Introduction
Physician-scientists have made an enormous contribution not only to clinical care but also to medical investigation, which has propelled the United States into its current global leadership position in biomedical research. These investigators are uniquely poised to make important scientific observations due to the “experiments of nature” they encounter while practicing medicine (1). However, this country’s biomedical research leadership status is threatened, in part, from the decline in its physician-scientist workforce. Of the roughly 1 million medical doctors in the United States today, just 2% are physician-scientists (2, 3). Of the 14,000 physician-scientists who make up this 2%, only 8200 are principal investigators (PIs) of National Institutes of Health (NIH) grants. This group is fairly evenly split between PIs holding MD degrees alone and those holding dual MD and PhD degrees (3). However, over the past 40 years, this workforce has become smaller in number and significantly older (2). Between 2003 and 2013, the number of physician-scientists over 61 years of age has risen, whereas the number of those 60 or younger has fallen (2). Although the number of nonphysician National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK) R01-funded investigators rose between 2011 and 2017 (4), the number of physician PIs and the number of kidney-focused NIDDK R01 applications have fallen (4).
Although we know that there are fewer NIDDK-funded physician-scientists, what is not known is whether this workforce is also aging, which portends a further decrease in their number. Also unknown is whether there have been changes in the represent ation of PIs doing basic science versus clinical research, the representation of men versus women, and the representation of investigators trained within versus outside the United States. The purpose of this study was to determine if there have been changes in age and sex among early-career and established physician- and nonphysician-scientists doing kidney research. We also explored changes in the representation of physician and nonphysician PIs who were educated in the United States and abroad, as well as changes in the relative numbers of basic versus clinical scientists.
Materials and Methods
Mining Principal Investigator Demographic Data
All information included in this study was public data obtained from the internet. We mined NIH RePORTER between March of 2020 and January of 2021 for kidney-focused R01s that were active in 1990, 2005, 2010, 2015, and 2020 and for K-series grants (i.e., K01, K08, K23, and K99) active in 2000, 2005, 2010, 2015, and 2020. To mine NIH RePORTER, we entered “kidney” in the search tab. We selected “The National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK)” as the NIH institute and entered each of the respective years listed above. Because both R01s and K-series awards are typically made for a period of 4–5 years, we mined data at 5-year intervals. Projects funded by other sources, such as the Veterans Administration, were not included in this analysis.
We used the internet to determine the degree type of the PI given for each grant listed on NIH RePORTER. PIs with MDs alone were combined with PIs holding joint MD and PhD degrees. PIs without MD degrees were collectively referred to as “nonphysicians” and were usually individuals with PhDs, although they sometimes held another advanced degree, such as a PharmD or a DVM. Physician-scientists were defined as individuals who graduated from either a US or international medical school. We did not discriminate between physicians who do and do not practice clinical medicine.
We used Google, LinkedIn, and doximity searches to determine the year in which PIs obtained their MD or graduate degree and the institution from which this degree was conferred. For individuals holding joint MD and PhD degrees, we reported just the year and institution from which they obtained their MD degree.
Criteria for Classifying Research as Kidney-Focused Basic or Clinical Research
We used the abstract and title provided by NIH RePORTER for each grant to ascertain whether its focus was primarily on the kidney and whether it involved clinical or basic science research. When this distinction could not be ascertained from the abstract or when the abstract was not available, it was determined on the basis of a PubMed search of the PI’s publications from about the time the grant was active or from the papers listed in NIH RePORTER that emanated from that grant. We considered clinical research to be studies of human subjects, human populations, or human databases that used widely available measurements, such as those performed by a clinical laboratory or an institutional core facility. This group included studies using computational biology, surveys, existing databases, or clinical trials. Basic science was deemed to be research utilizing cell systems, animals, or human tissue that was heavily dependent on biochemistry, molecular biology, physiology techniques, etc., that are available only in a basic science research laboratory. For each year studied, each PI was counted just once, even if the PI held multiple awards that year. If a PI held multiple R01s within a specific year, one or more of which were classified as “clinical” and one or more of which were classified as “basic science,” that investigator was classified as a basic scientist.
