A nationwide assessment of perceptions of research-intense academic careers among predoctoral MD and MD-PhD trainees

Introduction: While previous studies have described career outcomes of physician-scientist trainees after graduation, trainee perceptions of research-intensive career pathways remain unclear. This study sought to identify the perceived interests, factors, and challenges associated with academic and research careers among predoctoral MD trainees, MD trainees with research-intense (>50%) career intentions (MD-RI), and MD-PhD trainees. Methods: A 70-question survey was administered to 16,418 trainees at 32 academic medical centers from September 2012 to December 2014. MD vs. MD-RI (>50% research intentions) vs. MD-PhD trainee responses were compared by chi-square tests. Multivariate logistic regression analyses were performed to identify variables associated with academic and research career intentions. Results: There were 4433 respondents (27% response rate), including 2625 MD (64%), 653 MD-RI (15%), and 856 MD-PhD (21%) trainees. MD-PhDs were most interested in pursuing academia (85.8%), followed by MD-RIs (57.3%) and MDs (31.2%). Translational research was the primary career intention for MD-PhD trainees (42.9%). Clinical duties were the primary career intention for MD-RIs (51.9%) and MDs (84.2%). While 39.8% of MD-PhD respondents identified opportunities for research as the most important career selection factor, only 12.9% of MD-RI and 0.5% of MD respondents shared this perspective. Interest in basic research, translational research, clinical research, education, and the ability to identify a mentor were each independently associated with academic career intentions by multivariate regression. Conclusions: Predoctoral MD, MD-RI, and MD-PhD trainees are unique cohorts with different perceptions and interests toward academic and research careers. Understanding these differences may help to guide efforts to mentor the next generation of physician-scientists.


Introduction
The Declining Physician-Scientist Workforce Physician-scientists are an important part of the academic medical enterprise. Combining skills and knowledge from medical and scientific training, physician-scientists are uniquely positioned to advance academic medicine by answering critical scientific questions related to human health and disease [1,2]. Despite this importance, the number of physician-scientists has been declining for decades [3][4][5]. Recognition of this decline has spurred a number of groups to examine how to rejuvenate physician-scientist training [6].
Recognition of diminishing physician-scientist participation in academic medical research is not new. An assessment of National Institutes of Health (NIH) grant applications and awards through the early 2000s revealed reduced success among MD applicants despite a stable number of grant applications from this group [7]. MD participation in NIH-funded medical research peaked in the late 1980s. This decline was interrupted only temporarily when the NIH budget was doubled at the start of the 21st century [8]. By 2011, the most dramatic decline in grants awarded to physician-scientists was MD applicants. MD-PhD applicants began to decline after a steady trend of increasing participation. Factors contributing to these changes include poor funding environment, competition, pressures for clinical productivity, lack of opportunity for professional advancement, and poor work-life balance [9][10][11]. To combat faculty attrition, a number of innovative institutions have developed programs including structured mentorship and earlier intervention by leadership [12][13][14]. However, these important interventions occur after the transition from trainee to junior faculty [15]. This has resulted in a failure to determine and address factors influencing attrition prior to attaining a faculty position. Understanding the earliest stages of training is one approach to combat this decline in physician-scientists.

Trainee Career Choices Beyond a Dual-Degree
Training pathways and career choices pursued prior to attaining a first faculty position have been investigated. Outcomes of MD-PhD training programs have been examined retrospectively by several groups to determine the trajectories of trainee careers [16,17]. An important and encouraging result of these studies is the high proportion of physician-scientists, including MSTP trainees, who choose to remain in academic medicine. However, these studies often do not examine how or why trainees seek and obtain faculty positions.

The Need for Further Investigation
Several recent studies have described the career trajectories and outcomes of MD-PhD program graduates [18][19][20][21][22]. However, there is less information on the attitudes, perceived obstacles, and career perspectives of current trainees. Much of the existing literature on trainee perspectives has been limited by focusing only on MD-PhD trainee cohorts at individual institutions [23][24][25]. Single-degree investigators are a significant and largely understudied proportion of the physician-scientist workforce; thus, an assessment of this trainee population is a necessity.
To this end, the central objective of the present study was to examine the factors associated with academic and research career interests in a nationally representative cohort of single-and dual-degree predoctoral physician-scientist trainees. This objective was pursued using a previously validated survey employed in a pilot study of a similar trainee population [26]. We hypothesized that MD, DO, and MD/DO-PhDs are unique cohorts with different perceptions and interests toward academic and research careers.

