The national problem of medically underserved rural populations is closely related to the traditions of medical education.1 Many studies support the affinity model where a student who is from a rural area and then is trained in less urbanized areas is more likely to return to a small town to practice.2 However, the majority of medical students’ experience is in larger, usually urban, environments.
Despite a recent slight increase in the number of primary care physicians trained by U.S. medical schools, the number of physicians practicing in nonurban areas has changed very little.3 A recent report noted that only 3% of current medical students plan rural practice careers,4 whereas 20% of Americans live in rural areas.5 This problem of physician maldistribution is especially large in Kentucky, with 59 of the 120 counties classified as rural6 and almost 68% of all 120 counties considered to be health professional shortage areas (HPSAs).7 Kentucky studies also support the affinity model.8,9 In fact, studies show that doctors from Kentucky are particularly influenced by a sense of place that draws them back to be near their childhood home.10 This also applies to some medical students, who report that even a ground travel time of two to three hours to reach their extended families had an effect on their choice of medical school campuses.11
The pipeline approach to attempt to increase the production of rural physicians begins with high school or earlier and continues through the retention of rural physicians in practice. This pipeline is at best very “leaky,” with many opportunities along the way for rural students to become attracted to big-city life during their education, causing “urban disruption” to the affinity model. Many medical education pipeline programs have been implemented to attempt to increase the number of rural students who successfully apply to and complete medical school, but much of this activity is paradoxically in urban areas because of the already-existent location of faculty, support staff, patients, and facilities, often including the parent university. More recently, some medical schools have placed clinical campus resources in smaller towns to try to take full advantage of the affinity model.
Published results from special rural programs include the University of Missouri School of Medicine Rural Track Pipeline Program, which allows rural track students to do up to three of the required third-year rotations in rural sites, with the majority of their training in Columbia (approximate population 85,000).12 Michigan State University manages a similar program in the Upper Peninsula. After completing their first and second years in Lansing (approximate population 90,000), medical students relocate to one of six community-based campuses. Up to eight students per year relocate to Marquette (population 22,000) in the Upper Peninsula, doing at least one eight-week rotation in towns of 1,929 to 13,140.13 A program in Minnesota has targeted rural admissions, with the first and second years in Duluth (approximate population 87,000), the clinical years in Minneapolis, and the option to select a nine-month elective in rural Minnesota during the third year.14 The nine-month option is termed the Rural Physician Associate Program.15
The Illinois Rural Medical Education program has a rural admissions initiative and provides a rural health curriculum across all four years of medical school.16 The first year is based in Urbana-Champaign (combined approximate population 104,000), and then students spend the second through fourth years in Rockford (approximate population 150,000), except for a 16-week fourth-year rural rotation in very small towns. A program at the State University of New York places third-year students with rural physicians for nine months, with the remainder at the Syracuse (approximate population 147,000) or Binghamton (approximate population 47,000) campuses.17 A very successful, well-established program at Jefferson Medical College includes targeted rural admissions, financial aid, active rural mentoring, and a six-week required rural clerkship.18 The students spend the remainder of their school time in Philadelphia.
One strategy for the educational pipeline approach is establishing regional medical school campuses that provide an opportunity for students to spend the entire last two years of clinical medical school training in nonurban environments. Most of these, however, are located in medium-sized communities. A recent survey showed that of 63 geographically separate medical school campuses, only 4 were in towns with populations smaller than 21,000.19 In 1998, the University of Louisville School of Medicine (ULSOM) established the Trover Campus (ULTC) in western Kentucky as an attempt to optimize the affinity model. In Madisonville, a town of 20,000 located 150 miles from Louisville, the Trover Foundation had developed a 40-year tradition of community training, hosting the first family medicine residency in the commonwealth.20,21
Rural Campus Description
The ULTC provides a small group of 6 to 10 medical students per year the opportunity to complete their third and fourth years of medical school in Madisonville. After completing their first two years of basic sciences in Louisville, ULTC medical students move to Madisonville and complete all their required clinical rotations, and most of their electives, there or in the smaller surrounding communities. The small class size was chosen to provide a personal connection between faculty and student22 and scaled appropriately to the number of faculty and patients available. It also matched the scale of the problem, as many of the Kentucky HPSA counties are underserved by only one to two physician full-time equivalents.7 So, a small program can, after a reasonable number of years, make an important impact on the problem. To address the affinity model, the ULTC pipeline programs were established beginning with high school and continuing through the second year, focused on recruiting students from rural western Kentucky and facilitating their academic success at each academic step while maintaining their connection to rural health.
