Preparing for the MD: How Long, at What Cost, and With What Outcomes?

Abstract

Purpose 

To assess educational and professional outcomes of an accelerated combined bachelor of science–doctor of medicine (BS–MD) program using data collected from 1968 through 2018.

Method 

Participants of this longitudinal study included 2,235 students who entered medical school between 1968 and 2014: 1,134 in the accelerated program and 1,101 in the regular curriculum (control group)—matched by year of entrance to medical school, gender, and Medical College Admission Test (MCAT) scores. Outcome measures included performance on medical licensing examinations, academic progress, satisfaction with medical school, educational debt, first-year residency program directors’ ratings on clinical competence, specialty choice, board certification, and faculty appointments.

Results 

The authors found no practically important differences between students in the accelerated program and those in the control group on licensing examination performance, academic progress, specialty choice, board certification, and faculty appointments. Accelerated students had lower mean educational debt (P < .01, effect sizes = 0.81 and 0.45 for, respectively, their baccalaureate debt and medical school debt), lower satisfaction with their second year of medical school (P < .01, effect size = 0.21), and lower global satisfaction with their medical school education (P < .01, effect size = 0.35). Residency program directors’ ratings in 6 postgraduate competency areas showed no practically important differences between the students in the accelerated program and those in the control group. The proportion of Asian students was higher among program participants (P < .01, effect size = 0.43).

Conclusions 

Students in the accelerated program earned BS and MD degrees at a faster pace and pursued careers that were comparable to students in a matched control who were in a regular MD program. Findings indicate that shortening the length of medical education does not compromise educational and professional outcomes.

In the United States, many medical students expect to spend and possibly borrow a considerable amount of money to be a physician. Thus, leaders of medical educational institutions have wrestled with the question of whether the length of time needed for preparing to be a physician may be shortened without compromising educational outcomes or professional competency. The concept of combined baccalaureate–doctor of medicine (MD) programs in the United States is not new, and in many aspects, is similar to the medical education system in the United Kingdom and many European and Asian countries.¹ In fact, the Association of American Medical Colleges (AAMC) has provided a list of 51 medical schools that offer combined baccalaureate–MD programs.²

In the 1960s, factors contributing to the development of combined baccalaureate–MD programs in the United States included the shortage of physicians in general^1,3 and a growing need for primary care physicians in rural and underserved communities in particular.^4–6 Further, during this era, the number of applicants to medical schools declined as academically talented students shifted their career interests to pursue disciplines other than medicine.^7,8

In 1963, Jefferson (currently Sidney Kimmel) Medical College and The Pennsylvania State University (Penn State) launched a joint bachelor of science (BS)–MD (BS–MD) curriculum to award both degrees in 5 years. After graduating from high school, program participants spent the first 5 terms (1 academic year [i.e., 3 terms], plus 2 summers) at Penn State, proceeded to medical school, and then returned to Penn State for 2 further summer terms (after the first and second years of medical school). The goal was to reduce the time devoted to undergraduate (baccalaureate) education by integrating the baccalaureate and medical school curricula. That is, the 4-year MD curriculum matched that of regular medical students, but the BS curriculum was accelerated. In 1985, the program expanded to a 6-year program.

We felt empirical evidence was essential to determine if it was indeed feasible to shorten the time needed to prepare for the practice of medicine without compromising educational effectiveness; therefore, we designed this study to compare the educational and professional outcomes of the accelerated Penn State–Jefferson combined BS–MD program with a matched control group who pursued the regular Jefferson MD curriculum.

The availability of pertinent data from more than 50 years through the Jefferson Longitudinal Study (JLS)^9,10 provides a unique opportunity to assess educational and professional outcomes of the accelerated program. The JLS, which has collected data for more than half a century, is the most comprehensive, extensive, and uninterrupted longitudinal study of medical education outcomes maintained by a single medical school.¹⁰ It contains pertinent outcome variables, particularly clinical competence assessments early in residency training, and updated professional outcomes.

We also compared our findings with those from previous research that assessed the Penn State-Jefferson accelerated BS–MD program so as to examine the extent to which the goals of the program have been achieved despite changes in program length and in the composition of medical school applicants.

Method

The sample of this longitudinal study included 2,235 students (1,428 men, 807 women) who entered Jefferson between 1968 and 2014. We compared the accelerated program participants (n = 1,134; 723 men, 411 women) with students who pursued the regular Jefferson MD program (the control group [n = 1,101; 705 men, 396 women])—matched by year of entrance to medical school, gender, and scores on the science subtest of the Medical College Admission Test (MCAT). We retrieved data from the JLS database^9,10 in 2018. The study was approved and given exempt status by Thomas Jefferson University’s institutional review board.

