Customize your JAMA Network experience by selecting one or more topics from the list below.
Andriole DA, Whelan AJ, Jeffe DB. Characteristics and Career Intentions of the Emerging MD/PhD Workforce. JAMA. 2008;300(10):1165–1173. doi:10.1001/jama.300.10.1165
Context MD/PhD program participants represent only a small proportion of all US medical students, yet they are expected to play a major role in the future physician-scientist workforce. The characteristics and career intentions of recent MD/PhD program and other MD program graduates have not been compared.
Objective To identify factors associated with MD/PhD program graduation among recent medical graduates.
Design, Setting, and Participants Deidentified data were obtained from a nationally representative sample of 88 575 US medical graduates who completed the national Association of American Medical Colleges Graduation Questionnaire from 2000-2006. Multivariate logistic regression was used to test responses to items regarding graduates' characteristics and career plans in association with program graduation, reporting adjusted odds ratios (ORs).
Main Outcome Measure MD/PhD program graduation.
Results Of the 79 104 respondents with complete data (71.7% of all 2000-2006 graduates), 1833 (2.3%) were MD/PhD program graduates. Variables associated with greater likelihood of MD/PhD program graduation included planned substantial career involvement in research (OR, 10.30; 95% confidence interval [CI], 8.89-11.93); lower educational debt (compared with ≥$150 000: $100 000-$149 999, OR, 1.85; 95% CI, 1.35-2.52; $50 000-$99 999, OR, 5.50; 95% CI, 4.14-7.29; $1-$49 999, OR, 17.50; 95% CI, 13.30-23.03; no debt, OR, 17.41; 95% CI, 13.22-22.92); and receipt of medical school scholarships or grants (OR, 3.22; 95% CI, 2.82-3.69). Compared with planned training in internal medicine, MD/PhD graduation was positively associated with planned training in dermatology, neurology, ophthalmology, pathology, pediatrics, or radiology. Variables associated with lower likelihood of MD/PhD graduation included female sex (OR, 0.68; 95% CI, 0.60-0.77); race/ethnicity underrepresented in medicine (OR, 0.64; 95% CI, 0.52-0.80); and, compared with internal medicine, planned training in emergency medicine (OR, 0.58; 95% CI, 0.40-0.84) or surgery (OR, 0.70; 95% CI, 0.57-0.85).
Conclusion Compared with graduates of other MD degree programs, MD/PhD graduates tend to be less demographically diverse, have a lower debt burden, favor different medical specialties, and have greater planned career involvement in research.
Joint MD/PhD programs are currently offered at nearly every US allopathic medical school in a wide array of fields.1 Many MD/PhD programs receive institutional support through the Medical Scientist Training Program (MSTP). The MSTP, established by the National Institutes of Health (NIH) National Institute of General Medical Sciences, currently funds 40 programs involving 45 degree-granting institutions.2 Approximately 75 US allopathic medical schools offer MD/PhD degree programs not supported by MSTP grants.2 Historically, the primary intent of these joint MD/PhD degree programs, regardless of their funding mechanisms, has been to produce highly trained physician-scientists who will engage in biomedical science research careers.
The medical education environment in which MD/PhD programs are offered has changed markedly in the past 15 years. The demographic characteristics of medical school graduates have changed; most notably, the proportions of women among matriculating medical student populations have steadily increased and are now approaching parity.3 Specialty choice preferences for graduate medical education (GME) among recent US allopathic medical graduates have shifted away from the generalist specialties of family medicine, internal medicine, and pediatrics as greater proportions of graduates have chosen to pursue GME in “controllable lifestyle” specialties.4,5 In addition, professional setting options for physicians have expanded well beyond the dichotomy of academic medicine or private clinical practice.6 Moreover, US allopathic medical graduates are faced with steadily increasing debt loads.7
The extent to which the characteristics and professional plans of recent US allopathic medical graduates may differ between MD/PhD program graduates and other MD program graduates has not been investigated. Therefore, we sought to identify factors associated with MD/PhD program graduation and, among MD/PhD program graduates, with the extent of their planned career involvement in research.
