The rural physician shortage, and its resultant impact on access to care, represents a long-standing and serious problem in the U.S. health care system.1–4 Although consensus exists regarding the need to increase the supply of physicians in rural areas, there is considerable controversy as to the relative importance of various policies to achieve this goal. The major factors related to whether or not physicians practice in rural areas can be categorized by whether they occur before medical school (e.g., background factors such as where students grew up, or their future career plans), during their training (e.g., medical school or residency), or after training (e.g., financial issues, practice characteristics, role of spouse or partner).1,5–10 Although the strongest evidence supports the significance of background factors,4,7,11,12 each of these areas has been found to be important. A serious limitation with almost all of the studies investigating postmatriculation factors predictive of rural practice, however, is their failure to account for the potentially critical role of physicians' background factors on rural practice outcomes. For example, studies looking at the effect of medical education or income variables on rural practice location invariably have failed to account for whether or not physicians were already predisposed to practice in rural areas. Yet, it is very difficult to accurately account for the importance of these background factors when studying various policies because there are few data sources where these factors have been prospectively collected.
The Jefferson Longitudinal Study of Medical Education (JLS) is one of the few long-standing and comprehensive databases available which contain the information necessary to help answer these questions. The JLS includes over six million pieces of data on background, performance, and other variables for all students and graduates of Jefferson Medical College (JMC) of Thomas Jefferson University beginning with the entering class of 1964.13 JMC has also had a long-standing commitment to producing rural physicians, which includes its rural Physician Shortage Area Program (PSAP), one of a small number of comprehensive medical school rural programs that has been successful in increasing the supply and long-term retention of rural physicians.14–17 Since 1974, the PSAP has preferentially admitted a cohort of medical school applicants with three factors related to their background and future career plans: growing up in a rural area, planning to practice in a rural area, and planning to practice the specialty of family medicine. These factors have been shown to be most important in the success of the PSAP, although the program also includes strong mentorship and required clinical curricular experiences for these students.6,14
To inform this issue and determine the relative importance of background factors in contributing to overall rural physician outcomes, we chose to examine the relationship between the background and career plans of all entering medical students from the JMC graduating classes of 1978–1982 and their subsequent rural versus nonrural practice locations 29 to 33 years later.
As part of the ongoing longitudinal study of JMC graduates,14,16,18,19 data were obtained on the three critical variables used to select students for the PSAP, all self-reported on the JLS Matriculation Questionnaire completed on entrance to medical school. These variables were growing up in a rural area (i.e., rural area or small town versus a nonrural area, i.e., city or suburb), entering medical school with plans to practice in a rural area (versus a nonrural area), and entering medical school with plans to be a family physician (versus all other specialties). These three variables have also previously been shown to be related to practicing rural primary care and to practicing in rural Pennsylvania.6,20 Because data for two of these variables (growing up rural and future plans for rural practice) were only collected for all JMC graduates from the classes of 1978–1982, we limited the study to those graduates.
As part of the longitudinal rural tracking of JMC graduates that occurs every five years, the 2007 practice locations for these graduates were obtained from the Jefferson Foundation (2007), which maintains an address dataset of JMC alumni that is frequently updated and has been found to be highly accurate.18 As in previous studies, office address (county) was used,14,16,18,19 and when this information was unavailable we used home address county assuming it was of similar rurality. As in our prior study,16 graduates were considered to be practicing in a rural county if their 2007 county was designated as rural on the basis of its Rural–Urban Density Typology (RUDT).21 Graduates in all other counties were considered urban.
We only included physicians with complete data available for all three factors and practice location in this study. A predictive score, defined as the number of positive predictors (0–3), was then determined for each physician, and the proportion of physicians with various predictive scores practicing in a rural area was analyzed. We then calculated relative risks (RRs) and 95% confidence intervals (CIs), and two-sided P values of less than .05 were considered to indicate statistical significance. We developed a logistic model and used the Hosmer-Lemeshow test to compare the predicted and observed outcomes. For all JMC graduates practicing in a rural area in 2007, the proportion with various predictive scores was also determined. We used SPSS version 16 (SPSS Inc., Chicago, Illinois) to perform our statistical analysis. This study was approved by the institutional review board of Thomas Jefferson University.
