The United States faces a shortage1 and maldistribution2,3 of primary care physicians. This deficit is of particular concern given that access to health insurance is slated to increase substantially, the elderly population continues to grow, and many rural, poor, and minority communities remain medically underserved.4–9 However, fewer medical school graduates are choosing primary care as a specialty today than in the past.10–12 Concerns have also been raised about the future supply of medical school faculty.13,14 To address projected physician workforce needs, in 2006 the Association of American Medical Colleges (AAMC) called for a 30% increase in U.S. MD-granting medical school enrollment by 2015.15 Currently, medical schools in the United States are in the midst of what has been called the “second great expansion of post-Flexnerian medical education.”16
Since 2002, U.S. MD-granting medical schools have increased their class sizes, and 16 new schools have matriculated their inaugural classes; collectively, expansion has yielded more than 3,500 additional students yearly.17 An additional four medical schools are in the early stages of the accreditation pipeline.18 Medical school expansion creates an opportunity to establish a physician workforce that meets the nation’s needs, but uncertainty remains about whether and how the composition of the physician workforce will change with expansion. Although admitting more students can increase the overall supply of physicians, simply training more physicians is not an effective way to solve primary care shortages, remedy distribution problems, or increase the number of physicians caring for underserved populations.19–21
A robust body of literature demonstrates that certain student characteristics—such as being older, female, a member of a racial/ethnic minority group, or from a rural area—can be reliable indicators of primary care specialty choice3,22–25 and practice location.3,26–30 If medical schools were to admit more students with these characteristics, expansion could enable schools to train a physician workforce that increasingly addresses societal needs.
Given that there is significant variation in expansion across medical schools, we examine in this study whether the schools that have expanded the most differ from other schools in ways that could influence the composition and career choices of the future physician workforce. Further, we compare the period just prior to expansion (1999–2001) to the period 10 years later (2009–2011) to assess whether the characteristics and interests of recent matriculants to medical school have shifted in ways that suggest that the future workforce may better address primary care, diversity, and/or maldistribution challenges. Finally, we compare newly established medical schools’ matriculants with matriculants of medical schools that existed in both periods to ascertain whether new schools may have a distinct impact on the future workforce.
We examined the workforce implications of medical school expansion from two vantage points:
1. A comparison of past workforce outcomes (specialty choice, practice location, faculty service) of the 1999–2001 graduates (n = 47,515) from the 124 U.S. MD-granting medical schools that existed then (termed “existing schools”) by the schools’ levels of expansion, as defined below.
2. A comparison of key characteristics of the 92,787 applicants and 48,862 medical school matriculants in the period before expansion (1999–2001, termed “preexpansion”) with those of the 109,907 applicants and 56,276 matriculants in the period 10 years later (2009–2011, termed “postexpansion” for ease of reference, although expansion of medical schools was not complete in 2011).
Between 2001 and 2011, 11 additional MD-granting medical schools were established in the United States and began enrolling students for the first time (termed “new schools”). Where possible, we considered these new schools separately from the existing schools in our analyses.
The analyses were carried out in fall 2012 using deidentified data; records could be linked by means of a research ID provided by the AAMC. The American Institutes for Research institutional review board deemed this project exempt from further review.
Categorizing schools by amount of expansion
We retrieved the matriculant counts for the two periods from the American Medical College Application Service (AMCAS). We calculated the change in matriculants for each of the 124 existing schools and created four “expansion groups” based on the magnitude of growth. Because class size for a particular school can fluctuate somewhat arbitrarily from year to year, we used three-year rather than single-year matriculant counts to compare the existing schools’ class sizes in the pre- and postexpansion periods. The 20 schools that had fewer matriculants in the latter period or had growth < 1% were considered to have had “no growth.” There was a fairly smooth continuum of growth for the remaining 104 schools, which we roughly divided into tertiles. The “low growth” category consisted of 35 medical schools with growth ≥ 1% but < 10% (mean = median = 5%); the “moderate growth” category included 36 schools with growth ≥ 10% but < 18% (mean = median = 14%); and the “high growth” category encompassed 33 schools with growth ≥ 18% (mean = 33.6%, median = 24.6%). Individual schools are listed with their assigned expansion groups in Supplemental Digital Appendix 1 (http://links.lww.com/ACADMED/A167); the new schools are also listed there.
