Michigan State University College of Human Medicine (MSU-CHM) was founded in 1966 with a vision and legislative mandate to serve the state of Michigan—that is, to “educate the students of Michigan in the communities of Michigan in order to practice in the state of Michigan.”1 Since then, MSU-CHM has employed a longitudinal, distributed model of clinical education, immersing students across the state for their third and fourth years of medical school.2
The diverse communities include major urban centers, remote rural areas, and both affluent and economically depressed regions. For example, the Upper Peninsula (UP) community campus is headquartered in Marquette, Michigan, a small city surrounded by a vast rural region that is comparable in size to many states; students learn in clinical settings throughout the UP. In contrast, the Grand Rapids campus is in the downtown center of Michigan’s second-largest city, and includes a network of tertiary referral hospitals.
Fifty years later, we have decided to retrospectively evaluate the effect of each MSU-CHM community campus on the physician workforce in the six communities where campuses were initially established, to better understand how the effect of each community campus varies with its educational focus and character. As other institutions are expanding their reach by developing new campuses, this evaluation can help leaders both in host communities and at parent institutions consider the potential effect of a regional campus on the local physician workforce.
Community-Based Medical Education
Educators have advanced many definitions of community-based medical education.3 Farnsworth and colleagues4 define the term as “activities that use the community extensively as a learning environment, where students, teachers, community members and representatives of other sectors are actively engaged throughout the educational experience in providing medical education that is relevant to community needs.” Here, we use the term to denote longitudinal education in distributed, regional medical campuses, reflecting the philo sophical premise that students best learn about community context from within communities.5 As U.S. medical schools expand, many are adopting longitudinal distributed models, often with the goals of graduating students who will practice primary care, serve the local community, and care for underserved populations.4 As of 2014, 45 medical schools were operating with regional campuses.6
Importantly, regional community-based campuses are dependent on local support. Medical educators rely on the goodwill of local physicians, hospitals, and patients to teach students. For the university system, community-based medical education can be cost saving, in part, because of the employment of existing local faculty and structural resources.4,7–9 For example, by embedding students in community hospitals and office practices, universities expand their supply of teaching physicians without directly employing these faculty. However, because the practice patterns of graduates cannot be assessed for years after institution–community partnerships have been established, new and expanding institutions may find that convincing local stakeholders to invest in undergraduate medical education (UME) is difficult.
Despite the expansion of community-based medical education, little evidence describes its effects on the local physician workforce. We are aware of no studies examining the relationship between students’ clinical education in distributed and varied communities and those students’ long-term practice locations, although related studies have been conducted. Recent studies have examined the effect of rural medical education. One study of an Australian medical school with a rural focus demonstrated that graduates were more likely to begin early practice near their clinical placement sites,10 and our previous study of a U.S. rural medical campus showed that a high percentage of graduates remained in the region over time.11 Other studies have highlighted the local, statewide, regional, and national benefits of both whole medical schools and graduate medical education programs.5,12–14 One study—of Indiana University School of Medicine—demonstrated that students who completed basic science education at a regional campus, rather than the urban Indianapolis campus, were more likely to practice in that rural region.15 A 2012 study exploring specialty choice demonstrated that graduates from regional medical campuses are more likely to match into family medicine than other graduates.16 Our study augments this previous work by examining the relationships between MSU-CHM graduates’ clinical educational campuses and their eventual practice locations and specialty choices. We intend for our study to provide insight into the long-term effect of community-based medical education.
MSU-CHM uses a longitudinal distributed medical education model that includes two basic science campuses (East Lansing, Grand Rapids) and clinical campuses located in various communities throughout the state. All students must meet the same admissions criteria; likewise, criteria for studying at different campuses do not differ. Following admission, students interested in the rural UP campus are interviewed to ensure program fit. All students indicate their preference for their initial basic science campus and their subsequent clinical campus; most receive their preferred placement. Following year two, students relocate to this clinical campus for the remainder of their UME.