Study Design
This study was reviewed and exempted by the Institutional Review Boards at the University of Illinois at Chicago and University of Pennsylvania.
In this analysis, MD refers to both MD and DO trainees. MD-RI refers to both MD and DO trainees with research-intense career plans. MD-RI status was defined by a self-reported career interest of at least 50% research and was not intended to identify trainees enrolled in formal research pathway programs or curricula. This minimum research interest was selected because 50% is often the contractual maximum research to clinical ratio for surgeonscientists [27]. MD-PhD refers to both MD-PhD and DO-PhD trainees.

Data Collection
A 70-item survey (Supplementary information) was designed with feedback from a survey design team at the University of Illinois at Chicago, as well as lessons learned from a pilot study using this survey at a group of representative American medical schools [26]. This survey is a validated instrument with internal reliability to assess factors deemed important by trainees for their future careers and anticipated career challenges. The survey was sent to 32 nationally representative institutions via representatives of the American Physician Scientists Association (APSA) and the American Medical Women's Association. Survey response variables, including answer choices for career intentions, career sector, types of research (basic vs. translational vs. clinical), and obstacles, were not defined for the respondents and were, therefore, left to their interpretation.
Data were collected using an online survey tool (SurveyMonkey, San Mateo, CA). The survey was sent in September 2012 via email to all MD and MD-PhD trainees at these universities through trainee list serves and institutional representatives of APSA. One osteopathic medical school with DO and DO-PhD trainees was included in this study. Data collection ended in December 2014. Participants had the option to enter an institutional email address for a randomly selected $50 Amazon gift certificate. Email addresses were kept separate from survey responses to maintain anonymity of responses.

Statistical Analysis
Survey results were analyzed to identify significant differences in perceptions of factors influencing career interests between MD, MD-RI, and MD-PhD trainees. Chi-squared tests were used to measure associations between categorical variables. When data violated minimum expected cell counts, Fisher's exact test was performed. Logistic regression was used to identify factors associated with the intention to pursue academic and research careers. All tests were performed using SPSS and were two-sided with p < 0.05 considered statistically significant.

Demographics
There were 4433 respondents to the survey, yielding a 27% response rate. Demographic characteristics of respondents segregated by MD, MD-RI, and MD-PhD training status are summarized in Table 1. Overall, there were significantly more female than male respondents (56.3% female). However, there were significantly fewer female than male MD-PhD respondents (47.6%). Although trainees from each stage of medical and graduate training were represented, first-year medical students comprised the largest contingent of respondents (28.2%). Table 2 displays the responses for career sectors and career intentions stratified by MD, MD-RI, and MD-PhD training status. Academia was the predominant intended career sector selected by MD-PhD trainees (85.8%). While the majority of MD-RI trainees were also interested in academic careers (57.3%), a greater proportion of MD-RIs than MD-PhDs selected private practice (17.6%) and hospitalist (12.8%) sectors. MD respondents expressed the greatest interest in private practice (36.2%) followed closely by academia (31.2%) and hospitalist (23.3%) careers. Career intentions were consistent with career sector interests. Most MD-PhD respondents indicated career intentions of basic research (22.6%), translational research (42.9%), and clinical duties (18.6%). A majority of MD-RI respondents were interested in clinical duties (51.9%), followed by clinical (14.5%) and translational research (10.7%). MD respondents were most interested in clinical duties (84.2%) followed by education (6.1%), clinical research (2.6%), and advocacy (2.6%) Table 1. Demographic characteristics of respondents by MD, MD-RI, and MD-PhD

Research Career Feasibility in Acute Care or Surgical Specialties
Perceptions on the feasibility of research-intense careers (defined as >70% of time dedicated to research practice) in acute care medicine and surgical specialties were compared. A greater proportion of MD-PhD and MD-RI trainees viewed balancing a researchintense career with acute care specialties (i.e., critical care, emergency medicine) as highly feasible or feasible than MD trainees (Table 3). With respect to surgical specialties, MD-PhD trainees were less likely than MD-RI and MD trainees to view balancing research-intense careers with surgical specialties as highly feasible or feasible (Table 3).