To compare past workforce production across the four expansion groups, we examined the following outcomes among the 124 existing schools: primary care specialty (including primary specialties of internal medicine, family medicine, pediatrics, internal medicine–pediatrics, geriatrics, and adolescent medicine), practice location (rural and/or underserved areas), and faculty service. Existing schools’ 1999–2001 graduates (n = 47,515) were linked to their corresponding records in the 2010 American Medical Association (AMA) Physician Masterfile to determine their type and location of practice (rural versus nonrural). We determined whether they practiced in underserved regions using a geocoded file that linked practice location in the 2009 AMA Physician Masterfile to federally designated Primary Care Health Professional Shortage Areas (HPSAs) and Medically Underserved Areas (MUAs). We used the AAMC Faculty Roster to determine whether these graduates had served as full-time faculty at any point after their graduation through 2010. These individual workforce outcomes were aggregated by school to the level of the expansion group and were then compared across expansion groups to determine whether schools that had the most robust expansion were distinct from those with less or no expansion. Given that the 11 new schools had few if any graduates in practice, we did not consider their past workforce output.
Trends in applicant and matriculant characteristics
To study the potential changes over time in the characteristics of applicants and matriculants, we examined trends across the 124 existing schools, which enrolled students in both the pre- and postexpansion periods, and considered separately the 11 new schools, which enrolled students only in the postexpansion period. For these analyses, we linked data from two sources at the level of the individual. Characteristics of applicants and matriculants to U.S. MD-granting medical schools were obtained from AMCAS, and future practice intentions were retrieved from the AAMC Matriculating Student Questionnaire (MSQ).
Individual characteristics included in the study were sex, age at application or matriculation, birth county location, parental education, race/ethnicity, Medical College Admission Test (MCAT) score, cumulative undergraduate grade point average (GPA), and ownership of medical school attended (public or private). We used U.S. Department of Agriculture rural–urban continuum codes to classify the birth county as nonrural (codes 0–5) or rural (codes 6–9). In 2002, AMCAS collection of race and ethnicity data was changed from a single variable to separate variables for race and ethnicity, which precludes creating mutually exclusive race and ethnicity groups. To accommodate the change in race/ethnicity measurement from the pre- to postexpansion period, we created a single indicator variable that identified individuals who self-identified as black, Native American, and/or Hispanic.
From the MSQ, we obtained information regarding entering students’ specialty preferences, plans to primarily serve minorities and the underserved, planned practice locations, and parental income data. In 2009, the AAMC revised the MSQ specialty preference question, a change that appears to have decreased the percentage of students indicating they were undecided and to have increased the percentage of students missing data on the item. Because both of these categories (undecided and missing data) reflect nonspecific specialty interest, they were combined for reporting purposes.
In addition, two medical schools changed public/private ownership status during the years under study. We classified these according to their status in September 2012.
We compared the applicant pool and rates of matriculation during the pre- and postexpansion periods, using data on individuals’ characteristics from AMCAS. For persons who applied to medical school in multiple years and had multiple records, only the last record was included in the applicant pool for the period. For each subgroup of interest, we divided the percentage of applicants who matriculated in the postexpansion period by the percentage who matriculated in the preexpansion period to obtain a relative rate of matriculation. For example, for women, the percentage of applicants matriculating postexpansion (26,606/53,222 [49.99%]) was divided by the percentage matriculating preexpansion (22,653/43,862 [51.65%]) for a relative rate of 0.97, which denotes a slightly reduced rate of matriculation in 2009–2011 relative to 1999–2001.
We used chi-square tests to compare the discrete individual characteristics (sex, birth county location, race/ethnicity, parental education level) and interests (specialty preference, practice plans) for the matriculants in three groups: matriculants of the 124 existing schools preexpansion, of these same 124 schools postexpansion, and of the 11 new schools. Analysis of variance was used to compare the continuous student characteristics (MCAT score, age at matriculation, GPA, and parental income) for the three groups.
