and conditions and of med- ical judgment as to which of these conditions can benefit from medical services, by amount and type. In contrast the BHM approach is an economic modeling effort that treats medical care as it would any other market commodity. The aim is to predict the economic demand for medical services and the number of physicians that would provide those services based on cur- rently observable patterns and trends in medical care consumption. Second, the GMENAC approach is a deliber- ate goal-setting effort based on the Committee's estimates of the numbers and types of physi- cians' services that should be provided or at least be available to the American citizen in 1990. Thus, the estimates GMENAC arrives at will be ”target numbers,” goals GMENAC will be recommending that governmental and pri- vate sector activities be directed toward realiz- mg. In contrast, the BHM projects current behav— ior trends into the future. It is not a goal- oriented modeling effort, and its estimates are therefore not intended to be target numbers for Ch.3—Requirements 0 73 Figure 16.—Consumer Satisfaction With Physician Services Among those who saw a physician in a 12-month period, percentage dissatisfied/satisfied with various aspects of the visit. Out-ot-pocket Dissatisfied 37% cost of care ~ 9 Time waiting to see the doctor Information given by doctor 18% about what was wrong ‘ ‘ Time between calling for and receiving appointment Time spent with the doctor Amount of concern doctor seemed to have Quality of care patient felt was provided Cost of traveling to the doctor’s office Time to travel to the 12% doctor‘s office Courtesy, consideration shown by receptionist Courtesy, consideration shown by doctor Courtesy, consideration shown by nurses The overall visit to the doctor Satisfied 63% To determine whether people are satisfied or dissatisfied with their medical care, the research group asked questions of 82% all persons who had visited a doctor during the past year. The questions probed 34% impressions of specific aspects of the visit. “Responses were very 84% skewed toward the positive end of the scale,” the group reports. For the total 87% population, cost is the greatest cause of 87% dissatisfaction. Overall, the chart reveals the US. population is generally 87% satisfied. 72 °/o 88 °/o 91 % 92 °/u 93 % 88 0/0 NOTE: Because of the large sample size, the “true" percentages of satisfied and dissatisfied consumers in the population are unlikely to vary by more than 1 percent. SOURCE: The Consumer Viewpoint: “What is Health Care and What do We Want?” & A Response, prepared for the Graduate Medical Education National Advisory Com- mittee, Washington, 0.0.: Health Resources Administration, DHEW publication No. (HFlA) 79-632. policymaking purposes. The BHM modeling ef- fort is probably best understood as an ongoing process of monitoring factors and trends that are or presently seem most likely to affect the future economic demand for medical services. Thus, BHM physician estimates represent the numbers of physicians that would satisfy speci- fied levels of future economic demand for physi- cian services. To say that the BHM modeling effort is not goal-directed is to say that the model itself does not recommend or even assume that it is desir- able or should be a policy goal to satisfy any particular level of economic demand for physi- cian services in 1990. The model simply gener— ates an estimate of how many physicians it would likely take to provide the medical serv- ices that are likely to be utilized by the Amer- ican population in 1990, if particular conditions and trends existing now are assumed to continue on into the future. The assumption that conditions and trends that characterize the present and recent past will continue on into the future is probably the single 74 0 Forecasts of Physician Supply and Requirements Figure 17.—Consumer Satisfaction With Physician Services, by Nature of the Experience Among those who saw a physician in a 12-month period, percentage dissatisfied with various aspects of the visit classified by the nature of the experience. Aspect of the visit " 4% Time to travel to the Less than 15 minutes“ doctor‘s office 15 to 30 minutes 30 minutes to 1 hour More than 1 hour Percentage dissatisfied While people generally express satisfaction with most aspects of their medical experiences, there comes a time for most when they cross a threshold of tolerance. At that point, satisfaction turns to dissatisfaction. Each variable—travel time, cost, amount of time spent Time between calling for an appointment and the appointment Up to 2 days 3 days to 2 weeks More than 2 weeks with the doctor, appointment waiting time—has its special threshold. This chart matches a range of experiences with a set of variables and shows levels of dissatisfaction. Overall, it captures some sense of the dynamics of the doctor Time waiting to see the doctor Up to 30 minutes 30 minutes to 1 hour More than 1 hour patient relationship. 85 % NOTE: Because of the small number of people sampled in these categories, the true percents may likely vary by as much as 10 percent for this subgroup in the popula- ' tion. Other figures in the table are unlikely to vary more than 3 points. SOURCE: The Consumer Viewpoint: “What is Health Care and What do We Want?" 8. A Response, prepared for the Graduate Medical Education National Advisory Com- mittee, Washington, 0.6.: Health Resources Administration, DHEW publication No. (HRA) 79-632. most important assumption made by the BHM model. This assumption might be characterized as a sort of ”law of societal inertia” which states that the future is going to look alot like the pres— ent and that any major differences between the present and the future are going to be the out- come of changes already underway that are ob- servable as ongoing trends. The major problem is that unforeseen events, developments, inter- ventions, decisions, etc., are quite common and are therefore quite likely to cause the future to deviate from both present conditions and from the outcome of currently ongoing trends. Taken together, the uncertainty factors that affect modeling efforts make it advisable to View the results generated as ”benchmark” esti- mates rather than hard predictions. Another way of viewing these estimates is to think of them as ”if-then" numbers, as in ”if Americans continued to utilize medical services in the cur- rent per capita amounts, how many physicians would be in demand in 1990?" It would also be