Of 1,111 JMC graduates from the classes of 1978–1982, 762 (69%) had complete data for all three predictive factors and 2007 practice address and were included in this study. Of these, 172 (23%) were practicing in rural areas. This is similar to the percentage of graduates practicing in rural areas who lacked complete data (but who had a 2007 location) and were, thus, not included in this study (21%, 55/263). The percentages of graduates with each of the three predictive factors are shown in Table 1.
A logistic regression showed that all three predictors were independently related to rural practice (P < .001). Of graduates with all three predictors, 45% (45/99; CI 35%–55%) practiced in rural areas; of those with two predictors, 33% (48/145; CI 25%–41%) practiced rural; of those with one predictor, 21% (42/198; CI 15%–27%) were rural; and of graduates without any predictors, only 12% (37/320; CI 8%–15%) practiced in rural areas (Figure 1). Compared with the reference group of graduates with no predictors, the RR of practicing rural was 3.9 (CI 2.7–5.7, P < .001) for those with three predictors, RR = 2.9 (CI 2.0–4.2, P < .001) for those with two predictors, and RR = 1.8 (CI 1.2–2.8, P < .01) for those with one predictor. To determine whether the number of background factors might be useful in making predictions for future physicians, a logistic model was developed. The predicted outcomes were similar with the observed outcomes (P = .86), indicating that the model was highly predictive. For physicians with each of the various combinations of predictors, Table 2 shows the proportions practicing in rural areas. Looking retrospectively, of all 172 physicians practicing rural in 2007, almost four-fifths had one or more predictors at matriculation; only 22% (37/172) had none of these three predictors (Figure 2).
Our findings show that three background and career plan variables that are known at the time of matriculation to medical school are strongly and independently related to rural practice three decades later. The relationship between the number of these predictors and rural practice appears to be linear, with the presence of more factors being much more predictive of rural practice. The magnitude of this relationship is substantial, similar to that for well-accepted multiple predictive risk factors for medical diseases (e.g., the 38% estimated rate of coronary heart disease in men across a 10-year period who had five risk factors: elevated systolic blood pressure, elevated total cholesterol, low high-density lipoprotein cholesterol, diabetes mellitus, and cigarette smoking).22 In addition, the 45% rural practice outcome of graduates in this study with all three factors is almost as great as the 53% to 64% rural outcomes from successful comprehensive medical school rural programs,17 further emphasizing the importance of these background factors. Significantly, very few students in our study without any of these three factors known at matriculation were actually practicing rural medicine 30 years later. In addition, when viewed retrospectively, relatively few currently practicing rural physicians lacked all three of these factors at matriculation. This is particularly important, given all of the other factors that occur after matriculation which are known to be related to rural practice. Our findings, however, suggest that only students who possess one or more of these three factors known on entrance to medical school are likely to practice in rural locations, irrespective of the future presence or impact of such things as a rural curriculum, residency location, income expectations, or spouse or partner. These same postmatriculation factors may also be related to why one-fourth of graduates practicing rural in 2007 did not have any of the three background factors we studied, although we did not address this question in the current study.
In analyzing data for students who had various combinations of predictors, those who both grew up rural and initially planned family medicine were similarly likely to practice rural (47%) as were those with all three factors (45%), although the number of these students who did not also initially plan a rural career was small.18 Also, students who only planned family medicine (but did not grow up rural or plan to practice rural) were not much more likely to practice rural (13%) than those without any predictors (12%).
These findings have major implications for the role of the medical school admissions process in producing rural physicians. Increasing the number of medical students who have one or more of these factors is likely to have a significant impact on the rural physician supply. Similarly, if students have none of these factors, little else may be very effective in achieving that goal. On the other hand, although admitting students predisposed to rural practice may be necessary for increasing the rural physician supply, it may not be sufficient. These students also need strong mentorship, rural curricular experiences, role modeling, and support, as suggested by the success of all of the comprehensive medical school rural programs.17 In addition, although of lesser priority, it is also important that some attention be focused on other students who are less likely to practice rural but express interest during medical school or residency, or have other rural predictors (e.g., a rural-raised spouse20).
Whereas the outcome used in this study was physicians in all specialties who were practicing in rural areas, we focused on matriculation career plans only for those planning family medicine. This was because freshman plans for family medicine have been shown to be predictive of rural primary care.6 In addition, because of the small populations often found in rural areas, family physicians form the core of the rural physician workforce and are essential to rural health care. Similarly, the decline in the number of physicians choosing family medicine represents a major challenge to the future rural physician supply.23 Nevertheless, additional information is needed regarding the rural outcomes of those planning other specialties, and we plan to address this question in a future study.