ULTC students are based within a rural integrated health system with a 400-bed hospital staffed by 80 physicians representing primary and secondary care specialties. The system also includes 10 satellite clinics within a 30-minute drive that are in towns of 4,000 to 8,000 that host portions of clinical rotations.
Trover students participate in the same classroom lectures as the Louisville-based students by simultaneous live video connection. All curriculum elements, teaching materials, and evaluation systems are identical at the two campuses. Clinical rotations on the ULTC provide the opportunity for one-on-one learning with an experienced clinician preceptor. On about one-third of clinical rotations, a family medicine resident is on the teaching service as well. All third-year students regardless of rotation meet for small-group, problem-based learning sessions using the iterative process23 twice a month, facilitated by the ULTC associate dean. A longitudinal teaching skills program is in place to assist the community-based faculty in guiding the students through their required clerkships.24
During the time period reported here, students indicated their interest in placement at the ULTC during the fall of the second year, applied, visited the campus, and were interviewed. The Madisonville selection committee ranked the candidates, and offers were made to the selected students. Shortly after the time period reported here, the decision was made to begin dedicated admissions to the ULTC. The selection process was moved to the point of application for admission, so the student was admitted with assignment to the Louisville or Madisonville campus. Also at this point the pipeline programs were limited to the Madisonville-assigned students, termed the Trover Rural Track.
Using nearly complete data from both campuses, we report 10 years of data with a focus on educational outcome, specialty choice, and rural practice choice.
We collected data for all students who graduated from ULSOM from 2001 through 2010. To describe the characteristics of all graduates of the third- and fourth-year programs at each campus, we chose gender, race, rural hometown (as defined by high school being located in a city with a population less than 30,000), mother and fathers’ educational level and occupation (medical versus nonmedical field), and a nonscience major. We chose academic outcome measures based on previously reported data from rural programs.17,25–28 We compared Medical College Admission Test (MCAT) and United States Medical Licensing Examination (USMLE) Step 1 and Step 2 Clinical Skills (CK) scores as academic measures at each level of training. We calculated and reported simple descriptive frequencies, percentages, means, and standard deviations. For specialty and practice choice outcomes, we used measures previously reported.4,12,29
To address the question of comparability of training outcomes, we chose a noninferiority model to compare students who completed their clinical years at the Louisville Campus with those who were at the ULTC for both clinical years. In traditional hypothesis testing, the null states that no difference exists between the two educational groups. In a noninferiority model, the null hypothesis is that students at the ULTC would have lower scores than those at Louisville, and the alternate hypothesis is that their scores are statistically the same. For example, one might test the null hypothesis that the difference between the Trover and the Louisville Campus on an outcome would not go beyond a value that is so small that the scores essentially are equivalent.
We chose MCAT scores and USMLE Step 1 scores as descriptors for the two campus cohorts before they began their clinical training. We chose the scores on USMLE Step 2 CK, and shelf exam scores for third-year students in obstetrics–gynecology, surgery, and pediatrics, as measures of performance at the conclusion of each portion of the third- and fourth-year curriculum at each campus. Before data analysis, we canvassed the third-year clerkship directors for consensus that if the ULTC Clinical Program outcome scores were within one-half standard deviation of the Louisville Campus scores, the Trover students’ scores would be assumed to be educationally noninferior. We used the analysis of covariance to test for noninferiority of means of the two campuses with outcomes being adjusted by Step 1 scores. We established statistical significance at P < .025, and all analyses were one-tailed tests as set by convention for noninferiority models.30 PROC Mixed, a statistical algorithm in SAS version 9.3 (SAS Institute, Inc., Cary, North Carolina), was used for this analysis.
We matched the ULSOM residency match list to ULSOM graduates by campus for the classes of 2001 through 2010. Specialty choice was coded based on primary care (family medicine, internal medicine, pediatrics, or medicine/pediatrics) and separately for family medicine. After data were matched using Microsoft Excel 2003, they were imported into SPSS v. 20 (IBM Corp, Armonk, New York). We also matched data from ULSOM graduates from the classes of 2001 through 2006 who would have established practice to the American Medical Association Physician Masterfile for location of practice based on primary office address. We then matched practice site location to Rural Urban Continuum Codes (RUCC),31 and we used nonmetropolitan RUCC codes as a surrogate to identify rural/small-town practice.
