More than 50 U.S. medical schools sponsor combined baccalaureate–MD programs (combined programs), which offer conditional acceptance into medical school to high school seniors or early college students before graduation from college and upon completion of program requirements.1 Combined programs have emerged for a variety of reasons, including to increase workforce diversity,2–4 promote primary care careers,2–4 address physician shortages,5,6 reduce the length of training,6,7 promote the development of physician–scientists,1 emphasize a liberal arts education,7 and meet the needs of underserved patient populations.3,8,9 Combined programs can vary considerably in terms of mission, length of training (usually between 6 and 8 years), admissions entry point (either immediately after high school graduation or before graduation from college), acceptance criteria, size, and curriculum.
Recent research has shown similar medical school performance of students enrolled in combined programs compared with those in traditional MD programs (traditional programs).10,11 The majority of published data on combined programs, however, has been descriptive, focusing on the characteristics of individual programs and reporting few outcomes.3–9 For example, little is known about the motivations, demographics, or career choices of students enrolled in combined programs. In a detailed analysis of combined program mission statements, 1 in 4 programs specifically targeted students from racial/ethnic groups that are underrepresented in medicine (URM).1 A similar proportion of combined programs’ missions encourage students to pursue primary care specialties or care for underserved populations.1
We sought to determine the demographics of combined program graduates as well as their intentions to work in primary care or to work with the medically underserved, as compared with graduates of traditional programs. Given the context mentioned above, we hypothesized that students in combined programs would disproportionately be members of URM groups, be more likely to intend to care for underserved patients, and be more likely to intend to enter a primary care specialty compared with students graduating from traditional programs. We further hypothesized that within primary care specialties, family medicine would be selected significantly more often by medical students who graduate from combined programs than by students graduating from traditional programs.
As a secondary focus, we hypothesized that graduating medical student characteristics (e.g., gender, debt at graduation, URM group status) would moderate the relationship between the type of medical degree program (combined vs traditional) and IPPC or with underserved patients. We further hypothesized that level of debt upon medical school graduation would mediate the relationship between type of medical education degree program and IPPC or with underserved patients.
The Association of American Medical Colleges (AAMC) Graduation Questionnaire (GQ) is a nationally representative annual survey of graduating medical students. Deidentified data from 2010 to 2017 were obtained from the GQ databank and merged across survey year by the AAMC before analysis was conducted. Response rates ranged from 78% to 83% during the study period.12,13 Data from the following categories were analyzed: year of graduation, demographic characteristics (gender, age at graduation, and race/ethnicity), type of medical degree program attended (combined vs traditional), planned practice specialty, intention to care for underserved populations and/or to practice in an underserved area, and total student debt upon medical school graduation. All data were confidential and anonymous; the Rhode Island Hospital and The Miriam Hospital institutional review boards exempted this study.
The data were recategorized before analysis. Specifically, medical degree program was coded as either traditional MD program or combined baccalaureate–MD program (e.g., BA–MD, BS–MD); students who reported graduating from other medical degree programs (e.g., MD–PhD) were not included in these analyses. URM status was coded from race/ethnicity; all students identifying as a member of a group other than White or Asian were coded as a URM group member (including students who identified as multiracial). Intention to practice in internal medicine, family medicine, obstetrics and gynecology, or pediatrics were coded as intention to work in primary care. As a subanalysis, we specifically examined family medicine as the intended practice specialty. Total student debt at graduation was coded into 5 levels: no debt; < $100,000; $100,000–$199,999; $200,000–$299,999; and ≥ $300,000. IWMU was coded as yes or no/undecided based on responses on 2 variables: intention to practice in an underserved area and intention to care for underserved populations. Any response of yes (i.e., to either variable or to both) was coded as IWMU and a no/undecided response to both questions was coded as no/undecided.