Specialty Intentions
MD-PhD and MD-RI trainees demonstrated similar specialty interests. MD-PhD respondents were most interested in pursuing internal medicine and related subspecialties (32.6%). This was followed by pediatrics (12.6%), neurology (10.2%), surgery and its subspecialties (9.7%), psychiatry (4.2%), and pathology (3.6%). MD-RI respondents were also most interested in Table 2. Career plans and intentions by MD, MD-RI, and MD-PhD  Table 3. Perceptions of career feasibility by MD, MD-RI, and MD-PhD  (Table 4). Significant differences were also found in the perception of the importance of mentorship and ability to identify a mentor between MD-PhD, MD-RI, and MD trainees (Supplemental Table 3). MD-PhD respondents were significantly more likely to identify a mentor who helped them progress toward and/or achieve career goals (91.5%) than MD (70.6%) or MD-RI (79.4%) respondents (p < 0.001), and MD-PhDs (59.2%) were also more likely to say mentorship was very important in their careers thus far vs. MD-RI (49%) vs. MD (36.5%) (p < 0.001).

Obstacles and Responsibilities
We then evaluated the experienced and predicted obstacles iden-  Table 5.
Foreseeable non-work-related responsibilities during and after residency significantly differed between MD-PhD, MD-RI, and MD cohorts (Supplemental Table 4). Although raising children was viewed as a major non-work-related responsibility for all trainees, a greater proportion of MD-PhD (77.1%) respondents identified this as an important factor during residency compared with MD-RI (59.3%) and MD (54.1%) respondents. Other responsibilities including taking care of elderly parents, being a caretaker, and providing financial support are listed in Supplementary  Table 4.

Factors Associated with Career Intentions
Multivariate logistic regression analyses were performed to identify factors independently associated with academic and research career intentions. After controlling for other demographic and career factors, MD-PhD (OR 5.1, 95% CI 3.5-7.4) and MD-RI (OR 1.6, 95% CI 1.3-2.0) training status were each significantly associated with academic career intentions compared to MD status. In addition to trainee status, the ability to identify a mentor  was also significantly predictive of the intention to pursue an academic career (OR 1.6, 95% CI 1.3-1.9). MD-PhD trainees were significantly more likely than MD-RI and MD trainees to intend careers in basic science and translational research even after controlling for other demographic, career, and specialty interests by multivariate regression (Supplementary Tables 6-7). Clinical research intention was not significantly different between MD, MD-RI, and MD-PhD trainees in our models. Interestingly, while the ability to identify a mentor was associated with increased odds of translational (OR 1.4, 95% CI 1.0-1.8) and clinical research intentions (OR 1.5, 95% CI 1.2-1.9), it was not predictive of basic research career intention (OR 1.4, 95% CI 0.9-2.1) (Supplementary Tables 6-8).

Discussion
The objective of this study was to examine the factors associated with academic and research career interests in a nationally representative cohort of predoctoral physician-scientist trainees. Using a previously validated pilot survey, we evaluated current trainee career intentions and perceived obstacles to those intentions. Our results demonstrate that significant differences exist between MD trainees, MD trainees with research-intense career intentions (MD-RI), and MD-PhD trainees. Given the long-standing decline of the physician-scientist population, these data provide important insights into the perceptions of research and academic careers among trainees early in the physician-scientist career path.

Demographics
Demographic characteristics of this study cohort are largely consistent with the expected demographics of current medical trainees in the United States. Although the proportion of women matriculating into medical school has steadily increased in recent years, Sexual harassment 3 (0.08%) 1 (0.12%) 2 (0.31%) 0 (-) <.0001 a female trainees remain relatively underrepresented among applicants and matriculants in MD-PhD programs. In the present study, female trainees comprised 59.7% of MD and 56.1% of MD-RI respondents but only 47.6% of MD-PhD respondents. An additional demographic consideration of this sample population is that most respondents were in the early stages of career training. First-and second-year trainees comprised 53.4% of all respondents. As expected, this bias toward earlier stages of training was driven largely by the MD (59.4%) and MD-RI (55.3%) groups, as more MD-PhD respondents were in either graduate training (47.2%) or later years of the medical school (17.8%).