Relative risks rates and confidence intervals were used to compare applicants who were and were not admitted in the two time periods. The Cochran–Armitage test for linear trend31,32 was used to compare the workforce outcomes across the four expansion groups. We used SAS version 9.2 (SAS Institute Inc., Cary, North Carolina) for all analyses.
The number of students entering U.S. MD-granting medical schools before 2002 was stable, ranging between 16,000 and 16,200 matriculants per year. Between 2002 and 2011, there was a 16.6% increase in the annual number of matriculating medical students, from 16,488 in 2002 to 19,230 in 2011. The growth in matriculants was more pronounced within public schools than private schools, as shown in Supplemental Digital Appendix 2 (http://links.lww.com/ACADMED/A167).
Figure 1 illustrates the baseline and net gain or loss of students from 1999–2001 to 2009–2011 across categories of expansion. Net matriculants decreased among the no-growth schools by 234 students (–2.7%) and increased at the low-growth, moderate-growth, and high-growth schools by 701 (5.0%), 2,219 (14.0%), and 3,169 (30.6%) students, respectively. The 11 new schools generated an additional 1,559 matriculants during 2009–2011.
Table 1 examines the differences in existing schools’ past workforce production across expansion groups. Schools with higher growth rates were generally more likely than schools with lower or no growth to have produced graduates practicing in primary care specialties, in rural areas, and in underserved areas. These differences were most pronounced between high-growth and no-growth schools, which produced, respectively, 30.5% versus 24.1% of graduates in primary care practice, 8.6% versus 5.6% of graduates in rural practice, and 32.4% versus 26.9% graduates practicing in underserved areas. The no-growth schools produced a higher percentage of graduates who served as full-time faculty (29.7%) than the schools in any of the other expansion groups (23.2%–23.9%).
Table 2 examines the likelihood of matriculation into medical schools in the postexpansion period compared with the preexpansion period. Total applicants increased more than the number of medical school positions did, such that both male and female applicants had slightly lower rates of matriculation in 2009–2011 relative to 1999–2001. Numbers of applicants and matriculants from rural backgrounds both declined over the study decade. While the rate of matriculation remained stable for rural applicants, it declined for the growing pool of nonrural applicants. Black, Native American, and Hispanic applicants had a stable matriculation rate in the postexpansion period relative to the period 10 years prior. Students with MCAT scores above 33 had a similar matriculation rate in both periods, but students with successively lower MCAT scores (grouped as >33, 31–33, 28–30, and ≤27) had steadily declining rates of matriculation.
Table 3 displays the changes in matriculating student characteristics from the pre- to postexpansion periods for existing medical schools as well as the characteristics of students matriculating at new medical schools in the postexpansion period. Overall, the number of women entering medical school increased by 3,953 (17.5%), while the number of men entering medical school increased by 3,461 (13.2%). The number of black, Hispanic, and Native American matriculants was 1,902 (27.9%) higher in the postexpansion period, and new schools attracted a more diverse student body than did existing schools. Despite expansion, the number of matriculants born in rural areas decreased. Students with rural backgrounds were rare among new schools, where an average of 1 rural student matriculated per school, per year. Existing schools matriculated 849 (31.5%) fewer students from rural backgrounds in 2009–2011 than in 1999–2001.
Mean MCAT scores and parental education levels increased among matriculants at existing schools in the postexpansion period, whereas median parental income decreased slightly (Table 3). Mean MCAT scores rose from 29.6 to 31.3 at existing schools; the new schools’ mean MCAT score was somewhat lower, at 28.7. Nearly 10% of new schools’ matriculants came from homes in which the highest parental education level was high school, a higher percentage than that of matriculants at existing schools in either the pre- or postexpansion period. In contrast, more than 35% of students at existing schools in the postexpansion period had at least one parent with a doctoral degree (e.g., MD, JD, PhD) in comparison with 30% of students at new schools. Median parental income decreased from $114,400 in the preexpansion period to $110,000 in the postexpansion period at existing schools; it was $100,000 at new schools.
