Predictors of full-time faculty appointment among MD–PhD program graduates: a national cohort study

Purpose The authors sought to identify variables associated with MD–PhD program graduates’ academic medicine careers. Methods We analyzed data for a national cohort of MD–PhD program graduates from 2000 to 2005, using multivariable logistic regression to identify independent predictors of full-time academic medicine faculty appointment through 2013. Results Of 1,860 MD–PhD program graduates in 2000–2005, we included 1,846 (99.2%) who had completed residency training before 2014. Of these 1,846 graduates, 968 (52.4%) held full-time faculty appointments. Graduates who attended schools with Medical Scientist Training Program (MSTP) funding (vs. no MSTP funding; adjusted odds ratio [aOR], 1.41; 95% confidence interval [CI], 1.14–1.74) and participated in ≥1 year of research during residency (vs. no documented research year; aOR, 1.85; 95% CI, 1.50–2.28) were more likely to have held full-time faculty appointments. Asian/Pacific Islander (aOR, 0.74; 95% CI, 0.60–0.93) and under-represented minority (URM; aOR, 0.68; 95% CI, 0.48–0.98) graduates (each vs. white graduates), graduates who reported total debt of ≥$100,000 (vs. no debt) at graduation (aOR, 0.58; 95% CI, 0.39–0.88), and graduates in surgical practice (aOR, 0.64; 95% CI, 0.48–0.84) and other practice (aOR, 0.66, 95% CI, 0.54–0.81) specialties (each vs. ‘medicine, pediatrics, pathology, or neurology’) were less likely to have held full-time faculty appointments. Gender was not independently associated with likelihood of full-time faculty appointment. Conclusions Over half of all MD–PhD program graduates in our study had full-time faculty appointments. Our findings regarding variables independently associated with full-time faculty appointments can inform the design of strategies to promote academic medicine career choice among MD–PhD program graduates. Further research is warranted to identify other factors amenable to intervention, in addition to those included in our study, which will foster the further development of a diverse academic medicine physician–scientist workforce nationally.

E nrollment in MDÁPhD joint-degree programs at Liaison Committee for Medical Education (LCME)-accredited US medical schools experienced a period of substantial growth over the past 20 years. From 1993 to 2013, the total number of students who entered medical schools increased by 17%, while the number of students who were enrolled in MDÁPhD programs increased by over 100% (1). In the academic year 2014Á2015, there were 616 MDÁPhD program graduates (2), comprising 3.3% of the 18,705 US LCME-accredited medical school graduates in the academic year 2014Á2015 (3). MDÁPhD programs are currently offered at 83.4% (121/145) of all US LCMEaccredited medical schools (4), and 45 of these 121 programs (37.2%) receive National Institute of General Medical Sciences (NIGMS) Medical Scientist Training Program (MSTP) support (5). MDÁPhD program enrollees receive substantial financial support from medical schools themselves. In the academic year 2013Á2014, the most recent year for which data are available, non-needbased MDÁPhD support comprised 17% ($111.9 million/ $669.4 million) of all grants and scholarships (including both need-based and non-need based) that were awarded Medical Education Online ae Medical Education Online 2016. # 2016 Dorothy A. Andriole and Donna B. Jeffe. This is an Open Access article distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), allowing third parties to copy and redistribute the material in any medium or format and to remix, transform, and build upon the material for any purpose, even commercially, provided the original work is properly cited and states its license. without a service commitment by US medical schools (6). Thus, all medical schools with MDÁPhD programs (regardless of MSTP funding), as well as other federal agencies and private organizations, provide substantial financial support to MDÁPhD programs and their enrollees (7Á9).
Historically, MDÁPhD program graduates, particularly those of MSTP-funded programs, predominantly pursued careers in academic medicine (10,11). However, the extent to which more recent MDÁPhD program graduates may have chosen to pursue academic medicine careers and factors associated with this career path have not been examined in a national cohort. Thus, we conducted a retrospective cohort study to examine the extent to which recent MDÁPhD program graduates received full-time academic medicine faculty appointments and to identify variables associated with full-time faculty appointment. A review of the literature identified institutional MSTP funding (11), specialty choice for residency training (10), participation in substantive research experiences during graduate medical education (GME) (12), and total debt (9, 13) as potential predictors of career paths pursued by MDÁPhD program graduates. We hypothesized that MDÁPhD program graduates from an MSTP-funded medical school and who participated in substantive research experiences during GME would be more likely to have held a full-time faculty appointment; we also hypothesized that graduates with high levels of debt at graduation and who entered GME training in specialties other than internal medicine, pediatrics, neurology, or pathology, which are historically the most commonly selected specialties chosen by MDÁPhD program graduates (10), would be less likely to have held a full-time faculty appointment.
