Physicians in the United States, like the rest of the population, change their residence and their location of work or practice from time to time.1 For the overall population, intercounty and interstate moves are tracked annually by the U.S. Census Bureau and the IRS.2,3 Census data show that younger, college-educated persons are much more likely to move to another county or state than the rest of the population. Between 1995 and 2000, 13.1% of all U.S. residents aged 25 to 39 moved from one to another county, and 12.4% moved to another state.4,5 The previously reported migration rates for physicians cover 10-year periods, and just less than 25% of practicing, posttraining physicians moved to another county or state in the decades of 1981 to 1991 and 1991 to 2001.1 However, the annual rate of relocation for the overall U.S. population has steadily declined from an immediate post-WWII peak of nearly 25% in a year to an all-time low in 2010 of 12.5%. This article reviews recent migration trends of physicians using linked data files of physician distribution at the start of 2006 and 2011 to answer the questions, Do physicians still relocate more often than the general population, and from where and to where do they move?
Physicians have been trained primarily in urban places where medical schools and hospitals are located.6 They then “diffuse” to other locations to offer their services according to market demand and to meet the needs of patients.7 This diffusion does not match the distribution of the overall population, as there are proportionately far fewer physicians in rural areas than urban, leading some to call this distributional process a form of “market failure.”8 This pattern of incomplete diffusion was illustrated in an analysis of practicing physicians who moved from one county to another during the period 1981 to 2001, which showed that physicians tended to move to places with fewer physicians but better economic conditions.1 Given the recent slowdown in overall interstate migration of the general population and an economic recession, patterns of physician migration may have changed since 2001.
The relative rural–urban supply of physicians has been a focus of study and of policy interventions for over 100 years.9 Migration of physicians into rural areas is a necessary precondition for rural populations to gain access to care as medical education and postgraduate training has been almost exclusively carried out in urban areas. Factors that influence the supply in rural places have been studied extensively, with a focus on medical school characteristics,10 graduate medical education location and specialty choice,11 and the location of trainees’ birth and upbringing.12
Changing specialties is also a factor in geographic migration as physicians who practice in more specialized fields tend to move to more urban places where focused practice is more economically feasible. However, the choice of specialty is not fixed at the end of initial training. A study of a random sample of U.S. medical school cohorts from 1968 found that 40% of general internists, 42% of family or general practitioners, 29% of surgical subspecialists, and 45% of internal medicine subspecialists changed their self-designated specialty of practice between 1971 and 1978; overall, 29% of that cohort changed the specialty in which they practiced.13 This is relevant to an analysis of geographic movement as physicians may be influenced to migrate to more urban areas as they choose to become more specialized over their careers; alternatively, physicians may tend to “despecialize,” preparing them for other opportunities in different practice locations.
There are differing categories of influences on migration that likely apply to physicians.14 Doctors respond to economic pressures, to personal preferences or events, and to professional opportunities. The patterns of movement may cause “hollowing out” in some places, where the physician supply follows general population outmigration; “continental drift,” where physicians move from regions with fewer recreational opportunities or harsher winters, for instance, to more desirable regions; or “escape” from places where the economy is less robust than other communities or the quality of life is perceived as lower.
We have long understood the connection of medical school15 and residency training location16 and the type of medical school attended17 with the location of initial practice. The location of a physician’s practice has been shown to be heavily influenced by the location of her or his graduate medical education as well as where the person grew up.18,19 This supports the “prior contact” hypotheses of physician location behavior posited by Ernst and Yett,12 where physical residence or subsequent tenure in an area during education and training is a strong predictor of physician practice location.
