As is now well known, black scientists are less successful than their white counterparts in obtaining support from NIH R01 awards as designated Principal Investigators (PIs) (see here and here). Though recent NIH efforts are showing promise to enhance diversity in the biomedical workforce (see this post), much work is still needed to address the funding gap.
In a paper recently published in Science Advances, we delved into the underlying factors associated with this funding gap. We identified three decision points where disparate outcomes arose between white and black researchers: 1) the decision to bring applications to discussion during peer review study section meetings; 2) impact score assignments for those applications brought to discussion; and (3) a previously unstudied factor, topic choice – that is what topic the investigators chose to study.
We analyzed 157,549 R01 applications, both new and renewals, from fiscal years 2011-2015. We confirmed previous findings that black researchers submit fewer applications as PIs than white researchers. Applications with black scientists as PIs were brought to discussion in peer review only 77% as frequently as applications from white researcher PI’s (Figure 1).
When applications from black researchers were discussed in study section, they received worse impact scores— 38.4 + 13.4 vs 35.2 + 12.6. Combining lower submission rates, lower discussion rates, and worse impact, black scientists receive R01 funding only half as often as their white peers (Figure 1).
We found a number of differences in the characteristics of applications according to the race of the designated PI. Black scientists were more likely to propose research involving human subjects and less likely to propose work involving animal models.
We next used an informatics method called “word2vec” to bin the 157,549 applications into 150 topic clusters, which roughly aligns with the number of standing study sections at NIH. We can describe these clusters by word chains like “retina photoreceptor retinal cone MeSH_Photoreceptor_Cells_Vertebrate rod photoreceptor cells retinal degeneration” or “practice provider clinician care education evidence-based healthcare recommendation medical psychosocial.” Figure 2 shows word clouds associated with the topic clusters with the highest number of applications from black PIs (panel A) and with clusters in which there were no applications from black PIs (panel B). Note that the words in panels A and B are clearly qualitatively different.
A closer look at Figure 2 shows that black applicants were more likely to be associated with topics like health disparities, disease prevention and intervention, socioeconomic factors, healthcare, lifestyle, psychosocial, adolescent, and risk (panels A and C). Generally speaking, applications with these terms were less likely to be funded than topics linked like neuron, corneal, cell, and iron (panel B).
Figure 3 summarizes the word2vec topic cluster data for all applications. Panel A shows the topics in descending order of proportion of applications designating black PIs. Panel B shows the number of applications for each topic (presented in the same order as in panel A). A third of the applications from black scientists mapped to only eight of the 150 topic clusters, which tended to experience lower success rates (panel C). These lower-success topic clusters tended to focus on community and population-level research. Of note, applications from white researchers in these lower-success topic clusters were also less likely to be funded, although not to the same degree as those from black researchers.
We performed a series of multivariable analyses and found that after controlling for variables like applicants’ prior success, topic selection accounted for 21 percent of the funding gap observed between black and white researchers.
We briefly discussed potential implications of our findings – including the need to encourage a more diverse applicant pool, the potential value of mentoring systems to help investigators navigate the NIH system, and the possibility for NIH institutes and centers to consider discretionary funding for topics that may be under-appreciated by review but align with strategic priorities.