Study variables

Study variables included the following (explained in greater detail in the paragraphs below):

• demographics (age, gender, race/ethnicity),
• performance on medical licensing examinations,
• academic progress (i.e., on-time vs delayed graduation, attrition),
• satisfaction with medical school,
• educational debt,
• competence ratings made by residency program directors,
• specialty choice,
• board certification rates,
• percentage with faculty appointments, and
• length of the program.

We used data from the Penn State–Jefferson combined BS–MD matriculating class of 1968, not the inaugural year (1963), because MCAT scores used to select students in the control group were not available for all students in the JLS before 1968. For the pre-1978 version of the MCAT exam, we used the “Science Achievement” subtest; for the 1978–1991 version, we used the “Science Problem Solving” subtest; and for the 1991–2014 version, we used the “Biological Sciences” subtest. Previous research has shown that among MCAT subtest scores, the science subtests were the best predictors of performance in Parts 1 and 2 of the National Board of Medical Examiners (NBME) exams and Steps 1 and 2 of the United States Medical Licensing Examination (USMLE).¹¹

Medical licensing exam scores.

We transformed the NBME and USMLE scores, which were reported on different scales, to a standard scale with a mean of 100 and a standard deviation (SD) of 10. This linear transformation allowed us to examine differences on the NBME and USMLE on a common scale and to judge the magnitude of group differences using comparable units of measurement and score distributions.

Satisfaction with medical school.

Students rated their satisfaction with each year of medical school at the end of the fourth year in an exit survey (response rates varied from 70% to 94%). These ratings were made on a 4-point scale. Ratings on students’ global (overall) satisfaction with their medical school education were made on a 7-point scale.

Educational debt.

Educational debt data were available for matriculants after 1975. We adjusted baccalaureate and medical school debt to constant dollars—base year 2018—using the annual consumer price index in the year reported.

Competence ratings.

We assessed postgraduate competence using a rating form completed by residency program directors at the completion of residents’ first year. Specifically, we used the 2 factors—“science of medicine” and “art of medicine”—that emerged from a factor analysis of this rating form.¹² In addition, the rating form included 4 global items, which program directors answered on a 4-point scale, for the assessment of “medical knowledge,” “data-gathering skills,” “clinical judgment,” and “professional attitudes.”

Specialty choice, board certification, faculty appointments.

Data for self-reported specialty were available from the American Medical Association Physician Masterfile. For more recent graduates (those graduating in or after 2015) who were still in postgraduate training when we gathered the data, we used their residency program as a proxy to determine their specialty interest. We used the board certification rates from the American Board of Medical Specialties for graduating classes before 2009 to allow sufficient time for completing postgraduate training.¹³ Data on faculty appointments were available from the AAMC medical faculty roster for graduates between 1979 and 2014.

Statistical analysis

To examine the statistical significance between the students in the accelerated program and those in the control group, we used the chi-square test for categorical variables and the t test for continuous measures. To address the possibility of age influencing ratings on postgraduate competence, we used analysis of covariance in which students’ age served as a covariate. The number of observations varied in different statistical analyses due to missing data. Because of the large sample size, negligible group differences or associations would become statistically significant, even if practically (clinically) unimportant; therefore, for determining the practical (clinical) importance of the findings, we calculated the corresponding effect sizes.^14,15 Based on Cohen’s operational definition,¹⁴ any effect size of 0.20 or less was considered negligible (clinically unimportant) even if statistically significant.

Results

Age, gender, and race/ethnicity

The mean age at entrance to medical school, as expected, was significantly lower for the accelerated program participants (mean = 19.57, SD = 0.80) compared with that of the control group students (mean = 23.48, SD = 2.69) (t₍₁₂₉₀₎ = 46.32, P < .01, effect size = 2.20). We did not expect, nor evaluate for, gender differences since the accelerated program participants and the control group students were matched by gender.

Distribution of race/ethnicity in the 2 groups showed a substantially lower rate of Caucasian students among the accelerated program participants compared with those in the control group. While 49% (561/1,134) of program participants reported their ethnicity as Caucasian, 85% (931/1,101) of students in the control group designated themselves as Caucasian. A larger percentage of program participants designated themselves as Asian (25%, 282/1,134), compared with the control group (only 6%, 63/1,101). The percentage of African American and Hispanic/Latino students varied from < 1% to 2% in both of the groups. Differences in ethnicity in the 2 groups were statistically significant (χ²₍₄₎ = 423.73, P < .001). The effect size estimate of the association is 0.43, indicating that the association with ethnicity is of practical importance. See Table 1.

