The Roles of Nature and Nurture in the Recruitment and Retention of Primary Care Physicians in Rural Areas

A Review of the Literature

Abstract

Purpose 

A systematic review of factors associated with recruitment and retention of primary care physicians in rural areas.

Method 

Using PubMed and Medline databases, 21 quantitative articles analyzing recruitment and retention of primary care physicians in rural areas from 1990 to 2000 were found. To assess the methodologic strengths of these articles, a formal evaluation was conducted based on study design, study population, response rate, years studied, data source, and statistical methods (total possible score = 60 points). Studies were grouped by whether the factors assessed were related to pre-medical school, medical school, or residency.

Results 

A total of six studies (score range: 30-52) analyzed pre-medical school factors, 15 (score range: 30-52) considered medical school factors, and six (score range: 20-52) analyzed residency factors related to rural recruitment and retention. Pre-medical school factors such as rural upbringing and specialty preference were most strongly correlated with recruitment of physicians to rural areas. Training factors such as commitment to rural curricula and rotations, particularly during residency, were most strongly correlated with retention in rural areas.

Conclusions 

Although important gaps exist, scientific studies available to health educators and policymakers show there are predictable factors that influence recruitment and retention in rural areas. Policies for staffing rural areas with primary care physicians should be aimed at both selecting the right students and giving them during their formal training the curriculum and the experiences that are needed to succeed in primary care in rural settings.

Accuse not Nature! She has done her part; Do thou but shine!

JOHN MILTON, Paradise Lost

While 20% of the population of the United States resides in rural areas, less than 11% of physicians practice in rural communities.¹ Despite the fact that the supply of physicians has increased over the last 20 years, the percentage of physicians practicing primary care specialties has declined. Although the overall number of physicians practicing in rural areas has increased, the proportion relative to urban areas continues to decline. In sum, it continues to be difficult to attract and retain physicians in rural areas, particularly primary care physicians.² Because medical schools vary greatly in the likelihoods that their graduates will practice primary care in rural areas,³ and because access to health care in rural areas has an impact on the community, understanding factors that contribute to the recruitment and retention of physicians in rural areas is crucial if the populations in these areas are to be effectively served.

The question of whether nature or nurture is more responsible for human behavior has been argued since long before the modern age and seems appropriate to apply to the ongoing debate about the most important factors in forming physicians' practice preferences for primary care in rural areas. In this case, nature might be best defined as those factors present before medical school (e.g., a candidate's upbringing, specialty preference, and demographics) that may contribute to the selection of candidates who have preferences for rural practice. Nurture would be those aspects of training (e.g., curricula, faculty, rotations, or specialty tracks) both in medical school and during residency that may increase the proclivity of the student to enter and stay in a rural area. This concept⁴ provides a framework to review and evaluate recent important data on recruiting and retaining physicians for rural primary care. Because a number of excellent reviews of the factors that can influence the production of primary care physicians in general are already available,^5–7 this report focuses on quantitative studies that specifically discuss rural primary care.

METHOD

For our search of the literature published between 1990 and 2000 on physicians in rural and underserved areas, we used the National Library of Medicine's databases (PubMed and Medline), and we found over 17,000 references. Narrowing the search to articles on primary care in rural areas since 1990 resulted in nearly 800 articles. A third search excluding mid-level practitioners narrowed the results to approximately 400 articles. We used bibliographies of selected articles and personal communication with selected authors to extend the search. We screened the selected articles to identify those that included original quantitative data analysis of the associations between premedical school, medical school, or residency factors on outcomes of recruiting and retaining rural physicians. Qualitative studies, such as narrative program descriptions without comparative data and literature reviews, were excluded from the formal part of our analysis.

The 21 articles that met the study's criteria were scored for methodologic strength by four of the authors working jointly with a system similar to that used by Campos-Outcalt and colleagues.⁵ The scoring system was based on six categories (study design, study population and number of responses, response rate, years studied, data sources, and statistical methods), as suggested by Pathman et al.,⁸ and each article could receive a maximum of ten points in each category for a maximum total score of 60 points. Scoring within categories was as follows:

1. Study design: Randomized experimental = 10, cohort with controls = 6, cohort without controls = 4, cross-sectional = 2. To meet the definition of a cohort study, data on practice location or preference must have been collected from two or more points in time. A cohort study received one additional point for individual measurements at two points in time, and two additional points for individual measurements at three or more points in time.
2. Study population and number of responses: If the study's participants were schools or programs, points were assigned according to the number of responses: 100+ responses = 10, 71–100 responses = 8, 41–70 responses = 6, 11–40 responses = 4, two to ten responses = 2, one response = 0. If the participants were individuals, points were assigned according to the number of responses: 801+ responses = 10, 401–800 responses = 8, 201–400 responses = 6, 101–200 responses = 4, 26–100 responses = 2, fewer than 25 responses = 0.
3. Response rate: 91–100% = 10, 81–90% = 8, 71–80% = 6, 51–70% = 4, 31–50% = 2, less than 30% = 0.
4. Number of years studied: Points were awarded separately for years spanned by the study database and for time comparisons in the analysis of the data: ten or more data years = 6, six to ten data years = 4, two to five data years = 2, one data year = 0; and additional points were assigned for time comparisons in the analysis: three or more points in time = 4, two points in time = 2, no time comparison in the analysis = 0.
5. Data source: National dataset concerning objective characteristics = 10, questionnaire concerning objective data or subjective data with established validity and reliability = 6, questionnaire of subjective data with some validity and reliability checks = 2, questionnaire with no validity or reliability checks = 0.
6. Statistical methods: Points were awarded separately for the type of analysis and for the number of control variables: multivariate analysis = 7, stratified analysis = 4, bivariate analysis = 2, univariate analysis = 0. Studies with five or more control variables = 3, three or four control variables = 2, one or two control variables = 1, and no control variables = 0.

For purposes of this analysis, the interpretation of “rural” was left to the discretion of the authors of each article. Recruitment was defined as the enrollment of primary care physicians into rural practice. Retention was considered to be length of time in either the original rural community or any other subsequent rural location.

RESULTS

Overall, 21 articles met the selection criteria. Their scores ranged from 20 to 52 points (mean = 41.6). Six articles assessed the role of pre-medical school factors in recruitment, and three of these also addressed retention. Fifteen studies analyzed the effects of medical school factors, and six analyzed the effects of residency factors. Of the articles about medical school, nine discussed recruitment and ten discussed retention. Of the six concerned with residency-related factors, five discussed recruitment and three discussed retention. All studies were done using cohort or cross-sectional methods. The strongest studies in each of the three areas were cohort studies with control groups and multivariate analyses.

Pre—Medical-school Factors: The Role of Nature

Six studies^9–14 analyzed pre—medical-school factors that related to rural recruitment and retention (see Table 1). One of the most thoroughly studied factors has been physicians' place of upbringing, which appears to be a key factor in the decisions graduates make about their initial practice sites.^9,11,13,14 These studies consistently found that rural upbringing was positively associated with physicians' practicing in rural communities. For example, Rabinowitz and colleagues,⁹ in a methodologically strong study of the factors related to the recruitment and retention of physicians in rural areas of Pennsylvania, found that growing up in a rural area was the most important independent predictor of rural practice, occurring with 29% of such individuals. The only other factor strongly associated with rural practice in their study was the student's expressed plan to eventually become a family physician. When combined, these two factors were associated with a 36% likelihood of a graduate's practicing in a rural area, compared with 7% for individuals without these characteristics. Interestingly, this study found that neither of these factors, nor any of the other pre—medical-school factors studied, was associated with retention once a physician set up practice in a rural area.

Table 1: Quantitative Studies Published between 1990 and 2000 That Analyze Pre—Medical-school Factors in the Recruitment and Retention of Physicians in Rural and Underserved Areas, by Descending Total Score^*

Demographic factors such as age, gender, and race or ethnicity have recently been analyzed for their impacts on rural practice.^9–13 In a cohort study of nearly 2,000 physicians, for instance, Horner and colleagues¹⁰ found that physicians who initially located in rural North Carolina were more likely to be men and slightly older, but without significant racial or ethnic differences from their urban counterparts. While age, gender, and race or ethnicity were not predictors of retention in the study, instate undergraduate medical training was associated with greater rural retention. Using a multiple logistic regression model of 93 variables related to students' backgrounds and preferences, Rabinowitz et al.⁹ found no predictive value of age, gender, or race or ethnicity on likelihood of rural practice. Similarly, Looney and colleagues found no predictive value of gender, age, or marital status on likelihood of practicing in rural Kentucky.¹¹ Fryer et al., on the other hand, used multivariate analyses and found that being a man, but not ethnicity, was significantly associated with rural practice.¹³ West and colleagues also found that women graduates were significantly less likely to choose rural practice locations.¹²