Federal policies to influence physicians to practice in underserved areas have been in place for over three decades and are generally included under Title VII of the Public Health Services (PHS) Act and include the National Health Service Corps (NHSC), arguably the largest program.20 One econometric study found that the NHSC likely increased overall supply in Health Professional Shortage Areas (HPSAs) (a federal government designation to identify places and populations that have very low access to primary care) by 10%.21 Variations in state policies are another potential cause of physician location preference or change; these include malpractice climate, Medicaid policies, and scope of practice for medical practitioners. One summary of state-level factors found the only consistent predictors to be medical school and graduate training locations,22 while another analysis identified malpractice policies as having an influence on surgeon migration.23
The economic influences that determine practice location have been described as “classical” market forces7,24 that would, if supply were increased sufficiently, diffuse physicians to underserved areas because competitive forces would force them to areas with fewer competitors. The role of competition among physicians might tend to “push” doctors to underserved areas25 as opposed to “pulling” them with other factors, including a preferred lifestyle or recreational amenities. The adequacy of the “push” diffusing force to meet rural community needs was not apparent in the late 1980s,26 and there remained substantial inequality in local access to physician care; the general geographic imbalance of physician supply remains essentially the same in the 2010s as it was in the 1980s.
The problem of the geographic maldistribution of physicians has fallen out into two somewhat separate issues: the rural-versus-urban gap in supply, and access to physician care in underserved areas in general including urban areas and inner cities. Programs to affect supply have similarly divided themselves. Rural-focused programs emphasize the special conditions of rural practice and attempt to maximize the number of entrants with rural backgrounds.27 Programs that emphasize underserved areas and populations without special regard to geography include the NHSC and other programs supported by Title VII of the PHS Act and a wide range of state-supported programs that offer loan repayment or some form of obligated subvention of the training and practice expenses of physicians and other practitioners.18 These programs usually prioritize or restrict their support to practitioners in HPSAs or some form of designation of areas or populations. Payment bonuses or add-on incentives to the Medicare program are tied to HPSA designations for primary care physicians and, for the period 2010 to 2015, general surgeons practicing in those HPSAs. There is little evidence of the effectiveness of these bonus payments.28,29
Another major effort to support the redistribution of physicians is the J-1 visa waiver programs that allow non-U.S. citizen international medical graduates (IMGs) to enter the United States for additional training with the caveat that they practice in an underserved area. The ability of the J-1 program to influence location in rural and other underserved areas has shown some mixed effects in empirical studies, with one analysis showing J-1 recipients staying in their placement communities close to two years beyond their commitments,30 but having lower comparative retention and integration in another.31 The overall distribution of IMGs was found to not favor underserved places,32,33 and commitment to practice in underserved areas varied according to the IMG’s place of birth.34
In this study, I ask whether the future supply of physicians will be distributed in such a way to meet population needs, given recent trends in physician migration. Ernst and Yett12 observed in 1985 that concerns with overall physician supply were less important than understanding the “functional distribution” of physicians among specialties and practice locations. Measuring local physician supply in the United States has proved challenging as the inventories of doctors are most often compiled at the county level; for example, county-level data are available from the Area Resources File, distributed by the Health Resources and Services Administration (HRSA), and the American Medical Association’s (AMA’s) Physician Masterfile summarized in the publication Physician Characteristics and Distribution in the United States. These sources allow for a general assessment of the long-term trends in physician supply but have been criticized for their accuracy in estimating contemporaneous, local supply.35
The determination of an ideal or even an adequate local supply of physicians has been fraught with difficulty; there are guidelines but no clear and firm measures of what supply ought to be or when there is a local shortage.36,37 In this analysis, I use numbers of licensed physicians, both doctors of medicine (MDs) and those with osteopathic training (DOs), in describing total “headcounts” of practitioners. The effective supply, or “functional distribution,” of physicians is dependent on many factors, including the volume and productivity of practices, the ability to work with other practitioners and professions, and the capacity of facilities and technical services, as well as the presence or absence of a doctor in a place. This analysis is thus limited to a rough estimate of the actual supply of services based on counts of physicians.