Descriptive statistics (gender, age at graduation, race/ethnicity) were generated reporting frequencies and percentages. Differences in frequencies between traditional and combined programs were assessed using the chi-square test, with additional statistical analysis for trends in difference in medical program type over time using the Mantel–Haenszel test. Logistic regression models were conducted to estimate the unadjusted and adjusted odds ratios for the effects of covariates (e.g., gender, total debt at graduation, URM group member, type of medical degree program) on the 2 primary outcomes: IPPC and IWMU. Year of survey completion (i.e., year of graduation) was added to the model to determine if there had been any significant independent effect of time on the outcomes after adjusting for the other variables in the model. Interaction effects between the type of medical degree program and other predictors were added to the logistic model to evaluate potential moderating effects of student characteristics on the effect of medical degree program type on the primary outcomes. We were unable to conduct an individual or regional program analysis using AAMC GQ data given the AAMC policy regarding member institution privacy. All statistical analyses were conducted using SAS version 9.4 (SAS Institute, Inc, Cary, North Carolina).
Across the 8 years of the study, there were survey data from a total of 116,256 respondents. Data from 109,028 (93.8%) respondents who met the inclusion criteria of either graduating from a traditional MD or a combined baccalaureate–MD program were included in the analyses. There were 3,182 students graduating from combined programs and 105,846 from traditional programs. Significantly more students graduated from traditional programs, averaging 97.1% of graduates from 2010 to 2017. The percentage of students graduating from a combined program did show a significant positive increasing trend over the 8 years of survey data (from 2.2% in 2010 to 3.5% in 2017; Mantel–Haenszel χ2(1) = 42; P < .001). The overall gender representation was significantly different between the traditional and combined programs, with a greater percentage of females in combined programs (1,813, 57.0% vs 52,013, 49.1%; χ2(1) = 76; P < .001; Table 1). Almost all students from combined programs were 29 years or younger at graduation (3,143, 98.8%); this was a significantly greater percentage than that of students from traditional programs (89,688, 84.7%; χ2(2) = 482; P < .001).
Intention to practice in primary care
Across the survey years, 90,978 students indicated their intended area of clinical practice. Of these, 39,724 (43.7%) students indicated they intended to pursue a career in a specialty related to primary care. A significantly higher percentage of students in combined programs intended to enter primary care specialties, with 1,302 (49.0%) combined program graduates indicating intention to practice in a primary care specialty compared with 38,422 (43.5%) of traditional program graduates (χ2(1) = 32; P < .001; Table 1). The most frequent specialty within primary care chosen by both groups was internal medicine (traditional program: 15,291, 39.8% vs combined program: 627, 48.2%).
Unadjusted logistic regression models were conducted to estimate the effect of each covariate (gender, year of graduation, age at graduation, medical degree program type, URM group status, IWMU, and total debt at graduation) separately on the odds of primary care specialty choice. All predictors were significant and were then entered in an adjusted logistic regression model. In the adjusted logistic regression model (as can be seen in Table 2), identification as a member of a URM group no longer significantly predicted odds of IPPC. All the other predictors remained significant, showing that graduating from a combined program, identifying as female, and IWMU predicting significantly greater odds of IPPC. The odds of IPPC also increased over the years that the survey was administered but decreased with an increasing level of debt and being older than 29 at graduation. Overall, the model fit was significantly better in the adjusted model compared with the unadjusted models.
Intention to practice family medicine
A subanalysis was conducted to focus on the identification of family medicine as the specialty choice of graduating medical students. Across the survey years, 6,743 (7.4%) of 90,978 graduating students indicated that their specialty career choice was family medicine. Family medicine specialty was indicated significantly more frequently by students identifying as females (3,929/44,059, 8.9%) than by those identifying as males (2,814/46,919, 6.0%; χ2(1) = 272; P < .001); by those in traditional (6,584/88,320, 7.5%) versus combined programs (159/2,658 6.0%; χ2(1) = 8; P = .01); and by medical students who identified as a URM (1,132/12,171, 9.3%) than by those who did not (5,549/77,491, 7.2%; χ2(1) = 55; P < .001). In total, 31,299 of 87,207 (35.9%) graduating medical students indicated an intention to care for underserved populations and/or practice in underserved areas. Indicating an IWMU was more frequently indicated by those intending a career in family medicine (4,615/6,118, 75.4%) compared with those not intending to pursue a career in family medicine (23,479/73,924, 31.8%; χ2(1) = 3,556; P < .001).
