Background: Substantial numbers of people are medically underserved because of rural residence and/or economic circumstances. The mission of many medical schools is service to this group, so the ability to identify applicants likely to serve this population is valuable.
Method: In 2009, the authors asked graduates from their medical school, class of 1997 and forward, if they practiced in a medically underserved community in the past year. Variables obtained from medical school applications and scores from a survey of attitudes toward the underserved measured at matriculation were analyzed using logistic regression.
Results: Of 244 practitioners, 35% reported working in an underserved community. Rural background, older age (25+) at matriculation, and being a member of an underrepresented minority were independent, statistically significant predictors of practice in an underserved community.
Conclusions: Schools wanting to increase the number of practitioners caring for the underserved could consider older as well as rural and minority applicants.
Correspondence: Sharon Wayne, MPH, Office of Program Evaluation, MSC 08 4550, University of New Mexico, Albuquerque, NM 87131-0001; e-mail: SWayne@salud.unm.edu.
Providing medical care for the underserved is a complex and long-standing problem in this country. In 2007, nearly 45 million persons in the United States lacked health insurance, more than 96 million people lived in Medically Underserved Areas, and nearly 64.5 million resided in a Health Professional Shortage Area.1 Sufficient workforce capacity is essential to providing health care in medically underserved communities (MUCs),1 and one key to increasing capacity is the ability to identify characteristics of physicians likely to work in underserved areas. If these characteristics can be identified as early as application to medical school, institutions whose missions include addressing the needs of MUCs would benefit.
In 2000, Rabinowitz et al2 reported the results of a national survey studying predictors of generalist physicians' care of underserved populations. They identified four independent predictors: (1) identifying oneself as a member of an underserved ethnic or minority group, (2) growing up in a rural or inner-city area, (3) strong interest prior to medical school in practicing medicine in underserved areas, and (4) participation in National Health Service Corps. Of these variables, only the last is not identifiable prior to an individual's enrollment in medical school. Although published in 2000, subjects in the Rabinowitz study graduated from medical school in 1983 and 1984. The present study sought to determine whether the factors identified by Rabinowitz et al continue to be related to care of the underserved in more recent graduates. We also wanted to expand on Rabinowitz and colleagues' study by including all practitioners, as opposed to studying only primary care practitioners, and to include age and number of years since graduation in the analysis. The purpose of this study was, therefore, to investigate the characteristics of medical students that are identifiable on entry to medical school that influence subsequent practice in MUCs.
This study was conducted at the University of New Mexico School of Medicine (UNMSOM), a publicly funded school that enrolls 75 students each fall. Beginning with 1993 matriculants, all students were asked to complete a survey entitled Medical Students' Attitudes Toward the Underserved (MSATU)3 on entry to medical school. The MSATU assesses attitudes about medical care and underserved populations; it has good internal consistency (alpha >.80)4 and high agreement among researchers testing its content validity.3 The survey contains 37 statements ranked on a scale from 1 (strongly disagree) to 5 (strongly agree); a higher score indicates a more positive attitude toward the underserved. The MSATU score was standardized to a T-score with a mean of 50 and standard deviation of 10.
We used medical school application data to obtain demographic and Medical College Admission Test (MCAT) information on our students. To categorize students as having a rural or urban background, we determined whether the population of the city in which they attended high school had 450,000 people or more (using 2000 U.S. Census data). We chose this number because the largest city in New Mexico, Albuquerque, is this size and a majority of students at UNMSOM (73%) attended high school in New Mexico. Albuquerque is the only large city in our state; the next most populous city is Las Cruces with 74,000 people.
Annually, in June, we send a letter to all graduates of UNMSOM inquiring about their current practice location, type of practice, and whether they self-identify as working in an MUC in the past year. Graduates are asked to return the form in an enclosed, stamped envelope. Nonrespondents are sent two reminder letters spaced six weeks apart. Data reported in this study come from the 2009 mailing of this graduate letter.