We obtained American Medical Association (AMA) Masterfile data from 2011 for all MSU-CHM graduates from 1972 through 2006 and cross-matched these data with graduate demographic data from internal records. We excluded physicians who were retired, inactive, or deceased. Collectively, the graduates had studied at all of MSU-CHM’s original six clinical campuses (Flint, Grand Rapids, Kalamazoo, Lansing, Saginaw, UP). No formal pipeline programs or end-of-medical-school recruitment efforts were present during the years we studied.
We classified graduates’ practice specialties as primary care, high-need specialty, or all other specialties. We defined primary care as a primary specialty of family medicine, pediatrics, internal medicine, internal medicine/pediatrics, or general practice. If graduates who listed one of these as their primary specialty also listed a secondary specialty, we excluded them from the primary care group, with the exception of those whose specialization was likely to be an extension of their primary care training, such as geriatrics, adolescent medicine, or sports medicine; and who also self-identified as practicing primary care. Based on our review of the literature,17–25 we included the following in our classification of “high-need specialties”: all primary care physicians (as defined above), pediatric subspecialists, general surgeons, and psychiatrists.
Location of practice
We mapped and classified graduates’ practice locations. We cross-checked the practice locations of any graduates listed as residents, and all 2004, 2005, and 2006 graduates, with Internet searches using Google (Mountain View, California) to ensure that the Masterfile data had captured a practice, rather than a residency site. Similarly, we searched Google for physicians who did not list a work address. We excluded those whose practice locations we could not confirm.
Using ArcGIS Desktop Geographic Information Systems Software Version 10.1 (Esri, Redlands, California), we geocoded practice locations to the ZIP Code level. Next, we used 2006 Rural–Urban Commuting Area Code (RUCA) data to determine rurality. We considered graduates’ practice locations to be rural if the ZIP Code corresponded to a RUCA value of 4.0 or greater, per RUCA Categorization C.26 We considered graduates to be practicing within a Health Professional Shortage Area (HPSA) if any portion of their practice ZIP Code contained a Primary Care HPSA (per the Health Resources and Services Administration 2011 HPSA shapefile).27
We used Fisher exact tests to compare the proportion of graduates from each campus practicing in primary care, in high-need specialties, within 50 miles of their UME campus, in rural communities, in HPSAs, and in Michigan. We compared the data from each campus against a composite of all other campuses. We excluded graduates who practiced in the ZIP Codes of MSU-CHM’s clinical campuses and their teaching hospitals for HPSA comparisons to minimize the distortion of data. Some clinical campus ZIP Codes include an HPSA, while others do not, and physicians practicing in ZIP codes with both an HPSA and a clinical campus typically had academic appointments. We were primarily concerned with practice patterns of graduates who had left their home academic institutions.
We further analyzed campus differences using binary logistic regression to determine whether differences remained significant after controlling for available demographic variables known to be predictive of outcomes of interest. For primary care and high-need specialties, we included the following variables in the analysis: clinical campus, age at matriculation, rural origin, and gender.28 For HPSA and rural practice locations, we included the following variables: clinical campus, gender,29 race/ethnicity,30 and rural origin. For the race/ethnicity variable, we grouped graduates dichotomously by whether or not data sources identified them as members of an underrepresented minority population in medicine or URM (i.e., African American/black, Hispanic or Mexican American, Native American or Alaskan Native, or Puerto Rican).31 We included state of origin in the logistic regression model for in-state practice. We also visually compared the county maps of the footprints of each campus.
Finally, to assess generalizability, we compared the footprints of all MD-granting, nonmilitary U.S. medical schools that have well-established regional campuses (present for more than a decade) versus those that do not have well-established regional campuses. For this comparison, we counted the number of counties containing at least 70% of graduates within each institution’s state or geographic region, using Med School Mapper as a data source.32 We excluded institutions that had graduated fewer than 500 total graduates.