Medical students who are enrolled in MDÁPhD programs are eligible to apply for individual F30/F31 predoctoral awards (14,15), but the impact of F30/F31 awards on MDÁPhD program students' academic medicine career paths has not been explored. Thus, we also sought to determine whether there was a relationship between F30/31 awards and MDÁPhD program graduates' full-time academic medicine faculty appointment. Finally, we also sought to determine if there was a relationship between MDÁPhD program graduates' academic medicine faculty appointment and each of gender and race/ethnicity, as a key recommendation of the National Institutes of Health (NIH) Physician Scientist Workforce Working Group Report was that NIH should intensify its efforts to diversify the physicianÁscientist workforce (16). In 2015Á2016, women comprised 39.1% of all MDÁPhD program matriculants (17), and in 2014Á 2015 individuals from historically under-represented racial/ethnic minorities in medicine (URM) comprised 9.1% of all MDÁPhD program graduates (2). However, the extent to which gender or race/ethnicity may be associated with full-time academic medicine faculty appointment among MDÁPhD program graduates has not been explored (10, 11).

Method
Following Institutional Review Board approval at Washington University School of Medicine, a database was constructed with individually linked, de-identified records for all 129,867 students who had matriculated at one of the 129 LCME-accredited medical schools in the United States from 1993 to 2000 and who graduated between 2000 and 2005. We restricted our study sample to graduates in 2000Á2005 because MDÁPhD program enrollees take 7Á8 years, on average, to complete the program (9, 10), and we allowed at least 8 years of followup after graduation to allow sufficient time for most graduates to complete GME training prior to follow-up through 2013 (data acquired on February 27, 2014). The database included selected data from each of the Association of American Medical Colleges' (AAMC) Student Record System (SRS), Graduation Questionnaire (GQ), GME Track, and Faculty Roster, and from the NIH Information for Management, Planning, Analysis, and Coordination (IMPAC II) database and the American Medical Association (AMA) Physician Masterfile.
SRS data included matriculation date, graduation date, students' self-identified sex, and race/ethnicity as was reported from a list of options on the American Medical College Application Service questionnaire. We categorized race/ethnicity as Asian/Pacific Islander, URM (including Black, Hispanic, and American Indian/Alaska Native), other/unknown (including matriculants who selfidentified as 'other', as multiple races, or did not respond to this question), or white (reference group).
We used GQ data to create a 5-category variable for total debt at graduation (no debt [reference group], $1Á49,999, $50,000Á99,999, $100,000 or more, and missing for graduates who did not respond to the GQ or to this GQ item).
The AAMC also provided data for medical schools' MSTP-funding status based on rosters of MSTP-funded institutions updated annually by the NIGMS, as previously described (9); of the 129 US LCME-accredited medical schools in 1993Á2000, 39 (30.2%) had MSTP funding. Data for receipt (yes vs. no) of individual, federal, predoctoral F30/F31 awards were obtained from the NIH IMPAC II database. The NIH and AAMC contracted with Net ESolutions Corporation in Bethesda, MD, to conduct the record match. Grants data linked with individuals in our cohort by the AAMC and were provided to us on August 26, 2014.
We used GME Track data to distinguish between graduates who had completed ]1 year of research during GME, as indicated by their program director on the GME Census, and graduates for whom there was no documentation of research during GME.
We used AMA Physician Masterfile data for type of practice to exclude from analysis those MDÁPhD program graduates still in GME at end of 2013, as trainees are ineligible for full-time faculty appointments, and to create a three-category variable for practice specialty, for graduates no longer in GME: 'internal medicine, pediatrics, neurology or pathology', the specialties historically chosen most frequently by MDÁPhD program graduates (10), 'surgery', including all surgical specialties recognized by the American College of Surgeons (18), and 'other', including all other specialties and no designated specialty.
Data for full-time faculty appointment at a US LCMEaccredited medical school were obtained from AAMC Faculty Roster records (19). We created a dichotomous variable for ever having a full-time faculty appointment (yes vs. no) through 2013. Among those graduates with full-time appointments, we also included data for department type (clinical, basic science, or other), rank (associate professor, assistant professor, instructor, or other), and track (tenure eligible, nontenure track, and tenure not available at institution) of initial full-time faculty appointment.