We divided respondents into two groups based on whether or not they reported working in an MUC in the past year. We compared the two groups with respect to demographic variables, MCAT scores and attitudes toward the underserved as measured by MSATU. Variables that were significantly associated with working in an MUC in unadjusted analyses were included in a logistic regression model to determine their independent association with working in an MUC. The MSATU and MCAT scores were analyzed as continuous variables; all other variables were categorical. Age and graduation year were categorized because of their skewed distribution; in categorizing these variables we selected a cut point that divided the sample into two approximately equal groups. SAS 9.2 was used to analyze the data (SAS Institute, Cary, North Carolina). This research was approved by the UNMSOM human research review committee.
In 2009 we sent follow-up letters to 1,779 graduates and received responses from 1,260 (71%). We excluded 741 graduates who matriculated prior to 1993 because we have no MSATU data prior to that date, and we also excluded 191 graduates who were still in residency or fellowship or not practicing medicine, leaving 328 graduates who were practitioners. Of these, 68 (21%) left blank the question about working in an MUC, and 16 were missing data on ethnicity or high school city, leaving a final sample size of 244.
Eighty-five of 244 graduates (35%) reported working in an MUC in the past year (Table 1). With the exception of gender, there were statistical differences (P ≤ .05) between the groups for all demographic variables: Those working in an MUC were more likely to be underrepresented minorities, to have started medical school at an older age, to have gone to a high school in a city with a population less than 450,000, and to have graduated more recently. Graduates who reported working in an MUC had significantly higher scores (more positive attitudes toward the underserved) on the MSATU at matriculation than those not working in an MUC. The MCAT score did not differ by MUC group.
All variables that were associated with working in an MUC (P ≤ .05) in unadjusted analysis were included in a logistic regression model in order to determine the relative importance of each variable, controlling for the others. The odds ratios and associated P values are shown in Table 2. The adjusted results show that three variables are independently associated with practicing in an MUC (P < .05): being a member of an underrepresented minority, beginning medical school at age 25 or older, and attending high school in a city with a population less than 450,000. Although not statistically significant, there is a suggestion that more recent graduates are more likely to practice in an MUC (P = .09). With all variables in the model, positive attitudes toward the underserved were no longer associated with working in an MUC. The Hosmer-Lemeshow goodness-of-fit test5 showed that observed values were not significantly different from expected values (P = .23), suggesting the model adequately fits the data.
Practice specialty reported by graduates was categorized into four groups: primary care (family medicine, general pediatrics, and general internal medicine), psychiatry, surgery/anesthesiology, and all other specialties. Although it was not a specific goal of our study to include specialty practice in our analysis (since our focus is on identifying factors that are predictive prior to entry into medical school), these findings are reported for interest. Graduates who reported working in an MUC were most likely to be practicing in primary care (n = 56; 66%) followed by “all other specialties” (n = 14; 16%), psychiatry (n = 9; 11%), and surgery/anesthesiology (n = 6; 7%); comparable numbers and percentages for the non-MUC group were 72 (45%), 61 (38%), 2 (1%), and 24 (15%).
The results of our longitudinal study suggest that three factors identifiable on admission to medical school are strong and independent predictors of subsequent practice in MUCs: rural background, older age, and being a member of an underrepresented minority. Practitioners who attended high school in a city with a population less than 450,000 were three times more likely to report working in an MUC than those attending high school in a larger city. Physicians who began medical school at age 25 or older were almost twice as likely to report working in an MUC than those matriculating at a younger age; a difference of similar magnitude was observed for physicians who were members of an underrepresented minority compared with those who were not.
Studies of MUCs are hampered by definitions. Underserved communities are those that lack sufficient numbers of physicians or those where obtaining care is difficult because of economic circumstances. Much research has equated rural with underserved, and although most rural communities are underserved, this definition excludes the substantial underserved populations living in urban areas. Among studies that include both urban and rural areas in their definition of MUC, it is still defined a variety of ways; some examples are a high percentage of medically indigent patients in one's practice; practice in an area with a high percentage of minority residents; or practicing in a large city or rural area as opposed to a suburb or town. Although different definitions of MUC have been used, there is some consistency in the results: Both rural/inner-city upbringing and underrepresented minority status are associated with practicing in an MUC.2,6–8 Our study, using MUC as defined by the individual practitioner, supports these findings.