Michigan State University’s Institutional Review Board deemed this study exempt from review. We used SPSS Statistics version 22 (IBM, Armonk, New York) for all statistical analyses.
We analyzed 3,107 of 3,309 graduates (94%) and have provided demographic information and campus-to-campus comparisons in Table 1. We noted significant differences among graduates of different clinical campuses in gender, age at matriculation, URM status, in-state residence at matriculation, and rural origin.
Graduates of each campus tended to cluster around the metropolitan center of their campus (Figure 1). Flint, Grand Rapids, Kalamazoo, Lansing, and Saginaw graduates clustered around the counties including each of these cities. UP graduates gathered predominantly around the city of Marquette, the central educational site of the campus. In total, 20% of graduates (635 of 3,107) were practicing within 50 miles of their campus in 2011 (Table 2). Grand Rapids graduates were most likely (227 of 848; 27%) while Kalamazoo (83 of 554; 15%) and Saginaw (53 of 436; 12%) graduates were least likely to practice within 50 miles of their UME campus.
We want to highlight three regional findings. The UP campus focuses on educating physicians who will serve rural Michigan. Our results indicate that, consistent with this mission, 45% of UP graduates (76 of 168) practiced in rural communities, compared with 16% of all MSU-CHM graduates (488 of 3,107; P < .01). Our results show that UP graduates were also more likely to choose primary care (52% [88 of 168] vs. 44% [1,363 of 3,107], P = .03) and high-need specialties (65% [109 of 168] vs. 57% [1,781 of 3,107], P = .045). The results of our logistic regression analysis confirm this last finding: UP graduates remained significantly more likely to practice in high-need specialties (P = .04). UP graduates have had the strongest geographic effect on the UP.
According to our results, Grand Rapids graduates are least likely to practice primary care (39% [327 of 848] vs. 44% [1,363 of 3,107], P ≤ .01) or to choose high-need specialties (52% [437 of 848] vs. 57% [1,781 of 3,107], P < .01). They are less likely to choose rural practice than other graduates (13% [110 of 848] vs. 16% [488 of 3,107], P = .01) but more likely to practice in HPSAs (46% [392 of 848] vs. 44% [1,374 of 3,107], P < .01). Although, according to our logistic regression analysis results, the differences in likelihood of practicing in HPSAs and primary care do not remain significant, the rural practice and high-need specialty choice differences do persist. Grand Rapids graduates are also most likely to practice within 50 miles of their community (27% [227 of 848], compared with all graduates (20% [635 of 3,107]).
Our results show that Flint graduates are the least likely to practice in rural communities, compared with graduates from other campus communities (11% [59 of 539] vs. 16% [488 of 3,107], P = .01), and this difference persists in logistic regression analysis. Both Flint and Saginaw graduates have a regional effect on Wayne County, which includes Detroit. In contrast, Kalamazoo graduates concentrate more in counties with midsized urban communities, including Kalamazoo, Grand Rapids, Lansing, and Southeast Michigan.
Overall, 43% (1,328 of 3,107) of MSU-CHM graduates from the years studied were actively practicing in Michigan in 2011. Kalamazoo and Saginaw students were less likely to stay in Michigan than all other graduates (respectively, 38% [208 of 554] and 36% [157 of 436], P = .01 and P < .01). These differences persisted in multivariate analysis. Figure 2 illustrates all graduates’ distribution through the state as a whole. Taken together, the proportions of MSU-CHM graduates in each Michigan county grossly reflect the state’s population.
Notably, our results indicate that graduates of every campus are more likely to practice in HPSAs than other physicians. Med School Mapper indicates that 25.5% of Midwest physicians practice in HPSAs,32 whereas, according to our results, 44% of all MSU-CHM graduates studied (1,374 of 3,107) practiced in HPSAs in 2011. Our results indicate that, overall, 44% of MSU-CHM graduates (1,363 of 3,107) practice primary care.