Specialty Intentions
Given the relationship between clinical specialty and the pursuit of research-oriented careers, we evaluated the specialty interests among predoctoral MD, MD-RI, and MD-PhD trainees. We found that specialty interests differed between these cohorts. While internal medicine and its subspecialties were the most frequently indicated specialties among MD-RI (28.3%) and MD-PhD (32.6%) trainees, surgery and its subspecialties were most indicated among MD-only (19.9%) trainees. Following medicine, MD-RIs were most interested in surgery and its subspecialties (16.4%), pediatrics (11.2%), emergency medicine (6.3%), and neurology (4.57%), while MD-PhD trainees were most interested in pediatrics (12.6%), neurology (10.2%), surgery and its subspecialties (9.7%), and psychiatry (4.2%). The strong interest in medicine-related specialties among research-oriented MD-RI and MD-PhD trainees is consistent with previous reports [2,16,[20][21][22]. One notable area of difference in specialty intentions of research-oriented trainees between this study and previous studies on career outcomes of MD-PhD program graduates is in pathology. Like internal medicine, pathology has generally been regarded as a conventional specialty destination for physicianscientists. This is supported by a recent AAMC National MD-PhD Outcomes study, which identified pathology as the second most common specialty among MD-PhD program graduates (13.2%) [20][21][22]. By comparison, only 1.7% of MD-RI and 3.6% of MD-PhD trainees expressed an interest in this specialty in the present study. As described by others, it is plausible that this discrepancy is part of a broader trend among recent graduates away from traditional physician-scientist specialties and toward other fields such as surgery [18,25]. However, given the crosssectional nature of this study, we are unable to conclude this. Additional longitudinal assessments aimed at understanding both the rationale underlying specialty interests and the eventual specialty destinations of predoctoral trainee cohorts are warranted.

Research Interest
In addition to differences in intended specialties, we also found that MD, MD-RI, and MD-PhD trainees demonstrated significant differences in research interest and commitment to research careers (Supplementary Table 1). Opportunities for research was Considering that single-degree physician-scientists constitute a significant proportion of the physician-scientist workforce, these data indicate that persistent efforts to expose and mentor singledegree physician-scientist trainees in areas of research may be crucial for maintaining the physician-scientist pipeline in the future. MD-PhD and MD-RI trainees also expressed diverging interests in the types of research they intend to pursue. Overall, 65.5% of MD-PhD trainees intend careers in basic and translational research. This interest is consistent with recent data from the AAMC, which demonstrated that the majority of MD-PhD program graduates actively participate in basic and translational research efforts [20]. In comparison, only 13.3% of MD-RIs indicated an interest in basic or translational research. Instead, a higher proportion of MD-RIs selected clinical research as their preferred research activity (14.5%). Given the greater level of commitment required to conduct basic and translational research compared to clinical research activities, these data may, at least in part, explain the differences in the perceived importance of research opportunities as an important career selection factor between MD-PhD and MD-RI groups.

The Balancing Act
A growing body of work suggests that physicians are increasingly dissatisfied with their work-life balance relative to the general US labor force [28]. Current trainees are likely aware of these trends, as medical school instructors/professors are members of this increasingly dissatisfied physician population. Previous studies have suggested that the growing emphasis on the importance of work-life balance is highlighted in specialty trends of medical trainees in the United States, as graduates are increasingly selecting specialties amenable to more favorable and controllable lifestyles [29][30][31]. In this analysis, MD, MD-RI, and MD-PhD trainees all indicated balancing family and work responsibilities as the primary predicted obstacle in their careers. As such, the results of this study support the concern with work-life balance in medical careers and suggest that this factor may serve as a top priority in career interest among current trainees.
Experienced and perceived family care obligations are a major consideration in the assessment of work-life balance. Previous studies have shown that these additional responsibilities are increasingly difficult to manage with the prolonged training periods required for medical careers in general and for physicianscientist careers in particular [32]. Therefore, while all trainee cohorts in this study reported concerns related to the care of children or elderly parents during and after residency training, it is unsurprising that a greater proportion of MD-PhD than MD-RI and MD respondents identified these as foreseeable responsibilities during residency. This is further supported by the demographic characteristics of respondents in this analysis, which demonstrate that predoctoral MD-PhD trainees tend to be older, in later stages of medical school training, and are more likely to have children during medical school compared with MD-RI and MD trainees (Supplementary Table 10). Finally, because the responsibility of child and elder care has traditionally fallen to women, previous reports suggested that these obstacles may disproportionately impact the development and advancement of women in academic and research-focused careers [32,33].