Intention to work with the medically underserved
The same methodological approach that was used on the first outcome variable, IPPC (see above), was also used on the second outcome variable, IWMU (Table 3). All predictors were significant in the unadjusted models, but in the adjusted model, medical degree program was no longer a significant predictor. Due to collinearity with the prior model, primary care as specialty choice was not used as a variable for this outcome. Identifying as a URM, identifying as female, and having debt predicted significantly greater odds of IWMU. However, the odds of intending to work with the medically underserved decreased over the years of survey administration. Overall, the model fit was significantly better in the adjusted model compared with the unadjusted models.
Moderators of outcomes
There were significant differences in the race/ethnicity of students in the different program types, with more students identifying as White in the traditional programs than the combined programs (66,791, 63.1% vs 1,125, 35.4%) but also with significantly more students identifying as a member of a URM group in the traditional programs than in the combined programs (14,757, 13.9% vs 276, 8.7%; χ2(1) = 70; P < .001; Table 1). Additionally, medical school debt levels were significantly higher for students in traditional programs compared with students in combined programs (χ2(4) = 305; P < .001). Interaction effects of type of medical degree program with the other covariates (age at graduation, gender, URM group status, total debt at graduation, and year of graduation) were entered in the adjusted logistic regression models for the 2 primary outcome variables (IPPC and IWMU). There were no significant interactions between medical degree program type and any other covariate, suggesting that the main effect of type of medical degree program on IPPC was not moderated by other student characteristics.
Our analysis shows that students graduating from combined baccalaureate–MD programs are more likely to be women, to not identify as a URM, and to be younger in age than students graduating from traditional MD programs. They are also more likely to graduate with less student debt and intend to work in a primary care specialty. However, students graduating from combined programs are not more likely to intend to care for underserved populations or practice in an underserved area or to identify as a member of a URM group.
Despite the mission of 1 in 4 combined programs being to promote racial and ethnic diversity,1 our analysis indicates that on a national level, they may not increase the number of URM medical school graduates as intended. Our findings indicate a disproportionate number of students in combined programs identify as Asian compared with traditional programs. It is unclear from our analysis why this is the case; data from other sources suggest that students belonging to this group are overrepresented compared with the general population in terms of both applicants and matriculants to medical schools in general.14
Importantly, it is unknown if individual programs are achieving the goal of increasing the pool of URM students. There is some evidence that some combined programs with this specific mission are successful, such as the Premedical Honors College, an 8-year combined program based in Texas, which disproportionately graduates physicians of Latinx descent.8 However, as a recent analysis by the AAMC points out, URM students continue to make up a disproportionally low number of medical students, and combined programs as a whole are not an exception to this trend.13 Low URM student enrollment in combined programs may also be explained by recruitment or retention processes that may disadvantage URM students (e.g., standardized testing score requirements), regional differences, or other structural inequalities.
Factors that have been found to be associated with choosing a primary care career include having less debt at graduation, curriculum influence, and identifying as a member of a URM group or as female.15,16 Our findings indicate that students in combined programs have significantly less debt and disproportionately identify as female. However, in our adjusted model, which took these factors into consideration, students in combined programs were still more likely to choose primary care. This suggests that the expansion of combined programs may serve as an additional mechanism to increase the primary care workforce. Our finding that students from combined programs are not more likely to choose to work with the medically underserved may at first seem contradictory to the finding that they were more likely to choose medical careers linked to primary care. However, it may be that these students have a more expansive view of what it means to work with medically underserved patients—one that is not constrained by specialty choice.
Interestingly, when we specifically examined family medicine as a career choice, combined program students were less likely to indicate this as their career choice, but URM students and those interested in working with the medically underserved were more likely to indicate this as their career choice. This suggests that family medicine as a career choice is more closely aligned with an intention to increase medical provision to underserved communities than to attending a combined program.
To our knowledge, this is the first analysis of key characteristics of students enrolled in combined programs at a national level over multiple years. As combined programs have emerged to address a number of challenges, our research should be considered in the context of other work that seeks to identify strategies to (1) improve the diversity of the physician workforce, (2) promote primary care as a specialty choice, and (3) promote an intention to care for underserved patients.