Statistical Analyses
Each “n” reported represents separate PIs. For simple comparisons of two means ± SD, we used a t test. For simple comparisons of two proportions, we used a z test. We used least squares regression analysis to determine if time-dependent changes were significantly different from zero. We then calculated a t statistic for the slope of the line relating to the parameter of interest since the initial year studied. For some comparisons, such as changes in percentage over the period studied, we used chi-squared analysis of percentages to determine the statistical significance of these changes. A probability (P) of 0.05 was considered significant.
Results
Changes in the Representation of Physician and Nonphysician Principal Investigators
We asked if the number of established and early-career physician and nonphysician PIs engaged in kidney-focused, NIDDK R01-funded research has changed over the past 2–3 decades (Supplemental Tables 1–3). Figure 1A shows that the number of physicians and nonphysicians holding K-series awards appeared to peak in 2010 and then decline. Although the total number of physician PIs holding R01s did not change significantly between 1990 and 2020 (Figure 1A), the percentage of all R01-funded PIs who are physicians fell from 72% in 1990 to 59% in 2020 (Figure 1B) (chi-squared analysis, P=0.002). A similar relative decline was also seen for K-series-funded physicians (Figure 1B). In contrast, NIDDK K series– and R01-funded nonphysicians have increased both in total number and as a percentage of all PIs funded by these respective awards (Figure 1, Supplemental Table 3).
Figure 1.: Physicians represent a smaller proportion of K- and R01-funded principal investigators (PIs) studying the kidney. (A) The total number of established R01-funded and early-career K series–funded physician or nonphysician PIs at various points over the past 20–30 years. (B) The percentage of all K-funded investigators who are physicians or nonphysicians at various points between 2000 and 2020. Also shown is the percentage of all R01-funded PIs between 1990 and 2020 who are physicians and nonphysicians.
Changes in the Age of Kidney-Focused Physician- and Nonphysician-Scientist Principal Investigators
As an age surrogate, we ascertained the number of years that have elapsed since a PI received his or her medical (physicians) or terminal graduate degree (nonphysicians). For all years studied, we were able to ascertain the graduation year for 80%–93% of physicians and for 68%–82% of nonphysicians (Supplemental Table 1). The median number of years elapsed since medical or graduate school and their respective confidence intervals are shown in Figure 2. In 1990, R01-funded physician PIs had been out of medical school for a median of 22 years, but by 2020, the median time since medical school was 31 years (P<0.001 by Mann–Whitney rank sum test) (Figure 2A, Supplemental Table 1). Thus, between 1990 and 2020, the median time elapsed since medical school graduation increased at a rate of 0.28±0.071 years per year for R01-funded physician PIs (P=0.03 by regression analysis).
Figure 2.: R01-funded physician scientists are aging. The median number of years since graduation of R01-funded (A) physicians and (B) nonphysicians. Boxes show the 25th and 75th percentiles, while the horizontal lines indicate median values. Vertical lines show the 5th and 95th percentiles. Filled gray circles show individual data points that fall outside this range. Means ± SD are given in
Supplemental Table 1.
Over the past 30 years, nonphysicians have also been out of school for a greater period (Figure 2B, Supplemental Table 1). In 1990, nonphysician R01-funded PIs had been out of graduate school for a median of 16 years but out by 19 years in 2020 (P=0.01 by Mann–Whitney rank sum test). Thus, for nonphysicians, the median number of years elapsed since graduation has increased by an amount that is small but statistically significant. Similar changes were observed when expressed as the mean years since medical school or graduate school (Supplemental Table 1). We conclude that the number of years since graduation has increased for nonphysicians but more significantly for physicians.
Representation of International Graduates in Both the Physician and Nonphysician Principal Investigator Workforce
We examined the representation of international graduates among both physician and nonphysician PIs (Figure 3, Supplemental Table 1). Over the past 30 years, the number of NIDDK R01-funded physician PIs who are US medical school graduates has remained fairly stable, although the number who are international graduates has increased (Figure 3A). Moreover, United States–trained physicians now make up a smaller percentage of all physician-scientists, whereas international medical graduates make up a greater percentage (Figure 3B). International medical graduates made up 28% of the R01-funded, kidney-focused physician PIs in 1990, while they represented 43% in 2020 (P=0.006 by z test).