A number of other pre—medical-school factors have been assessed for their predictive value during the last decade, although these factors have been less thoroughly studied. In the cohort study by Rabinowitz and colleagues, for example, univariate analysis demonstrated that the size of the student's undergraduate college and the level of his or her father's education were related to the initial selection of rural practice.⁹ In addition, specialty plans as a freshman and the size of the community the student planned to work in were predictors. A number of other variables, such as attending a public college, science grade-point average, verbal Medical College Admission Test (MCAT) score, and reading MCAT score were all positively associated with students' choosing rural practice and were statistically significant, although of low effect size. Again, it appeared by multivariate analysis that none of these factors was associated with longer retention once the graduate entered practice.

Medical School Factors: The Role of Nurture

Fifteen studies published within the last decade analyzed factors related to medical schools that affect rural practice decisions of students (see Table 2).^{3,9,10,14–25} The strongest were cohort studies with controls and stratified or multivariate analysis. Several studies looked at whether the type, size, location, funding, and emphasis of medical schools might impact rural primary care. In 1992, Rosenblatt and colleagues published the results of a cross-sectional study of U.S. medical school graduates from 1976 to 1985 who had subsequently completed residency, in which they found that a medical school's location in a rural state, public ownership, production of family physicians, and smaller amounts of funding from the National Institutes of Health (NIH) were each strongly associated with the production of more physicians practicing in rural communities.³ In 1998, Basco et al. found that public medical school ownership and active recruitment of students interested in careers as generalists were significantly associated with students' intentions to practice in rural areas.²² In contrast, in a recent survey by Pathman et al.¹⁷ of a group of physicians practicing in rural areas of the United States, none of the five medical schools' characteristics (public school, emphasis on rural medicine, emphasis on underserved-area health care, percentage of graduates trained in family practice, and NIH-funding dollar amount) or medical school training experiences were associated with longer retention in rural areas.¹⁷ Similarly, Horner et al.'s study of primary care physicians in North Carolina (noted above for pre—medical-school factors) also found that attending an in-state medical school did not correlate with a higher likelihood of rural practice.¹⁰

Table 2: Quantitative Studies Published between 1990 and 2000 That Analyze Medical School Factors in the Recruitment and Retention of Physicians in Rural and Underserved Areas, by Descending Total Score^*

Table 2: Continued.

Another variable within the control of a medical school is its commitment to curricula and rotations that focus on rural primary care. Many of the early studies of rural curricula and rotations demonstrated a higher proportion of graduates in rural practice. For instance, Verby and colleagues,²⁵ reporting on the Minnesota Rural Physician Associate Program, found that an intensive third-year medical school curriculum that focused on community teaching and preceptorship resulted in a 59% recruitment rate to rural areas among participants, compared with only 18% of peers who did not participate in the program. In Colorado, a voluntary rotation of students in rural preceptorship teaching has been shown by Fryer et al. to result in a higher likelihood of rural practice as well, although even with such rotations during medical school the eventual difference between those choosing rural practice and those choosing other practices was relatively small (16.4% versus 9.6%).¹⁸ This lower number of graduates in rural practice in Colorado likely reflects the shorter, less intense community experience compared with the Minnesota-type curriculum described by Verby et al.

More recent studies of medical schools' curricula and rotations have controlled for pre—medical-school factors and have produced important results. In Rabinowitz et al.'s longitudinal study of 20 years of graduates of the school's Physicians Shortage Area Program (PSAP), the authors used univariate analysis and found that participation in the PSAP program or in a junior- or senior-year rural curriculum correlated with current rural practice.⁹ However, when a logistic model was used to separate out rural background and specialty interest, rural curriculum was not an independent predictor of current rural practice. This finding was of particular interest because it helped explain the findings of two other studies of the same program that had found the PSAP students to be more likely to practice in rural areas.^16,19 In one of these articles, the authors studied 22 years of graduates of the PSAP program, and found that 103 (52%) of 200 students had gone into family medicine, compared with 13% of their non-PSAP peers.¹⁹ Of these, 42 (21%) were practicing in rural areas. The authors noted, however, a trend toward fewer family-medicine—trained physicians in the most recent classes.

