The National Institutes of Health has several programs focused on improving diversity and inclusion at all levels of the biomedical research enterprise. Diversity in the professoriate is of significant concern, and a recent analysis suggests that the lack of faculty diversity is due to university hiring biases rather than a lack of qualified candidates. This has prompted the National Institute for General Medical Sciences to release a request for information for ideas to strengthen the postdoctoral scholar to faculty transition.
The NIH’s K99/R00 Pathway to Independence program, launched in 2007, is the agency’s only grant program specifically designed to facilitate the postdoc to faculty transition. K99 awards to postdocs support two years of mentored work and can be converted to an R00 once the postdoc starts a faculty position. About 90 percent of K99 awardees convert their award to an R00 indicating the program is largely successful in supporting the postdoc to faculty transition.
Given the success of the K99/R00 program but the lack of improvement in faculty diversity over the past ten years, I asked how K99 and R00 awards were distributed across racial groups. I submitted a Freedom of Information Act request to the NIH for the racial and ethnic breakdown of K99 applicants and awardees and also R00 recipients between 2007 and 2017. The NIH reported eight racial categories: American Indian or Alaska Native, Asian, Black or African American, Native Hawaiian or Other Pacific Islander, White, Person reporting more than one race, unknown and withheld. The NIH also reported two ethnicity categories: Hispanic and non-Hispanic.
Overall, there were 8,408 K99 applications from 2007 through 2017 and 7,023 applications (84 percent) came from Asian, Black or African American (hereafter Black), White, and persons reporting more than one race (hereafter Multiracial). Values for American Indian or Alaska Native and Native Hawaiian or Other Pacific Islander applicants were suppressed by the NIH due to privacy concerns regarding the very small number of applicants. It is unclear how to interpret data regarding the unknown and withheld categories, and these were not included in this analysis.
White scientists comprised nearly 2/3 of the K99 applicant pool, and Asian applicants were nearly another third. Black and Multiracial applicants each made up roughly 2 percent of the population.
Race |
K99 Applicants |
K99 Awardees |
R00 Awardees |
White |
4,460 |
1,384 |
1,065 |
Multiracial |
130 |
40 |
32 |
Asian |
2,279 |
610 |
464 |
Black |
154 |
25 |
15 |
Ethnicity |
K99 Applicants |
K99 Awardees |
R00 Awardees |
Hispanic |
387 |
104 |
80 |
Non-Hispanic |
6,448 |
1,948 |
1,501 |
The K99 success rate, defined as the number of awardees divided by the number of applicants, of White applicants was 31 percent. For the other populations (p values as compared to White applicants):
- The success rate of Multiracial applicants was 30.7 percent (p = 0.95).
- The success rate of Asian applicants was 26.7 percent (p = 0.00028).
- The success rate of Black applicants was 16.2 percent (p < 0.0001).
In other comparisons, the success rate of Black applicants was significantly different from Multiracial (p = 0.0038) and Asian (p = 0.004) applicants. The success rates of Multiracial and Asian applicants were not significantly different (p = 0.32).
Converting the K99 to an R00 is triggered when the K99 awardee moves to a faculty position. There were 2,059 K99 awards made to Asian, Black, White and Multiracial applicants from 2007 to 2017 and 1,576 were converted to R00 awards. White awardees comprised just over 2/3 of the R00 pool, Asian awardees were just under 30 percent, Multiracial awardees were 2 percent and Black awardees comprised just under 1 percent.
The R00 conversion rate, defined as the number of R00s awarded divided by the number of K99s awarded, was 80 percent for Multiracial awardees, 77 percent for White awardees, 76 percent for Asian awardees and 60 percent for Black awardees. The only difference that achieved statistical significance was the comparison of White and Black awardees (p = 0.0466).
The percentage of K99 awardees converting their award to an R00 reported here is lower than previous reports. A check of NIH RePORTER indicates that this is due to the inclusion in the present analysis of 2016 and 2017 K99 awardees who have not yet had a chance to find a faculty position and convert their award to an R00.
I also received data for Hispanic K99 applicants. Hispanic applicants comprised 5.6 percent of the applicant pool. The success rate for Hispanic applicants was 26.9 percent and was not significantly different from the non-Hispanic pool (p = 0.16). Similarly, 77 percent of the Hispanic and non-Hispanic populations converted their K99s to R00s.
Discussion
These data indicate a clear difference in the success rates of K99 applicants based on race. The success rate for Black applicants was nearly half that of White or Multiracial applicants and 60 percent less than Asian applicants. Furthermore, Asian applicants had a statistically significant lower success rate than White and Multiracial applicants. These data are consistent with other analyses of bias against Black applicants in other NIH grant programs.
The source of this bias is not clear, although study section evaluations are an obvious possibility. For K99s specifically, the NIH will likely need to conduct its own investigation as the agency can conduct internal analyses without running afoul of privacy concerns.
An interesting analysis would be to parse the data presented here by gender. Without the lens of race, men and women have roughly equivalent K99 success rates and both convert their K99s to R00s at nearly the same rate. A combined race and gender analysis for Black and Multiracial K99 awardees is likely not possible because splitting these cohorts by gender would reduce the size of the population to the point that the information would be suppressed due to privacy concerns. However, an analysis of the differences between white men and women and Asian men and women could address whether a double-bind exists for Asian women in K99 awards as it does in other programs.
Conversion of a K99 to an R00 requires the K99 holder to be hired as a faculty member at a research institution. The data here indicate that a significantly smaller percentage of Black K99 awardees convert their grant to an R00 relative to White, Asian or Multiracial K99 awardees. There are several potential explanations for these data, but they are consistent with prior results that put the onus on universities to improve their hiring practices with regard to underrepresented minorities.
The data presented here underscore the importance of diversity-focused programs such as the BRAIN Initiative’s diversity K99/R00 award and the Howard Hughes Medical Institute’s Hanna H. Gray Fellows program. However, these programs are not permanent solutions to systemic funding disparities. Biases against underrepresented minorities in grant making decisions have been demonstrated for several NIH mechanisms, and the federal funding community must counteract these biases to successfully diversify the biomedical research enterprise.
Reader says
Thank you for the analyses. The first step towards fixing the system is to know the problem. Seeing the biases make me wonder about other agencies, in particular NSF, in which applicants get only binary outcome and don’t get any numerical scores. The potential for implicit bias should be higher.
Beth Schachter says
I wonder how much of this difference by race in K99 success rates has to do with relative strength of the mentoring team and and training plan, not just the applicant and the proposed science. It’s not just the applicant and their proposed research that is evaluated for the K99; the training experience and overall record of the mentor(s) are key factors in the K99 evaluations.