accurate to characterize the BHM projections as providing a baseline or yardstick against which the comparative size and impact of particular sorts of changes—especially deliberate policy interventions—can be measured. The BHM model also includes separate “con- tingency" estimates. These are intended to gauge the probable impact on future demand for physician services of major changes that appear either likely or quite possible. Current contin- gency modeling efforts focus mainly on predict- ing the impact of alternative NHI plans. This points out that the political process and its pol- icy outcomes are among the major uncertainty factors affecting predictive modeling. The fact that the accuracy of current predictive modeling efforts is highly dependent on the unknowable outcomes of political decisions yet to be made Ch.3—Requirements 0 75 simply underscores the point made earlier that these estimates should be viewed as ”bench- mark" or baseline estimates. As a goal-directed modeling effort, GMENAC's most important core assumptions are that reasonable estimates of appropriate utilization of medical services and the numbers and types of physicians needed to provide those services can be derived from a combination of empirical data and professional judgment con- cerning ”medical need.” The standard of med- ical need being applied to the determination of appropriate utilization of services requires a somewhat stronger presumption of linkage be- tween medical service and improvements in health outcome than a standard based on a vol- ume of services provided, not because they are expected to produce beneficial effects in most in- stances, but because in some instances they might yield improvements and in the rest of cases are believed to do no harm. The standard being used would tend to be more conservative in estimating the medical need for such margin- ally beneficial services, especially where the medical problem or illness is a nonserious con- dition. An illustration of how this theoretical dif- ference translates into practice can be seen in the deliberations of GMENAC’s dermatology panel on the treatment of acne. In its first round, the panel’s initial estimate of medical need assumed that every case of acne should be seen by a der- matologist, and that a typical case would re— quire six visits annually (which assumes medica- tion and the need to monitor its effects). In subsequent rounds of discussion, however, the panel determined that this number was exces- sive and revised their initial estimate downward to reflect a more conservative definition of need. In lowering their original estimate of need for dermatologists to treat acne, the panel took into account such factors as the nonserious, self- limiting character of a high proportion of acne cases, the fact that medical treatment is most likely to produce improvement in severe cases and, finally, the fact that nonserious cases of acne can be treated as safely and efficaciously by appropriately trained generalists as by spe— cialists in dermatology. As a goal-directed model more concerned with defining what should be rather than fore- casting what is likely to be, the GMENAC effort need not be as concerned with the problems posed by uncertainty factors as the BHM trend projection and contingency models must be. It is also relevant that the goals being formulated are for the comparative near—term future. Thus, GMENAC's definitions of medical need are based on the assumptions: 1) that Americans will continue to have the same medical prob- lems at the same demographic (e.g., age, sex, social class) rates in 1990 as they do now, and 2) that there will be no major breakthroughs in medical knowledge or technology that will seri- ously alter current medical practice. While these assumptions are probably reasonable for a peri- od covering little over a decade, they might well become more doubtful if the time span were ex- panded. More problematic for a near-term, goal-di— rected model than the uncertainties of the future are the reality constraints of the past and pres- ent. What we mean by this is that societies are not like marching bands: policymakers cannot blow the whistle and expect society (or one of its major subunits such as the health system) to ex- ecute a 90-degree turn in formation. Yet in a sense that is what efforts to establish and achieve collective goals frequently assume can be accomplished. Clearly, the more greatly a goal-directed estimate differs from anticipated supply and the shorter the time period avail- able, the more we must implicitly assume that 90—degree societal pivots are possible; at least if the goal is taken seriously as one we ought to try to achieve. A much more serious reality constraint is that there may be insufficient ”play” or “slack” in the system to permit actual attainment of a physi- cian “requirements” estimate that deviates dras- tically from the currently projected 1990 supply. At issue is what policy researchers term the rela- tive ”malleability" of key variables. The possi- bility of attaining a goal within a given period of time is dependent on the malleability of supply factors. Supply factors, however, are not highly malleable. The reason for this is, that, as 1990 is only 10 years away and physician training re- 76 ° Forecasts of Physician Supply and Requirements quires a long leadtime, most of the 1990 supply is already locked in. And even though 40 per- cent of the physicians in practice in 1990 will have completed training since 1979 (Jacoby, 1980), major changes in graduate medical edu- cation cannot be expected to take place and have a significant impact on the specialty distri- bution of the 1990 physician supply. Perhaps fu- ture goal—oriented modeling efforts should pay explicit attention to the relative “malleability” of key variables. In so doing, they might pro- vide alternative estimates, signaling, on the one hand, goals that are capable of attainment with— in the allotted time span and, on the other hand, goals that are considered desirable but would re- quire a longer time frame and are thus best con- sidered as signaling the appropriate direction for deliberate change but not taken as immediate targets. So far we have discussed only one set of reali- ty constraints (i.e., limitations on the mallea- bility of the health manpower supply) that im- pinges on the feasibility of a goal-oriented model of physician requirements. There are other factors as well. One is the difference be- tween the types of conditions