Developing Physician-Scientists: A Perspective

INTRODUCTION

Giving this talk with Andy Schafer sitting in the audience may define “chutzpah.” In his book The Vanishing Physician-Scientist? (1), he provides a comprehensive collection of thoughtful essays on the topic of training physician-scientists and I recommend it to you.

As I begin, I must admit to several biases. The first is the belief that a scientist who is also a physician is uniquely poised to make discovery based on his or her exposure to the experiments of nature in the practice of medicine. The identification of H. Pylori as the cause of gastritis and peptic ulceration by Barry Marshall and Robin Warren provides a classic example. In 1984, Marshall and Warren published a seminal paper describing the identification of curved bacilli in the stomach of patients with gastritis and peptic ulceration (2). Their drive to pursue this observation and to test the hypothesis that ingestion of the bacillus led to gastritis distinguished them as scientists as well as observant physicians (3).

Conversely, I hold that physicians who think scientifically or critically provide the best patient care. Here I will cite the example of April Pettit, a young physician-scientist at Vanderbilt who recently detected the index case of a now well-known outbreak of fungal meningitis resulting from an injection with a contaminated steroid preparation (4). Dr. Pettit's experience as a scientist studying tuberculosis led her to recognize that the patient for whom she was caring had an unusual form of meningitis with significant arachnoiditis, to consider pathogens beyond the usual bacteria, and to sound the alarm when it became clear that an immunocompetent host with a recent history of a spinal injection had fungal meningitis.

Beyond these biases, I will elaborate several assumptions regarding the training and development of physician-scientists. The first is that all physician-scientists require a firm grounding in the principles of basic science—molecular biology, genetics, systems biology—as articulated in the report of the American Association of Medical Colleges-Howard Hughes Medical Institute report, Scientific Foundations for Future Physicians (5). Second, the mentored research apprenticeship forms the foundation of training physician-scientists. Third, these physician-scientists, whether they are clinical or translational investigators or molecular biologists, require protected time for research to be successful. Lastly, success requires exposure to a critical mass of smart people. In the context of these assumptions, in the remainder of this essay I will provide one person's perspective on how we attract and train future physician-scientists, how we maximize the yield on our training investment (and reduce attrition), and how we increase the productive lifespan of the physician-scientist.

HOOKING THE PHYSICIAN-SCIENTIST

The fundamental question of how one “hooks” a young person to be interested in a career as a physician-scientist lies beyond the scope of this talk but undoubtedly involves exposure to role models and to compelling stories of discovery. Reports by the National Research Council provide in-depth recommendations as to how undergraduate exposure to science should be structured (6).

To attract medical students to careers as physician-scientists, programs such as the Howard Hughes Medical Institute–National Institutes of Health (NIH) Research Scholars Program (Cloister Program), the Howard Hughes Medical Institute Medical Fellows Program, the Pfizer-funded Clinical Research Training Program, and the Doris Duke Clinical Research Fellowship Program were designed to provide students with an immersion experience in research during medical school. Fang and Myer reported outcomes of the 1987 to 1995 graduation cohorts of the Cloister Program and the 1991 to 1995 graduation cohorts of the Medical Fellows Program in 2003 (7). Although a larger percentage of graduates were likely to have been appointed to faculty positions with research responsibility compared to non-awardees or other US medical school graduates, the yield of graduates with faculty appointments with research responsibility was but 16% to 20%. In the summer of 2012, the NIH merged the Clinical Research Training Program with the Howard Hughes Medical Institute-NIH Research Scholars Program, replacing the programs with the Medical Research Scholars Program co-sponsored by the NIH and the Friends of the NIH.