Since 1978, the Association of American Medical Colleges (AAMC) has annually administered the national Medical School Graduation Questionnaire (GQ) to all graduating US allopathic medical students in the spring of their final year.8 Items included on the GQ vary somewhat each year but cover a broad range of topics important to medical students and medical educators. Until 2005, the GQ was administered in its entirety on a confidential but identifiable basis. Since 2005, the GQ has been administered as a 2-part questionnaire. The GQ Program Evaluation Survey is administered anonymously; after completing this section, students then have the option to complete the supplementary GQ Student Survey on Priorities in Medical Education: Diversity, Career Aspirations, and Indebtedness, in which respondents are identifiable.9
We analyzed data for individual respondents who completed the national GQ from 2000 through 2006 (including, in 2005 and 2006, the supplementary GQ Student Survey on Priorities in Medical Education). These GQ records were linked to the AAMC Matriculating Student Questionnaire (MSQ) records, which allowed us to identify the degree program at enrollment reported by those GQ respondents in our sample who also had completed the MSQ. Because all of these existing data were deidentified by the AAMC prior to our analysis, the institutional review board at Washington University School of Medicine approved this study as non−human-subjects research.
Degree-program enrollment at matriculation and degree program completed at graduation were obtained from student responses to parallel items on the MSQ (“Type of degree program in which you are enrolled”) and GQ (“Type of degree program from which you are graduating”). For analysis, we created a 3-category variable for program enrollment at matriculation, comprising joint MD/PhD programs, all other degree programs (including MD/other advanced-degree programs and MD-only programs, which included BA/MD or BS/MD programs), and unknown (for those GQ respondents who did not complete the MSQ). We created a 2-category variable for program graduation, comprising joint MD/PhD program graduation and all other MD programs (including MD/other advanced-degree programs and MD-only programs, which included BA/MD or BS/MD programs).
We examined 11 GQ items in association with the first dependent variable of interest (MD/PhD graduation) in a logistic regression model including our entire study sample; we examined 9 GQ items in association with the second dependent variable of interest (planned substantial career involvement in research) in a logistic regression model including only MD/PhD program graduates.
Demographic variables were year of graduation, sex (men vs women), age at graduation (≤29 years vs 30-32 or ≥33 years) and self-identified race/ethnicity, which students reported from a list of options on the GQ (categorized as white vs Asian/Pacific Islander; other or unknown race/ethnicity; or racial/ethnic groups considered underrepresented minorities in medicine relative to their numbers in the general population, including black, Hispanic, and American Indian/Alaska Native). We included GQ respondents' self-identified race/ethnicity because of national concerns about biomedical research workforce diversity.10,11 Respondents who chose not to answer the race/ethnicity item comprised the unknown group; we combined this group with other race/ethnicity for analysis, since we were unable to determine race/ethnicity for either group.
We included 2 items pertaining to student financing for medical school: total educational (both premedical and medical school) debt at graduation (≥$150 000 vs $100 000-$149 999, $50 000-$99 999, $1-$49 999, or no debt) and receipt of scholarships or grants for medical school (no vs yes).
We included 3 career-intention variables: specialty choice for specialty-board certification, planned career involvement in research, and career setting preference. From the 31 response options for specialty choice on the GQ, we created 14 specialty-choice categories based on preliminary logistic regression models of associations between MD/PhD graduation and GQ specialty choice. We combined specialties having similar effects (ie, in the same direction and of approximately the same magnitude when each specialty was compared with internal medicine as the reference category). Specialty categories used in subsequent analyses included dermatology, emergency medicine, family medicine, internal medicine, neurology, obstetrics and gynecology, ophthalmology, pathology, pediatrics, psychiatry, radiology, surgery (including all surgical specialty choices), all other GQ specialty choices, and no specialty selected. Internal medicine was chosen as the reference category because it has remained the most frequently selected specialty choice among recent graduates of US allopathic medical schools (both MD/PhD and MD) and accounts for the largest proportion of all Accreditation Council for Graduate Medical Education–accredited GME positions in the United States.12,13
Response choices for planned career involvement in research used a 5-point scale (1 = “not involved,” 2 = “involved in a limited way,” 3 = “somewhat involved,” 4 = “significantly involved,” and 5 = “exclusively involved”). For the purposes of analysis in the regression models, we created a dichotomous variable for planned career involvement in research to distinguish between planning not substantial (including “not involved,” “involved in a limited way,” and “somewhat involved”) and substantial (including “significantly involved” and “exclusively involved”) involvement in research. For analysis, we also created 6 career setting preference choices from all the career setting preferences included in the GQ: “full-time (nonuniversity) clinical practice,” “full-time university faculty: basic science teaching/research,” “full-time university faculty: clinical science teaching/research,” “nonuniversity research scientist,” “other” (including state/federal or medical/health care administration), and “undecided.” We included 2 variables for characteristics of the medical school attended: ownership of school (public vs private) and school region (Northeast vs Midwest, South, or West).