that ought to be seen by particular types of physicians and actual patterns of physician use. For example, a non- trivial portion of the current caseload of gener- alist physicians is composed of nonserious, self- limiting conditions that medical treatment can do little to cure or ameliorate (e.g., “colds”). These cases would tend to be discounted based strictly on medical need. There is little reason to expect however, that, in reality, patients would rapidly be reoriented to stop bringing such com- plaints to physicians or that physicians would refuse to see patients with such complaints. Thus, GMENAC’s normative model accounts for some need to provide services to the ”wor- ried well,” and the BHM estimates include pres- ent use of medical services by the “worried well” in its projections. The need to pay attention to both goals and reality leads naturally to a consideration of the complementarity of goal—driven and trend pro- jection modeling. Because each of these two ma— jor modeling efforts is oriented toward different purposes and focuses on rather different vari- ables, they are, in truth, more complementary than competing. As such, each model's results can aid our interpretation of the other's. The GMENAC model focuses on translating a normative definition of medical need into ap- propriate rates of utilization of medical services, while the BHM model looks on medical care as a “consumer good" and treats empirical trends in utilization of medical services as a proxy for economic demand. If the BHM demand esti- mates should prove significantly greater than the GMENAC estimates, this would suggest that there are powerful factors at work that are pushing utilization of medical services beyond the level medically necessary and appropriate for ”good" care. This would then raise the policy question of what percentage—if any—of the projected future economic demand for medi— cal services over and above the professional judgment-based estimates of medical need should be considered legitimate. Conversely, if the BHM demand estimates should prove sig- nificantly less than the GMENAC estimates, this would suggest that there remain and will remain in the near future significant barriers to obtain— ing medically necessary care. Finally, if the BHM and GMENAC demand estimates prove to be in rough parity, this would suggest that the economic demand for services is more or less in line with professional estimates of the medical need for physician services. Obviously, since the GMENAC model has yet to generate any numbers, it is impossible to say at the present time which of these three alternatives will prove to be the case. We can say, however, that the most likely occurrence would appear to be rough parity or a BHM de- mand estimate that is significantly greater than the GMENAC aggregate estimate. The major reason for anticipating that the BHM estimate will most likely prove greater than or at least equal to the GMENAC estimate is that one of the major variables in the BHM model is a pro- jected trend toward rising per capita utilization rates for medical services, independent of demo- graphic changes and projected changes in price. In contrast, the GMENAC model assumes no major changes in medical need apart from changes in medical need induced by demograph- Ch.3—Requirements 0 77 ic shifts (e.g., an aging population), between now and 1990; hence no medical rationale for steadily rising per capita utilization of services. Thus, in order for the GMENAC estimate to logically come out larger than the BHM esti- mate, one would need to assume that there is currently such a large unmet medical need for services, that, despite the trend of rising per capita utilization rates, assumed in the BHM model, considerable unmet medical need will re- main in 1990. What is a reasonable estimate of requirements from the BHM economic model which might ap- proximate aggregate adjusted need from the GMENAC modeling effort? Recall that BHM now projects demand at approximately 600,000 physicians in 1990, or what the supply will be. We saw (table 37) that this represented an in- crease of 217,841 over the 1975 figure of 378,376; 35,960 was due to an increasing and changing population, and 181,881 due to pro- jected increases in per capita utilization trends. That is, without increasing per capita utiliza- tion, demand in 1990 would be roughly 415,000. We also saw (figures 10 through 13) that the large increase attributed to rising per capita utilization would nearly disappear if pre—1970’s data were deleted from the trend base. But we would not want to discount this increase entire- ly for several reasons: 1) the possibility of NH1, 2) possible decreases in the average physician's PRODUCTIVITY Both BHM's and GMENAC'S modeling ef- forts emphasize the amount of medical services that either will (based on predictions of trends) or should (based on normative determinations of medical need) be used in the future. However, estimates of the number of physicians required are derived by dividing projected use by physi— cian productivity. With the exception of task delegation to physicians' assistants and nurse practitioners, which would enhance physician productivity and thereby reduce aggregate phy- sician requirements, neither modeling effort ex- plores possible changes in productivity and their effects on requirements estimates. Rather, both workweek, and 3) increasing the time physi- cians spend with patients. The possible effects of an NHI program have previously been summarized. Physicians cur- rently average longer workweeks than most of the rest of the working population. Bringing the physician workweek more into line with the present patterns of work productivity of the labor force in general would lower productivity. Alternatively, the cushion of excess physicians might enable physicians to see fewer patients and spend more time with each one. According to the National Center for Health Statistics, about half of all office visits to physicians in both 1973 and 1977 lasted 10 minutes or less. With smaller patient loads, doctors might be able to use the additional time to provide pa- tients with more information, education, and counseling. For these reasons, it is difficult to estimate physician requirements. If one takes projected population changes alone, requirements in this model would be for 415,000 physicians. Some contingency is necessary to account for such fac- tors as NHI, decreased working hours for physi- cians, and more time spent with patients