We conducted cross-tabulations with chi-square analysis to examine the relationship between specialty choice and campus. We calculated relative risk in order to assess the likelihood of choosing a specific residency. We also conducted cross-tabulations with chi-square and risk ratios to examine the relationship of the two campuses to metropolitan versus nonmetropolitan practice site location.
Demographics comparison by campus
The population was 1,331 students from the Louisville Campus and 60 students from the ULTC, or all of the graduates during this time period. Table 1 describes the demographic data for the two cohorts. ULTC students were almost twice as likely to be African American as Louisville students, 13% (8/60) versus 7% (94/1,331), and a greater percentage of Trover students came from rural hometowns than did the Louisville students, 68% (41/60) versus 41% (544/1,331). The parents of Trover students had fewer years of school than their Louisville counterparts, and the Trover students’ fathers were almost three times less likely to work in the medical field, 7% (4/60) versus 20% (271/1,331).
MCAT, Step 1, and Step 2 CK comparisons
We compared the Trover and Louisville campuses on their MCAT, USMLE Step 1, and USMLE Step 2 CK scores. Trover students had lower scores on all MCAT components. Trover students scored lower on their USMLE Step 1 and Step 2 CK scores as well. However, the mean differences between the two campuses actually decreased from 7.83 (Step 1) to 5.50 (Step 2 CK), indicating that the ULTC students tended to close some of the gap after their clinical training when compared with their Louisville counterparts (Table 2).
Step 2 and shelf scores by campus
We then evaluated the Shelf and USMLE Step 2 CK scores using a noninferiority paradigm in which the difference between the Trover and Louisville campuses was compared with a predetermined, educationally critical difference level based on one-half standard deviation. If the adjusted mean difference score (see Table 3) was less than or equal to the critical difference level, then there was no true difference between the two campuses’ scores after controlling for USMLE Step 1 scores. The results of the noninferiority analysis indicate that for all three shelf exams and the USMLE Step 2 exam, the ULTC scores achieved equivalence with the Louisville Campus, P < .001. Only a difference of 0.64 or less was found between the two campuses on all shelf exams. The USMLE Step 2 CK difference was only 0.07.
As shown in Table 4, graduates who completed their clinical years at the ULTC were 40% (32/57 versus 518/1,294) more likely to choose a specialty in primary care (P = .016). Table 4 also examines those ULSOM graduates who chose family medicine as their residency of choice. Graduates from the ULTC were over four and a half times more likely to choose a family medicine residency than those graduates who remained in Louisville (21/57 versus 104/1,294). Thisassociation was significant at a level of P< .001.
Practice site choice
Table 5 shows that, for those physicians in practice from the graduating classes of 2001 through 2006, graduates from the ULTC were over six times more likely to choose a nonmetropolitan practice site than their Louisville Campus counterparts. Fifty-five percent (18/33) of the ULTC graduates chose nonmetropolitan practices compared with only 9% (66/759) of Louisville graduates (P < .001). Sixty one percent (20/33) of the ULTC graduates chose Kentucky practices compared with 49% (375/759) of the Louisville Campus graduates (relative risk 1.22, P = .208).
Our findings from the first 10 years of activities from a small rural regional medical school campus support the value of such campuses in addressing the maldistribution of physicians while providing a comparable academic experience. The key difference from many similar established programs is that almost all of the learning during both clinical years is in a rural area. The students who were chosen to complete their clinical rotations at the ULTC were more likely to be from rural areas and to have been the first in their family to have chosen a medical career. Although the ULSOM admissions committee during this time period did include among their priorities selection for those more likely to practice in an underserved area, they made no special effort to choose applicants more likely to choose the ULTC.
The finding that a higher percentage of African American students trained at the ULTC might at first seem paradoxical, as the majority of Kentucky’s underrepresented minority students reside in urban areas. However, ULSOM has an active recruiting effort that seeks URM students from other states, and in fact five of these eight students were from rural backgrounds and five were also from out of state. In the early years of the ULTC when there were not enough candidates from rural Kentucky to fill a class, these out-of-state students were drawn to the small class size and close contact with faculty that characterize the ULTC, and clearly flourished in the apprenticeship model. Their numbers were not large enough to significantly affect the other quantitative associations seen in our findings.
As they began their clinical years, the ULTC students also had lower quantitative measures of academic performance as measured by MCAT and USMLE Step 1 scores, as has been found in previous studies of rural medical students.17,18,25,26,32 However, also as seen in most of these previous publications, these differences faded by the time of the measures of clinical shelf exams and USMLE Step 2 CK. This supports the notion that a small environment with much patient contact and close relationships with faculty can provide a comparable clinical experience for these rural students and in fact allow them in some cases to “catch up” and even surpass the academic performance of their urban counterparts as measured on these examinations.