For this analysis I combined individual-level AMA Physician Masterfile data for the beginning of 2006 and 2011. The analysis data set included individual-level variables describing the age, gender, specialty, whether the individual graduated from a U.S. or Canadian medical school or not, the individual’s practice location by ZIP code, activity status, practice type, and whether the individual was in residency training or not. These were combined with data that described whether the location was in an HPSA as defined by the HRSA or a metropolitan or nonmetropolitan (rural) county as defined by the Office of Management and Budget. For the final data, I excluded physicians in residency training in 2006 or 2011, those in federal employment in either year, those 80 years or older in 2006, and any physician not practicing in 1 of the 50 states in both years. In initial analysis, I compared the practice location for all active physicians in the two years; if the ZIP code location for a physician indicated that he or she had moved to a different county, then the physician was considered a “mover.” I created bivariate tables comparing movers with nonmovers, the interstate movement of physicians, and an analysis of the movement into and out of HPSAs. A logistic regression model using the logit function in Stata statistical software (Stata Inc., College Station, Texas) was run, which examined the simultaneous effects of characteristics of the physicians who moved compared with nonmovers. For this combined analysis, I compared all DOs, surgeons, and primary care physicians with all other specialties. This study was considered exempt by the institutional review board of the University of North Carolina at Chapel Hill.
Number of movers
There were 796,496 total physicians in the analysis data set, and of those, 620,976 (77.9%) met the initial criteria of being located in the 50 states and active in both years. A summary of the proportion of certain categories of physicians who moved is included as Table 1.
Excluding residents and those 80 years or older resulted in a total of 489,971. Of that total, 97,361 (19.9%) physicians moved from one county to another in the five-year period. By contrast, of the larger number that included residents in 2006, 150,140 (25.9%) moved from one county to another. Of the 97,361 posttraining county-to-county movers, 43,366 (44.5%) moved from one state to the next. The most common interstate moves were from or to New York and New Jersey: 693 physicians moved from New York to New Jersey and 576 in the opposite direction, and 525 New York physicians moved to Florida and 522 to California. There were more physicians who moved from an urban (metropolitan) county to a rural (nonmetropolitan) county, 10,171 (10.4% of those who moved), than in the reverse direction, 9,183 (9.4%). These urban–rural moves were only 18.4% of all moves; more than 80% of moves were from urban to urban (70,829, or 72.7%) places or rural to rural (7,178, or 7.3%). The net migration between states is illustrated in Figure 1. The midwestern and mid-Atlantic states as well as those in the lower Mississippi valley experienced outflows, while the western and remaining southern states had net gains.
The average move for all U.S. physicians who relocated from one U.S. county to another was 537 miles, but removing moves to and from Hawaii reduces that average to 146 miles. These moves were substantial, but the approximately two-hour travel time for the average move could be equated to be on the order of moving from one tertiary service area, for example, from one of the Dartmouth Hospital Referral Regions, to an adjacent one or “next-over” region.
Movers by specialty
Among the specialties, family medicine physicians had almost exactly the same overall proportion of county-to county movers as all physicians (15,797 of 77,214, or 20.5%). Higher county-to-county rates were found for neuroradiologists (531 of 1,602, or 33.1%), vascular and interventional radiologists (498 of 1,544, or 32.3%), hospitalists (1,184 of 3,844, or 30.8%), and emergency medicine specialists (7,935 of 26,416, or 30%). Lower rates were found for gynecologists (211 of 2,139 or 9.9%), pulmonary medicine (624 of 4,970, or 12.6%), plastic surgery (731 of 5,568, or 13.1%), and gastroenterology (1,655 of 10,640, or 15.6%). Internists (15,763 of 83,721, or 18.8%) and pediatricians (7,657 of 44,732, or 17.1%), two of the three largest groups of specialists, had slightly lower-than-average rates of moving. Doctors of osteopathic medicine were more likely to move to another county (8,825 of 35,590, or 24.8%) than MDs (88,536 of 454,381, or 19.5%).