Figure 3.: The representation of physician and nonphysician PIs who are international graduates is increasing. (A) The number of PIs who graduated from either US or international medical or graduate schools who were funded between 1990 and 2020. (B) The percentage of total physicians and the percentage of total nonphysicians who are either US or international graduates.
The representation of international graduates among nonphysicians has also increased. Figure 3A shows that the number of R01-funded nonphysicians who are international graduates has increased. Moreover, they represent a larger component of all R01-funded nonphysician PIs (Figure 3B). In 1990, 12% of nonphysician R01-funded, kidney-focused PIs attended international graduate schools, but by 2020, 35% had attended such schools (P=0.008 by z test). We conclude that individuals with graduate education outside the United States are increasingly represented among both physician and nonphysician NIDDK R01-funded PIs.
Changes in the Representation of Principal Investigators Doing Basic and Clinical Research
To determine if basic science research participation has changed among early-career investigators, we compared the number of physician and nonphysician PIs with active K-series funding (i.e., K01, K08, K23, and K99) over the past 20 years (Figure 4, A and B, Supplemental Table 3). Clinical research was taken to be studies of human subjects, human populations, or human databases that use widely available measurements, such as those performed by a clinical laboratory or an institutional core facility. Basic science was deemed to be research utilizing cell systems, animals, or human tissue that was heavily dependent on biochemistry, molecular biology, physiology techniques, etc., that are available only in a basic science research laboratory. As shown (Figure 4A), K series–funded physician PIs doing clinical research are both greater in number and represent a greater percentage of all K series–funded physician PIs (Figure 4, A and B). Whereas physicians doing clinical research made up 24% of total K series–funded physician-scientists in 2000, they represented 50% in 2020 (P=0.02 by z test). This greater relative representation of physician PIs doing clinical research has been accompanied by a significant reciprocal fall in basic science participation (Figure 4B). Thus, the relative number of early-career physician-scientists engaged in kidney-focused basic science has fallen, whereas the relative representation of those doing clinical research has risen.
Figure 4.: Among R01-funded basic scientists, the relative representation of physicians is falling. (A) The number of K series–funded PIs at various points between 2000 and 2020 who are physician or nonphysician basic and clinical scientists. (B) The percentage of K series–funded physicians who are basic and clinical scientists at various points over this 20-year period. Also shown is the percentage of K series–funded nonphysicians who are basic scientists. (C) The number of physician and nonphysician R01-funded PIs at various times between 1990 and 2020 who are basic and clinical scientists. (D) The percentage of all R01-funded physicians who are basic or clinical scientists at various points over this 30-year period. Also shown is the percentage of R01-funded nonphysicians who are basic scientists.
Among nonphysicians, the number of K series–funded basic scientists increased between 2000 and 2020 (Figure 4A, Supplemental Table 3). However, because both nonphysician total PI number and basic science PI number have increased to a similar extent (Figures 1A and 4A), the percentage of all nonphysician PIs who are basic scientists has remained roughly stable (Figure 4B). However, this group’s sample size is relatively small.
We explored whether either the total or relative number of established PIs doing basic science research has changed over the past 30 years (Figure 4, C and D, Supplemental Table 2). Figure 4C shows that the number of R01-funded physician PIs doing basic research has been relatively stable over the past 30 years, whereas the number of physician PIs doing clinical research has increased. Figure 4D shows that physician PIs doing basic research now represent a smaller fraction of all physician PIs (Figure 4C) (chi-squared analysis, P=0.004). Whereas 88% of physician PIs with active R01 funding were doing basic science research in 1990, 74% were doing so in 2020. We conclude that the relative representation of physician-scientists doing basic research is falling.
Among R01-funded nonphysicians, total PI number (Figure 1A) as well as the number of those doing either basic or clinical science increased over the last 30 years (Figure 4C). Thus, the percentage of all R01-funded nonphysician PIs who are basic scientists has remained roughly stable (Figure 4D).
Representation of Women among Both Early-Career and Established Kidney-Focused Principal Investigators
Among physicians, the representation of women with R01 funding has increased both in total number and as a percentage of all physician PIs (i.e., men plus women) (Figure 5, A and B, Supplemental Table 2). In 1990, women represented 8% of NIDDK R01-funded, kidney-focused physician PIs, but they increased significantly to 25% by 2020 (P<0.001 by z test). This increase in women physician PIs has been relatively constant at 0.59%±0.11% per year. Similar trends were observed among early-career, K series–funded physicians (Supplemental Table 3).