Despite a decline in the MSQ response rate, the number of matriculants who indicated interest in primary care specialties increased from the pre- to postexpansion period; the rate was slightly higher at existing schools than new schools (Table 4). The numbers and percentages of students who reported plans to work primarily with the underserved and minorities, as well as of students who were undecided about these plans, were higher for both existing schools and new schools in the postexpansion period than in the preexpansion period. While the percentage of matriculants who were undecided about practice location grew, the percentage who indicated they planned to work in rural areas declined.
From 2002 to 2012, MD-granting medical schools in the United States increased their output of new physicians substantially, representing the largest expansion in the production of U.S.-trained physicians in more than three decades. As others have noted,33 expansion presents an opportunity for medical schools to target growth to achieve greater alignment between the physician workforce and the needs of society. To do so most effectively, medical schools should articulate and embrace the workforce outcomes they seek to achieve. Whether these outcomes include more primary care physicians, more providers practicing in rural and underserved communities, more faculty clinicians and investigators, or other worthy ends should be a matter of active discussion and planning within each medical school.
Despite the tremendous growth in medical school enrollment assessed in this study, expansion had only hit midstride as of 2011 according to estimates provided by medical school deans.17 Further expansion planned through 2017 is projected to result in a total of 21,434 matriculating students annually, representing another 11% overall increase.17 Additional expansion is not without challenges, however. The growing number of learners across all health professions, coupled with increased clinical pressures on faculty, has raised concerns about the number of clinical training opportunities available to students.17 In addition, ongoing medical school growth coupled with limited concurrent growth in residency positions may create a graduate medical education bottleneck in which an increasing number of students struggle to find a residency position.
While the ultimate impact of expansion on the emerging physician workforce will not be known for a number of years, this study examined several factors that provide insight about the workforce implications of expansion to date. Collectively, the medical schools with the most expansion over the study decade produced the highest proportion of primary care doctors as well as the highest proportion of physicians practicing in underserved and rural areas. Given long-standing workforce shortages in these areas, disproportionate expansion by these schools could yield outsized benefits in meeting societal needs. On the other hand, the schools that produced the highest proportion of faculty were disproportionately those with little to no expansion. This may extend the challenges of maintaining an adequate workforce of clinician–investigators and clinician–educators well into the future.34,35
Remarkably, from 1999–2001 to 2009–2011, the pool of applicants increased even more than the number of medical school positions, making entry more competitive. It appears that the applicant’s MCAT score may be an increasingly decisive tool in the selection of students, as applicants with lower scores have become increasingly unlikely to matriculate. Applicants with high MCAT scores, from racial/ethnic minority groups, from rural areas, and those older than 24 years of age were among the few applicant subgroups for whom the relative rate of matriculation did not decline over the period of expansion. This distinction may reflect that schools have increasingly come to value the diversity of experiences these students bring to the classroom and the workforce. The racial and ethnic diversity of matriculants has increased modestly, with new medical schools contributing disproportionately to this gain. Yet, diversity in the matriculating student body continues to lag behind that of the general population. Greater increases are still needed to match the increasing diversity of the population.
Medical student age, sex, parental education level, and parental income have been associated with specialty choice in previous research.22–24,36 If medical schools were seeking students predisposed to careers in primary care, for instance, one might expect a shift toward matriculants who were older, female, and from households of lower socioeconomic status. We found limited evidence that expansion of medical schools overall was accompanied by a meaningful shift in matriculating students according to age or sex. Parental socioeconomic status, however, shows an interesting divide between existing and new schools, with new schools more likely than existing schools to matriculate students from homes with lower parental education levels and lower parental income. This, along with new schools’ higher percentage of racial and ethnic minority matriculants, may reflect a directed effort by these schools to address the need for a more diverse workforce.
We found that there were significantly fewer matriculants with rural backgrounds in the postexpansion period than in the preexpansion period. Research27,37,38 suggests that physicians with rural backgrounds are far more likely than other physicians to practice in the rural United States, a sector of the nation that struggles perennially to attract a sufficient physician workforce. Comparing 2009–2011 with 1999–2001, the pipeline of applicants from rural backgrounds declined by an average of 500 per year, whereas the overall rural population increased, albeit more slowly than the population in urban areas.39 Whether this trend reflects limited educational opportunities, limited role models, or other factors, additional efforts are needed to ensure a robust pipeline for the rural medical workforce.