Movers by age or gender
As might be expected, younger physicians are more likely to move; 36% (21,138 of 58,639) of active physicians between the ages of 30 and 39 moved across counties between 2006 and 2011. That percentage dropped to 22.5% (35,219 of 156,330) for those between 40 and 49, then to 16.3% (26,297 of 161,698) for physicians aged 50–59; 13.2% (12,092 of 91,495) for those aged 60–69; and only 11.8% (2,534 of 21,440) for those over 70. Overall, women were more likely to move than men, 22% (31,073 of 141,138) versus 19% (66,288 of 348,833), but this varied by age as more younger men aged 30–39 were likely to move (16,115 of 41,247, or 39.1%) than women (13,021 of 35,091, or 37.1%), and more older women (older than 60) (3,208 of 21,904, or 14.6%) were likely to move than men older than 60 (15,892 of 124,300, or 12.8%). Slightly more women aged 50–59 (8,951 of 53,308, or 16.8%) moved from one county to another than men in the same age group (24,447 of 148,549).
Movers and underserved areas
The federal government uses the HPSA designation to identify places and populations that have very low access to primary care. Counties may be designated as either whole- or part-county geographic HPSAs or populations in the whole- or part-county HPSAs designated. The former are termed area or geographic HPSAs, and the latter are termed population HPSAs. These designations make the counties or organizations and practitioners in the counties eligible to benefit from multiple programs to support primary care, including placements or loan repayment from the NHSC and payment bonuses. Of the physicians who moved and who were located in a non-HPSA county (either part- or whole-county designated) in 2006, only 5% (1,511 of 60,946) moved to a whole-county HPSA, and 28.4% (17,296 of 60,946) moved to a part-county HPSA—a total of 18,807 of 97,361, or 19.3%, of all movers. In contrast, 21,676 of 36,415 (59.5%) physicians in an HPSA county in 2006 moved to non-HPSAs in 2011. There were net outflows among the movers, 2,682 from part-county (33,546 in 2011 versus 36,415 in 2006) and 187 from whole-county HPSAs (2,782 in 2011 versus 2,969 in 2006), that were made up by new entrants into HPSA counties. In 2006 there were 4,481 primary care and 8,435 total physicians in whole-county HPSAs; by 2011, the numbers increased to 4,955 and 9,683.
Summary characteristics of movers versus nonmovers
The logistic regression analysis (see Table 2) included specialty, age, gender, and whether the physician was in a metropolitan county in 2006. The analysis identified osteopaths and females as more likely to move and surgeons and primary care physicians as less likely when compared with other specialists. Older and urban physicians were less likely to move.
The turnover for each county can be calculated as a percentage of the total physicians in a county in 2011 who moved into the county between 2006 and 2011. There are 3,035 counties or county equivalents in the United States that had at least one physician in both years; 55 county-equivalent jurisdictions had none in either or both years. The modal decile of turnover was between 20% and 30%, with just over one-quarter of counties experiencing that degree of change in the five-year period (see Figure 2).
Counties with higher rates of turnover tended to have fewer physicians. The correlation between total supply of physicians in 2011 and percent turnover is slightly negative (−0.144, P = .0000).
One overarching characteristic of the overall supply of physicians in the United States for the past 30 years has been the consistency in the patterns of migration and the resulting geographic distribution of practitioners. Physicians continue to migrate over their working lives more often than the U.S. population as a whole. My findings from a recent five-year period are consistent with earlier analyses of the rates of migration for doctors. There is constant geographic change in the local supply of practitioners as younger physicians move away from practice locations they may have deemed less attractive than the place where they are going; older physicians may have different reasons for moving and do so at a slower pace. Broader theories of population migration can also be applied to physicians: They move in the same general directions as the overall population (“continental shifts” from the Northeast and Midwest toward the West and South). These trends follow shifts of the economy that “pull” people to new locations as jobs and people move to places with more opportunities. Economic considerations specific to medical markets may play a “push” role in the decision to change location as physicians choose to move away from dense concentrations of competitors (in the Northeast and Midwest) to places where they can practice effectively and meet their career expectations and have less competition for patients.