Nationally, in 2016, 22 MD-granting U.S. medical schools had established regional campuses (present for more than a decade); 89 did not. Those with regional campuses had a larger geographic footprint within their state or region (mean 31 counties vs. 21 counties; P < .01).
This study adds to current knowledge about medical education by demonstrating that a community-based, longitudinal distributed model can benefit the regional and statewide physician workforce. Perhaps most important, every MSU-CHM campus had a clear effect on its regional physician workforce. Although MSU-CHM is only one medical school, our analysis of other institutions suggests that our findings are generalizable: medical schools with regional campuses have broader, long-term, positive geographic effects compared with single-campus institutions.
The findings are important for communities because access to local physicians, especially in primary care, is essential for public health. Physicians also make significant contributions to local economies, particularly in underserved communities, by staffing hospitals and outpatient care facilities, generating local employment.33–36 The study demonstrates that regional campuses not only support primary care3,5,16 but also support local and regional physician retention. The time and funds that a community spends on teaching medical students are a good investment; both help to build the community’s future.
The results of our study are also important for medical schools. Community-based regional campuses require enormous institutional will and investment to create and maintain.5 Institutional leaders may find these results useful for demonstrating the value of regional campuses to community partners, funding organizations, private donors, and states.
Many MSU-CHM campuses are in high-need areas of the state, and these communities face significant challenges. MSU-CHM has invested in these communities, and in turn, the communities have invested in the growth of the college. For example, the entire UP constitutes a rural, underserved region. Flint, Michigan, which has recently emerged from a financial crisis, is consistently ranked among the most dangerous U.S. cities and is in the midst of a public health crisis resulting from lead contamination of the city water supply.37 The number of students completing their UME at both campuses has increased over time, and the two regional campuses are now central locations for unique student programs.11,38 Immersion of students in underserved communities such as these likely contributes to the high proportion of MSU-CHM graduates who now practice in HPSAs and practice primary care.
Some major findings are consistent with the distinct local cultures and geographic features of each campus. Here we discuss the most important campus characteristics in relation to our results and consider their implications.
The UP campus has been successful in its mission of educating rural physicians who will choose high-need specialties and serve Michigan. We have previously published a comprehensive evaluation of the geographic effect of UP campus graduates,11 which, along with the results of this current study, adds to previous literature demonstrating that students educated in rural communities are more likely to practice rurally, practice primary care, and remain in the local community.10,15,39–41
Flint is geographically close to Detroit and shares many characteristics of this urban core. Both are large cities that grew during the prosperous years of the automobile industry and now face significant economic hardship. Saginaw, a smaller city, also shares a common history of economic development based on automobile manufacturing, followed by economic decline. Interestingly, our results show that Flint and Saginaw graduates are as likely to practice in Wayne County, which includes Detroit, as they are to practice in Genesee and Saginaw Counties (which include, respectively, Flint and Saginaw). We believe this finding is due to the similar urban, underserved characteristics of these three cities.
Grand Rapids is Michigan’s second largest city, and home to MSU-CHM’s largest clinical campus. It is the only campus with a tertiary referral center, which includes a children’s hospital, a cardiovascular subspecialty hospital, a cancer center, and a research institute. This medical complex has grown substantially in the last two decades, offering many opportunities for physician employment. The high local retention of MSU-CHM graduates in Grand Rapids aligns with this growth. The campus graduates’ lower rates of rural practice and high-need specialty practice reflect its subspecialty orientation.
Interestingly, the communities with enhanced graduate retention are also the communities that have expanded over time. Like Grand Rapids, the Flint campus has retained more MSU-CHM graduates within its 50-mile radius than average. Further, MSU-CHM has substantially increased its clinical, educational, and research presence at the campuses in both of these communities in the last decade. In contrast, both the Kalamazoo and Saginaw campuses have fewer practicing graduates than the MSU-CHM average. In the last decade, both of these campuses have ended their relationship with MSU-CHM. These campus–community relationships might have survived if local graduates had had a more visible effect. We believe that the presence of alumni may be important for both sustaining this community-based clinical model and facilitating growth. Although we did not measure whether our graduates serve as faculty at any campus, we believe many do, and that their work in these roles strengthens MSU-CHM’s presence in and relationships with communities.
