Bennett, Keisa L.; Phillips, Julie P.
Section Editor(s): Barr, Michael S. MD, MBA
Primary care educators in the United States face a critical challenge. Researchers have demonstrated that health systems based on primary care produce better health outcomes at less cost and have more capacity to reduce health disparities than specialty-centered care systems.1–4 Primary care physicians, especially family physicians, also disproportionately serve where access needs are greatest.5 With the current focus on expanding access to health insurance, the primary care workforce could become more strained as demand for care increases.6,7 Although advanced practice nurses, physician assistants, international medical graduates (IMGs), and osteopathic physicians have potential to reduce this gap, their valuable contributions will be inadequate.6 With the possible exception of nurse practitioners, all of these providers are specializing more, and their absolute numbers are small in comparison with the primary care shortage.6,8–10 Allopathic family physicians continue to provide the majority of care to underserved populations, especially in rural communities.6,8 If we are to realize our national vision for health—the provision of timely, appropriate, comprehensive, patient-centered care for all people at a cost we can afford—we must begin to rebuild the primary care physician workforce.
Over the last four decades, researchers have produced a rich literature on specialty choice. Effective strategies to recruit students into primary care careers, however, have remained elusive. In 2008, only 2% of internal medicine residents planned generalist careers,11 and only 7.4% of U.S. seniors matched in family medicine in 2009.12 Solutions to the primary care shortage that are proposed are often not supported by the literature.
We hypothesize that knowledge of the factors associated with primary care choice can be more effectively translated into interventions to build the primary care workforce, but this requires a focus on the process of career decision making, from early premedical considerations through postresidency choices. Few studies have used a conceptual model of career choice to guide their research and interventions, and these models have generally not been tested systematically. The purpose of this study was to develop a new conceptual model of physician primary care specialty choice using a critical evaluation of the literature. We aim to demonstrate more clearly gaps in current knowledge and identify specific points for future research.
Although this analysis is primarily a qualitative examination, we systematically searched the literature in order to build a robust, comprehensive conceptual model. There is a large volume of scholarship related to primary care specialty choice. Many studies were published during the 1990s, when primary care was becoming more popular with students and managed care emerged as a dominant market force. It is unclear how much relevance those studies have in today's market and culture, but several quality reviews helped summarize that wealth of information. Accordingly, we searched first for reviews published from 1990 to 2000. Beginning in 2001, which marked the last data included in a major review on specialty choice, as well as the approximate end of the primary care resurgence, we considered either original studies or reviews for inclusion.
We searched MEDLINE and EMBASE using the terms “primary care” AND “career” restricted to studies in English. In the first search for reviews only, EMBASE produced three results, while MEDLINE produced 84 articles. In the second search for all articles since 2001, EMBASE and MEDLINE both produced nearly 200 articles. One author scanned the titles and abstracts of all articles and eliminated those that did not concern the United States, were clinical in nature, did not apply to physicians, and (in the first search) were not review articles. The second author reviewed the list for agreement of relevance. Additional articles were identified by scanning reference lists of more relevant articles and repeating MEDLINE searches using newly identified key words. Because of the evolving nature of the literature and the limited precision of health science databases in performing nonclinical searches, it is difficult to document the organic course we pursued in accessing the breadth and depth of this literature. We continued to evaluate abstracts and articles until we felt a saturation point in new information had been met. Most of these hundreds of articles were applied to the development of the conceptual model, although many are not specifically referenced because of duplication of information or representation by a review article. In addition, although it is not peer reviewed, we included the 2009 report of the Robert Graham Center13 because of its quality and comprehensiveness. Finally, we specifically looked for sources germane to formulating a conceptual model of specialty choice, citing several nonreview articles published prior to 2001 and a number of editorials or commentaries that were particularly applicable to decision making. In the evaluation of these papers, we strongly considered applying a “strength-of-evidence” taxonomy to guide both our inclusion criteria and the readers' understanding of each study's quality. The available taxonomies, however, are designed for clinical quantitative studies and are particularly difficult to apply to a body of health services research that uses multiple methodologies including qualitative and theoretical work. We therefore chose to focus on applicability rather than strength of evidence in our search. Although most included studies focused on allopathic medical students, those concerning osteopathic students were also included. Studies specific to nurse practitioners, physician assistants, and IMGs were not included.