χ2 Tests were used to measure the significance of associations between categorical variables. Analysis of variance was used to measure the significance of associations between continuous variables and categorical variables. Unadjusted logistic regression models were used to test the significance of associations between each predictor variable and the 2 dependent variables of interest: MD/PhD program graduation (MD/PhD program graduates vs all other MD-degree graduates) and extent of planned career involvement in research (“substantial” vs “not substantial”). We report descriptive statistics for each variable of interest as well as adjusted odds ratios and 95% confidence intervals from the multivariate logistic regression models. All tests were performed using SPSS version 14.0.2 (SPSS Inc, Chicago, Illinois). All P values were 2-sided; P < .05 was considered significant.
A total of 88 575 US allopathic medical students completed the GQ (in full or in part) from 2000 through 2006, representing 80.3% of all 110 253 US allopathic medical school graduates in these classes.8 Overall annual response rates varied from 60.8% in 2005 to 91.7% in 2000. Of these 88 575 respondents, 9471 (10.7%) were excluded from our analysis because they did not respond to 1 or more of the GQ items of interest for our study. Thus, our analysis included individualized records for 79 104 respondents with complete data for all GQ items of interest. These 79 104 respondents represented 89.3% of all 88 575 GQ respondents and 71.7% of all 110 253 US allopathic medical school graduates in these classes.8,14 Of these 79 104 respondents, 1833 (2.3%) were MD/PhD program graduates. The proportion of graduates in each year who were MD/PhD program graduates ranged from 2.0% to 2.5%, with no significant change over time (P = .52).
Table 1 shows the descriptive statistics for each categorical variable grouped by degree program at graduation. Univariate associations between each variable shown in Table 1 and MD/PhD program graduation were all statistically significant at P < .001. Of the 1859 MD/PhD program enrollees at the time of matriculation to medical school, 1330 (71.5%) reported graduating from MD/PhD degree programs and 529 (28.5%) did not. The proportions of these 1859 MD/PhD program enrollees at matriculation who ultimately graduated from MD/PhD degree programs differed significantly by sex (men, 916/1245 [73.6%]; women, 414/614 [67.4%]; P = .006) and by race/ethnicity (white, 891/1224 [72.8%]; Asian/Pacific Islander, 339/475 [71.4%]; other/unknown, 4/13 [30.8%]; underrepresented minorities, 96/147 [65.3%]; P = .002). Most of the MD/PhD program graduates planned to enter careers involving research; 83.8% (1537/1833) planned full-time university clinical or basic science teaching and research, and 81.2% (1489/1833) planned substantial career involvement in research.
A multivariate logistic regression model identified predictors of MD/PhD program graduation compared with all other MD program graduates (Table 2). Graduates were more likely to be MD/PhD program graduates if they were older; had less than $150 000 of debt; had received scholarships or grants for medical school; planned specialty training in dermatology, neurology, ophthalmology, pathology, pediatrics, or radiology, compared with internal medicine; planned substantial career involvement in research; planned full-time university faculty careers, nonuniversity research scientist careers, or were undecided about career setting preferences; or graduated from privately owned medical schools. Graduates were less likely to be MD/PhD program graduates if they were women; of underrepresented minority or other/unknown race/ethnicity; planned specialty training in emergency medicine or surgery specialties, compared with internal medicine; or graduated from schools in the South or West. The variables in this model together accounted for 51.1% of the variance of MD/PhD program graduation.
Table 3 shows the descriptive statistics for categorical variables grouped by the extent of planned career involvement in research among the 1833 MD/PhD program graduates. In addition, graduates in more recent years were more likely to plan substantial career involvement in research (P = .02). Since the response choices for preference of career setting specified types of research careers as options (the word “research” was included in some response choices), and this variable was strongly associated with and accounted for 31.2% of the variance of planned substantial career involvement in research in an unadjusted logistic regression model, we did not include the variable for career setting preference in the multivariate model.
