The American Medical Association has estimated that more than 35 million people live in medically underserved areas in the United States.1 Association of American Medical Colleges (AAMC) projections indicate that the United States will face a shortage of 91,500 physicians by 2020, which will disproportionately affect underserved populations.2 While demand for care is increasingly outpacing the supply of primary care providers and other physicians practicing in underserved areas, the elderly population and the burden of chronic illness continue to grow, as does the proportion of the U.S. population that consists of ethnic and racial minority groups.3 Furthermore, health care outcomes in the United States are strongly associated with the availability of primary care physicians.4
Strategies and proposals to alleviate the physician shortage have focused mainly on the recruitment of a medical school applicant pool with specific individual characteristics shown to be associated with career choice, including race/ethnicity that is underrepresented in medicine (URM) and a background of poverty.5 More recently, studies evaluating the impact of targeted programs to alleviate physician shortages in medically underserved areas, such as the Physician Shortage Area Program and the Charles Drew/UCLA Medical Education Program, have shown that a higher proportion of graduates of these programs are more likely to go on to practice in medically disadvantaged areas compared with nonparticipants.5,6 These findings are encouraging, but the broader impact of medical school experiences, outside of participation in special programs, on medical students’ intention to practice in underserved areas has not been examined.
Even though a significant proportion of students change their practice intentions during medical school, the factors associated with this change have not been investigated systematically.7 According to most career development theories, career outcomes are determined by the dynamic relationship between persons and their experiences.8 The social learning theory of career decision making (SLTCDM) framed by Krumboltz emphasizes that prior learning experiences (both direct and vicarious) affect people’s career decisions, interests, and values.9 According to SLTCDM, learners develop positive and negative attitudes and beliefs about career paths through their encounters with a broad array of external stimuli. Educational researchers also suggest that it is essential to consider the impact of the curriculum and educational experiences as well as individual characteristics when examining student outcomes.10–12 Examination of curricular content, including a focus on how much exposure students have had to specific skills and experiences during medical school, may help explain the etiology of changes in student career choice.
In this study, we used data from a national sample of students across 113 U.S. MD-granting medical schools to examine the association of individual student characteristics and educational experiences delivered in the context of the regular curriculum with changes in practice intention from matriculation to graduation. Specifically, we explored the types of educational experiences that are associated with positive changes in and with reaffirmation of medical students’ intention to practice in underserved areas at matriculation and at graduation. Given the various learning opportunities afforded within medical school curricula, we anticipated that students’ educational experiences would have a significant impact on their career intentions.
The AAMC administers annually the Web-based Matriculating Student Questionnaire (MSQ) and Graduation Questionnaire (GQ), respectively, to all first-year students and all graduating students enrolled in U.S. MD-granting medical schools. These questionnaires collect information on student background, practice plans, and educational debt; the GQ also asks students to evaluate the medical school experience.
For this study, the sample dataset was based on the population of 13,259 students from 125 medical schools who matriculated in the 2006–2007 academic year and graduated in the 2009–2010 academic year. Among these individuals, 7,793 responded to both the 2006 MSQ and 2010 GQ. (Students who participated in both surveys but who did not graduate within four academic years were excluded from the study population.) These 7,793 respondents came from 121 medical schools and exclude 26 students who responded to both surveys but whose schools differed at matriculation and graduation.
Because of concerns about data privacy, the sample was further reduced by 84 students because those students graduated from one of seven medical schools where ≤ 20 students responded to both the 2006 MSQ and 2010 GQ. In addition, 348 students who had missing data on the practice intention question (see below) were excluded from the analysis. The final study sample consisted of 7,361 students from 113 medical schools.
Sample weights were calculated based on the population and included in the analysis to minimize sampling bias. All the data were matched and deidentified by the AAMC. The institutional review board at the University of California, San Francisco approved the study design.
We measured change in practice intention on the basis of responses to the following question asked in both the 2006 MSQ and 2010 GQ: “Do you plan to locate your practice in an underserved area?” The three response options were “no,” “undecided,” and “yes.” In our analyses, we collapsed the response categories of “no” and “undecided” because our outcome of interest was whether or not students expressed intention to practice in an underserved area.
Change in practice intention.
To capture positive change, we identified the students who changed their practice intention from “no” or “undecided” at matriculation to “yes” at graduation; we compared these students with those who responded “no” or “undecided” at both matriculation and graduation.
Reaffirmation of practice intention.