Statistical analysis
We used chi-square tests to describe associations between two categorical variables. We report adjusted odds ratios (aOR) and 95% confidence intervals (CI) from a multivariate logistic regression model to identify independent predictors of full-time faculty appointment among all MDÁPhD program graduates in our study sample. All tests were performed using IBM SPSS Statistics, version 22 (IBM Corporation, 1989, 2013); two-sided P values B0.05 were considered significant.

Results
Of the 129,867 medical-school matriculants in our database, there were 3,420 MDÁPhD program graduates, including 1,860 who had completed the dual-degree program in academic years 1999Á2000 (2000) through 2004Á2005 (2005). Of these 1,860 MDÁPhD program graduates, we excluded 14 graduates who were still in GME at the end of 2013. Our final study sample of 1,846 MDÁPhD program graduates (99.2%) included 968 (52.4%) MDÁPhD program graduates who had held full-time faculty appointments by the end of 2013.
Descriptive statistics for the 1,846 MDÁPhD program graduates in our sample, grouped by full-time faculty appointment, are shown in Table 1. Full-time faculty appointment was significantly associated with each of medical-school MSTP funding, total debt at graduation, F30/F31 award receipt, practice specialty, and research during GME. Associations with gender, race/ethnicity, and graduation year were not significant. Table 2 shows the results of the regression model identifying variables independently associated with fulltime faculty appointment. Graduates who were Asian/ Pacific Islander and URM (each compared with white), who reported total debt of ]$100,000 at graduation (compared with no debt), and who were in surgery practice or other practice (each compared with 'internal medicine, pediatrics, neurology, or pathology') were less likely to have held a full-time faculty appointment. Graduates who attended medical schools with MSTP funding (compared with no MSTP funding), who received F30/F31 awards and who had participated in ]1 year of research during GME (compared with no documented research year(s) during GME) were more likely to have a full-time faculty appointment.
As shown in Table 3, 94.2% of faculty appointments were in clinical science departments, and 54.0% of faculty were initially appointed to assistant professor and 40.8% to instructor positions. Primary department type and rank of initial appointment did not differ significantly by gender or by race/ethnicity.

Discussion
Of all 1,846 MDÁPhD program graduates in our cohort, 52.4% had been appointed to a full-time, academic medicine faculty position. This proportion was substantially greater than the proportion of MD graduates without PhD (18%) who had full-time faculty appointments in the larger cohort, as previously reported (20,21). We also found that, compared with graduates of non-MSTPfunded schools, graduates of schools with MSTP funding were more likely to have held full-time faculty appointments. The overall proportion of MDÁPhD program graduates with a full-time faculty appointment that we observed was somewhat lower than the proportions observed among earlier cohorts of MDÁPhD program graduates, most of whom had been followed for much longer periods of time (11). In this earlier NIGMS study, faculty appointments among MDÁPhD program graduates from 1971 through 1990 were described; across all years through follow-up in 1995, 68% of non-MSTPfunded MDÁPhD program graduates at non-MSTP schools, 78% of non-MSTP-funded MDÁPhD program graduates at MSTP-funded schools, and 84% of MSTPfunded MDÁPhD program graduates had been employed in academic positions (11). Although the lower percentage of full-time faculty appointees among the MDÁPhD program graduates in our more recent cohort might increase with longer follow-up, our observations also might reflect an increase in the extent to which more contemporary MDÁPhD program graduates pursue career paths outside academic medicine, including research-related careers in nonmedical-school-affiliated research institutes (e.g., the NIH and industry) or full-time clinical practice in nonacademic settings. That 94.2% of initial primary appointments in our cohort were in clinical rather than basic science departments is consistent with the NIH Physician Scientist Workforce Report (16).
Our observation that receipt of F30/F31 predoctoral fellowship awards was associated with greater likelihood of full-time faculty appointment among MDÁPhD program graduates informs the evidence base for the role these predoctoral fellowship programs can play in the development of the physicianÁscientist workforce (16).
Concerns have been raised that the increasing debt load among medical school graduates might negatively impact physicianÁscientists' career paths; thus, the goal of the NIH Extramural Loan Repayment Program (LRP) is 'to attract and retain early career health professionals in biomedical and behavioral research careers (22)'. Two national studies of MD degree graduates (both excluded MDÁPhD program graduates) reported no significant relationship between debt and full-time, academic medicine faculty appointments, in general (21), or between  (Figure 19, p. 25). The lower likelihood of faculty appointment for heavily indebted MDÁPhD program graduates in our cohort provides support for the recommendation that the amount of loans forgiven in NIH LRPs should be increased to more realistically reflect the debt burden of current trainees (16). After controlling for total debt at graduation and other variables of interest, MDÁPhD program graduates in the surgery and the other practice categories were less likely than graduates in the 'internal medicine, pediatrics, neurology, and pathology' practice category to have had faculty appointments. Lengthier GME requirements for general certification by member boards of the American Board of Medical Specialties might explain these observations in part for graduates in surgery (24). However, our findings also may reflect choices among MDÁPhD program graduates in some specialties other than internal medicine, pediatrics, neurology and pathology to pursue career paths that diverge from more traditional researchrelated, academic medicine career paths (10).