This analysis has several limitations. First, combined programs have variable missions and different strategies for accomplishing their aims. Second, compared with traditional programs, the number of students graduating from combined programs still remains relatively small. Third, we were unable to conduct an individual or regional program analysis given the AAMC policy regarding member institution privacy. The variability of combined programs and the relatively small number of graduating students from them limit the utility of nationwide outcomes-based analysis. The inability to compare program outcomes with program missions at an individual program level limits our ability to craft strategies to refine combined programs.
In conclusion, our analysis shows that combined baccalaureate–MD programs are more demographically diverse than traditional MD programs with a higher percentage of students identifying as Asian and female. Students in combined programs are significantly more likely to intend to pursue careers in primary care but are no more likely to intend to care for underserved populations or practice in an underserved area or to identify as a member of a URM group than students in traditional programs. Further research into why students in combined programs are more likely to intend to work in a primary care specialty and how combined programs can attract more students from URM groups will inform medical school officials responsible for programming and have broader implications for the health care workforce.
1. Eaglen RH, Arnold L, Girotti JA, et al. The scope and variety of combined baccalaureate-MD programs in the United States. Acad Med. 2012;87:1600–1608.
2. Roman SA Jr, McGanney ML. The Sophie Davis School of Biomedical Education: The first 20 years of a unique BS-MD program. Acad Med. 1994;69:224–230.
3. Roman SA Jr. Addressing the urban pipeline challenge for the physician workforce: The Sophie Davis model. Acad Med. 2004;79:1175–1183.
4. Thomson WA, Ferry P, King J, Wedig CM, Villarreal GB. A baccalaureate-MD program for students from medically underserved communities: 15-year outcomes. Acad Med. 2010;85:668–674.
5. Drees BM, Arnold L, Jonas HS. The University of Missouri-Kansas City School of Medicine: Thirty-five years of experience with a nontraditional approach to medical education. Acad Med. 2007;82:361–369.
6. Callahan C, Veloski JJ, Xu G, Hojat M, Zeleznik C, Gonnella JS. The Jefferson-Penn State B.S.-M.D. program: A 26-year experience. Acad Med. 1992;67:792–797.
7. Blaustein EH, Kayne HL. The accelerated medical program and the liberal arts at Boston University. JAMA. 1976;235:2618–2620.
8. Thomson WA, Ferry PG, King JE, Martinez-Wedig C, Michael LH. Increasing access to medical education for students from medically underserved communities: One program’s success. Acad Med. 2003;78:454–459.
9. Cosgrove EM, Harrison GL, Kalishman S, et al. Addressing physician shortages in New Mexico through a combined BA/MD program. Acad Med. 2007;82:1152–1157.
10. George P, Park YS, Ip J, Gruppuso PA, Adashi EY. The association between premedical curricular and admission requirements and medical school performance and residency placement: A study of two admission routes. Acad Med. 2016;91:388–394.
11. Green MM, Welty L, Thomas JX Jr, Curry RH. Academic performance of students in an accelerated baccalaureate/MD program: Implications for alternative physician education pathways. Acad Med. 2016;91:256–261.
12. Grischkan J, George BP, Chaiyachati K, Friedman AB, Dorsey ER, Asch DA. Distribution of medical education debt by specialty, 2010-2016. JAMA Intern Med. 2017;177:1532–1535.
13. Association of American Medical Colleges. Medical School Graduation Questionnaire: 2017 All Schools Summary Report. https://www.aamc.org/system/files/reports/1/2017gqallschoolssummaryreport.pdf
. Published July 2017. Accessed May 29, 2020.
14. Lett LA, Murdock HM, Orji WU, Aysola J, Sebro R. Trends in racial/ethnic representation among US medical students. JAMA Netw Open. 2019;2:e1910490.
15. Schieberl JL, Covell RM, Berry C, Anderson J. Factors associated with choosing a primary care career. West J Med. 1996;164:492–496.
16. Phillips JP, Petterson SM, Bazemore AW, Phillips RL. A retrospective analysis of the relationship between medical student debt and primary care practice in the United States. Ann Fam Med. 2014;12:542–549.