Practice location choice
When compared with the students who studied at the Louisville Campus, the ULTC students were clearly more likely to choose small-town practice. This was a dramatic difference, with 55% (18/33) compared with 9% (65/758) of those who completed their clinical years in Louisville choosing small-town practice. Although the absolute numbers are small, these 18 students are enough of a cohort to change 6 to 10 rural Kentucky countiesfrom being classified as HPSAs to better-served categories. And although our data have not been accumulating long enough yet to address retention, other studies have shown that students from rural-focused education programs tend to remain much longer in their chosen rural sites than other students, who may initially choose rural and then move back to an urban area.33 ULTC students tended to be more likely to choose to practice in Kentucky, 61% (20/33) compared with 49% (375/759) of those from the Louisville Campus. This difference did not reach statistical significance. This could be because a small true difference will require a larger sample size to discern or that in fact this outcome is the same for the two campuses.
Family medicine is the only specialty that distributes as the American population does, both with about 20% to rural areas.34 General internists seek rural practice at about 10%, and general pediatricians at about 9%.34 ULTC graduates chose family medicine at a rate almost five times higher than Louisville Campus graduates. They also chose primary care residencies at a slightly higher rate than Louisville graduates. The effect of the ULTC on the actual practice of primary care may be greater than that we report because of a classification bias. To have a large enough data set for analysis, we included graduates from both campuses in the specialty category that they chose at the time of the match. For family medicine, this classification as primary care would be stable into practice. For internal medicine, pediatrics, and combined medicine/pediatrics, some proportion of those will ultimately choose fellowships and subspecialty practice. Nationally, this percentage has ranged from 40% to 70% choosing subspecialty fellowships,35,36 which would probably apply to the Louisville graduates. Although the numbers are small, only 2 of 11 (18%) of the ULTC graduates who initially chose one of these three residency types have chosen subspecialty training.
As with all nonrandomized medical education interventions, all of the students included in this study were to some degree self-selected for their site of training. What can be concluded is that this rural campus was attractive to the target students and that their activities there maintained or increased their interest in rural practice, as supported by previous studies of the affinity model. The absolute number of students at the rural campus is small, but it is scaled to regional need and local resources. The design of the program can be useful to those implementing similar activities, but whether they will be similarly effective in different environments will have to be proven. The practice site database has some missing data, and we have designed a process where a staff member updates our campus graduate list annually, using Web-based searches and personal communication.
Important next steps for our campus development include enlarging class size and continuing innovative methods of faculty development for our actively practicing community faculty. Careful attention to outcomes of our pipeline programs will be our guide as to how we can enlarge class size while still attracting rural students who can succeed in our environment.
For those considering such a program, takeaway points we have discerned include the value of choosing a site with an existing tradition of medical education, including residency training. Clear commitment and leadership from main campus champions make the building of a new regional campus easier, and having a clinician known to local faculty serving as the academic officer at the regional campus has been critical to our success.
Few publications include outcomes of regional clinical campuses in any-size city, and the outcomes reported here are, to our knowledge, the first from a small rural campus. This report is also strengthened by the availability of main campus outcomes for comparison. We recommend consideration of including a small, rural, regional medical school campus among the important strategies to address the current maldistribution of physicians. We continue to collect data prospectively on our pathways participants, and as more enter practice, we can further evaluate the effects.
Acknowledgments: The authors thank Pam Carter and Shelia Justice, Trover Campus, and Emily Carr, Office of Medical Education–Louisville Campus, for their many hours of data retrieval and entry.