We formulated the model through a back-and-forth effort with mutual revisions and an informal cumulative process of consultation with colleagues. We honed the model through a modified Delphi approach in which the model's figure and table were sent to 11 volunteers from various training levels and specialties. Five of them returned anonymous feedback. This process was repeated (with four respondents) after the model was improved based on their suggestions. We also presented the model at professional meetings and an institutional meeting, where we received helpful insights and feedback. All of these processes contributed to the final model until we felt that saturation in professional peer input had been reached.
Review articles 1995-present
Six relevant review articles were included, the most recent published in 2003. All the reviews described inconsistent quality and heterogeneity of methods in the primary literature. Despite these limitations, some consistencies are evident (List 1). The most widely reported factors associated with primary care specialty choice were female gender, attendance at a publicly funded medical school, rural background or plan for a rural career, and lower expected income. Several components of the medical school curriculum have been reported in fewer reviews but are still consistently associated with primary care choice (List 1). The concept of a “hidden curriculum” that subtly discourages primary care choice through the culture of the academic health center, the example of role models, and curricular elements was often addressed in these reviews. It is a difficult concept to measure and one that overlaps with many of the more objective factors, but it is nonetheless an important global impression of the effect of the medical school experience and deserves further study.14–19
Primary studies since 2001
Results of these studies are summarized in List 2. On the whole, they strengthen the evidence that students who choose primary care are less interested in prestige or research.18,20,21 Several studies confirm the association between exposure to programs funded by Title VII of the Public Health Act (a federal grant program that supports educational efforts in primary care) and a higher percentage of primary care graduates.13,22 Several studies have attempted to generate new personality, interest, or values measures that would predict specialty choice at the beginning of medical school without any consistent success.23,24 One study, however, had promising results using a measure of “social compassion or consciousness,”5 confirming other research suggesting that students choosing primary care are likely to have altruistic personal values and a commitment to service.25–28
Finances also matter. A given specialty's residency fill rates have been linearly correlated with that specialty's expected income.29 This relationship is confirmed by the Robert Graham Center, which found that lifetime return on investment has substantial power in explaining specialty choice trends.13 Although medical student debt is frequently cited as a deterrent from primary care specialty choice, the above studies suggest that gross income may be more influential.30 In fact, most studies have shown no linear association between debt and primary care specialty choice, and some studies have indicated that students who choose primary care are more likely to have some amount of debt than those choosing other specialties.31,32 However, three large studies that used sophisticated statistical analysis and multivariate modeling to control for potentially confounding influences found that particularly high levels of debt are associated with a decreased likelihood of choosing a family medicine career.33–35 If a threshold of debt that deters students from choosing family medicine does exist, it is likely to become more relevant as both debt levels and the need for primary care physicians continue to increase.