In addition to rural/inner-city upbringing and minority status, Rabinowitz et al also found that those with a strong interest in practicing in an underserved area prior to attending medical school were significantly more likely to provide care to the underserved.2 McDougle et al8 found similar results in a more recent study of 73 practitioners. Although we measured attitudes toward the underserved, not interest in practicing in underserved areas, we had expected attitudes to be associated with eventual practice, but in our study they were not. Keeping in mind that attitudes and interest are not the same, one possible explanation is that the other two studies were cross-sectional—they asked practitioners to recall their interest in the underserved at entry to medical school which was many years in the past. Our study benefits from its longitudinal design; although most variables we assessed do not change over time, attitudes may change, and in our study they were measured at entry to medical school as opposed to being recalled at a later date. Another possible explanation for the lack of association between attitudes and work in an MUC is the time at which attitudes are measured. Chirayath9 suggests that care of patients is influenced by attitudes, and it is possible that attitudes later in a student's education are more closely linked to caring for the underserved than attitudes at matriculation to medical school.
Another contribution of our study is the inclusion of all practitioners instead of limiting the analysis to primary care providers. We wanted to include all specialties because many physicians who were not working in an MUC were primary care practitioners and, conversely, about one-third of physicians who reported working in an MUC were not primary care practitioners. Our study also benefits from the diversity of graduates from our medical school in both age and ethnicity. Previous studies have not reported an association between age and MUC practice. Because many schools have small variability in age of students, these schools may not be able to assess the association between MUC and age.
There are several limitations to this study: Working in an MUC was self-defined, and the data are from one medical school and, therefore, may not generalize to other schools. Further, our method for defining rural background may be specific to states in the mountain west—states with moderate-sized urban areas, few suburbs, and large areas that are very sparsely populated. Using a definition of rural or inner-city to describe background does not apply because in New Mexico we have no large urban, inner-city areas.
The ability to identify applicants who are likely to practice in MUCs is valuable. Our study confirms that two of three factors identified in earlier research (rural background and minority status) remain strong predictors of practicing in an underserved area. Our findings also suggest that older-than-average applicants who meet other admission criteria may be good choices for admissions committees interested in increasing the number of their graduates practicing in underserved communities. A next step in increasing the number of physicians treating the underserved, or increasing the number of underserved patients treated by individual physicians, would be a better understanding of the rewards of and barriers to treating this group. We suggest that schools develop or expand tracking systems of their graduates. Periodic short surveys of graduates could increase our understanding of factors other than demographics that figure into treatment of the underserved.
This research was approved by the human research review committee at the University of New Mexico School of Medicine.
1Rosenbaum S, Jones E, Shin P, Ku L. National Health Reform: How Will Medically Underserved Communities Fare? Washington, DC: Geiger Gibson/RCHN Community Health Foundation Research Collaborative; July 9, 2009. Policy Research Brief No. 10.
2Rabinowitz HK, Diamond JJ, Veloski JJ, Gayle JA. The impact of multiple predictors on generalist physicians' care of underserved populations. Am J Public Health. 2000;90:1225–1228.
3Crandall SJ, Volk RJ, Loemker V. Medical students' attitudes toward providing care for the underserved. Are we training socially responsible physicians? JAMA. 1993;269:2519–2523.
4Crandall SJ, Volk RJ, Cacy D. A longitudinal investigation of medical student attitudes toward the medically indigent. Teach Learn Med. 1997;9:254–260.
5Hosmer D, Lemeshow S. Applied Logistic Regression. 2nd ed. New York, NY: Wiley & Sons; 2000.
6Brooks 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.
7Rabinowitz 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.
8McDougle L, Way DP, Rucker YL. Survey of care for the underserved: A control group study of practicing physicians who were graduates of The Ohio State University College of Medicine premedical postbaccalaureate training program. Acad Med. 2010;85:36–40.
9Chirayath HT. Who serves the underserved? Predictors of physician care to medically indigent patients. Health (London). 2006;10:259–282.