Table 1: Race/Ethnicity of Participants in the Penn State–Jefferson Combined BS–MD Degree Program and of Students in the Matched Control Group^a

Licensing examinations

We compared the performance of the 2 groups on Parts I, II, and III of the NBME and on Steps 1, 2, and 3 of the USMLE. Table 2 presents the means and SDs of the transformed scores, along with the summary results of our statistical analyses. We found no significant differences between accelerated program participants and the control group students on any of the medical licensing examinations.

Table 2: Mean Performance on Medical Licensing Examinations of Participants in the Penn State–Jefferson Combined BS–MD Degree Program and of Students in the Matched Control Group^a

Academic progress

We examined the percentage of on-time graduation, delayed graduation, and attrition for the 2 groups. On-time graduation was slightly lower for the accelerated program participants (86%, 981/1,134) than for the control group students (91%, 999/1,101). Although the difference was statistically significant (z test for proportion = 3.13, P < .01), the effect size (and therefore the practical effect) was negligible (effect size = 0.16). Detailed analysis showed that the difference in delayed graduation for academic remediation between the 2 groups was not significant (4% [48/1,134] of the program participants and 3% [32/1,101] of the control group students). The delayed graduation for nonacademic reasons for program participants was 7% (75/1,134), compared with 3% (31/1,101) for control group students (z test for proportion = 4.28, P < .01). However, the corresponding effect size did not reach the 0.20 mark (effect size = 0.19), indicating that differences were not of practical importance. The most frequently reported reasons for nonacademic delayed graduation among the program participants were medical or personal (59%, 44/75); the corresponding rate for the control group students was 42% (13/31). Attrition rate (due to dismissal/withdrawal/transfer/other reasons) was 3% (30/1,134) among program participants and 4% (39/1,101) in the control group.

Satisfaction with medical school

We compared satisfaction ratings between the accelerated program participants and the control group students for each of the 4 years of medical school and overall (global satisfaction). Summary results of statistical analysis, reported in Table 3, indicated no significant difference between the 2 groups on satisfaction ratings in the fourth year of medical school. However, program participants were less satisfied with their first year (t₍₁₄₆₅₎ = 2.84, P < .01), second year (t₍₁₄₆₅₎ = 3.97, P < .01), and third year (t₍₁₄₆₄₎ = 3.75, P < .01). Also, average global satisfaction ratings on whether students felt medical school prepared them for a career in medicine were lower for program participants than for the control group (t₍₉₈₄₎ = 5.51, P < .01); however, the effect size estimates for years 1 and 3 were negligible (d < 0.20). The magnitude of the effect size for the global satisfaction ratings with medical school education was 0.35, which approaches a moderately important difference.^14,15

Table 3: Comparisons of Participants in the Penn State-Jefferson Combined BS–MD Degree Program, and Students in a Matched Control Group, on Satisfaction with Education in Medical School, Ratings of Competence by Program Director Upon Completion of the First Year of Residency, and Educational Debt^a

Educational debt

We examined the percentage of students among the program participants and in the control group who accrued educational debt during their undergraduate (baccalaureate) education and/or during medical school. The percentage of program participants with debt was lower than that of control group students (see Table 3). The percentage of those who accrued debt during their baccalaureate education was 24% (215/892) of the program participants and 32% (282/881) of the control group students; the proportion who accrued debt during medical school was, respectively, 66% (587/892) and 74% (653/881).

For accelerated program participants, the mean debt accrued during their baccalaureate education at Penn State before medical school in 2018 dollars was approximately $10,400. Control group students accrued an average of $19,000 of baccalaureate debt before medical school. The difference between the 2 groups was not only statistically significant (t_(439.1) = 9.80, P < .01) but also practically important (effect size = 0.81).

We observed a similar pattern for debt accrued during medical school. To illustrate, the average debt in medical school for the program participants was $113,000, compared with $149,000 for control students. The difference was statistically significant (t₍₁₂₃₅₎ = 7.98, P < .01) and practically important (effect size = 0.45). These findings confirm that accelerated program participants accrued substantially less debt to earn their MD (see Table 3).

Clinical competence in residency

Data for competence ratings in postgraduate medical education programs were available for graduates who granted written permission (approximately 70%) to collect such data from program directors; however, completed rating forms were available for 60% (n = 676) of the program participants and 62% (n = 688) of the control group. The 2 groups were compared (means and SDs) on ratings given by program directors in 6 areas of competence (see Method for description and Table 3 for statistical results and analysis).