Although we found limited increases in the matriculation of students whose backgrounds suggest they may be predisposed to careers in primary care and/or to practicing with underserved populations, this should not be interpreted as indicating an absence of effective efforts within individual medical schools. Indeed, several medical schools have demonstrated that targeted admissions and training experiences can produce impressive outcomes toward meeting workforce challenges. For example, graduates of the Physician Shortage Area Program (PSAP) at Jefferson Medical College are approximately 10 times more likely to practice rural family medicine and 12 times more likely to practice any specialty in rural areas within the state than non-PSAP graduates of the school.40 The Rural Medicine Program in the University of New Mexico’s Department of Family and Community Medicine recruits rural and underrepresented minority students, who are more likely to remain in-state and practice in underserved communities.41 The University of California, Los Angeles/Charles R. Drew University Medical Education Program has a long-standing history of selecting medical students based on a demonstrated commitment to serving underserved populations. A study of this program found that, as of 2003, over half of Drew’s graduates were practicing in underserved areas designated as high-minority areas, high-poverty areas, rural areas, HPSAs, or MUAs.42 In addition, several new medical schools are seeking to address social need in their communities through recruitment efforts and innovative curricula (e.g., exposing students early in medical school to longitudinal clinical experiences tailored to the local community).43,44 Although the outcomes of these programs in new medical schools remain to be seen, they have clearly targeted important workforce needs.
It is important to note that this study was restricted to U.S. MD-granting medical schools. Enrollment at DO-granting medical schools has risen rapidly and is also contributing to increases in the physician workforce, meriting further research.17 Other limitations of this study include the self-reported nature of data in the MSQ, the decreasing response rates to the MSQ over time (from 87% for 1999–2001 to 75% for 2009–2011), and missing data on individual MSQ items. The possibility remains that nonresponding students differ from respondents.
As medical school expansion continues to unfold, it will be important for schools to consider how their plans for expansion align with the actual workforce needs of the communities they serve. Perhaps not all expansion is warranted; perhaps more is needed in some settings. The need for primary care physicians and physicians who are dedicated to caring for rural and underserved populations is acute and widespread. In the end, the success of medical school expansion in the United States will not be measured by the net increase in new physicians trained but by how well the next generation of physicians serves society by improving the health of all. Targeting the output of physicians toward identified workforce needs is an important first step.
Acknowledgments: The authors thank Casey Tilton of the Association of American Medical Colleges Center for Workforce Studies for administrative assistance with the manuscript.
1. Petterson SM, Liaw WR, Phillips RL Jr, Rabin DL, Meyers DS, Bazemore AW. Projecting US primary care physician workforce needs: 2010–2025. Ann Fam Med. 2012;10:503–509
2. Bodenheimer T, Pham HH. Primary care: Current problems and proposed solutions. Health Aff (Millwood). 2010;29:799–805
4. U.S. Department of Health and Human Services. Health Resources and Services Administration, Bureau of Health Professions. The Physician Workforce: Projections and Research Into Current Issues Affecting Supply and Demand. 2008 http://bhpr.hrsa.gov/healthworkforce/reports/physwfissues.pdf
. Accessed September 4, 2013
5. Kirch DG, Henderson MK, Dill MJ. Physician workforce projections in an era of health care reform. Annu Rev Med. 2012;63:435–445
7. Bach PB, Pham HH, Schrag D, Tate RC, Hargraves JL. Primary care physicians who treat blacks and whites. N Engl J Med. 2004;351:575–584
8. Gazewood JD, Rollins LK, Galazka SS. Beyond the horizon: The role of academic health centers in improving the health of rural communities. Acad Med. 2006;81:793–797
9. Shi L, Macinko J, Starfield B, Wulu J, Regan J, Politzer R. The relationship between primary care, income inequality, and mortality in US States, 1980–1995. J Am Board Fam Pract. 2003;16:412–422
11. West CP, Dupras DM. General medicine vs subspecialty career plans among internal medicine residents. JAMA. 2012;308:2241–2247
12. Smith SR. A recipe for medical schools to produce primary care physicians. N Engl J Med. 2011;364:496–497
13. Bickel J, Brown AJ. Generation X: Implications for faculty recruitment and development in academic health centers. Acad Med. 2005;80:205–210
14. Holmes DE, Lindeman LM Proceedings of the 9th Congress of Health Professions Educators: Faculty Shortages Across the Health Professions. 2003 Washington, DC Association of Academic Health Centers eds.