For those students who develop an early interest in becoming physician-scientists, MD-PhD training programs provide an obvious pathway. In 2010, Brass et al published an analysis of the outcomes of 24 MD-PhD programs (8). They reported that 66% of graduates remained in academia, 6% joined research institutes, 8% were engaged in research in industry, and only 16% had entered private practice. The vast majority of those who had entered academia had joined departments in the so-called “cognitive specialties,” including Internal Medicine, Pediatrics, Pathology, and Neurology. Forty percent dedicated 75% of their time or more to research and 64% dedicated at least 50% of their effort to research. These data suggest that MD-PhD programs are achieving their intended outcome. On the other hand, from 1965 through 2007, the number of graduates choosing the aforementioned specialties decreased, whereas the number of graduates choosing specialties not traditionally associated with research careers (dermatology, ophthalmology, radiation oncology, and surgery) increased, raising concerns that the yield of MD-PhD graduates engaged in research might decrease in future years.

DELAYED GRATIFICATION

One motivation for attracting students to the physician-scientist pathway early in their training has been the perception that doing so could impact on the trend toward the ever-increasing age of investigators at the time of first “independent” funding. Today, physician-scientists (either MD or MD-PhD) achieve independent funding at a mean age greater than 44 years, compared to 42 years for PhD scientists (9).

The data would suggest, however, that MD-PhD training has not accelerated the time-to-independence. Why is this? Although the timing of PhD training in traditional MD-PhD programs serves to engage students earlier in their career, an unintended consequence of this timing is the subsequent interruption of the research career for a minimum of 5 years while trainees complete clinical training. During this time, a scientific field may evolve dramatically and a trainee may find himself or herself starting from scratch when he or she reenters research training during fellowship, thus prolonging the time required to achieve research independence. An alternative strategy for shortening time-to-independence while increasing the yield on investments in training programs may be to change the sequencing of research training. Many medical schools are developing curricula designed to increase the opportunity to experience research in a concentrated way later during medical school. At Vanderbilt, we are refocusing the pre-clinical science curriculum and moving the core clerkships into the second year of medical school. The intention is to allow for immersion in scientific endeavors in the third and fourth years when students are more mature and have begun to differentiate. In addition, we have created an alternative track in our MD-PhD program that permits MD graduates to complete a PhD during their subspecialty fellowship training. The original intent was to create a clinical investigator track in which degree candidates could conduct patient-oriented research after they had achieved sufficient clinical training. In addition, this creates time efficiency in that the PhD research then serves as the basis for future grants. Importantly, this program has been supported by the basic science departments.

“MIND THE GAP”

The early career of the physician-scientist may be marked by gaps in funding as well as gaps due to clinical training. The first gap in funding comes at the start of a physician-scientist's career, at the transition from the training grant to a career development award. The new faculty member may need an additional 1 to 2 years to develop a research program before applying for a career development award. In the past, academic departments have supported physician-scientists during this period. Small departments may not have sufficient revenues to support many physician-scientists, however, and with shrinking clinical margins, even larger departments may have trouble. In addition, clinical departments may be tempted to ask physician-scientists to spend more time in clinic to cover salary, eroding protected time for research.

To address this gap at Vanderbilt, then Associate Dean for Physician-Scientist Development, Jeff Balser, MD-PhD, created the institutionally funded Vanderbilt Physician-Scientist Development (VPSD) Program in 2000. The program provides modest salary support to physician-scientists at the assistant professor level. Awardees must work within the research program of an NIH-funded mentor. The selection process is competitive and candidates are selected based on the quality of the applicant, the mentor, and the research proposal. The clinical chair must guarantee 75% protected time and provides $25,000 to support research in the mentor's lab. The chair also articulates a future commitment to start-up resources and research space as the scholar emerges from mentorship. Perhaps most important, the Associate Dean responsible for the VPSD program monitors the progress of each faculty member. Participation in the second year is contingent upon evidence of scientific progress, including the submission of career development award application. Based on the success of the VPSD program, in 2002 when we received funding from the National Center for Research Resources for an institutional career development program in patient-oriented research, we modeled the Vanderbilt Clinical and Translational Research (VCTR) Scholars Program after the VPSD, incorporating the competitive selection process, centralized oversight, and required milestones for continued participation.