Results showed no statistically significant difference between the program participants’ and control group students’ ratings in the areas of medical knowledge, data-gathering skills, and clinical judgment, nor on their ratings on the factor science of medicine. However, accelerated program participants obtained significantly lower average ratings in the area of professional attitudes (t₍₁₀₇₀₎ = 2.91, P < .01, effect size = 0.18) and on the factor art of medicine (t₍₁₃₀₉₎ = 2.30, P < .05, effect size = 0.13), both of which are related to professionalism in medicine.

Using analysis of covariance in which students’ age served as a covariate, results showed that the statistically significant group difference on ratings of professional attitudes remained statistically significant (adjusted F_{(2, 1087)} = 6.07, P < .01, adjusted effect size = 0.22). However, the difference between the 2 groups on the factor of the art of medicine became nonsignificant after adjusting for the effect of students’ age (adjusted F_{(2, 1329)} = 2.67, P = .10, adjusted effect size = 0.13).

Specialty choice

The distribution of specialty choice between the program participants and the control group students was close (see Table 4). Graduates of the accelerated program were slightly more likely to choose internal medicine (29% [322/1,118]) compared with control group students (23% [245/1,083]) but slightly less likely to choose family medicine (7% [80/1,118] of program graduates and 11% [119/1,083] of control group students). Although the association was statistically significant (χ²₍₉₎ = 43.89, P < .01), the magnitude of the corresponding effect size estimate was negligible (0.14), indicating no practical or clinical difference.

Table 4: Specialty Choice for Participants of the Penn State–Jefferson Combined BS–MD Degree Program and for Students in a Matched Control Group^a

Board certification and faculty appointments

Board certification rates for graduates of the accelerated program and control group students were, respectively, 94% (839/897) and 96% (804/839), which is not statistically significant. Program participants were more likely to hold full-time faculty appointments (32%, 315/995) compared with the control group (24%, 233/962). While the association was statistically significant (χ²₍₁₎ = 13.07, P < .01), the magnitude of the effect size estimate was negligible (0.08), rendering the practical importance of the difference inconsequential.

Length of the program

We investigated the effects on outcomes of expanding the program from 5 to 6 years, replicating the statistical analyses for those who matriculated before and after the expansion. We found no substantial change in the pattern of findings with one notable exception: The number of Asian students enrolled in the combined degree program increased in the 6-year program.

Discussion

Published studies assessing the accelerated Penn State–Jefferson combined BS–MD program have reported varying results. For example, in their analysis of the early years of the program, Herbut and colleagues¹⁶ observed no statistically significant difference between the program participants and regular students in either attrition or withdrawal rates. Gottheil and colleagues³ found no significant differences in accelerated program students’ performance in medical school or on licensing examinations compared with the students pursuing the regular program. Callahan and colleagues¹⁷ compared performance of the program participants with both (1) a control group (matched by SAT scores) and (2) the rest of their medical school classmates. The findings showed that the program participants performed as well as those in the control group, but better than the rest of their classmates on most performance measures; however, the delayed graduation rate for nonacademic reasons was slightly higher among program participants.¹⁷

With the exception of the Callahan et al study,¹⁷ no other studies have used a matched control group for comparisons. Compared with Callahan et al,¹⁷ the present study benefits from 3 advantages. First, we matched our control group with the program participants not only by gender and year of entrance to medical school but also by MCAT scores, a more relevant measure for medical school admission than SAT scores. Second, we explored whether substantial changes in the educational outcomes might result from changes in either the demographic composition of medical school applicants or the expansion of the program from 5 to 6 years. Third, in addition to using inferential statistics to test the statistical significance of the differences between the program participants and the control group students, we also calculated the effect size estimates (standardized differences).

Prior empirical studies have addressed the educational outcomes of combined baccalaureate–MD programs in medical schools other than Jefferson’s. A shortcoming of studies of combined baccalaureate–MD programs has been comparing the program participants with their regular classmates,¹⁸ not with a matched control group. To our knowledge, the only exception is the Callahan et al study described above,¹⁷ in which SAT scores were used to select the control group. Of course, this study is also on the combined baccalaureate–MD program at Jefferson.

Lanzoni and Herbert¹⁹ reported that students in the combined degree program at Boston University performed better than regular students on medical licensing examinations, but they detected no significant difference between the 2 groups in class rank, attrition rate, honors at graduation, clerkship grades, board certification status, or career choices. Jacobs and colleagues²⁰ found that students in the University of Miami combined degree program performed at an academic level in medical school equal to or better than that of their classmates in the regular MD program.