Figure 5.: While the number of women physician-scientists is rising, they are increasingly more likely to do clinical than basic science research. (A) The number of men and women physician and nonphysician PIs at various points between 1990 and 2020. (B) The percentage of physician and nonphysician PIs who are women relative to the total (men and women) holding the same degree type. (C) The number of R01-funded women who are physician or nonphysician clinical or basic science PIs focused on kidney at various points over this 30-year period. (D) Similar data for men. (E) The percentage of all men physician-scientists who are basic scientists, with analogous data for women.
The representation of women among nonphysicians has also increased over the past 30 years both in total number (Figure 5A, Supplemental Table 2) and as a percentage of all PIs with the same degree type (Figure 5B, Supplemental Table 2, last column). In 1990, women made up 19% of the kidney-focused, NIDDK R01-funded nonphysician PI workforce, which increased to 34% by 2020 (Figure 5B, Supplemental Table 2, column 6) (P=0.05 by z test), although the percentage of nonphysician PIs who are women has not changed significantly since 2005 (Figure 5B, Supplemental Table 2, column 6) (P=0.91 by z test). Nonetheless, over the past 30 years, the relative representation of women has been consistently greater among nonphysician than physician PIs (Figure 5B). Similar trends are seen among early investigators. Since 2000, the number of K series–funded women nonphysicians has increased (Supplemental Table 3).
Representation of Men and Women in the Basic and Clinical Sciences
We asked if the choice to do basic versus clinical science research has differed between men and women. Figure 5C (Supplemental Table 2) shows that over the past 30 years, the number of both basic and clinical science women physician PIs has increased. However, among women physician PIs, there has been a greater increase in clinical than basic scientists (Figure 5C), although the difference was not statistically significant (P=0.66). Although the number of women physician basic scientists has increased over this period, the relative representation of women basic scientists has fallen (P=0.006) (Figure 5E, Supplemental Table 2). Therefore, although the number of women physician-scientists has risen, those who enter this workforce are less likely to do basic science and more likely to do clinical research.
Over the past 30 years, the number of men who are physician clinical science PIs has remained stable or has risen slightly (Figure 5D, Supplemental Table 2). However, the representation of men physician PIs who are basic scientists has remained roughly stable or is slightly lower over this period both in total number (Figure 5D) and as their percentage of all men physician PIs (basic and clinical scientists) (Figure 5D). However, among nonphysician PIs, there has been a substantial increase in the number of both men and women basic scientists (Figure 5, C and D, Supplemental Table 2).
Discussion
This study demonstrates increasing representation of women and international graduates within the physician- and nonphysician-scientist workforce doing kidney research. However, the physician-scientist workforce is declining in relative number (4), getting older, and less focused on basic science research.
We calculated an average age of 50 for R01-funded nonphysicians in 1990 and 53 for those funded in 2020. These estimates were made on the basis of our observation that R01-funded nonphysicians had been out of graduate school for an average of 19 years in 1990 and for 22 years in 2020. We also assumed that they received their PhD at the age of 31 (5). For R01-funded physicians, we assumed an age of 28 upon medical school graduation (6) and that they had been out of school on average for 22 years in 1990 and out for 30 years in 2020. Using these assumptions, we calculated a mean age of 50 among R01-funded physician-scientists in 1990 and 58 in 2020. By comparison, Blish (7) reported a mean age of 43 for all R01-funded PIs with joint MD and PhD degrees across all NIH institutes in 2003 but 52 in 2012. Although our calculations are rough estimates, they show that among R01-funded investigators, physician-scientists are slightly older than nonphysician-scientists and that NIDDK R01-funded PIs are getting progressively older, particularly among physicians.
The longer time now typically needed to obtain R01 funding has contributed to the increased average age of kidney-focused, R01-funded physician and nonphysician PIs. Among all NIH institutes, the average age at which physicians obtain their first R01 has increased from 37 in 1980 to 45 in 2016 (8). Over this 36-year period, the average nonphysician age increased from 36 to 42 (8). Therefore, from 1980 to 2016, the age to first R01 has increased by 6 years for nonphysicians and by 8 years for physicians. There are likely many causes for the greater age at which R01 funding is initially obtained. First, to obtain a US medical license, international medical graduates often must repeat their clinical training in the United States. As such, greater time is spent in clinical training prior to starting one’s research program. Second, NIH funding rates have fallen (9), which generally increases the number of submissions needed to obtain a fundable score. These issues contribute to the increased average age of funded PIs.