In 2008, we reported the outcomes of these two career development programs (10). We found that participants in the VPSD program achieved independent funding at an earlier age than faculty who did not participate in these programs and had obtained mentored K-award funding from 2000 to 2006. Since 2000, we have realized a 4.4-fold increase in career development funding compared to a 2-fold increase nationally over the same period. The financial return on an approximately $10 million investment in the VPSD over the first 11 years has been $88.3 million in funding. VCTR scholars have garnered $29.8 million in funding over the period from 2002 through 2011.

The second major potential gap in support for a young physician-scientist looms at the transition from career development award to first independent funding. If all goes according to plan, the faculty member's research program succeeds, he or she submits an application for an independent award by the end of the third year of career development funding, and this is favorably reviewed within the first or second submission. All too often, however, research proceeds at a slower than expected pace, or the faculty member and mentor lose track of time, or a grant is not well-reviewed and the faculty member finds him or herself without funding.

Based on our early perception that participants in the VPSD and VCTR programs fared better than others in obtaining independent funding, in 2006 we created a centralized infrastructure, the Newman Society, for all faculty with any career development funding whether through the VPSD, through a growing number of institutional K12 programs, or through an individual career development award such as a K08, K23, or VA Career Development Award. Members of the Newman Society meet at least annually with the Associate Dean to review their progress, to ensure that they are meeting regularly with their mentorship committee, to ensure that they are getting adequate protected time, and to identify any concerns. In addition, members of the Newman Society participate in a monthly career development series, have access to a repository of successful grants, and are encouraged to participate in an internal study section before submitting grants externally. The net impact of this program is that our faculty members enjoy an unusually high “K-to-R” conversion rate, and that the mean interval from start of the career development award to independent award is now 4.3 years.

An important aspect of the Newman Society infrastructure is that it is designed to supplement, not supplant, the critical roles of the individual mentor, the division director, and the department chair in promoting the careers of our faculty. Two years ago, when I transitioned from the position of Associate Dean to the role of Chair of a large clinical department, I came to appreciate the value of the Newman Society from a different perspective, as I realized that during the previous 4 years we had developed a cadre of sophisticated physician-scientists who formed the foundation for internal growth within the department. It also became evident to me that there are other gaps in the career trajectory that are paramount to address.

HIDDEN GAPS

The garnering of the first R01 or independent funding is an important milestone for the physician-scientist. It also ushers in a particularly vulnerable time in the career of a scientist, one that we have too often neglected. With the transition from career development to independent funding, the faculty member's protected time can decrease from 75% to something more like 40% on a typical R01 if the faculty member does not garner adequate support through other grants. There may be pressures to increase clinical activity or to provide administrative service. All too often, the “independent” faculty member stops looking for mentorship or seeking input on grants and papers. The rate of publication may slow if the faculty member loses track of time, becomes paralyzed by the need for perfection, or becomes isolated. To address this period of vulnerability, we have created the Neilson Society in the Department of Medicine for tenure- track faculty with independent funding who have not yet achieved tenure. The Neilson Society is led by a Vice Chair for Faculty Development who periodically meets with faculty to review their curricula vitae and communicates closely with division directors.

Today we are once again facing a gap that threatens to increase the rate of attrition among mid-career and senior investigators when they may be at their peak of productivity—the gap in funding for an established investigator. With pay lines for NIH grants now in single digits and opportunities to resubmit grants reduced, the likelihood that any one investigator will experience a period of diminished funding is high. It will be incumbent on academic institutions to develop alternative sources of funding, developing methods of selecting faculty and research meriting investment, and providing bridge funding. Vanderbilt has developed a robust institutional bridge funding program, with a high return on investment. Faculty members are required to submit a competing renewal of their grant on time and must adhere to other guidelines including an internal peer review.

GENDER, RACIAL, AND GENERATIONAL GAPS AMONG PHYSICIAN-SCIENTISTS

The under-representation of women and minority groups among physician-scientists represents yet another type of gap. With women comprising at least 50% of our medical school classes, for example, attrition among women on the physician-scientist pathway threatens this workforce. Timothy Ley and Barton Hamilton published an article in Science that highlighted the issue of attrition among women scientists applying for funding from the NIH (11). The authors reported a statistically significant, but minor, difference in success rates between men and women applying for independent funding. The attrition rate among women applying for independent versus career development awards was dramatic, however. What accounts for this attrition?