Table 4 shows the results of the multivariate regression analysis across all career setting preferences to identify variables independently associated with planned substantial career involvement in research. Graduates were more likely to plan substantial career involvement in research if they had graduated in more recent years; were older; had less than $100 000 of debt; had received medical school scholarships or grants; or planned specialty training in neurology, compared to internal medicine. Graduates were less likely to plan substantial career involvement in research if they were women or if they planned specialty training in emergency medicine, family medicine, ophthalmology, psychiatry, radiology, or surgery or had selected no specialty. The variables in this model together accounted for 19% of the variance in planned substantial career involvement in research among MD/PhD program graduates.
These findings should be considered in the context of 4 recent trends in US medical education: (1) increasing demographic diversity of the US medical student populations, (2) a shift in preferences of GME specialty among US medical graduates away from generalist specialties, (3) an expansion of the range of career setting options for US medical graduates, and (4) a steady increase in the debt burden of recent graduates.
MD/PhD program graduates represented a less demographically diverse group compared with the overall non–MD/PhD population of recent graduates of US allopathic medical schools. Although the proportion of women medical graduates in the United States has steadily increased in the past 10 years, women remained less likely to have graduated from MD/PhD programs, reflecting both the relatively low proportion of women among MD/PhD program enrollees at matriculation and women's overrepresentation among MD/PhD program noncompleters in our sample. Among MD/PhD program graduates, women were also less likely than men to plan substantial career involvement in research.
The substantial increase in numbers of MD/PhD program enrollees nationally at US allopathic medical schools since 2000 has largely been due to increased numbers of women enrollees; the number of men among MD/PhD program enrollees has remained relatively stable.15 Because women comprise nearly 40% of current enrollees in MD/PhD programs, the proportion of women among MD/PhD program graduates in the biomedical research workforce could increase in coming years.16 However, our findings suggest that these expectations may not be fully realized. As the number of women enrollees increase, attrition from MD/PhD programs may also increase. Concerns about combining a physician-scientist's career with childbearing and family life as well as concerns about lack of physician-scientist role models have been cited as reasons for the ongoing underrepresentation of women among physicians pursuing research careers.17 Our finding of an association between female sex and lower likelihood of planning substantial career involvement in research among MD/PhD program graduates, along with that of a recent single-institutional study of MD/PhD program enrollees in which a lower proportion of women than men identified research as a primary professional goal,18 suggests that sex differences in career intentions among MD/PhD program graduates also may contribute to ongoing disparities in the biomedical research workforce.
While we observed differences in MD/PhD program graduation on the basis of race/ethnicity, this was not a predictor of planned substantial career involvement in research among MD/PhD program graduates. This observation suggests that strategies that successfully increase the number of underrepresented minority graduates of MD/PhD programs should benefit diversity in the biomedical research workforce, an issue of ongoing national concern.10,11
MD/PhD program applicants represent a group with strong academic qualifications. However, due to the high level of competition for entry into MD/PhD programs, not all MD/PhD program applicants receive interview offers. In 2007, more than one-third of MD/PhD program applicants who received interview offers did not receive any program acceptances.16 Given this selectivity for MD/PhD program participation and the level of financial and institutional support provided to many MD/PhD program enrollees, our finding that more than 1 of 4 students enrolling in MD/PhD programs at the time of medical school matriculation did not complete the MD/PhD program seems high and warrants further investigation.