Finally, there is a need for medical schools to strengthen the community-based model by further developing the scholarship of community-engaged medical education.5 A recent Best Evidence Medical and Health Professional Education synthesis describes the literature on community–medical school relationships as “heterogenous and largely idiographic,” characterized primarily by descriptive studies.3 Our study contributes to this scholarship by examining specific workforce outcomes, but is limited by the inclusion of only a few variables. Likely, many unexamined factors, including location of residency training,12 have contributed to our successes (and failures). Still, we know that MSU-CHM students have lived, studied, cared for patients, and volunteered in communities for decades. In the years ahead, it will be useful for medical educators to continue to measure and evaluate the best ways to learn from, respond to, and collaborate with communities where students become physicians.
Because we performed many statistical comparisons, we suggest caution to avoid overinterpreting results. Our study analyzes a single medical school, which limits generalizability; however, we addressed this by adding a small national evaluation. Our analysis used self-reported AMA Masterfile data, which could raise concerns regarding validity. To minimize errors, we confirmed practice locations of recent graduates.
We defined a practice as a HPSA practice if any portion of the practice ZIP Code included a Primary Care HPSA. This definition may be less stringent than other measures of underserved practice, limiting comparability.
For our study, we used a more stringent definition of primary care than some others have used. Specifically, we excluded graduates who began their careers in a primary care residency, but then later completed a subspecialty fellowship. We believe our narrower definition more accurately reflects the physician workforce,42 but it may limit comparability to other published outcomes. Our definition also included hospitalists, who constitute an increasing proportion of general internal medicine physicians.43
Although the study demonstrates a relationship between students’ medical school campus and eventual practice location, we cannot infer causality. As discussed above, MSU-CHM students indicate their preferred clinical community. We have not assessed students’ communities of origin, and many may have returned to practice in places where they grew up. Residency location is strongly associated with eventual practice location,12 but we did not assess or control for this factor in this descriptive study.
Consistent with other research, this study demonstrates that students’ long-term practice choices—that is, their specialties and practice locations—reflect their educational communities. Thus, establishing new medical schools—or expanding existing ones—in dispersed communities may have significant long-term implications for the local physician workforce. Community-based medical education may help increase the proportion of graduates who practice primary care, practice other high-need specialties, and serve urban and rural underserved populations. Thus, further developing community-based and community-engaged medical schools can help advance the social mission of medical education.44,45
The authors wish to thank the Robert Graham Center for Policy Studies in Family Medicine and Primary Care and Dr. Aron Sousa.
1. Mavis B, Sousa A, Osuch J, et al.The College of Human Medicine at Michigan State University: Expansion and reinvention. Acad Med. 2012;87:1705–1709.
2. Cheifetz CE, McOwen KS, Gagne P, Wong JLRegional medical campuses: A new classification system. Acad Med. 2014;89:1140–1143.
3. Ellaway RH, O’Gorman L, Strasser R, et al.A critical hybrid realist-outcomes systematic review of relationships between medical education programmes and communities: BEME guide no. 35. Med Teach. 2016;38:229–245.
4. Farnsworth TJ, Frantz AC, McCune RWCommunity-based distributive medical education: Advantaging society. Med Educ Online. 2012;17:8432.
5. Strasser R, Worley P, Cristobal F, et al.Putting communities in the driver’s seat: The realities of community-engaged medical education. Acad Med. 2015;90:1466–1470.
6. Association of American Medical Colleges. Official list of regional medical campuses (as of Sept 2014). https://www.aamc.org/members/grmc/resources/
. Published September 2014. Accessed April 21, 2017.
7. Norris TE, Coombs JB, House P, Moore S, Wenrich MD, Ramsey PGRegional solutions to the physician workforce shortage: The WWAMI experience. Acad Med. 2006;81:857–862.