In a slightly different context, in 2005, at a Conference on Diversifying the Science and Engineering Workforce, then President of Harvard Lawrence Summers posited three hypotheses to explain why women are under-represented in tenured positions in science and engineering: 1) the high-powered job hypothesis, 2) different availability of aptitude at the high end, and 3) different socialization and patterns of discrimination in a search (12). Summers was widely criticized for his remarks, but I would like to consider whether either of the first two of these hypotheses pertains to attrition among tenure-track women physician-scientists, sharing studies by leaders at Vanderbilt's Peabody School of Education.

With respect to differences in availability of aptitude in math and science at the high end, data support a relationship between math ability within the top 1% and the likelihood of obtaining a doctorate, publishing a peer-reviewed paper, or achieving tenure (13). Such differences cannot explain, however, why women who have already obtained a doctorate and are publishing are less likely to achieve tenure in academic medicine. Relevant to the applicability of the high-powered hypothesis to attrition among women in the physician-scientist workforce is an article on gender differences in life values among those gifted or profoundly gifted in math and science (14). The authors report that women, but not men, increasingly valued limiting work to 50 to 60 hours per week and flexibility in the work schedule as they progressed from their mid-20s to their mid-30s.

My own interpretation of these data is that Larry Summers was partially correct when he posited the high-powered career hypothesis: women do worry about the time investment required to succeed in science. This is not because women are not willing to work hard, but rather because women experience a conflict of commitments during a phase of their lives when they often have young children and are facing the hurdles of obtaining independent funding. If we can help women physician-scientists (and other scientists) to develop adequate support systems, we can reduce rates of attrition among our junior faculty.

At Vanderbilt, there exists a grassroots organization called “Women on Track,” that was started, with the support of our Associate Dean for Faculty Affairs, by a group of five women scientists after they returned from an Association of American Medical Colleges Early-Career Women Professional Development Seminar. As I have met with this group over the years, I have been impressed by their motivation and I have been impressed that the group discusses the same career development topics as do faculty within other career development groups at Vanderbilt, with one exception: child care. The women are often putting in long hours, but something as simple as a sick child can disrupt work life. Tasks we take for granted like traveling to attend national meetings, so important for developing collaborations and stature, can wreak havoc for a tenuously balanced two-career family. We need to innovate here if we are to retain women physician-scientist and prevent the loss of this human capital. These issues are more and more likely to affect young male faculty members as more men and women share family responsibility.

A 2011 article in Science highlighted attrition among PhD scientists from under-represented minority groups (15). The authors reported that R01 applications submitted by Blacks and Asians were less likely to be awarded than those submitted by Whites and Hispanics. For investigators born in the United States, R01 grant success rates were similar among Asians, Whites, and Hispanics, but remained decreased for Blacks. Some of the most important data in this article appear in the supplementary tables. For example, Figure S4 showed the impact of the environment on the likelihood of grant success. For the full sample, grants submitted by an investigator at a top 30 NIH-funded institution were more likely to be scored than those submitted by investigators from other institutions, but for Blacks, the effect of institutional environment was dramatic. Grants submitted by Black applicants from top 30 NIH-funded schools fared better, but those submitted from schools ranking below 30^th fared poorly. The same group subsequently reported similar data for physician-scientists (16). The authors also reported that applications from Black physician-scientists appointed to the faculty of a medical school fare better than applications from Black physician-scientists appointed at other institutions. An impact of environment on the time course for conversion from K award funding by R01 funding has been observed by others. The exaggerated effect of environment on the success of R01s from Black applicants suggests that strategies that optimize the research environment may be most successful in reducing attrition among under-represented minority physician-scientists. Programs such as the Harold Amos Minority Faculty Development Program have successfully adopted the strategy of connecting historically disadvantaged scholars with senior faculty mentors who are at medical schools with a strong track record of research and mentorship. At Vanderbilt, our Associate Dean for Diversity has been particularly effective in assisting chairs in identifying talented residents and fellows from under-represented minority groups who are candidates for these programs.