In our second set of analyses, we focused on students who indicated intention to practice in underserved areas at matriculation. For these analyses, we compared the students who reaffirmed their practice intention at graduation (i.e., those who responded “yes” at both matriculation and graduation) with the students who changed their response from “yes” at matriculation to “no” or “undecided” at graduation.
The independent variables were individual student characteristics including sociodemographic variables, educational experiences, and school characteristics.
Individual student characteristics.
The sociodemographic information was based on student self-reported data from the 2010 GQ:
- Age: We included age of the respondent at the time of graduation as a continuous variable.
- Race/ethnicity: We categorized students into three mutually exclusive groups based on self-reported race/ethnicity: (1) white, (2) non-URM minority, and (3) URM. The URM group consisted of students who reported their race/ethnicity as black or Hispanic. The non-URM minority category consisted of students who reported Asian and “other” race/ethnicity. (Although Native American students should have been included in the URM category, the 2010 GQ categorized Native Americans into the “other” category. Thus, it was not possible to separate out the Native American students.)
- Total educational debt: We categorized reported total educational debt into tertiles as low (≤ $90,000), medium ($90,001–179,000), and high (≥ 179,001).
Educational experience variables.
We identified all of the 2010 GQ survey items that directly assessed the relevant educational experiences during medical school; these appeared in the General Education section of the questionnaire.
Curricular activities items related to increased exposure to cultural awareness and health disparities were as follows:
Indicate the activities that you will have participated in during medical school on an elective (for credit) or volunteer (not required) basis:
– Field experience in community health (e.g., adult/child protective services, family violence program, rape crisis hotline)
– Experience related to cultural awareness and cultural competence
– Learned another language in order to improve communication with patients
– Experience related to health disparities
Response options related to participation in curricular activities were dichotomous (“yes” versus “no”).
The following items measured perceptions on diversity using a five-point Likert-type scale (strongly disagree, disagree, neutral, agree, strongly agree):
Based on your experiences, indicate whether you agree or disagree with the following statements:
My knowledge or opinion was influenced or changed by becoming more aware of the perspectives of individuals from different backgrounds.
The diversity within my medical school class enhanced my training and skills to work with individuals from different backgrounds.
We dichotomized the response options into two categories (“agree/strongly agree” versus “strongly disagree/disagree/neutral”).
In addition to the MSQ and GQ student response data, we included two school characteristic variables: (1) school type (private versus public) and (2) the school’s social mission index score as developed by Mullan et al.13 Using 2008 AMA Physician Masterfile data, Mullan et al took the percentages of graduates who practiced primary care, worked in Health Professional Shortage Areas, and were underrepresented minorities and combined them into a composite social mission score. We included this variable because Mullan et al reported significant association between a school’s social mission score and its production of primary care physicians practicing in underserved areas. We dichotomized the schools in our study into those included in Mullan et al’s list of the 20 schools with the highest social mission scores (top 20) versus those not listed in the top 20.
We performed a descriptive analysis to examine the characteristics of the students and schools in the study sample. We conducted two different sets of analyses to address our two research questions. First, to determine which factors are associated with positive change in intention to practice in underserved areas, we conducted multilevel logistic regression analyses to examine the relationships between predictor variables (individual student characteristics, educational experiences, and school characteristics) and the outcome variable (positive change in intention) among students who responded either “no” or “undecided” at matriculation and “yes” at graduation. Second, to examine the factors associated with reaffirmation of intention to practice in underserved areas, we conducted multilevel logistic regression analyses to examine the relationships between predictor variables (individual student characteristics, educational experiences, and school characteristics) and the outcome variable (reaffirmation of practice intention) among students who responded “yes” at both matriculation and graduation.
For both sets of analyses, we examined the added value of educational experience and school characteristic variables by comparing models with just the individual student characteristic variables and the full models with inclusion of educational experience and school characteristic variables. We report descriptive statistics for each variable as well as adjusted odds ratios (ORs) and 95% confidence intervals (95% CIs) from the multilevel logistic regression models. All P values were two sided; P < .05 was considered statistically significant. We used Stata 12.0 (StataCorp, College Station, Texas) for all statistical analyses.
The characteristics of the 7,361 medical students who responded to both the 2006 MSQ and 2010 GQ and graduated in the 2009–2010 academic year are provided in Table 1. Of these respondents, 16% were categorized as URM. Respondents’ mean age was 27 years (SD 3 years), and their mean total educational debt was $137,008 (SD $89,277), with 6,405 (87%) reporting debt over $1,000 at graduation. Among the 113 schools in our study sample, more than half (n = 65; 58%) were designated as public and 14 (12%) were in Mullan and colleagues’13 list of the top 20 by social mission score.