Finally, the recent study conducted by the Robert Graham Center, which assessed actual (rather than intended) practice specialty as its outcome, strengthened the evidence that women, public medical school students, older and/or married students, those with exposure to Title VII programs in medical school, and those from rural backgrounds preferentially select primary care careers. This study also found that participation in rural electives, attending a school with more community linkages, giving higher ratings to the medical school primary care experiences, and interest in serving the underserved are all correlated with primary care choice. The study also confirmed that primary care physicians are more likely to graduate with debt and generally have more debt than their specialty colleagues.13
Innovative studies and development of the conceptual model
The purpose of this review of the literature was to integrate the information into a conceptual model that bridges established scholarship with opportunities for new understanding. The only prior true conceptual model was formulated by Bland et al14 in 1995. The Bland model improved on previous literature by emphasizing students' need to satisfy societal pressures and expectations of self and others. It also clarified that student perceptions of specialties, rather than actual characteristics, are what drive choices.12 However, the Bland model may have overemphasized the importance of formal medical education in shaping choices, and underemphasized longitudinal student interaction with the health care system.18,36,37 Students also highly value matching their interests and lifestyle goals with specialty characteristics, a concept which is not adequately captured in the Bland model.38,39
In working to address these limitations, we discovered evidence that students who eventually choose different specialties experience the career development process differently. Several studies strengthened prior observational evidence36,40,41 that students bound for primary care are more likely to have chosen their specialty before medical school. Compton et al20 surveyed a large number of students in 15 schools at matriculation, entry to wards, and just prior to residency match. In this study, family medicine was actually one of the most stable choices, with 23% of those originally most interested in the specialty maintaining that choice at all three time points, and another 8% choosing it both at the beginning and end of medical school. A more recent study found that about 20% of students described choosing the specialty they were matching into before medical school. Although the authors did not differentiate which specialties were more likely to be in this 20%, they determined that those students choosing a surgical specialty found medical training itself critical to their specialty choice, whereas nonsurgeons found it less important. Students headed to primary care residencies rated social consciousness values much higher in influence.5 Both of these studies also confirmed that many students who begin medical school undecided or with a another specialty preference eventually choose primary care, demonstrating that medical school experiences can indeed be powerful for certain students.5,20
Burack et al42 explored the issue of differential experience using qualitative methods. The authors noticed a striking difference in how primary-care-bound versus non-primary-care students defined a good “lifestyle,” with the former valuing a surplus of work opportunities and job flexibility, and the latter placing more emphasis on control of daily work hours and the ability to separate work and home life. Finally, Burack and colleagues found that 30% of students choosing primary care identified role models as an important influence, while only 5% of those choosing other specialties did so, a result consistent with other small studies.38,39,42
The health care environment is also important in shaping student specialty choices. Several studies done at the height of primary care popularity noted that job opportunities had emerged as a factor for those choosing primary care.39,43 In one of the few studies specific to this topic, primary care residency match rates were associated with increased public and decreased private health care funding over time.44 Other aspects of the health care environment that also may be relevant but have not been studied significantly include payer oversight versus physician autonomy, paperwork burden, risk of being sued, and the cost of malpractice insurance.
Taken together, these studies show that applying one conceptual model to all students misrepresents the decision-making process and may lead to incorrect assumptions and conclusions. With these principles in mind and the vast literature on associations available, we constructed the conceptual model seen in Figure 1. This model depicts in its center the student's course from matriculation to specialty choice, emphasizing four major, distinct pathways: those who matriculate and graduate committed to primary care, those who are interested in primary care but not committed and may go to either specialty group eventually, those who are truly undecided, and those matriculating and remaining committed to non-primary care. In this model, students can only be identified with one of the pathways in retrospect, meaning that we only know who was truly committed to a specialty track after the choice is made late in the fourth year (or after). Therefore, these pathways are consistent through time, and students cannot change from one to another.
The factors that influence student decisions over time surround the central process and are divided into demographics and predisposition, financial and lifestyle considerations, health care environment, choice process and identity development, student interests relative to perceived specialty characteristics, and curriculum and school experience. A well-established “pipeline” into the primary care pathway composed of students with more rural or underserved origins is acknowledged. Table 1 then explains how each category of influence may affect students in each pathway. We included in this table factors that have evidence in the literature, but did not limit it to established factors. One purpose of the table is to highlight factors that deserve further study and analysis.
To increase medical students' choice of primary care, we must understand that students are not a homogenous group. Neither the choice process nor the effect of a given intervention will be the same for different students. Previous descriptive studies demonstrate that students predisposed to primary care, those inclined toward non-primary care, and those who are undecided or with very malleable preferences experience medical school differently. Our conceptual model further characterizes these groups into four distinct categories, thereby drawing attention to separate areas for study and intervention. Those groups are primary care committed, primary care positive, undecided, and non-primary-care committed.