per visit. How much of a contingency is necessary is a matter of judgment, and the reader can come to his or her own conclusion on what it should be over and above the increase in requirements due to population growth. models basically assume that physician produc- tivity will remain constant through 1990. The BHM model does this by assuming that the ratio of practicing physicians to total output of physi- cian services will be the same in 1990 as it was in 1975. In the case of GMENAC, the main effort has been toward choosing the most reliable and accurate measures of current productivity as re- flected in various empirical studies. Its modeling effort makes explicit assumptions about the average workweek, patient visit rates, etc., by each medical specialty. But physician requirements estimates are highly sensitive to changes in productivity 78 ' Forecasts of Physician Supply and Requirements (Reinhardt, 1975). As an illustration of this sen- sitivity, if we were to postulate that the appro- priate ratio of physicians-to-population in 1970 was 185 per 100,000 (the actual ratio in the best— endowed areas) and percent growth in produc- tivity kept pace with percent growth in per capi- ta use of physician services, then the physician- to—population ratio would remain constant in- definitely. However, as Reinhardt points out: If in 1970 a set of policies could have been im- plemented such that average annual growth in physician productivity during the following two decades were one percentage point higher than the annual growth in the per capita utilization of physician services, then the required ratio at the end of the forecast horizon would have been only 151 physicians per 100,000. Relative to a forecast based on maintenance of the base-year ratio of 185 per 100,000 and for a population of roughly 250 million in 1990, this turn of events would have led to a reduction of about 85,000 in the number of MDs that would otherwise have been “required." The corresponding number for 1980, based on a projected population of 225 million, is 40,500. These figures must surely strike one as significant, especially if held up against the annual number of medical graduates (between 15,000 and 16,000) likely to be pro- duced during the next several decades (Rein- hardt, 1975). Reinhardt's analysis was primarily concerned with the impact of substantial gains in produc- tivity that might occur as a result of organiza- tional changes in physicians’ practices and of task delegation to nurse practitioners and physi- cians' assistants. His research also provides some data suggestive of a possible relationship between growth in physician supply and de- creases in physician productivity. Table 46 (re- produced from Reinhardt, 1975) provides some data on relationships between physician supply, physician productivity, and financial factors such as average visit fee and average annual physician income. If physician supply increases (item #1) but demand for services (per capita use) remains constant (item #9), then produc- tivity (measured in patient visits per MD) will drop (item #3). The data suggest that physicians then may charge more per visit (item #7), though not necessarily enough for physicians' incomes to reach the same level as in areas with fewer physicians and/or greater demand for services (item #8). A different approach to the question of physi- cian productivity and its relationship to require— ments estimates is to examine trends in physi- cian productivity. According to Medical Eco— nomics magazine’s Continuing Survey, compar- ing workweeks in 1965 versus 1976, office—based physicians spent 2 hours less with office pa- tients, 2 hours less on housecalls, and 1 hour less on hospital rounds and consultations in 1976. Median time spent on all professional activities (including activities other than patient care) in a typical workweek fell from 64 hours in 1965 to 60 hours in 1976 (Owens, 1977). More recently, Medical Economics reported that the number of office visits has continued to decline. Over the period 1974-78, office-based physicians were seeing 8 fewer patients per week in 1978 as compared to 1974, a median weekly number of 126 rather than 134 (Owens, 1979). Generally, two hypotheses are given to ex- plain recent decreases in productivity and to predict further decreases. One hypothesis is that physicians, as most other Americans, would prefer to work less and enjoy more leisure time. The other hypothesis is if growth in physician supply outpaces growth in the demand for phy- sician services, then declines in physician pro- ductivity may occur as a means of bringing sup- ply and demand into balance. In the latest survey (Owens, 1979) 57 percent of the office-based physicians stated that they believed they were practicing at full capacity. Of this 57 percent, a minority (amounting to 12 percent of the entire sample of physicians sur— veyed) said they would prefer not to practice at peak productivity. Of those surveyed 43 percent said that they were not practicing at peak pro- ductivity. Of these, a majority said they did not want to practice at full capacity. Of the entire sample 18 percent, however, stated that they were not practicing at full capacity but would prefer to do so. Among the specialties, 31 per- cent of urologists, 24 percent of general sur— geons, and 24 percent of otolaryngologists said that they were not practicing at peak productivi- ty and would prefer to do so. ChJiRequirements 0 79 Table 46.—Regiona| Differences in Certain Health-Care Statistics, United States, 1969-70 Census divisions East-North East-South item No. Year New England Central Central 1. Number of active MDs involved in patient care as their primary ac- tivity, per 100,000 population ................................ 1970 161 115 95 (1.00)8 (0.71) (0.59) 2. Average annual number of hours worked per MD a) Total practice hours ...................................... 1969 2,504 2,495 2,568 b) Hours of direct patient care ............................... 1969 2,128 2,151 2,303 3. Average annual number of patient visits per MD a) Total patient visits ....................................... 1969 4,808 6,611 8,408 (1.00)8 (1.38) (1.75) b) Office visits only ........................................ 1969 3,384 4,799 6,052 (1.00)8 (1.42) (1.79) 4. Total visits per hour a) Total visits per practice hour .............................. 1969 1.92 2.65 3.27 b) Total visits per hour of patient care ......................... 