Another issue that may affect whether students choose rural, primary care practice is the amount of debt they incur during medical training. Approximately 80% of medical students take out loans, and 25% have service obligations to repay scholarships or loans.²³ Some data suggest that a higher debt load upon leaving training is inversely related to the likelihood of practicing in a rural area.⁹ Pathman and colleagues found, however, that a greater need for financial assistance may actually increase the chance of rural practice by motivating more physicians into contracts with state and federal loan repayment and financial incentive programs.²³ One of the most important of these programs is the National Health Service Corps (NHSC). Studies throughout the 1990s have shown that, although this program has brought physicians into rural and underserved areas in high numbers, the turnover of physicians once they complete their obligation is high.^20,21,24 The 1992 cohort study of NHSC physicians by Pathman and colleagues showed that retention rates dropped dramatically after the first three to four years of service in the original rural practice, and were only 14.7% when assessed after an average of ten years.²⁰ Rosenblatt and colleagues' cohort study of NHSC assignees trained in family practice showed similar local retention rates after serving the obligated time, but when a broader definition of retention in underserved areas outside the original one was used, the retention rates were much higher (60%).²⁴ Further work from this group has shown factors associated with a longer retention of NHSC physicians include family practice specialty, longer required obligation, more recent graduation from medical school, and private school training.¹⁵

Residency Training Factors: Nurture Continued

Most data support the fact that the postgraduate residency experience is also important to the recruitment and retention of physicians in rural areas (see Table 3).^{10,17,21,26–28} As early as the 1970s, research showed a clear propensity for graduates of family medicine residencies, for instance, to practice in rural settings at a higher rate than other specialties.²⁹ Both the content and the location of training experiences appear to be important contributors to the likelihood of practicing in rural communities. By surveying a cohort of randomly selected physicians practicing in rural areas across the country in 1991 and again in 1996–1997, Pathman et al. found that participation in a rural rotation as a resident and graduation from a residency that emphasized health care in underserved areas were the only factors that predicted retention in rural areas.¹⁷ In this same study, the sense of “feeling prepared” for small-town living was more important for retention than was the feeling of preparedness for the practice of rural medicine itself. When asked in the 1996-1997 questionnaire to name the one training experience that had best prepared them for living in a rural community, participation in a rural health care experience as a medical student or resident was most often chosen (43%). A previous study by two of these authors had found no relation between participation in a rural residency and rural retention.²¹ However, the earlier study had looked at retention after ten to 20 years and used a sample of physicians working at externally-subsidized rural practices. This sample may not be representative of the same population in the more recent study. Also, other factors may affect retention more strongly than the residency experience after longer time periods, so no effect of residency could be detected in the data.

Table 3: Quantitative Studies Published between 1990 and 2000 That Analyze Residency Factors in the Recruitment and Retention of Physicians in Rural and Underserved Areas, by Descending Total Score^*

By gathering data from nearly all of the 367 family practice residency programs in the United States from 1994 to 1996, Bowman and Penrod found that programs that graduated more rural physicians tended to have (1) more required rural and obstetrical training months, (2) full or partial rural missions, (3) locations in states that were more rural, (4) emphasis on procedural training, and (5) the program director designated as the rural contact.²⁶ In this study, the population of the program site (city or metropolitan), the presence of a public hospital sponsor, and the percentage of faculty with rural practice experience were not statistically associated with the likelihood of rural practice. In their same study examining medical school factors, Horner et al. found that physicians who moved to North Carolina after completing residencies in other states were more likely to settle and remain in rural practice locations than were those with in-state residencies.¹⁰ Whether this result would apply to other states is unknown.

Several more recent approaches to increasing family medicine residents' interest in rural practice have been attempted. A particular type of family practice program called the One—Two Rural Residency Track (RRT) deserves specific note here.^{27, 28} The program, which requires residents to complete the first year of training in an urban center and years two and three in a distant, rural community, offers comprehensive, ongoing training in rural areas. Rosenthal and colleagues have surveyed the 13 RRT programs across the United States.²⁷ Seventy-six percent of the 99 graduates of these programs were practicing in rural areas, compared with 30% of all U.S. family practice graduates in 1996. In a small, cross-sectional follow-up study of graduates of the RRT programs between 1988 and 1997, 76% again were found to be practicing in rural locations, and 61% of these practiced in Health Professional Shortage Areas.²⁸ Seventy percent were in their original practice sites, and of those who had moved, 63% had gone from one rural location to another. Importantly, 72% of these respondents indicated their intentions to stay in their current locations indefinitely.