For students truly committed to primary care, we must focus on recruitment and retention by developing the premedical pipeline and academic support. Because rural-born, lower-socioeconomic-status, and older and married students are more likely to enter primary care specialties, we must work with communities and undergraduate institutions, including community colleges, to find, support, and recruit these students.45,46 Experts have suggested that recruitment through the premedical pipeline begins very early in the educational process because broad career categories are chosen by the middle-school years.47–49 There is also concern that the same medical school aspirants most likely to choose primary care may not have the academic background to be competitive for admission when medical schools' reputations rest on Medical College Admissions Test (MCAT) and United States Medical Licensing Exam scores.45 Several studies have suggested that the tactic of blinding interviewers to GPA and MCAT scores for all candidates achieving a threshold level can be effective in changing recruitment patterns and would be an obvious area for further research and likely increased primary care recruitment.50,51 Medical schools also should strongly consider these issues in developing admissions policies. Finally, these future primary care physicians must be supported academically, and care should be taken to match them with mentors who can help them thrive.
Conversely, those students who matriculate already committed to non-primary-care specialties could be approached differently. Rather than trying to convince these students to become primary care physicians, educators should use evidence to demonstrate the value of primary care and teach interdisciplinary collaboration. Because these students are apt to identify specific features of experiences and mentors across specialties that shape their future practice plans, they should be connected with primary care mentors who demonstrate skills that interest them, such as business ownership, clinical procedures, or community leadership. This approach could help build effective professional relationships between primary care and other specialists in joint patient care.
Interventions focused on students with positive but uncommitted attitudes toward primary care and those who are truly undecided at matriculation carry perhaps the highest immediate potential for building the primary care workforce. Unfortunately, it is not known how to accurately identify these students early on, much less how to steer them toward primary care. The increase in primary care choice in the 1990s demonstrates that a significant number of students can be influenced in their specialty choices and that both market forces and curricular innovations likely affect their decisions.37,43 Title VII's proven effect is likely attributable to influencing uncommitted and undecided students, and could theoretically become a more efficient and valuable program if targeted more specifically. Payment reforms that value care coordination and health outcomes rather than volume might convince more of these students that primary care can be rewarding both financially and emotionally. Similarly, although research is lacking on the effects of other aspects of the health care environment, including physician autonomy, malpractice conditions, and uncertainty about health care reform and the economy, these issues may also be very important to the “malleable” student, and some are amenable to intervention.
One of the benefits of this model is the breadth of research questions it highlights. For example:
* How can primary-care-committed, primary-care-positive, undecided, and non-primary-care-committed students be identified in the admissions process or early in medical school?
* Would changes in relative incomes of primary care and specialist physicians or changes in the health care system affect the decision of “malleable” students? If so, how much and what kind of changes are needed?
* Which students are influenced by proven programs such as Title VII and rural or underserved tracks, and what aspects of these programs are most important?
* Does anticipated high debt prevent certain groups of students from considering medical school, and if so, are those students the ones who were more likely to choose primary care?
* Does very high debt deter a certain group of “malleable” students from primary care?
* What accounts for the success of some “pipeline” programs?
* How can we best use students' specialty and personal interests to match them with appropriate primary care mentors?
* If graduate medical education positions do not increase with the expansion of medical school enrollment, will a greater proportion of students not committed to primary care match into primary care programs? If so, will these physicians eventually exhibit different practice patterns than traditional primary care, and what impact would these patterns have on underserved communities?