Nearly 23% of respondents reported an intention to practice in an underserved area at matriculation, and this increased to 28% at graduation. Of the 23% who reported initial intent, 61% remained consistent in their response at graduation. As shown in Chart 1, 19% of the students who reported at matriculation that they were undecided or did not plan to practice in an underserved area indicated at graduation that they did intend to practice in an underserved area.
Factors associated with positive change in practice intention
In the logistic regression analyses, age, gender, URM status, educational experiences, and school characteristics were all significantly associated with positive change in practice intention (Table 2). Because of a multicollinearity issue between the experience related to health disparities variable and the experience related to cultural awareness and competence variable, the experience related to health disparities variable was omitted in the final models.
In terms of individual student characteristics, students who were older (OR: 1.05 per year; 95% CI: 1.03, 1.08), female (OR: 1.43; 95% CI: 1.24, 1.66), and categorized as URM (OR: 1.84; 95% CI: 1.49, 2.27) were all more likely to change their response concerning intention to practice in underserved areas from “no” or “undecided” at matriculation to “yes” at graduation.
All but one of the educational experience variables were significantly associated with positive change in practice intention. After controlling for individual student characteristics, students who had field experience in community health (OR: 1.36; 95% CI: 1.18, 1.57), had learned another language to improve communication with patients (OR: 1.41; 95% CI: 1.22, 1.63), had experience related to cultural awareness and competence (OR: 1.38; 95% CI: 1.21, 1.58), and reported becoming more aware of the perspectives of individuals from different backgrounds (OR: 1.24; 95% CI: 1.04, 1.48) were all more likely to indicate positive change in their practice intention.
At the school level, institution type (private versus public) was not associated with change in students’ practice intentions. However, students from schools in the top 20 for social mission score were more likely to indicate positive change compared with students from non–top 20 schools (OR: 1.66; 95% CI: 1.28, 2.16). These findings take into account both sociodemographics and educational experiences.
Factors associated with reaffirmation of practice intention
For students who indicated intention to practice in underserved areas at matriculation, gender, URM status, educational experiences, and school characteristics were all found to be significantly associated with reaffirmation of this practice intention at graduation (Table 2).
Consistent with the previous model, students who were female (OR: 1.57; 95% CI: 1.31, 1.89) and categorized as URM (OR: 2.12; 95% CI: 1.59, 2.82) were more likely to have consistency in their intention to practice in underserved areas. Age and amount of total educational debt were not significant factors for reaffirmation of practice intention.
In terms of educational experiences, students who had field experience in community health (OR: 1.24; 95% CI: 0.99, 1.53) or had learned another language to improve communication with patients (OR: 1.29; 95% CI: 1.01, 1.65) were more likely to reaffirm their intention to practice in underserved areas. After taking into account both the individual student characteristics and educational experiences, students from schools in the top 20 for social mission score were more likely to reaffirm their practice intention compared with students from non–top 20 schools (OR: 1.62; 95% CI: 1.09, 2.43).
In this study, we found that more-adaptable variables such as educational experiences can have a significant reinforcement effect as well as a positive influence on the practice intentions of medical students as they progress from matriculation to graduation. In addition to individual student characteristics, participation in community health, cultural awareness/diversity, and language-learning educational experiences were associated with intention to practice in underserved areas. One notable difference in the predictors between the students with positive change in intention to practice in underserved areas and the students with reaffirmation of this intention is the number of educational experiences associated with the outcome: Four of five educational experience variables were associated with positive change compared with two for reaffirmation. Although we acknowledge that causal inferences are not possible with the current set of analyses, one possible explanation for this result could be that students who enter medical school with the intention to practice in underserved areas are more likely than other students to be already informed and shaped by their background and earlier experiences with cultural awareness and diversity issues and thus may not be as influenced by educational experiences that facilitate these skills and attitudes. However, those students who initially indicated no intention to practice in underserved areas may have formed that decision based on only limited past experiences; thus, participation in educational experiences that facilitate increased cultural awareness may be more likely to influence such students’ career intention at graduation.