16. Mallon WT. Medical school expansion: Déjà vu all over again? Acad Med. 2007;82:1121–1125
19. Goodman DC. The physician workforce crisis: Where is the evidence? Health Aff (Millwood). 2005;suppl Web exclusives:W5–W108
20. Lakhan SE, Laird C. Addressing the primary care physician shortage in an evolving medical workforce. Int Arch Med. 2009;2:14
21. Lindsay D, Bazemore AW, Bowman R, Petterson S, Green LA, Phillips RL. Will medical school expansion help diversify the physician workforce? Am Fam Physician. 2007;76:38
22. Bland CJ, Meurer LN, Maldonado G. Determinants of primary care specialty choice: A non-statistical meta-analysis of the literature. Acad Med. 1995;70:620–641
23. Lawson SR, Hoban JD. Predicting career decisions in primary care: A theoretical analysis. J Cont Educ Health Prof. 2003;23:68–80
24. Senf JH, Campos-Outcalt D, Kutob R. Factors related to the choice of family medicine: A reassessment and literature review. J Am Board Fam Pract. 2003;16:502–512
25. Xu G, Veloski J, Hojat M, Politzer RM, Rabinowitz HK, Rattner SL. Factors influencing primary care physicians’ choice to practice in medically underserved areas. Acad Med. 1997;72(10 suppl 1):S109–S111
26. Chen F, Fordyce M, Andes S, Hart LG. Which medical schools produce rural physicians? A 15-year update. Acad Med. 2010;85:594–598
27. Rabinowitz HK, Diamond JJ, Hojat M, Hazelwood CE. Demographic, educational and economic factors related to recruitment and retention of physicians in rural Pennsylvania. J Rural Health. 1999;15:212–218
28. Rabinowitz HK, Diamond JJ, Veloski JJ, Gayle JA. The impact of multiple predictors on generalist physicians’ care of underserved populations. Am J Public Health. 2000;90:1225–1228
29. Saha S, Shipman SA. Race-neutral versus race-conscious workforce policy to improve access to care. Health Aff (Millwood). 2008;27:234–245
30. Komaromy M, Grumbach K, Drake M, et al. The role of black and Hispanic physicians in providing health care for underserved populations. New Engl J Med. 1996;334:1305–1310
31. Cochran WG. Some methods of strengthening the common χ2 tests. Biometrics. 1954;10:417–451
32. Armitage P. Tests for linear trends in proportions and frequencies. Biometrics. 1955;11:375–386
33. Hyer J, Bazemore AW, Bowman R, Zhang X, Petterson S, Phillips RL. Medical school expansion: An immediate opportunity to meet rural health care needs. Am Fam Physician. 2007;76
34. Rosenberg L. Physician–scientists—endangered and essential. Science. 1999;283:331–332
35. Neilson EG. The role of medical school admissions committees in the decline of physician–scientists. J Clin Invest. 2003;111:765–767
36. Scott I, Gowans M, Wright B, Brenneis F, Banner S, Boone J. Determinants of choosing a career in family medicine. CMAJ. 2011;183:E1–E8
37. Brooks RG, Walsh M, Mardon RE, Lewis M, Clawson A. The roles of nature and nurture in the recruitment and retention of primary care physicians in rural areas: A review of the literature. Acad Med. 2002;77:790–798
38. Rabinowitz HK. Recruitment, retention, and follow-up of graduates of a program to increase the number of family physicians in rural and underserved areas. N Engl J Med. 1993;328:934–939
40. 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
41. Pacheco M, Weiss D, Vaillant K, et al. The impact on rural New Mexico of a family medicine residency. Acad Med. 2005;80:739–744
42. Ko M, Heslin KC, Edelstein RA, Grumbach K. The role of medical education in reducing health care disparities: The first ten years of the UCLA/Drew Medical Education Program. J Gen Intern Med. 2007;22:625–631