Faculty development strategies need also take into account generational differences in motivation, summarized in an article by Janet Bickel and Ann Brown (17). Whereas “Boomers” (born from 1945 to 1962) are more likely to work hard out of loyalty, members of “Generation X” (born from 1963 to 1981) are more likely to require work-life balance. From my perspective, many of the characteristics that define members of Generation X may be ideally suited for a career of inquiry, such as a certain comfort with risk and a tendency to question authority. One need only read the report card of recent Nobel laureate Gurdon to appreciate this, “I believe he has ideas about becoming a Scientist; on his present showing this is quite ridiculous, if he can't learn simple Biological facts … (18)”

TRUE THREATS TO THE DEVELOPMENT OF PHYSICIAN-SCIENTISTS

Until now, I have considered the challenges in mentoring and providing support for the career of the physician-scientist. In truth, struggles to provide financial support for the development of the physician-scientist do not concern me as much as the insidious threats to the attraction of and development of our best and brightest as physician-scientists, and I would like to close by sharing my view of these threats and how we are addressing them.

A fundamental concern is the inadequate exposure of our residents and fellows to physician-scientists. As NIH funding lines and clinical margins have decreased, more and more institutions have adopted explicit clinician-educator and physician-scientist tracks to allow faculty to make wise time investments. With a tremendous growth in the clinical mission of academic medical centers, our residents are less and less frequently exposed to the successful physician-scientist role models. In the Vanderbilt Department of Medicine, we have adopted a number of strategies to increase the exposure of are interns and residents to scientists. This includes the introduction of an elective in clinical research that incorporates a number of informal, 1-hour “Meet the Researcher” sessions, the invitation of physician-scientists to Morning Report, and the use of physician-scientist “ringers” during Professor's Rounds, a weekly session in which the Chair works through a case with the residents.

Related to this is what I would call the indiscriminate application of “evidence-based medicine,” as opposed to “science-based medicine.” I certainly do not mean to imply that we should not treat our patients using the best available evidence. The evidence, however, is only as good as the science and is often applied without an understanding of the generalizability or lack of generalizability. In 1974, just 10 years before Marshall and Warren published their seminal article in Lancet (2), the New England Journal of Medicine published a study of the relationship among smoking, alcohol, coffee and peptic ulcer disease (19). Physicians of the time, applying the best evidence, advised their ulcer patients regarding lifestyle modification. It took science and unconventional thinking to reveal pathophysiology and lead to better treatments. Similarly, the application of systems approaches and technology to medical care can have an important impact on improving healthcare along the slope of Moore's law (a law developed by former chairman of Intel Corporation Gordon E. Moore that refers to the rate at which data density grows) but it is still the discontinuous change of a discovery that “shifts the curve.”

Lastly, we conflate technology with science. As a scientific community we spend entirely too much time training future scientists in the methodological means to an end—molecular medicine, translational medicine, evidence-based medicine, team science. What we really need to teach them is how to formulate a hypothesis, how to ask the great question. This requires an in-depth understanding of what is known and the ability to recognize outliers. In my mind, this is what makes not only a great scientist, but also a great physician. Cook, Irby, and O'Brian said it well in their book, Educating Physicians: “It is not enough to have time-proven and reliable approaches to routine problems; every physician requires a depth of understanding that allows him or her to respond to unusual clinical problems with original rather than habitual approaches.” (20)

CONCLUSION

I have shared one institution's strategy and programs for the development of physician-scientists. With respect to the timing of advanced scientific training, there is no “one-size-fits all” solution. Earlier immersion may “hook” students who would otherwise be lost, but later immersion can enhance efficiency and shorten the time-to-independence. There is no doubt that having a competitive selection process increases yield on early internal investment. Centralized oversight of the career trajectories of physician-scientists can enhance success across departments. In lean times, we need to shepherd our investment beyond “independence.” Lastly, “we get what we ask for,” and we must consider the unintended consequences of our economic decisions on the future development of physician-scientists.