Grant and colleagues²¹ assessed the combined degree programs at 4 schools (Boston University School of Medicine, City College of New York School of Medicine, University of Michigan Medical School, and University of Missouri–Kansas City) and concluded that participants in the combined programs had lower attrition rates compared with regular students; in contrast, Drees et al¹ reported a significantly higher attrition rate (20.6%) among students in the University of Missouri–Kansas City School of Medicine’s combined degree program than those of traditional four-year medical schools. Green and colleagues²² reported no significant difference on performance measures between students in the accelerated program and their classmates at the Northwestern University School of Medicine.

Our findings regarding the satisfaction ratings do not align with those reported by Gottheil et al³ or Daubney et al.¹⁸ Gottheil and colleagues reported that combined degree program participants expressed more satisfaction with medical school than their regular classmates. Similarly, in their literature review, Daubney et al¹⁸ reported that the combined baccalaureate–MD programs were successful based on students’ reported satisfaction. To further examine our own findings regarding satisfaction, we conducted additional analyses, comparing the satisfaction scores of Asian students with those of the rest of their classmates—both among the program participants and within the control group. We found no statistically significant differences, indicating that Asian ethnicity was not a confounding factor in students’ satisfaction ratings for different years of medical school or overall (globally). We speculate that workload, academic pressure, and a lack of leisure time at the beginning of college life could contribute to the program participants’ lower satisfaction. More research is needed to confirm this speculation.

Empirical findings on specialty choice of participants in combined programs are mixed. We found no substantial difference in specialty choice between program participants and control group students, which is consistent with the findings of Lanzoni and Herbert.¹⁹ Yet, 2 other studies—by Grant et al²¹ and by Green et al²²—report that a greater percentage of graduates of combined degree programs chose primary care compared with graduates of the regular medical curriculum. Roman and McGanney²³ report that the Sophie Davis School of Biomedical Education combined BS–MD program broadened medical career access among underrepresented minority (URM) students and increased the number of URM students practicing primary care specialties at a higher rate than the national average. Bloom et al⁴ reported that more graduates of the Gannon-Hahnemann Family Medicine combined BS–MD program, designed to train family physicians for medically underserved areas, pursued a family medicine residency compared with regular admission students.

To our knowledge, no prior research has addressed debt in accelerated baccalaureate–MD programs. Our findings, consistent with a previous study from our team,¹⁷ confirm that accelerated programs can mitigate the fear of accumulating educational debt in pursuing a medical career.^4,17,24 Notably, we expected the accelerated program participants to have lower educational debt compared with the control group students due to their shorter duration at Penn State. Baccalaureate debt accrued by students in the control group came from both public and private colleges, where the cost could have differed from that of Penn State. Thus, the observed difference in baccalaureate debt could be not only a function of duration but also of variation in cost at different undergraduate colleges.

Exploring reasons for the difference in educational debt accrued during medical school between the 2 groups, despite their comparable time actually in medical school, calls for further research. One speculation is that the amount of money saved by program participants during their baccalaureate education was spent on medical school tuition, leading to reduced medical school debt. Another speculation may be related to a perceived cultural factor; monetary debt is viewed especially negatively in Asian families, who may consider students’ debt an indication of brushing aside parental financial responsibilities. A higher proportion of Asian students among program participants could, at least partially, explain the lower educational debt observed in program participants. This cultural factor might also explain the larger number of Asian students among program participants. For example, Xu and colleagues²⁵ reported that Asian American medical students are strongly encouraged by their parents to take advantage of every opportunity to pursue higher education, particularly in medical fields. All of these speculations await empirical scrutiny.

Eaglen and colleagues⁵ have reported that, in addition to reducing the costs, combined baccalaureate–MD programs mitigated certain adverse events or possibilities in medical education, including protracted schooling and unwanted redundancies in learning. They also suggest that the programs helped reduce the competitive pressure associated with preparing for medical school admission, alleviated the so-called “premedical syndrome” (including excessive sense of competitiveness),²⁶ and provided a faster transition to an income-generating career and patient care.⁵

Conclusions

We assessed decades’ worth of educational and professional outcomes of an accelerated combined BS–MD program that began more than 50 years ego. Results provide evidence that it is feasible to reduce the cost and time required to obtain an MD degree without compromising educational outcomes and professional competence. Additional benefits include avoiding redundancies in the baccalaureate and medical school curricula and graduating younger physicians with longer career spans serving the public greater good.

Acknowledgments:

The authors would like to thank Pamela Walter and Jennifer Wilson for their editorial assistance.