Using graduation year as an age surrogate has limitations. For example, one’s age at graduation is typically younger for students from international than from US medical schools. For such individuals, our approach overestimates PI age. However, other issues lead to an underesimation of age when using this approach. It is increasingly common for young people to take gap periods between college and either medical school or graduate school (101112–13). In so doing, they become older at the time they enter and then graduate from medical or graduate school. Moreover, because this study specifically explored NIDDK-funded kidney-related research, we cannot exclude the possibility that demographic data differ among PIs supported by other sources, such as those funded by other NIH institutes, the Veterans Administration, etc.
Our data are consistent with previous reports citing a fall in the representation of physician PIs holding NIDDK R01s. Although physician PIs held 52% of R01 awards in 2011, they held only 29% of these awards in 2017 (4). Similarly, our data show a decline in the number of early-career physician PIs with K awards over the past 10 years, and are, therefore, consistent with NIH-wide data showing a fall in R01 applications by early established investigators (9).
The decline in the physician-scientist workforce extends across all subspecialties of medicine and has many causes, including changes in medical student specialty selection. From 1965 to 2007, MD/PhD program graduates were progressively less likely to choose specialties more commonly associated with research activities, such as internal medicine, pediatrics, pathology, and neurology and more frequently chose specialties such as dermatology, ophthalmology, radiation oncology, and surgery, which are less commonly associated with research, particularly in the basic sciences (1).
Recruiting physicians into nephrology either as clinicians or physician-scientists has been particularly challenging. In the 2016 National Resident Matching Program, only 61% of US nephrology fellowship positions filled (14), although 73% filled in 2021 (15,16). To identify the cause(s) of this applicant shortage, Nair et al. (14) surveyed 4199 medical students and internal medicine residents in the United States and observed that 92% of applicants cited a lack of interest as the basis of their decision not to pursue nephrology. Seventy percent of respondents attributed this lack of interest to the absence of mentors, while 66% cited an absence of exposure to the field. Conversely, an interest in kidney physiology and a positive interaction with a nephrology mentor were commonly cited reasons for considering a nephrology career (14).
To address the general physician-scientist workforce shortage, organizations, such as the Alliance for Academic Internal Medicine, have studied this problem and made recommendations (2), which include streamlining physician-scientist clinical training, increasing the number of foreign medical graduate physician-scientists, increasing research funding, forgiveness of student loan debt, and assistance with work-life balance issues. The physician-scientist pipeline is also enhanced by early exposure of future physicians to cutting-edge research, formalized mentoring, and exposure to sex-diverse and racially diverse role models (2,3,14).
We conclude that over the past 30 years, there have been greater numbers of women and international graduates in the physician- and nonphysician-scientist workforce doing NIDDK-funded, kidney-focused research. Although there are greater numbers of women physician-scientists doing basic science research, women physician PIs are increasingly more likely to pursue clinical than basic science research. Overall, the physician-scientist workforce doing kidney-related research is getting older and represents a smaller proportion of all PIs, particularly among those doing basic science. This trend raises serious concerns as to whether there will be a sufficient workforce to drive future progress in kidney-related research.
Disclosures
D.C. Eaton reports receiving honoraria from the National Institutes of Health and the University of Pittsburgh; patents and inventions with Monograph “Vander's Renal Physiology” (McGraw Hill Publishing); and serving as a scientific advisor or member of the editorial boards of American Journal of Physiology-Renal Physiology and APSselect. S.M. Wall reports owning stock in Abbott, Becton Dickinson, Johnson and Johnson, Merck, Roche, and Thermo Fisher. S.M. Wall does not have any stock options. S.M. Wall reports serving as an American Physiological Society representative to the Federation of American Societies for Experimental Biology Finance Committee. All remaining authors have nothing to disclose.
Funding
This work was supported by National Institute of Diabetes and Digestive and Kidney Diseases grants DK110409 (to D.C. Eaton) and DK119793 (to S.M. Wall).