Our finding that older age was associated with increased likelihood of graduation from an MD/PhD program is not surprising, because MD/PhD students typically take 7 to 8 years to complete their dual-degree programs.16 Older age was also associated with increased likelihood of planned substantial career involvement in research among MD/PhD program graduates. This association could reflect a longer, more comprehensive period of research to fulfill PhD thesis requirements. It also might reflect differences among MD/PhD program graduates' experience in research before entering medical school: the MD/PhD Students' Attitudes, Goals, and Education survey report documented that approximately 30% of MD/PhD program respondents had entered medical school with at least 1 year of full-time research experience before matriculation.19
Over the past 10 years, increasing proportions of US allopathic medical graduates have entered GME in nongeneralist specialties, including those characterized as controllable-lifestyle specialties.4,5 Historically, MD/PhD program graduates typically pursued GME training in internal medicine, neurology, pathology, or pediatrics. However, students in contemporary MD/PhD programs are reportedly interested in a much broader range of specialties, including both surgical and controllable-lifestyle specialties.19 While surgery was the specialty choice for 1 of 8 (12.1%) MD/PhD program graduates in our study, surgery as a specialty choice was associated with a lower likelihood of graduation from an MD-PhD program. Some controllable-lifestyle specialty choices, such as dermatology, ophthalmology, and radiology, which are competitive for entry, were associated with a greater likelihood of MD/PhD program graduation.12,20 Ophthalmology and radiology were among the specialty choices associated with a lower likelihood of planned substantial career involvement in research among MD/PhD program graduates.
For some specialties, the specialty choices associated with graduation from an MD/PhD program likely reflect not only the preferences of MD/PhD program graduates but also those of residency program directors for US medical school applicants with PhD degrees. In the 2006 match, for example, the National Resident Matching Program reported that US senior applicants graduating with PhD degrees who ranked positions in highly competitive specialties were nearly 3 times as likely to match to these positions compared with US senior applicants graduating without PhD degrees.12
Although the range of career setting options for US medical school graduates has expanded, most MD/PhD program graduates in our sample indicated career plans that aligned with MD/PhD programmatic goals: more than 80% of these graduates planned full-time academic medicine careers involving teaching and research and planned substantial career involvement in research. Despite fluctuations in levels of NIH funding for biomedical research during the period of our study, graduation in more recent years was among the variables associated with an increased likelihood of planned substantial career involvement in research among MD/PhD program graduates.21 However, only 5.3% (98/1833) of the MD/PhD program graduates in our sample planned careers exclusively devoted to research, and most graduates who planned careers exclusively devoted to research were not MD/PhD program graduates (68.7% [215/313]). Our findings suggest that many contemporary MD/PhD program graduates may not plan to pursue predominantly research-based careers. This was similarly suggested by the findings of the Students' Attitudes, Goals, and Education survey, in which just over half of the MD/PhD student respondents indicated research as their desired primary professional activity.19
Our results should be considered in the context of the increasing debt burden among recent US medical graduates. Among MD/PhD program graduates in our study, receipt of a medical school scholarship or grant was associated with a greater likelihood of planned substantial career involvement in research, independent of the relationship between lower debt and greater likelihood of planned substantial career involvement in research. A wide range of scholarship and grant support is available for selected MD/PhD program participants (including institutionally awarded MSTP grants and individually awarded MD/PhD fellowships available to students through numerous NIH institutes, among other scholarships or grants).2,22-24 It was therefore not surprising that receipt of a scholarship or grant was associated with an increased likelihood of graduation from an MD/PhD program. However, the association of support with an increased likelihood of planned substantial career involvement in research among MD/PhD program graduates suggests that important differences exist among MD/PhD program participants on the basis of support that likely reflect characteristics of the students as well as the institutions at which they conduct their PhD research. Medical Scientist Training Program and other institutional NIH funding, as well as many individual MD/PhD fellowships, are selectively awarded on a competitive basis. This selectivity likely affects MD/PhD program outcomes in ways that reflect MD/PhD program objectives to train and produce a cadre of physicians who will pursue basic science or clinical research careers after GME training.25
Our finding of much lower debt associated with MD/PhD program graduation was also expected, given the range and magnitude of financial support available for selected MD/PhD program participants. Because MD/PhD program participants typically take about twice as long as their MD-degree counterparts to complete program requirements for graduation, thus delaying their entry into the employed-physician workforce, financial support for MD/PhD program participation seems justified to minimize their debt load at graduation. However, more than 10% of MD/PhD program graduates in our study reported debt loads of at least $100 000. Among graduates with any educational debt, only debt loads less than $100 000 were associated with an increased likelihood of planned substantial involvement in research. Therefore, loan repayment programs designed to encourage heavily indebted physician-scientists to enter research careers may be of particular value.26
