The biomedical research enterprise strives to be a meritocracy, but structural inequities and the reliance on nonscientific proxies such as journal impact and university prestige present roadblocks. Attempts to alleviate these problems are numerous— for example, ASAPbio, the grant-support index, the National Research Mentoring Network, and the Building Infrastructure Leading to Diversity—and have met with varying degrees of success. However, the depth to which these structural and prestige issues permeate the system must be fully understood in order to overcome them.
In computer science, business and history, between 70 and 86 percent of all tenure-track faculty are produced by only 25 percent of institutions. One conclusion from this study is that institutional prestige, rather than the consistent production of exemplary scholars, is responsible for this skew in future faculty production. The concern is that high quality scholars from good, yet less prestigious, institutions are overlooked to the detriment of the system as a whole. Similar studies have not been published for biomedical sciences, but such a skew in future faculty production would represent a threat to the sustainability of the research enterprise.
To determine if such a skew exists in biomedical research, I analyzed the institutional distribution of K99 and R00 National Institutes of Health awards. A K99 is a grant made to a postdoc for one to two years to complete his/her postdoctoral studies. The K99 is then converted into an R00 once the postdoc attains a permanent faculty position. Not every new faculty member has a K99, but following the institutions receiving a K99 and R00 awards are an important proxy for understanding the flow of people from postdoc positions to faculty positions.
K99 awards were first made in 2007, and I downloaded all K99 and R00 data between 2007 and 2015 from NIH RePORTer. During this time, the NIH awarded 1,811 K99s. Of these, 1,419 have been converted to R00 awards. This is a 78.3 percent conversion rate, which is lower than expected but likely due to postdocs who received a K99 in 2014 or 2015 who have not yet converted their award to an R00.
I then matched the PIs of the 1,419 K99 awards with their subsequent R00s to follow a person’s path from postdoc to faculty member. Of these, 1,067 awards were made to people whose K99 institution was different from their R00 institution. This indicates 1,067 K99/R00 awardees attained a faculty position at an institution different from their postdoc institution. For the K99/R00 grantees that did not change institutions, most of these awardees worked in clinical departments or hospitals, possibly reflecting a difference in hiring practices for clinical departments versus basic research departments.
I next analyzed the institutional distribution of the 1,067 K99s compared to the same distribution of the matched R00s. Between 2007 and 2015, 160 institutions received at least one K99. Stanford University (70) and Harvard (63; a combination of Harvard University, Harvard Medical School and Harvard School of Public Health) received the most K99 awards (Fig. 1). Across all 160 institutions receiving at least one K99, there was an average of 6.7 K99s per institution and a median of 3 awards. The K99s from these 160 institutions were converted to R00s at 247 institutions (Fig. 2). The University of Michigan (28) and the University of North Carolina-Chapel Hill (23) received the most R00 awards. Across the 247 institutions receiving at least one R00, there was an average of 4.3 R00s per institution and a median of 2 awards.
To determine if a subset of institutions provide a majority of faculty members for other institutions, I examined the difference between K99s and R00s received by each institution. 83 institutions received more K99s than R00s, and 17 institutions, or 11 percent of all K99 recipient institutions, received at least 10 more K99s than R00s over this time frame (Fig. 3). These 17 institutions accounted for 472, or 44 percent, of all K99 awards that had been converted to R00s between 2007 and 2015. These 472 K99s were converted to R00s at 155 other institutions, or 63 percent of all R00 recipient institutions (Fig. 4).
Do institutions preferentially hire postdocs with K99/R00 awards over those who do not? As monitoring K99/R00 awards does not capture all faculty hires between 2007 and 2015, this question cannot be answered fully by the data presented here. However, of the nearly 250 institutions that received at least one R00, 31 institutions averaged between one and two R00 awards per year since 2007, four institutions averaged between two and three R00 awards per year and one institution averaged more than three R00 awards per year. While we are not able to fully dismiss other options, a plausible interpretation of these data is that some institutions prefer hiring postdocs with a K99/R00 award over those that do not.
Diversity of backgrounds and viewpoints is necessary to maximize the speed and efficiency of solving scientific problems. However, the data presented here indicate a handful of institutions are having an outsized effect on faculty composition at other universities. Nearly half of all K99 awards were concentrated at 17 institutions, and these awardees then became faculty at over 60 percent of all R00 recipient institutions. Furthermore, the above data suggest that some institutions may preferentially hire K99/R00 recipients over postdocs without such funding.
Biomedical research strives to award grants and faculty positions based solely on meritorious ideas and track records. However, one interpretation of the data presented is that institutional prestige biases the awarding of K99 awards, which has a direct effect on the composition of faculty across the research enterprise. This potential bias requires further examination such as a survey of faculty search committees to understand the depth to which bias for postdocs from a select few institutions or with a K99 award affects interviewing and hiring decisions.