1. Geyman JP, Hart LG, Norris TE, Coombs JB, Lishner DM. Educating generalist physicians for rural practice: How are we doing? J Rural Health. 2000;16:56–80
2. 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
3. Blackman JGeyman JP, Norris TE, Hart LG. Predoctoral education for rural practice. Textbook of Rural Medicine. 2001 New York, NY McGraw-Hill:359–367
4. 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
5. 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
6. Price M. Kentucky population growth: What did the 2010 Census tell us? Kentucky Annual Economic Report. 2012 Lexington, Ky Center for Business and Economic Research, Gatton College of Business and Economics, University of Kentucky:17–25 cber.uky.edu/Downloads/CBERAnnRpt12.pdf
. Accessed April 17, 2013
8. Elam CL, Rosenbaum ME, Johnson MM. Geographic origin and its impact on practice location in Kentucky. J Ky Med Assoc. 1996;94:446–450
9. Looney SW, Blondell RD, Gagel JR, Pentecost MW. Which medical school applicants will become generalists or rural-based physicians? J Ky Med Assoc. 1998;96:189–193
10. Cutchin MP. Physician retention inrural communities: The perspective ofexperiential place integration. Health Place. 1997;3:25–41
11. Crump WJ, Barnett D, Fricker S. A sense of place: Rural training at a regional medical school campus. J Rural Health. 2004;20:80–84
12. Quinn KJ, Kane KY, Stevermer JJ, et al. Influencing residency choice and practice location through a longitudinal rural pipeline program. Acad Med. 2011;86:1397–1406
13. Brazeau NK, Potts MJ, Hickner JM. The Upper Peninsula Program: A successful model for increasing primary care physiciansin rural areas. Fam Med. 1990;22:350–355
14. Boulger JG. Family medicine education and rural health: A response to present and future needs. J Rural Health. 1991;7:105–115
15. 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
16. Stearns JA, Stearns MA, Glasser M, Londo RA. Illinois RMED: A comprehensive program to improve the supply of rural family physicians. Fam Med. 2000;32:17–21
17. Smucny J, Beatty P, Grant W, Dennison T, Wolff LT. An evaluation of the Rural Medical Education Program of the State University of New York Upstate Medical University, 1990–2003. Acad Med. 2005;80:733–738
18. Rabinowitz HK, Diamond JJ, Markham FW, Hazelwood CE. A program to increase the number of family physicians in rural and underserved areas: Impact after 22 years. JAMA. 1999;281:255–260
19. Mallon WT, Liu M, Jones RF, Whitcomb M Mini-Med: The Role of Regional Campuses in U.S. Medical Education. 2003 Washington, DC Association of American Medical Colleges
20. Polk HC Jr. Can AHES really influence the distribution of physicians? J Med Educ. 1977;52:633–638
21. Crump WJ, Fricker RS, Wiegman DL. A 10-year evaluation of the Trover Campus: Lessons learned for addressing the need for more rural physicians. J Ky Med Assoc. 2010;108:137–143
22. Mumford DB. Clinical academies: Innovative school–health services partnerships to deliver clinical education. Acad Med. 2007;82:435–440
23. Kassirer JP. Teaching clinical reasoning: Case-based and coached. Acad Med. 2010;85:1118–1124
24. Crump WJ, Hughes Miller K, Fricker RS, Patel P, Ostapchuck M. An innovative technique for faculty development at a rural Kentucky clinical campus. J Ky Med Assoc. 2012;110:274–282
25. Zink T, Power DV, Finstad D, Brooks KD. Is there equivalency between students in a longitudinal, rural clerkship and a traditional urban-based program? Fam Med. 2010;42:702–706
26. Verby JE, Newell JP, Andresen SA, Swentko WM. Changing the medical school curriculum to improve patient access to primary care. JAMA. 1991;266:110–113
27. Schauer RW, Schieve D. Performance of medical students in a nontraditional rural clinical program, 1998–99 through 2003–04. Acad Med. 2006;81:603–607
28. Waters B, Hughes J, Forbes K, Wilkinson D. Comparative academic performance of medical students in rural and urban clinical settings. Med Educ. 2006;40:117–120
29. Barrett FA, Lipsky MS, Lutfiyya MN. The impact of rural training experiences on medical students: A critical review. Acad Med. 2011;86:259–263
30. Lesaffre E. Superiority, equivalence, and non-inferiority trials. Bull NYU Hosp Jt Dis. 2008;66:150–154
32. 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
33. Rabinowitz HK, Diamond JJ, Markham FW, Rabinowitz C. Long-term retention of graduates from a program to increase the supply of rural family physicians. Acad Med. 2005;80:728–732
34. U.S. Department of Health and Human Services. . Agency for Healthcare Research andQuality. Primary Care Workforce Facts and Stats No. 3: Distribution of the US Primary Care Workforce. http://www.ahrq.gov/research/pcwork3.htm
. Accessed April 17, 2013
35. West CP, Popkave C, Schultz HJ, Weinberger SE, Kolars JC. Changes in career decisions of internal medicine residents during training. Ann Intern Med. 2006;145:774–779
36. Freed GL, Dunham KM, Jones MD Jr, McGuinness GA, Althouse LResearch Advisory Committee of the American Board of Pediatrics. . General pediatrics resident perspectives on training decisions and career choice. Pediatrics. 2009;123(suppl 1):S26–S30