One of the benefits of using a conceptual model to guide research is the ability to test specific elements of the model against other models or to complement other models. For example, the theory of planned behavior52 might engender a hypothesis that social norms and self-efficacy are enough to predict specialty choice, whereas our model posits that social norms may be defined or interpreted very differently based on a student's predisposition toward primary care or non-primary care. If a validated measure for medical student self-efficacy and perception of norms could be developed, it could be measured in a study designed to then differentiate student scores on this measure by specialty choice. Similarly, Bland et al14 assume a large role for faculty and admissions committee composition. Because our model assumes that different groups of students have different perceptions of role models, it follows that simply changing faculty composition to include more primary care is not as important as the quality of mentorship for the students, that is, matching them appropriately with role models and mentors. Within our model, simply increasing primary care faculty could cause non-primary-care-predisposed students to feel excluded and pressured to change their minds, resulting in a “backlash” against primary care, an effect that was evident in some schools after the institution of programs to increase primary care graduates39 (Noel M, senior associate chair for academic affairs, Department of Family Medicine, Michigan State University College of Human Medicine, personal communication, May 2010).
The chief limitation of this study is its theoretical nature and the qualitative method of incorporating prior research into the model. Health systems research is much more difficult to access with traditional search mechanisms than clinical inquiries. Database searches were supplemented by cross-checking references and applying more targeted keyword searches; however, it is conceivable that some studies were missed. The process and context of specialty choice also change over time, so it is unclear how adequate older articles are in informing a contemporary conceptual model. Finally, there is no standard strength-of-evidence taxonomy in evaluating this literature, and the quality of evidence is mixed. Including only high-quality evidence would limit the theoretical possibilities of the model. Readers should understand that many factors cited in the model need further research.
In conclusion, despite our extensive knowledge of associations with primary care career choice, we have not been able to affect student choices to any large degree. Much remains unknown about the methods that different students use in making this complex decision, and there is even less research into the decision-making process prior to medical school. Our model, which divides students into primary care committed, primary care oriented, undecided, and non-primary-care oriented, forms a conceptual basis for more efficiently targeted research. We are optimistic that focused inquiries and interventions can be successful in rebuilding the primary care physician workforce.
The authors wish to acknowledge the staff of the Robert Graham Center, the policy research center of the American Academy of Family Physicians, who were instrumental in the formulation of the concept for this paper and in their logistical and moral support: Robert Phillips, Andrew Bazemore, Martey Dodoo, Stephen Petterson, Imam Xirali, and Bridget Teevan. We also acknowledge those who agreed to formally review and critique the conceptual model: Teresa Anderson, Jennifer Bruekner, Matthew Burke, Carol Elam, Anne Gaglioti, Daniel Gibbs, Bridget Harrison, Winston Liaw, Robert Rayson, Kelechi Uduhiri, and Christina Warner, as well as the many people who offered informal feedback.
The authors have no conflicts of interest to disclose. This study was not funded but was supported in-kind by Dr. Bennett's time (July 2008 to June 2009) as policy research fellow at the Robert Graham Center for primary care research in conjunction with the Georgetown University School of Medicine Department of Family Medicine.
The opinions expressed in this article are those of the authors alone and do not reflect the views of the University of Kentucky, Michigan State University, Georgetown University, or the Robert Graham Center.
Preliminary versions of this conceptual model were presented as a poster at the Society of Teachers of Family Medicine annual conference in April 2009 and as a 10-minute presentation at the North American Primary Care Research Group annual conference in November 2009.
1Mancinko J, Starfield B, Shi L. The contribution of primary care systems to health outcomes within Organization for Economic Cooperation and Development (OECD) countries, 1970-1998. Health Serv Res. 2003;38:831–865.
2Ferrer RL, Hambidge SJ, Maly RC. The essential role of generalists in health care systems. Ann Intern Med. 2005;142:691–699.
3Starfield B, Shi L, Grover A, Mancinko J. The effects of specialist supply on populations' health: Assessing the evidence. Health Affairs (Millwood). January-June 2005(suppl Web exclusives):W5-97–W5-107.
4Baicker K, Chandra A. Medicare spending, the physician workforce, and beneficiaries' quality of care. Health Affairs (Millwood). January-June 2004(suppl Web exclusives): W-184–W-197.
5Bazargan M, Lindstrom RW, Dakak A, et al. Impact of desire to work in underserved communities on selection of specialty among fourth-year medical students. J Natl Med Assoc. 2006;98:1460–1465.