This analysis depends on the accuracy of the location data for physicians. Physicians may actually practice in more than one location—across ZIP code and county boundaries—complicating any geographic analysis. This may be made even more problematic as physicians are more and more becoming employees of large systems that may move physicians across practice locations as part of their normal staffing process. No adjustment has been made in these data to account for multiple practice locations for physicians.
The many programs and policies intended to redistribute the physician supply, including funding for expansion of Federally Qualified Health Centers and many parts of the Affordable Care Act, have not been able to significantly change the general trends of physician migration and distribution to provide an appropriate number of practitioners in many rural places. Given that there is an active cycling of physicians into and out of rural and underserved places, those programs should anticipate these flows and attempt to influence those who are likely movers to select places where there is more need for their services. The good news is that there is a slight positive net flow toward nonmetropolitan counties that suggests that these redistributive policies are having some effect, especially among younger physicians. There is also a slight trend toward diffusion into HPSA counties, both urban and rural, with overall increases in the numbers of all specialty as well as primary care physicians. However, this is not enough to bring the overall distribution of physicians to a level that matches population needs in the foreseeable future. Policy makers should consider expanding current programs as they appear to have positive net effects but to carefully assess the performance of the existing policies against the background of a dynamic physician workforce and consider that they are likely to be more successful by influencing flows of physicians who are likely to move anyway as well as supporting the retention of physicians who may consider relocating away from underserved places.
Acknowledgments: The author appreciates the work of Jennifer Groves and Randy Randolph in managing the data required for the analysis.
1. Ricketts TC, Randolph R. The diffusion of physicians. Health Aff (Millwood). 2008;27:1409–1415
4. Schachter J Geographical Mobility: 2002 to 2003. March 2004 Washington, DC U.S. Department of Commerce, Economics and Statistics Administration, U.S. Census Bureau
5. Franklin RS Migration of the Young, Single, and College Educated: 1995 to 2000. November 2003 Washington, DC U.S. Census Bureau, Population Distribution Branch
6. Burfield WB, Hough DE, Marder WD. Location of medical education and choice of location of practice. J Med Educ. 1986;61:545–554
7. Rosenthal MB, Zaslavsky A, Newhouse JP. The geographic distribution of physicians revisited. Health Serv Res. 2005;40(6 pt 1):1931–1952
8. Newhouse JP, Williams AP, Bennett BW, Schwartz WB. Does the geographical distribution of physicians reflect market failure? Bell J Econ. 1982;13:493–505
9. Rosenblatt RA, Hart LGRicketts TC. Physicians and rural America. Rural Health in the United States. 1999 New York, NY Oxford University Press:38–51
10. Rosenblatt RA, Whitcomb ME, Cullen TJ, Lishner DM, Hart LG. Which medical schools produce rural physicians? JAMA. 1992;268:1559–1565
11. Seifer SD, Vranizan K, Grumbach K. Graduate medical education and physician practice location. Implications for physician workforce policy. JAMA. 1995;274:685–691
12. Ernst RL, Yett DE Physician Location and Specialty Choice. 1985 Ann Arbor, Mich Health Administration Press
13. Holden WD, Levit EJ. Migration of physicians from one specialty to another. A longitudinal study of US medical school graduates. JAMA. 1978;239:205–209
14. Hernandez-Murillo R, Ott LS, Owayang MT, Whalen D. Patterns of interstate migration in the United States from the survey of income and program participation. Fed Reserve Bank St. Louis Rev. 2011;93:169–185