1969 2.25 3.07 3.65 5. Average number of auxiliary personnel employed per physician . .. 1967 1.3 1.8 2.1 6. Percentage of physicians in group practice .................... 1969 9.3 % 17.4 % 19.4 % 7. Average fee fora routine followup office visit: a) General practice ......................................... 1969 $6.79 $6.29 $5.21 b) Internal medicine ........................................ 9.40 8.05 7.20 c) Pediatrics .............................................. 7.53 6.94 5.40 d) General surgery ......................................... 9.76 7.76 6.85 e) Obstetrics/gynecology ................................... 9.77 9.32 7.60 8. Average net income (all specialties) ........................... 1970 $38,019.00 $47,000.00 $41,963.00 9. Reported no. of physician-patient visits per capita: a) Based on survey of physicians 1969 —total patient visits ..................................... 7.7 7.6 8.0 —office visits only ....................................... 5.4 5.5 5.8 b) Based on household surveys .............................. 1970 4.4 4.0 4.1 10. Infant mortality rate: a) White .................................................. 1968 19.2 19.4 20.9 b) Nonwhite .............................................. 31.8 35.4 40.5 11. Personal per capita income ................................. 1970 $4,469.00 $4,306.00 $3,146.00 aFigures in parentheses are indexes with New England set at 1.0. SOURCE: U. Reinhardt, Physician Productivity and the Demand for Health Manpower, An Economic Analysis, Cambridge, Mass: Ballinger, 1975 (table 26, revised). According to Medical Economics, the finding that almost one-fifth of the physicians surveyed felt that they were not practicing at full capacity but would prefer to do so “suggests that maldis- tribution of medical manpower plus the grow— ing number of new doctors—nearly 35,000 have joined the ranks of office-based MDs over the past 5 years—may already have left physicians short of patients in some areas” (Owens, 1979). In sum, the available evidence suggests that both physician preferences and the increasing number of physicians are contributing to declin- ing productivity. Yet, some physicians still feel overworked, which suggests that maldistribu- tion remains. Decreased physician productivity, it is impor- tant to note, is not necessarily undesirable. If a physician is practicing in an underserved area, then high productivity is likely to reflect over- work. Under these conditions of chronic over- work, decreased productivity would probably represent increased quality. As the physician supply increases and unmet demand slackens, then decreases in productivity would, at some point, begin to represent not better quality care, but inefficiency. Table 47 summarizes one effort to quantify this relationship for primary care physicians’ services (Walker and Armondino, 1977). Additional research that would increase our understanding of this relationship would be important because of the cost implications. The possibility of further decreases in physi- cian productivity has important, though largely unexplored, implications for the problem of the 80 0 Forecasts of Physician Supply and Requirements Table 47.—Shortage, Adequate, and Surplus Levels of Primary Carea Physicians Population potentially served Average primary Office visits in 5.5 Average annual per full-time care visits Workload patient contact Visits per 240-day primary care visits equivalent Criteria per hour evaluation hours per day year(supp|y) per person (need) physician designation 6 Too high 33 2.5 3,200 Shortage 4 Ideal 22 2.5 2,100 Adequate 2 Too low 11 2.5 1,100 Surplus aGeneral and family practice, internal medicine, and pediatrics. SOURCE: J. E. C. Walker and N. L. Armondino. The Primary Care Physician: Issues In Distribution. Lawrence, Kan.: Connecticut Health Services Research Series, No. 7, 1977. locational maldistribution of physicians. Argu- ments about the likely effects on physician shortage areas of increasing the aggregate sup- ply of physicians have tended to focus on two alternative hypotheses. One hypothesis is that, ultimately, the law of supply and demand will force physicians to move into what are current- ly shortage areas as long as growth in supply outpaces growth in demand for services in areas that already have high physician-to-population ratios. The alternative hypothesis is that physi- cians have the capability to generate demand for their services, and, if exercised to any significant degree, this capability would decrease the pres- sure on physicians to move out into less attrac- LOCATIONAL REQUIREMENTS The application of both the BHM and GMENAC models has relevance primarily at the national level. Shortages will always remain in specific service areas no matter how “correct” the balance between national supply and re- quirements are and even if supply exceeded re- quirements substantially. Yet locational esti- mates must be made at the national level: 1) to plan for the National Health Service Corps (NHSC) to meet some part of this requirement and 2) to provide guidelines and eligibility cri- teria for Health Manpower Shortage Area (HMSA) designations. Consequently, estimates of the requirements for physicians are used to determine need and serve as the starting point for shortage area designations, augmented by other criteria that represent barriers between the physician and the population he/ she is expected to serve. tive areas as aggregate physician supply in- creased. Whatever the case, enormous increases in the aggregate physician supply cannot be assumed to guarantee that an eventual solution to the problem of locational shortages will “naturally” occur. An attractive metropolitan area where the physician-to—population ratio is high enough to satisfy the highest reasonable levels of medical need or consumer demand for care can nevertheless continue to absorb many additional physicians if individual productivity decreases. Otherwise stated, it is quite likely that we could have a large ”oversupply” of physicians in the aggregate in future years and still have shortages in particular locations. We have seen that future supply for loca- tional distribution is estimated in similar fashion as national supply. Subtraction of the estimated supply from estimated requirements equals total unmet need. Need in 1990 has been estimated at 16,400 pri- mary care physicians and psychiatrists (USDHEW, 1979d). Primary care is defined as non-Federal MDS and DOS providing direct pa- tient care who practice principally in general or family