6Bodenheimer T, Pham HH. Primary care: Current problems and proposed solutions. Health Aff (Millwood). 2010;29:799–805.
7Colwill JM, Cultice JM, Kruse RL. Will generalist physician supply meet demands of an increasing and aging population? Health Aff (Millwood). 2008;27:W232–W241.
8Colwill JM, Cultice JM. The future supply of family physicians: Implications for rural America. Health Aff (Millwood). 2003;22:190–198.
9Hooker RS, Cawley JF, Leinweber W. Career flexibility of physician assistants and the potential for more primary care. Health Aff (Millwood). 2010;29:880–886.
10Salsberg E, Grover A. Physician workforce shortages: Implications and issues for academic health centers and policymakers. Acad Med. 2006;81:782–787.
11Hauer KE, Durning SJ, Kernan WN, et al. Factors associated with medical students' career choices regarding internal medicine. JAMA. 2008;300:1154–1164.
12Pugno PA, McGaha AL, Schmittling GT, et al. Results of the 2009 National Residency Matching Program: Family medicine. Fam Med. 2009;41:567–577.
13What Influences Medical Student and Resident Choices? Washington, DC: Robert Graham Center; 2009.
14Bland CJ, Meurer LN, Maldonado G. Determinants of primary care specialty choice: A non-statistical meta-analysis of the literature. Acad Med. 1995;70:620–641.
15Campos-Outcalt D, Senf JH, Watkins AJ, Bastacky S. The effects of medical school curricula, faculty role models, and biomedical research support on generalist physician careers: A review and quality assessment of the literature. Acad Med. 1995;70:611–619.
16Meurer LN. Influence of medical school curriculum on primary care specialty choice: Analysis and synthesis of the literature. Acad Med. 1995;70:388–397.
17Reed VA, Jernstedt GC, Reber ES. Understanding and improving medical student specialty choice: A synthesis of the literature using decision theory as a referent. Teach Learn Med. 2001;13:117–129.
18Lawson SR, Hoban JD. Predicting career decisions in primary care medicine: A theoretical analysis. J Contin Educ Health Prof. 2003;23:68–80.
19Senf JH, Campos-Outcalt D, Kutob R. Factors related to the choice of family medicine: A reassessment and literature review. J Am Board Fam Pract. 2003;16:502–512.
20Compton MT, Frank E, Elon L, Carrera J. Changes in U.S. medical students' specialty interests over the course of medical school. J Gen Intern Med. 2008;23:1095–1100.
21Chang LL, Grayson MS, Patrick PA, Sivak SL. Incorporating the fourth year of medical school into an internal medicine residency: Effect of an accelerated program on performance outcomes and career choice. Teach Learn Med. 2004;16:361–364.
22Rittenhouse DR, Fryer GE, Phillips RL, et al. Impact of Title VII training programs on community health center staffing and National Health Service Corps participation. Ann Fam Med. 2008;6:397–405.
23Markert RJ, Rodenhauser P, El-Baghdadi MM, et al. Personality as a prognostic factor for specialty choice: A prospective study of four medical school classes. Medscape J Med. 2008;10:49.
24Beach RA, Eva KW, Reiter HI. Can self-declared personal values be used to identify those with family medicine career aspirations? Adv Health Sci Educ Theory Pract. 2008;13:193–202.
25Newton DA, Grayson MS, Whitley TN. What predicts medical student career choice? J Gen Intern Med. 1998;13:200–203.
26Senf JH, Campos-Outcalt D, Kutob R. Family medicine specialty choice and interest in research. Fam Med. 2005;37:265–270.
27Scott I, Wright B, Brenneis F, et al. Why would I choose a career in family medicine? Can Fam Physician. 2007;53:1956–1957.
28Schubot DB, Cayley W Jr, Eliason BC. Personal values related to primary care specialty aspirations. Fam Med. 1996;28:726–731.
29Ebell MH. Future salary and US residency fill rate revisited. JAMA. 2008;300:1131–1132.