A limitation of this study is that it involved a single medical school. However, the substantial literature regarding the importance of rural background and career choice in rural practice, and the successful outcomes of a number of comprehensive medical school rural programs that are based on these factors, provide support for the generalizability of our findings. Because no standard definition of rural exists, we used the RUDT classification system to determine whether graduates were located in rural practice. The RUDT has been used nationally, it directly measures rurality (which most other commonly used classifications no longer do), and it has been shown in our prior studies to result in similar patterns of rural outcomes as other definitions.16,21 Finally, although the three factors we analyzed in this study are straightforward, easily collected, and strongly related to rural practice, other measures and alternative models might also be studied.
In summary, we have shown that three factors related to medical students' background and career plans that are known at the time of entrance to medical school are strongly related to practicing in a rural area three decades later. The importance of these factors suggests a much more limited role for other factors which occur after medical school matriculation and raises important questions regarding research that does not specifically account for the effect of background factors on rural practice outcomes.
The authors would like to thank the Jefferson Center for Research in Medical Education and Health Care, for access to the Jefferson Longitudinal Study of Medical Education; the Jefferson Foundation, for access to the address files of Jefferson Medical College alumni; and the Jefferson Medical College Office of Admissions, for their long-standing support of the Physician Shortage Area Program.
3. Institute of Medicine Committee on the Future of Rural Health Care. Quality Through Collaboration: The Future of Rural Health. Washington, DC: National Academies Press; 2005.
4. Rosenblatt R. Commentary: Do medical schools have a responsibility to train physicians to meet the needs of the public? The case of persistent rural physician shortages. Acad Med. 2010;85:572–574.
5. Cullison S, Reid C, Colwill JS. Medical school admissions, specialty selection, and distribution of physicians. JAMA. 1976;235:502–505.
6. Rabinowitz HK, Diamond JJ, Markham FW, Paynter NP. Critical factors for designing programs to increase the supply and retention of rural primary care physicians. JAMA. 2001;286:1041–1048.
7. 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.
8. Rosenthal TC, McGuigan MH, Anderson G. Rural residency tracks in family practice: Graduate outcomes. Fam Med. 2000;32:174–177.
9. Rosenblatt RA, Saunders G, Shreffler J, Pirani MJ, Larson EH, Hart LG. Beyond retention: National Health Service Corps participation and subsequent practice locations of a cohort of rural family physicians. J Am Board Fam Pract. 1996;9:23–30.
11. Ernst RL, Yett DE. Physician Location and Specialty Choice. Ann Arbor, Mich: Health Administration Press; 1985.
12. Wilson NW, Couper ID, De Vries E, Reid S, Fish T, Marais BJ. A critical review of interventions to redress the inequitable distribution of healthcare professionals to rural and remote areas. Rural Remote Health. 2009;9:1060. http://www.rrh.org.au/articles/subviewnew.asp?ArticleID=1060
. Accessed December 14, 2011.
13. Gonnella JS, Hojat M, Veloski J. AM last page. The Jefferson Longitudinal Study of Medical Education. Acad Med. 2010;86:404.
14. 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.
15. 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.
16. 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. Acad Med. 2011;86:264–269.
17. 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.
18. Rabinowitz HK. Evaluation of a selective medical school admissions policy to increase the number of family physicians in rural and underserved areas. N Engl J Med. 1988;319:480–486.
19. 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.
20. 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.
21. Isserman AM. In the national interest: Defining rural and urban correctly in research and public policy. Int Reg Sci Rev. 2005;28:465–499.
22. Kannel WB. Blood pressure as a cardiovascular risk factor: Prevention and treatment. JAMA. 1996;275:1571–1576.
The project described was supported in part by grant number D56HP08346 from HRSA / HHS. Dr. Rabinowitz was also supported in part by the Eakins Legacy Fund of Jefferson Medical College.
This study was approved by the institutional review board of Thomas Jefferson University.
The contents of this report are solely the responsibility of the authors and do not necessarily represent the official views of the HRSA / HHS.
This study was presented in part at the Association of American Medical Colleges Seventh Annual Physician Workforce Research Meeting, National Harbor, Maryland, May 6, 2011.