8. Rackleff LZ, O’Connell MT, Warren DW, Friedland MLEstablishing a regional medical campus in southeast Florida: Successes and challenges. Acad Med. 2007;82:383–389.
9. Schwarz MRThe WAMI program: 25 years later. Med Teach. 2004;26:211–214.
10. Woolley T, Sen Gupta T, Murray RJames Cook University’s decentralised medical training model: An important part of the rural workforce pipeline in northern Australia. Rural Remote Health. 2016;16:3611.
11. Wendling AL, Phillips J, Short W, Fahey C, Mavis BThirty years training rural physicians: Outcomes from the Michigan State University College of Human Medicine Rural Physician Program. Acad Med. 2016;91:113–119.
12. Reese VF, McCann JL, Bazemore AW, Phillips RL JrResidency footprints: Assessing the impact of training programs on the local physician workforce and communities. Fam Med. 2008;40:339–344.
13. Association of American Medical Colleges. 2015 State Physician Workforce Data Book. https://www.aamc.org/data/workforce/reports/442830/statedataandreports.html
. Published November 2015. Accessed April 21, 2017.
14. Rosenblatt RA, Chen FM, Lishner DM, Doescher MPThe future of family medicine and implications for rural primary care physician supply. http://depts.washington.edu/uwrhrc/uploads/RHRC_FR125_Rosenblatt.pdf
. Published August 2010. Accessed April 21, 2017.
15. Brokaw JJ, Mandzuk CA, Wade ME, et al.The influence of regional basic science campuses on medical students’ choice of specialty and practice location: A historical cohort study. BMC Med Educ. 2009;9:29.
16. Liaw W, Cheifetz C, Luangkhot S, Sheridan M, Bazemore A, Phillips RLMatch rates into family medicine among regional medical campus graduates, 2007–2009. J Am Board Fam Med. 2012;25:894–907.
17. Dill MJ, Salsberg ESThe Complexities of Physician Supply and Demand: Projections Through 2025. Washtington, DC: Association of American Medical Colleges Center for Workforce Studies. https://members.aamc.org/eweb/upload/The%20Complexities%20of%20Physician%20Supply.pdf
. Published November 2008. Accessed April 21, 2017.
18. Basco WT, Rimsza MECommittee on Pediatric Workforce; American Academy of Pediatrics. Pediatrician workforce policy statement. Pediatrics. 2013;132:390–397.
19. Petterson SM, Liaw WR, Phillips RL Jr, Rabin DL, Meyers DS, Bazemore AWProjecting US primary care physician workforce needs: 2010–2025. Ann Fam Med. 2012;10:503–509.
20. Kim WJAmerican Academy of Child and Adolescent Psychiatry Task Force on Workforce Needs. Child and adolescent psychiatry workforce: A critical shortage and national challenge. Acad Psychiatry. 2003;27:277–282.
21. Etzioni DA, Finlayson SR, Ricketts TC, Lynge DC, Dimick JBGetting the science right on the surgeon workforce issue. Arch Surg. 2011;146:381–384.
22. Lynge DC, Larson EH, Thompson MJ, Rosenblatt RA, Hart LGA longitudinal analysis of the general surgery workforce in the United States, 1981–2005. Arch Surg. 2008;143:345–350.
23. Institute of Medicine; Board on Health Care Services; Committee on the Future Health Care Workforce for Older Americans. Retooling for an Aging America: Building the Health Care Workforce. 2008.Washington, DC: National Academies Press;
24. Scully JH, Wilk JESelected characteristics and data of psychiatrists in the United States, 2001–2002. Acad Psychiatry. 2003;27:247–251.
25. Colwill JM, Cultice JM, Kruse RLWill generalist physician supply meet demands of an increasing and aging population? Health Aff (Millwood). 2008;27:w232–w241.