15. Chen F, Fordyce M, Andes S, Hart LG. Which medical schools produce rural physicians? A 15-year update. Acad Med. 2010;85:594–598
16. Rosenblatt RA, Schneeweiss R, Hart LG, Casey S, Andrilla CH, Chen FM. Family medicine training in rural areas. JAMA. 2002;288:1063–1064
17. Rosenblatt RA, Whitcomb ME, Cullen TJ, Lishner DM, Hart LG. The effect of federal grants on medical schools’ production of primary care physicians. Am J Public Health. 1993;83:322–328
18. Phillips RL, Dodoo M, Patterson S, et al. Specialty and Geographic Distribution of the Physician Workforce: What Influences Medical Student and Resident Choices?. March 2, 2009 Washington, DC Robert Graham Center
19. Rabinowitz HK, Diamond JJ, Markham FW, Santana AJ. Increasing the supply of rural family physicians: Recent outcomes from Jefferson Medical College’s Physician Shortage Area Program (PSAP). Acad Med. 2011;86:264–269
20. Rittenhouse DR, Fryer GE Jr, Phillips RL Jr, et al. Impact of Title VII training programs on community health center staffing and national health service corps participation. Ann Fam Med. 2008;6:397–405
21. Holmes GM. Increasing physician supply in medically underserved areas. Labour Econ. 2005;12:697–725
22. Henderson T, Farmer C, Szwarc S Practice location of physician graduates: Do states function as markets?. January 2003 Washington, DC
23. Chou CF, Lo Sasso AT. Practice location choice by new physicians: The importance of malpractice premiums, damage caps, and health professional shortage area designation. Health Serv Res. 2009;44:1271–1289
24. Newhouse JP, Williams AP, Bennett BW, Schwartz WB. Where have all the doctors gone? JAMA. 1982;247:2392–2396
25. Jiang HJ, Begun JW. Dynamics of change in local physician supply: An ecological perspective. Soc Sci Med. 2002;54:1525–1541
26. DeFriese GH, Ricketts TC. Primary health care in rural areas: An agenda for research. Health Serv Res. 1989;23:931–974
27. Rabinowitz HK, Diamond JJ, Markham FW, Wortman JR. Medical school programs to increase the rural physician supply: A systematic review and projected impact of widespread replication. Acad Med. 2008;83:235–243
28. Shugarman LR, Farley DO. Shortcomings in Medicare bonus payments for physicians in underserved areas. Health Aff (Millwood). 2003;22:173–178
29. Chan L, Hart LG, Ricketts TC 3rd, Beaver SK. An analysis of Medicare’s incentive payment program for physicians in health professional shortage areas. J Rural Health. 2004;20:109–117
30. Kahn TR, Hagopian A, Johnson K. Retention of J-1 visa waiver program physicians in Washington State’s health professional shortage areas. Acad Med. 2010;85:614–621
31. Crouse BJ, Munson RL. The effect of the physician J-1 visa waiver on rural Wisconsin. WMJ. 2006;105:16–20
32. Mick SS, Lee SY, Wodchis WP. Variations in geographical distribution of foreign and domestically trained physicians in the United States: “Safety nets” or “surplus exacerbation”? Soc Sci Med. 2000;50:185–202
33. Fink KS, Phillips RL Jr, Fryer GE, Koehn N. International medical graduates and the primary care workforce for rural underserved areas. Health Aff (Millwood). 2003;22:255–262
34. Mertz E, Jain R, Breckler J, Chen E, Grumbach K. Foreign versus domestic education of physicians for the United States: A case study of physicians of South Asian ethnicity in California. J Health Care Poor Underserved. 2007;18:984–993
35. Staiger DO, Auerbach DI, Buerhaus PI. Comparison of physician workforce estimates and supply projections. JAMA. 2009;302:1674–1680
36. Ricketts TC, Goldsmith LJ, Holmes GM, et al. Designating places and populations as medically underserved: A proposal for a new approach. J Health Care Poor Underserved. 2007;18:567–589
37. Department of Health and Human Services. Negotiated Ruling. Committee on the Designation of Medically Underserved Population and Health Professional Shortage Areas, Final Report to the Secretary. October 31, 2011 Washington, DC Department of Health and Human Services