30Bodenheimer T, Berenson RA, Rudolf P. The primary care-specialty income gap: Why it matters. Ann Intern Med. 2007;146:301–306.
31Kassebaum DG, Szenas PL. Relationship between indebtedness and the specialty choices of graduate medical students. Acad Med. 1992;67:700–707.
32Bernstein D. Medical student indebtedness and choice of specialty. JAMA. 1992;267:1921.
33Rosenthal MP, Marquette PA, Diamond JJ. Trends along the debt-income axis. Acad Med. 1996;71:675–677.
34Colquitte WL, Zeh MC, Killian CD, Cultice JM. Effect of debt on US medical school graduates' preferences for family medicine, general internal medicine, and general pediatrics. Acad Med. 1996;71:399–411.
35Rosenblatt RA, Andrilla HA. The impact of U.S. medical students' debt on their choice of primary care careers: An analysis of data from the 2002 Medical School Graduate Questionnaire. Acad Med. 2005;80:815–819.
36Senf JH, Campos-Outcalt D, Watkins AJ, Bastacky S, Killian C. A systematic analysis of how medical school characteristics relate to graduates' choices of primary care specialties. Acad Med. 1997;72:524–533.
37Retchin SM, Boling PA, Nettleman MD, Mick SS. Marketplace reforms and primary care career decisions. Acad Med. 2001;76:316–323.
38Kutob RM, Senf JH, Campos-Outcalt D. The diverse functions of role models across primary care specialties. Fam Med. 2006;38:244–251.
39Kuzel AJ, Moore SS. Choosing a specialty during a generalist initiative: A focus group study. Fam Med. 1999;31:641–646.
40Ellsbury KE, Burack JH, Irby DM, et al. The shift to primary care: Emerging influences on specialty choice. Acad Med. 1996; 71(10 suppl):S16–S18.
41Campos-Outcalt D, Senf JH, Kutob R. A comparison of primary care graduates from schools with increasing production of family physicians to those from schools with decreasing production. Fam Med. 2004;36:260–264.
42Burack JH, Irby DM, Carline JD, Amrozy D, Ellsbury KE, Stritter FT. A study of medical students' specialty-choice pathways: Trying on possible selves. Acad Med. 1997;72:534–541.
43Prislin MD, Saultz JW, Geyman JP. The generalist disciplines in American medicine one hundred years following the Flexner Report: A case study of unintended consequences and some proposals for post-Flexnerian reform. Acad Med. 2010;85:228–235.
44Hueston WJ, Bradford WD, Sheperd TM. Family medicine and hospital specialty match rates: Does the economy have anything to do with it? Fam Med. 2004;36:265–269.
45Barr DA, Gonzalez ME, Wanat SF. The leaky pipeline: Factors associated with early decline in interest in premedical studies among underrepresented minority undergraduate students. Acad Med. 2008;83:503–511.
46Freeman J, Ferrer R, Greiner KA. Developing a physician workforce for America's disadvantaged. Acad Med. 2007;82:133–138.
47Rao V, Flores G. Why aren't there more African-American physicians? A qualitative study and exploratory inquiry of African-American students' perspectives on careers in medicine. J Natl Med Assoc. 2007;99:986–993.
48Winkleby MA. The Stanford Medical Youth Science Program: 18 years of a biomedical program for low-income high school students. Acad Med. 2007;82:139–145.
49Gottfredson L. Gottfredson's theory of circumscription, compromise, and self-creation. In: Brown D, ed. Career Choice and Development. 4th ed. San Francisco, Calif: Jossey-Bass; 2002:85–184.
50Cohen JJ. Our compact with tomorrow's doctors. Acad Med. 2002;77:475–480.
51Albanese MA, Farrell P, Dottl S. Statistical criteria for setting thresholds in medical school admissions. Adv Health Sci Educ Theory Pract. 2005;10:89–103.
52Ajzen I, Fishbein M. Understanding Attitudes and Predicting Social Behavior. Englewood Cliffs, NJ: Prentice Hall; 1980.