About 30% of all MDÁPhD program graduates in our study were documented by their residency training program directors to have spent ]1 year in research during GME; by comparison, only 9% of MD-degree graduates in this national cohort were documented to have spent ]1 year in research during GME (21). Options for substantive research experiences during GME may be particularly attractive to MDÁPhD program graduates aspiring to physicianÁscientist careers. In single-site studies of residency training programs offering substantive research experiences, high proportions of residents are MDÁ PhDs (25,26). In addition, results of a survey conducted by the American Board of Internal Medicine of individuals who completed Research Pathway residency training in internal medicine indicated that 240 (64.7%) of 371 respondents had met the requirements for a doctoral degree (e.g., PhD) prior to the start of Research Pathway training (12). MDÁPhD program graduates in our cohort who participated (vs. did not participate) in ]1 year of research during GME were more likely to have held a fulltime faculty appointment. As we controlled for specialty category in the model, this finding suggests that MDÁPhD program graduates with substantive research experience during GME do not have a greater propensity for academic careers simply because they tend to be in specialties that offer more opportunities for research during GME and a stronger pipeline toward faculty appointments. The proportion of female MDÁPhD program graduates has more than doubled over the past 20 years; women comprised less than 20% of MDÁPhD program graduates in 1995 (11) and nearly 42% of all MDÁPhD program graduates in 2011, the most recent year for which these data are available (27). That we found no significant association between gender and full-time faculty appointment suggests that increases in the number of female MDÁPhD program graduates should result in greater numbers of female MDÁPhDs in academic medicine. We also observed gender parity with regard to the rank of initial faculty appointment ( Table 3).
The racial/ethnic diversity of MDÁPhD program graduates also has increased over the past 20 years. In 1995, 75% (236/313) of MDÁPhD program graduates were white; in academic year 2014Á2015, this percentage declined to 61% (376/616) (2, 28). URM graduates and Asian/Pacific Islander graduates in our cohort were less likely than white graduates to have held full-time faculty appointments. We did not, however, observe significant associations between race/ethnicity and each of department type and rank at initial faculty appointment. As there remains a critical need for greater academic medicine workforce diversity in the United States (29Á31), particularly regarding greater inclusion of URM graduates in academic medicine, our findings suggest that continued targeted efforts to identify and recruit URM MDÁPhD program graduates to academic medicine careers are warranted.
Our study had a number of strengths. We included information for several different variables that have been identified in the literature as possible predictors of MDÁPhD program graduates' career paths but have not previously been explored for their associations with academic medicine careers, and we had data for a national cohort of MDÁPhD program graduates, all with a minimum follow-up of 8 years after graduation. Unlike previous studies of academic medicine faculty appointment among MDÁPhD program graduates that utilized data obtained from curricula vitae provided by graduates themselves (11) or data from MDÁPhD program directors (10), we obtained AAMC Faculty Roster data for all individuals in our cohort with faculty appointments in US LCME-accredited medical schools.
Our study also has several limitations. Because we did not have programmatic information for the MDÁPhD programs in which the graduates in our sample had been enrolled, it is possible that outcomes for graduates of specific MDÁPhD programs may be different from the findings we report here. We did not have information about graduates' PhD degree fields. MDÁPhD programs offer PhD in a wide range of fields other than biomedical and physical sciences (7, 32), but it remains unknown whether and to what extent there may be a relationship between MDÁPhD program graduates' PhD degree fields and the likelihood of an academic medicine faculty appointment. We also lacked information about whether MDÁPhD program graduates participated in GME programs that offered research experience in an integrated manner throughout residency (26,33), rather than as a designated elective research year(s).
Despite these limitations, our observations may be of interest to many agencies and organizations that provide funding to MDÁPhD programs and those that provide financial support for LRPs in which MDÁPhD program graduates may participate. Our findings regarding variables associated with initial full-time faculty appointment among MDÁPhD program graduates also can inform the  efforts of US medical schools seeking to recruit a diverse and highly qualified academic medicine physicianÁ scientist workforce well-positioned to advance our national biomedical-research agenda.