The association between educational experiences and practice intention may be attributed to increased competencies around cultural awareness in students who participate in these types of activities.14,15 Godkin and Savageau,16 for example, reported that preclinical medical students who participated both in domestic community service and in a six-week foreign language program abroad had greater knowledge and ease in working with patients from other cultures. A potential explanation for the association we found between cultural competency training—including learning another language, working in community health, and education on cultural awareness—and intention to practice in underserved areas may be related to increased self-efficacy around working with culturally and ethnically diverse patients. There is a growing body of literature to support the positive impact of cultural competency training on intermediate outcomes such as the knowledge, attitudes, and skills of health professionals.14
Given the current trends in the medical curriculum and the potential impact of cultural competency training on physicians, our findings may support increasing educational opportunities around cultural awareness. In addition, the association we found between the medical school’s commitment to social mission and students’ intention to practice in underserved areas provides support for school-level initiatives aimed at helping alleviate the physician shortage in those areas.
Individual student characteristics such as gender, race/ethnicity, and age also have significant effects on practice intention. Consistent with previous research, we found that female students, older students, and students with self-identified URM race/ethnicity were more likely than other students to indicate intention to practice in underserved areas. A previous study by Moy and Bartman17 found that minority physicians were more likely than other physicians to care for minority, medically indigent, and sicker patients.
In this study, we also found that total educational debt amount was not significantly associated with intention to practice in underserved areas, after controlling for individual student characteristics and educational experiences. These findings suggest that individual student characteristics such as race/ethnicity and socioeconomic status are significantly associated with debt amount and that debt may be a confounding variable. Although contradictory to some initial studies on this topic, this result also has been seen at a broader level in other studies that have reported that the amount of educational debt does not seem to be a significant factor in student career choice after controlling for other factors.18,19 This confounding may explain the inconsistent results in the literature on the association between debt and career choice.
On the basis of our findings, we propose that alleviating the physician shortage in underserved areas requires a two-pronged approach at the medical school level. Given the significant association we found between educational experiences and practice intention, we suggest that medical schools should provide more educational opportunities that expose students to cultural awareness and diversity issues. In addition, the significance of the effect of individual student characteristics on practice intention in this study emphasizes the importance of increasing minority student recruitment. Previous studies have outlined several strategies for improving this pipeline. For example, Alexander et al20 pointed out the importance of robust interventions at the college level to support URM student performance in gateway courses to increase the URM medical school applicant pool. Beacham et al21 suggested that the recruitment strategy needs to include making connections with URM students as early as middle school and high school. Winkleby’s22 report on the long-term outcomes of the Stanford Medical Youth Science Program for low-income high school students illustrated the potential impact of strong mentoring, college admissions preparation, and long-term career guidance on program success.
This study has several limitations. First, the sample was limited to those students who graduated in the 2009–2010 academic year and completed both the 2006 MSQ and 2010 GQ, or about 60% of the total population of eligible medical students. However, to correct for this sampling bias, we incorporated sample weights into the analyses. Second, the data are based on self-report, which is often linked to inaccuracy because of the social desirability effect. Also, “underserved area” was not defined in the questionnaires, which may have resulted in various interpretations of the term. However, statement of behavior intention has been shown to be highly predictive of future behavior and outcome,23 and previous research has shown that medical students who indicated intention to practice in underserved areas on the GQ were more likely to practice in underserved areas after graduation.6 Third, our use of archival data originally collected for other purposes may have limited the scope of our analysis, but we are confident that the significant alignment between our research questions and the available data provides adequate justification of our study method. Finally, we acknowledge the limitation of using the social mission score developed by Mullan et al13 to characterize a medical school’s commitment to social mission. As others have pointed out,24 this score focuses on workforce issues only and thus is very narrow in scope and definition. Also, one of the factors considered in this score is the percentage of graduates working in Health Professional Shortage Areas, and so it is not clear whether the score reflects the influence of a school’s social mission milieu on students’ intentions or is a proxy for schools that have a higher percentage of graduates working in underserved communities. Nevertheless, we decided to include the score in our analysis to empirically examine the association between certain school characteristics and the practice intentions of students in our study sample.
In summary, the factors associated with medical students’ intention to practice in underserved areas are multifaceted and include individual student characteristics as well as more-adaptable variables such as educational experiences and school characteristics. Our finding of a lack of association between total educational debt and practice intention suggests a need for further studies to improve understanding of the complex relationship between career choice and financial considerations. Additional research is also needed to gain a better understanding of specific curricular strategies that will provide the optimal educational experiences. This study suggests that, through targeted curriculum interventions and recruitment, medical schools can play an active role in alleviating the physician shortage in underserved areas.