practice, general internal medicine, gen- eral pediatrics, and obstetrics-gynecology. At current budget levels, NHSC scholarship recipi- ents now in the pipeline will result in 3,950 NHSC physicians in the field by 1990. Through NHSC scholarships and combining 1,150 physi- cians expected to volunteer with 900 midlevel practitioners (assumed to each equal 0.5 physi- Ch. l—Summary 0 81 cians), 34 percent of need will be met. Assuming a 10-percent conversion rate from NHSC to pri- vate practice in underserved areas, 1,000 physi- cians, or 6 percent of need, will be met. Finally, assuming current levels of 2,000 physicians in federally funded health centers, another 12 per- cent of need will be met in 1990. Together, these sources are expected to provide 52 percent of the projected need of 16,400 in 1990. These projec— tions are summarized in table 48. “Need" as defined for purposes of projecting future HMSAs should not be confused with the need for physicians based on estimates of a given population’s economic demand or medical need for services, as described in the analysis of the BHM and GMENAC modeling efforts. In the case of shortage area projections, two physi- cian-to-population ratios are used as criteria to determine the level of need for primary care physicians in an area: ' Designation ratio—The actual minimum ratio of active, non-Federal, patient care physicians engaged in primary care to the civilian population of an area below which an area is considered to have a shortage of health manpower sufficient to justify its being counted as a shortage area in the model. 0 Staffing ratio—The theoretical maximum ratio of active non-Federal, patient care physicians engaged in primary care to the civilian population of an area used as a standard above which an area is considered to have adequate health manpower so that additional Federal intervention with NHSC staffing is no longer necessary (USDHEW, 1978a). ”Need" is the number of physicians required to reach the staffing ratio for all designated areas. The designation ratio is based on equity considerations and reflects that quarter of the United States having the least number of pri- mary care physicians. It has been set at 1:3,500. The staffing ratio establishes a limitation upon the extent of Federal involvement and specifies the relationship between the service demands of the population and the primary care physicians available to provide those services. It has been set at 1:2,000. The designation ratio of 1:3,500 means that areas with smaller ratios would not be included, including areas with ratios between 1:3,500 and 1:2,000. However, because criteria for making HMSA designations were expanded under the 1976 Act to include, in addition to manpower-to-popula— tion ratios, other indicators of need such as in- fant mortality rates, access to health services, health status, income level, and the number of foreign medical graduates practicing in the area, the method for projecting future shortage areas and their physician needs has been adjusted in the following way: Comparison of projected areas with actually designated areas showed that the projection model missed part-county areas designated upon factors other than strict physician to population ratios. The physician to population ratios, strict- ly determined, fell within a range from 1:2,000 to 1:3,500. Therefore, the unmet need for coun- ties with ratios between 1:2,000 to 1:3,500 is used as a proxy for part-county rural areas. (USDHEW, 1978a). Table 48.—Need for Primary Care3 Physicians and Psychiatrists in 1990 Current NHSC Total scholarship Conversion from NHSC to private practice in Current level of physicians in federally Unmet need recipients Volunteers underserved areas funded centers need 16,400 3,950 1,600b 1,000 2,000 7,850 100% <———-———-—-—~_ 8,550 (52%) = 48% 3General and family practice, general internal medicine, general pediatrics, and obstetrics-gynecology. bIncludes 900 midlevel practitioners, each equal to 0.5 physician. SOURCE: “Outyear Size of the National Health Service Corps (NHSC)—DECISION MEMORANDUM," from the Assistant Sec- retary for Health and the Acting Assistant Secretary for Planning and Evaluation to the Secretary for Planning and Evaluation to the Secretary, Washington, DC, Spring 1979. 82 ' Forecasts of Physician Supply and Requirements The result is that, taking the year 1980, unmet need from reducing the ratio from 1:3,500 (or more) to 1:2,000 would be 5,659 primary care physicians, with an additional 3,037 from the proxy measure for part-county rural areas. These designation and staffing ratios are ap- plied to metropolitan and nonmetropolitan areas. The staffing criteria for correctional insti- tutions were partly based on the needs identified by the Federal Bureau of Prison's Medical Direc- tor's office. DHHS’s Alcohol, Drug Abuse, and Mental Health Administration provided the 600 workload unit estimates and the 1:20,000- 30,000 ratio. The Indian Health Service esti— mates were based on an expected increase for primary care and psychiatric physician needs of 3 percent yearly. These criteria are summarized in table 49. These designation and staffing ratios were used to arrive at the estimated need in 1990 for 16,400 primary care and psychiatric physicians. These projections have been used to plan for future staffing of NHSC. The great majority of future NHSC physicians will come from medical school scholarship recipients obligated to serve year-for-year in the Corps. The emphasis is therefore on recruiting first-year medical stu- dents, as the total obligation will be 4 years. Table 49.—Criteria tor Unmet Need Calculation by Area Designation Staffing Area ratio ratio Nonmetropolitan ......... 123,500 1:2,000 Metropolitan ............. 123,500 12,000 Correctional institutions Primary care ........... 1:1 ,000 1:500 Psychiatry ............. 1:2,000 1:1,000 State mental hospitals (psychiatrists) .......... 600 workload 600 workload unitsa/FTE units/FTE Community mental health centers (psychiatrists). . . 