Our study has notable strengths and limitations. A strength was the inclusion of a large, nationally representative sample of MD/PhD program and all other MD program graduates in the 2000-2006 US allopathic medical school graduating classes. Graduates of MD/PhD programs have comprised approximately 2.5% of all US allopathic medical school graduates,27 and 2.3% of our sample were MD/PhD program graduates. The sample was representative of all graduates in terms of sex and race/ethnicity.3,14 The proportions of graduates from public and private medical schools in our study sample reflected the ownership status of all US allopathic medical schools during the study period (approximately 60% public and 40% private ownership).3 The relatively low proportions of graduates in our study sample from the West, compared with the proportions of graduates from all other geographic regions, is consistent with the geographic distribution of US allopathic medical schools. The associations we observed between graduation from an MD/PhD program and school ownership and geographic region are consistent with the distribution of MSTP-funded programs among US medical institutions.28,29
Although we analyzed GQ data over a 7-year period, a limitation was that these data were cross-sectional, representing US medical graduates' perspectives at the time of their graduation only. We cannot make causal inferences from our analyses, because we were unable to validate plans at time of graduation with future behavior. Career plans might change during GME or fellowship training; debt, family status, and other experiences also might influence change in career plans after graduation. Nevertheless, intention to engage in a behavior is highly predictive of future engagement in that behavior,30 and at least 1 study reported that graduates who indicated on the GQ that they intended to practice in underserved communities were more likely to actually do so.31 The predictive validity of responses to the career intention items included in our analysis of career setting preferences and planned career involvement in research has not been established; such validity is the subject of future research.
Our finding that the variables in our model accounted for only 19% of the total variance in planned substantial career involvement in research among MD/PhD program graduates was not unexpected. Many unmeasured variables are likely to be associated with this outcome measure, but this information is not collected on the GQ. The graduates' planned career involvement in research may be influenced by duration of PhD thesis research, research field (MD/PhD students now have opportunities to pursue PhD research in an ever-widening array of fields), location for thesis research, student academic performance, type and amount of financial support, research productivity, and qualities of the student's research mentor.
Most of the graduates in our study will complete GME and enter their chosen careers within the next few years. It is feasible for researchers at MD/PhD programs across the country to link these critical types of student data, accessible only to individual MD/PhD program directors, with individualized GQ responses for their graduates and with information regarding career setting and research involvement. Such a study could determine the predictive validity of the career-intention items on the GQ among MD/PhD program graduates. Perhaps more important, it could assess the extent to which MD/PhD programs at medical schools across the country are fulfilling their missions and goals in the current medical education environment.32
In summary, contemporary MD/PhD program graduates differed in many respects from their MD-degree counterparts. The MD/PhD program graduates in our sample represented a heterogeneous group in many regards, including specialty choice, debt load, and planned career involvement in research. The number of MD/PhD program graduates is likely to increase in the next few years, reflecting the steadily increasing number of MD/PhD program enrollees among students who matriculated in US allopathic medical schools from 2000 to 2006. Graduates of MD/PhD programs accounted for 3.1% of all medical graduates in 2007, and this proportion may increase further as current MD/PhD program enrollees complete the 7 to 8 years of education required for most MD/PhD program participants to complete their dual-degree requirements.16 The results of this study could inform a research agenda to more fully explore the extent to which MD/PhD program goals are indeed being met in the US allopathic medical education system.
Corresponding Author: Dorothy A. Andriole, MD, Washington University School of Medicine, 660 S Euclid Ave, Campus Box 8210, St Louis, MO 63110 (firstname.lastname@example.org).
Author Contributions: Dr Jeffe had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Andriole, Jeffe.
Acquisition of data: Andriole, Jeffe.
Analysis and interpretation of data: Andriole, Whelan, Jeffe.
Drafting of the manuscript: Andriole, Jeffe.
Critical revision of the manuscript for important intellectual content: Andriole, Whelan, Jeffe.
Statistical analysis: Jeffe.
Financial Disclosures: None reported.
Funding/Support: This study was not supported by any external funds.
Disclaimer: The conclusions of the authors are not necessarily those of the Association of American Medical Colleges (AAMC) or AAMC staff.
Additional Contributions: We thank our AAMC colleagues Rajeev Sabharwal, MPH, Jason Cantow, MS, and David Matthew, PhD, for provision of the data and assistance with coding. These individuals received no financial compensation for their assistance.