Financial (In)security and Other Obstacles
Interestingly, MD-RI respondents cited financial security as an important factor in career selection more frequently than MD-PhD and MD respondents. Previous studies have shown that research-intense careers tend to provide lower financial compensation than purely clinical careers. Owing to the fact that MD-RI trainees are generally responsible for paying medical school tuition, the financial implications of pursuing less-well-compensated, research-intense careers is likely to disproportionately impact MD-RI trainees than MD-PhD trainees [34,35]. As the burden of physician debt continues to grow, MD-RI trainees may feel less secure navigating a successful research career compared to MD-PhD trainees due, in part, to a lack of sponsored formal training provided by the PhD [36]. To combat this problem, the NIH Physician-Scientist Workforce Working Group has recommended expansion of the NIH Loan Repayment Program [37]. Other options to address this concern include increased access to scholarships for MD-RI trainees and more structured research training programs to provide adequate experience for future research careers.

Mentorship
The value of mentorship for physician-scientist trainees in this study is clear. MD-PhD trainees attribute greater importance to the mentorship they have received compared with MD-RI and MD trainees. This is possibly the result of the mentorship relationship that develops during PhD training [3,26]. In support of this notion, the ability to identify a mentor was significantly associated with academic career intentions and with interest in translational and clinical research by multivariate regression analysis. In addition to cultivating research interest, previous studies have argued that early and persistent mentorship may help address specific concerns from trainees regarding physician-scientist identity formation, as well as the challenges faced by physician-scientists within academia [38]. In this regard, a diverse and dedicated cohort of physician-scientist mentors appears to be a critical determinant of the success of efforts to cultivate a diverse next generation of physician-scientists [39].

Limitations
A major limitation of this study is that it is a cross-sectional analysis of career intentions and interests of predoctoral trainees, which did not allow for a causative understanding of the factors that predict outcomes in this cohort. Moving forward, it will be important to do a follow-up study to see whether these differences in intentions/interests at the predoctoral level translate to differences in career outcomes and to identify what factors allow trainees to ultimately succeed in academic, research careers. Another limitation is the 27% response rate. However, this is consistent with expected response rates of social sciences surveys. Finally, several years have passed from the date of survey completion. There may have been changes in research interest, attitudes, and biomedical research policies/environment that may influence this predoctoral cohort's career intentions. However, it is reassuring that the number of MD-PhD students and NIH funding have increased over this period, suggesting a growing interest in physician-scientist careers at the predoctoral level, and there is continued scientific support. The challenge is how to prevent attrition as physician-scientists advance in their training from predoctorates to postdoctorates, an area of active scrutiny and investigation.

Conclusion
Significant differences in the perceptions of academic and research careers exist between predoctoral MD trainees, MD trainees interested in research careers (MD-RI), and MD-PhD trainees. Although both MD-RI and MD-PhD trainees express an interest in research, the two groups vary in the type of research they intend to pursue. A greater proportion of MD-PhD trainees intend careers in basic and translational research, while a greater proportion of MD-RI trainees intend clinical research. Outside of research interest, MD-RI and MD-PhD trainees share many of the same perspectives on training and career obstacles. However, financial concerns tend to be more prevalent among MD-RIs, and family care responsibilities during training tend to be more prevalent among MD-PhDs. It is reassuring to see increasing interest in academic career intentions from earlier training stages to later training stages. Research interest also remains stable across training stages. Overall, this study provides important insights into trainee perceptions of academic and research career pathways. These insights can be leveraged by policy-making bodies and institutions to shape policies and practices that would help retain physician-scientists.