1:30,000 1:30,000 lndian Health Service ..... all officially 3-percent recognized yearly tribes increase aTotal workload units = average daily inpatient census + 2x (number of inpa» tient admissions per year) + 0.5x (number of admissions to day care and out- patient services per year). SOURCES: Memorandum from the Chairman, NHSC Needs Task Force A, to the Director, Bureau of Community Health Services, Health Services Administration; the Deputy Director, Bureau of Health Manpower, Health Resources Administration; and the Chairman, NHSC Needs Task Force, Washington, DC, May 26, 1978; and 42 CFR sec. 5. Currently, an option has been adopted whereby the Corps will consist of 8,300 physicians and 2,800 physician extenders in 1990, with the understanding that the target could be read- justed to 15,000 physicians and 2,800 physi- cians’ assistants and nurse practitioners after a 3-year study. The latter would meet almost all projected need, but not until 1995. The 8,300- physician target would require about 13 percent of each medical school class through 1983. The 15,000 physician target would require recruiting 25 percent of each class by 1986 (USDHEW, 1979d). The primary care physician-to-population designation ratio of 1 to 3,500 is employed as a major criterion in the process of determining whether a particular area qualifies for official designation as an HMSA and thus eligible for NHSC placements and other aid. It is not, how- ever, the only criterion employed, and areas with lower physician-to-population ratios may qualify for designation under certain condi— tions. This is best illustrated by the criteria for geographic areas. A sample of the specific meth- odologies for meeting these criteria should illus— trate the point. Detailed criteria can be found in the Code of Federal Regulations, title 42, section 5, appendix A. Three criteria must be met for designation: 1. The area is a rational area for the delivery of primary medical care services. 2. One of the following conditions prevails within the area: (alfhe area has a population-to-primary care physician ratio of at least 3,500:1, or (b)The area has a population—to-primary care physician ratio of less than 3,500:1 but greater than 3,000:1 and has either unusually high needs for primary medi- cal care services or insufficient capacity of existing primary care providers. 3. Primary medical care manpower in con- tiguous areas is overutilized, excessively distant, or inaccessible to the population of the area under consideration (42 CFR sec. 5, app. A). Ch.3—Requirements 0 83 “Rational area for the delivery of primary care" includes: i. A county, or a group of contiguous coun- ties whose population centers are within 30 minutes travel time of each other. ii. A portion of a county, or an area made up of portions of more than one county, whose population, because of topogra- phy, market or transportation patterns, distinctive population characteristics, or other factors, has limited access to contig- uous area resources, as measured general- ly by a travel time greater than 30 minutes to such resources (42 CFR sec. 5, app. A). “Insufficient capacity of existing primary care providers” will be met if at least two of the following criteria are documented: a. More than 8,000 office or outpatient visits per year per FTE primary care physician serving the area. b. Unusually long waits for appointments for routine medical services (i.e., more than 7 days for established patients and 14 days for new patients). c. Excessive average waiting time at primary care providers (longer than 1 hour where patients have appointments or 2 hours where patients are treated on a first-come, first-served basis). 01. Evidence of excessive use of emergency room facilities for routine primary care. e. A substantial portion (two—thirds or more) of the area’s physicians do not accept new patients. f. Abnormally low utilization of health serv- ices, as indicated by an average of 2.0 or less visits per year on the part of the area’s population (42 CFR sec. 5, app. A). Several points should be noted in comparing the actual criteria used for HMSA designation with the methods and assumptions used to mod- el and project the number of shortage areas an- ticipated in future years. First, the model for predicting future shortages uses the county as the geographic base to calculate physician-to- population ratios. The HMSA designation proc- ess uses a ”rational service area” as the geo- graphic base. A “rational service area,” as we have seen from the regulations, could be a coun- ty or it could be an area larger or smaller than a county. In sum, the definition of a ”rational service area” contains considerable flexibility to permit responsiveness to local conditions in making the actual HMSA designations. Additional flexibility to respond to local con- ditions is introduced by permitting areas to qualify as HMSAs if they have a physician-to- population ratio lower than 1:3,500 but greater than 1:3,000 as long as they can show either un- usually high needs for primary medical care services or insufficient capacity of existing pri- mary care providers. One final factor that differentiates the meth- ods used in projection of future shortage areas from those used in the actual official designation process is that, in order for an area to actually receive official designation as an HMSA, a re- quest for designation must come from the local level. Thus, a request for HMSA designation serves as a preliminary indicator that there is in- terest at the local level in obtaining NHSC physicians. However, remember that HMSA designation is necessary not only for assignment of NHSC physicians, but also that such desig- nated areas: would be areas in which students who borrowed money under health professions student loan programs could practice in lieu of repaying the loans in money; would be eligible for grants in various health manpower training programs; would be eligible or given preference for grant funds for several Bureau of Communi- ty Health Services programs such as the urban and rural health initiatives; and would be the only areas in which rural health clinics could be certified for reimbursement of nurse practitioner and physicians’ assistant services under Medi- care and Medicaid. So HMSA designation does not necessarily mean that NHSC physicians will be recruited to provide services in these areas. The model had predicted a need for 14,033 primary care physicians in 1979, 8,839 in non- metropolitan areas, and 5,194 in metropolitan areas (USDHEW, 1978a). The actual number of HMSAs designated in 1979 was 1,711, with a total primary care physician need of 11,336. Of the HMSAs, 1,226 were in nonmetropolitan 84 ' Forecasts of Physician Supply and Requirements areas, with a need for 5,368 physicians, and 485 were in metropolitan areas, with a need for 5,968 physicians (Reid, 1980). Thus, there was an overestimate of physician need in nonmetro- politan areas and an underestimate of need in metropolitan