DISCUSSION

As Geyman and colleagues have recently stated, “the educational pipeline to rural medical practice is long and complex.”³⁰ Some of those involved with policy relative to rural physician recruitment and retention believe that the preference among physicians for rural practice “appears more highly associated with physician characteristics than by any aspect of their medical education,”¹¹ and that “increasing the number of physicians who grew up in rural areas is not only the most effective way to increase the number of rural physicians, but any policy that does not include this may be unsuccessful.”⁹ On the other hand, most controlled studies suggest that the value of pre—medical-school demographic data is limited^{9, 10} and that the experiences students have in medical school and especially residency training have a significant role in their decisions to practice or remain in rural areas.^{13, 17, 27, 29}

In an excellent review of the state of the subject in 1996, Pathman highlighted the fact that the failure of most early studies to control for students' characteristics before entering medical school has made it difficult to ascertain the role these factors have played in comparison with medical school family practice and rural curricula.⁴ While there have been several strong recent studies that have tracked sizeable cohorts over long periods of time and used sophisticated statistical techniques to analyze the data, many published papers have used smaller study groups or less robust designs. Although we feel these papers add important information to the literature on this subject, these limitations impede the ability to generalize their findings and create the opportunity for continued research to fill gaps in our knowledge base. In recent years, several articles have used multivariate methods and have been helpful in distinguishing the effects of nature and nurture on practice decisions.

Although the studies we reviewed varied in their designs and abilities to control for pre—medical-school back-grounds and attitudes, overall it appears that the training environments in medical school and residency are important in the students' eventual decisions about specialty and practice community. To support this fact, we think it is important to note that, although students who initially express a desire to go into family practice have a much higher likelihood of doing so, more than half of all students who end up in family practice residencies did not list family practice as their specialty choice when admitted to medical school.^{31, 32} This suggests the importance of medical school experiences in students' eventual specialty choices.

Clearly, medical schools that focus on family practice have higher numbers of graduates who go into the family practice specialty.³² Since a higher percentage of family medicine (residency) graduates go into practice in rural areas, it can be inferred that emphasizing general family practice, separate from rural curricula and rotations in medical school, would increase the supply of rural physicians. Although this inherently seems to make sense, the data to support a link between family practice rotations in medical school or family practice residencies and eventual rural practice is mostly indirect, and it merits additional research.

Studies do support the finding that residency programs that focus on family medicine with an integrated rural health component have more graduates who go on to practice in rural areas.²⁷ Programs with rural “missions,” located in more rural states, and with longer rural rotations, seem to produce more rural physicians. Among these factors, the presence and duration of rural rotations appear to be the best predictors of retention in rural areas, a finding that is likely due to students' being better prepared for what awaits them in rural practice.

We believe the above findings are important to health educators and policymakers, because it is likely that the challenge of providing physician coverage in rural areas will continue into the future. Factors associated with the initial recruitment and the retention of physicians in these areas seem to be different. Selection of candidates who are more inclined to practice in rural areas must be considered by medical school admission committees if we are to close the gap between the supply of and the demand for physicians in rural areas. Pre-admission surveys of students' attitudes and specialty interests could help direct the selection of medical students. Once students are selected, a nurturing atmosphere in primary and rural care should be initiated. Both rural-specific tracks, and attention to rural health issues in the traditional track can help prepare students for later training and practice experiences. Emphasis by family practice residency programs on the rural health experience has been generally even more effective than such rotations in medical schools. Such experiences should be encouraged by educators and sought by students interested in rural health.

Although our review focused on the issues of selection of candidates and their formal training, we also recognize the important role played by post-residency factors in the battle to recruit and retain rural physicians. Local factors, such as professional, lifestyle, and financial issues, play increasingly important roles in retention once the physician and his or her family reach the new community. Problems with professional isolation, long work hours, lack of specialty support, lack of educational opportunities, and lower financial reimbursement rates all have important impacts on the longevity of rural practice. Although many of the solutions to these issues are outside the purview of medical educators, recent studies show the importance of integration—of both the physician and the family—into the community.³³ Specific formal training related to living in rural communities³⁴ and in the unique needs of rural health and better instruments to measure true job satisfaction³⁵ should be studied and supported by educators. Internet-specific training during medical school and residency should be encouraged,³⁶ since telemedicine may assist with issues such as professional isolation and the need for continuing education.³⁷

Through attention to the needs of the populations in rural areas, and the physicians who desire to serve them, educators and policymakers have the opportunity to improve the quality of health care to a large and important part of our country's population. Solutions must be aimed at both selecting the right medical students and giving them during medical school and residency the content and rural-setting experiences that are necessary to both introduce them to and train them in rural primary care.