26. WWAMI Rural Health Research Center. Rural–Urban Commuting Area Codes (RUCAs). http://depts.washington.edu/uwruca
. Accessed April 21, 2017.
27. U.S. Departement of Heatlh and Human Services. Health Resources and Services Administration. Shortage areas. http://datawarehouse.hrsa.gov/topics/shortageAreas.aspx
. Accessed April 21, 2017.
28. Bennett KL, Phillips JPFinding, recruiting, and sustaining the future primary care physician workforce: A new theoretical model of specialty choice process. Acad Med. 2010;85(10 suppl):S81–S88.
29. Chen F, Fordyce M, Andes S, Hart LGWhich medical schools produce rural physicians? A 15-year update. Acad Med. 2010;85:594–598.
30. Grumbach K, Hart LG, Mertz E, Coffman J, Palazzo LWho is caring for the underserved? A comparison of primary care physicians and nonphysician clinicians in California and Washington. Ann Fam Med. 2003;1:97–104.
31. Association of American Medical Colleges. Underrepresented in medicine definition. https://www.aamc.org/initiatives/urm/
. Published June 2003. Accessed April 21, 2017.
32. HealthLandscape. Med School Mapper. http://www.medschoolmapper.org/
. Accessed May 10, 2017.
33. Reeves A, Basu S, McKee M, Meissner C, Stuckler DDoes investment in the health sector promote or inhibit economic growth? Global Health. 2013;9:43.
34. Brooks L, Whitacre BECritical access hospitals and retail activity: An empirical analysis in Oklahoma. J Rural Health. 2011;27:29–38.
35. Doeksen GA, Loewen RA, Strawn DAA rural hospital’s impact on a community’s economic health. J Rural Health. 1990;6:53–64.
36. Ona L, Davis AEconomic impact of the critical access hospital program on Kentucky’s communities. J Rural Health. 2011;27:21–28.
37. Hanna-Attisha M, LaChance J, Sadler RC, Champney Schnepp AElevated blood lead levels in children associated with the Flint drinking water crisis: A spatial analysis of risk and public health response. Am J Public Health. 2016;106(2):283–290.
38. Wendling A, Schnuth R, Mavis B, Rappley MPreparing medical students to care for the underserved: Michigan State University’s Leadership in Medicine for the Underserved Program. Paper presented at: Society of Teachers of Family Medicine Conference on Medical Student Education; January 29, 2016; Phoenix, AZ.
39. Zink T, Center B, Finstad D, et al.Efforts to graduate more primary care physicians and physicians who will practice in rural areas: Examining outcomes from the University of Minnesota–Duluth and the Rural Physician Associate Program. Acad Med. 2010;85:599–604.
40. Rabinowitz HK, Diamond JJ, Markham FW, Paynter NPCritical factors for designing programs to increase the supply and retention of rural primary care physicians. JAMA. 2001;286:1041–1048.
41. Barrett FA, Lipsky MS, Lutfiyya MNThe impact of rural training experiences on medical students: A critical review. Acad Med. 2011;86:259–263.
42. Petterson S, Burke M, Phillips R, Teevan BAccounting for graduate medical education production of primary care physicians and general surgeons: Timing of measurement matters. Acad Med. 2011;86:605–608.
43. Kamerow D, Petterson S, Wingrove P, Bazemore AOnce a hospitalist, always a hospitalist? Paper presented at: North American Primary Care Research Group Annual Meeting; October 26, 2015; Cancun, Mexico. http://www.napcrg.org/Conferences/PastMeetingArchives/2015AnnualMeetingArchives/SearchEducationalSessions?m=6&s=15133
. Accessed May 16, 2017.
44. Morley CP, Mader EM, Smilnak T, et al.The social mission in medical school mission statements: Associations with graduate outcomes. Fam Med. 2015;47:427–434.
45. Mullan F, Chen C, Petterson S, Kolsky G, Spagnola MThe social mission of medical education: Ranking the schools. Ann Intern Med. 2010;152:804–811.