areass. However, a request for designation must come from the local level, so the difference between the predicted need for 14,033 primary care physicians and the actual need for 11,336 is not surprising. How well do the HMSA designation and staffing criteria relate to the use of primary care physician services? Some physician capacity uti- lization surveys have recently been made avail- able. Salient findings from the surveys are: The analysis also examined the influences of exogenous factors on practice characteristics of HEW-designated physician shortage area coun- ties—defined primarily in terms of physician- population ratios. In practical terms, none of the fully or partially designated shortage counties studied gave evidence of excess demand in the traditional economic sense. Physicians’ office hours in these shortage areas were about the same as those of physicians in nonshortage areas. Nor were patient waiting times, for ap- pointment and in the office, significantly dif- ferent from those in nonshortage areas. How- ever, a decreasing waiting time to appointment as the supply of general practice physicians in- creases is some indication of excess demand in shortage areas. It was found, in fact, that short- age area physicians had slightly fewer patient visits than physicians at large. (This result is substantiated by a recent study of the Health Service Administration that reported similar observations of productivity from NHSC physi- cians placed in shortage areas.) A control for population density (used as a proxy for travel distance to see a physician) made little difference in these results (reference and footnotes omitted) (USDHEW, 1978b). The NHSC experience, however, could be re- versed over time. Table 50 disaggregates physi- cian encounters/physician by: 1) self-support ratio categories, 2) initial staffing years, and 3) sites with and without midlevel practitioners. Bearing in mind that the NHSC experience was only 5 years old at the time of the study, the table shows that: 1) patronage of NHSC sites builds over time, patient demand being positive- ly and significantly related to the number of years the sites have been in operation, 2) the more mature sites tended to have higher produc- tivity per provider, and 3) sites that were ap- proaching the capability to be financially self- supporting showed higher productivity levels per provider. Table 50.—Physician Encounters per Physician and Physician and Physician Encounters per Physician Hour by Selected Cohorts, National Health Service Corps (FY 1976) Self-supporta ratio categories Initial staffing year Sample sites 1 2 3 1972 1973 1974 1975 1976 Physician encounters/physician 4,664 7,092 4,568 3,524 6,144 5,164 5,140 3,912 2,804 (all sites) .................. (130) (30) (52) (48) (25) (11) (27) (64) (2) Physician encounters/physician 4,428 6,420 4,048 3,440 5,544 5,780 5,392 3,792 2,804 (sites with no PEsb) ......... (94) (23) (40) (31) (12) (6) (18) (56) (2) Physician encounters/physician 5,272 9,304 6,296 2,888 6,700 4,420 4,644 4,772 — (sites with PEs) ............ (36) (7) (12) (17) (13) (5) (9) (8) — Physician encounters/ physician hour ............. 2.4 3.3 2.3 1.8 3.0 2.8 2.6 2.0 2.1 (all sites) .................. (130) (30) (52) (48) (25) (11) (27) (64) (2) Physician encounters/ physician hour ............. 2.6 5.2 2.9 1.4 3.3 2.6 2.8 1.1 — (sites with PEs) ............ (36) (7) (12) (17) (13) (5) (9) (8) — Physicians encounters/ physician hour ............. 2.2 2.8 2.1 2.0 2.7 2.9 2.4 1.0 2.0 (sites with no PEs) .......... (94) (23) (40) (31) (12) (6) (18) (56) (2) aSelt-support ratios measure the relation between the total revenues from all sources and total costs experienced at sites at a given time. Category 1 sites are the most self-supporting, diminishing to category 3. Physician extenders, or midlevel practitioners. SOURCE: H. L. Heaton, et at, Comparative Cost and Financier/Analysis oIAmbu/atory Care Providers, GEOM ET. Inc., report No. HF-SGO, DHEW contract No. HSA-74A68, 1976. "Aiken, L H. et a1 1979, “The! ;:tribution of Underserved?” 121147-162 . Bure Labor Statistics, U. S. Depar Labor, 1979, Occupational Project: ~ . .Department of staff paper in preparation for _ ole, R., 1980, personal co I, the Division of Manpowe of Health ' ower,,w , 'sonal communication from anpower Analysis, Bureau . foice of ' Physi- oyections," Inquiry . L. et al., 1976, Comparat" I Providers, GEOME " , _ ting the Need for Primary Care Physicians," ports 94: 3 10. U. S Department of Heal Welfare, nology Assessment. , 1975, Impact of Natiori ance~The Case of the , , ‘ance Plan, An Issue Pape ton, D. C.: Bureau of Health V'Health Resources Adminis- (HRA) Medical Education National Ad tee to the Secretary, Dep _ alth, Education, and Welfare . VI 'ttee Washington, D. C. “ Manpower, Health Re- _, istration, 79-633. Congress on the Status of Hea2th Professions Personnel in the United States Washingto D C4811 of Health Manpower, Health ministration DHEW publica- Needs Task ,_ 1978c,Phy51 ts, GMENAC I Underserved Areas, 1973-19 the People on the National 3 Staff Papers #2, *Washingto ‘of Health Manpower, Health ,lication No. (HRA) 78-12. : 1980 0 - 60—618 Office of Technology Assessment The Office of Technology Assessment (OTA) was created in 1972 as an advisory arm of Congress. OTA’s basic function is to help legislative policymakers anticipate and plan for the consequences of technological changes and to examine the many ways, expected and unexpected, in which technology affects people’s lives. The assessment of technology calls for exploration of the physical, biological, economic, social, and po- litical impacts which can result from applications of scientific knowledge. OTA provides Congress with independent and timely information about the potential effects—both beneficial and harmful—of technological ap- plications. Requests for studies are made by chairmen of standing committees of the House of Representatives or Senate; by the Technology Assessment Board, the governing body of OTA; or by the Director of OTA in consul- tation with the Board. The Technology Assessment Board is composed of six members of the House, six members of the Senate, and the OTA Director, who is a nonvoting member. OTA currently has underway studies in 10 general areas—energy; materials; international security and commerce; food and renewable re- sources; genetics and population; health; space, telecommunication, and information systems; oceans; and transportation. OTA-H-113 APRIL 1980