CHICAGO (Reuters) - Black researchers are significantly less likely to win grant funding from the National Institutes of Health than white applicants, according to a study published on Thursday, and the director of the agency says it will take action to address the potential for "insidious bias" in the review process.
An NIH-backed study, published in the journal Science, found a 10 percentage point difference in the number of grants awarded to black researchers compared with white researchers with similar credentials.
"This situation is not acceptable. The data is deeply troubling," NIH Director Dr. Francis Collins told reporters on a conference call.
"It indicates that we have not only failed to recruit the best and brightest minds from all of the groups we need to be coming to us, but for those who have joined us, there is inequity in their ability to achieve funding from the NIH, which in many instances is essential to academic success in biomedical research."
Blacks make up 10.2 percent of the U.S. population, but only 1.2 percent of the principal investigators of biomedical research studies are black.
Collins said concerns over this and anecdotal evidence prompted NIH, which is the largest source of funding for medical research in the world, to take a deeper look.
They commissioned a study led by University of Kansas economics professor Donna Ginther. Ginther and colleagues from Discovery Logic, a unit of Thomson Reuters, and the NIH analyzed data from 2000 to 2006.
They studied success rates of applicants of different races with similar research records and affiliations at being awarded a new NIH research project grant.
The raw data showed a 13 percentage point gap in the success rates between black and white applicants. It also showed a 4 percentage point gap in the success rates of white over Asian applicants. There were no apparent differences in success rates among Hispanic and white applicants.
The team next spent more than two years looking for possible explanations for the differences. Among Asian applicants, differences in success rates disappeared when they looked only at applications from U.S. residents, suggesting that language among nonnative speakers may play a role.
Among black applicants, differences in education, country of origin, training, employer characteristics, previous research awards, and publication record accounted for 3 percentage points of the difference.
But that still left a 10 percentage point spread.
That suggests that for every 100 applications from white applicants, 30 percent were funded, but among every 100 applications from blacks, only 20 percent were funded.
The study did not look at gender.
Ginther said the findings suggest one of two possibilities. Either black applicants had not been afforded the same types of "cumulative advantages" as whites -- including education but also mentoring or serving on an NIH grant review panel -- or there was bias in the NIH grant review system.
NIH grant applications are reviewed by a panel of peer scientists that considers the application's scientific significance, but also takes into account the researcher's qualifications and research environment.
When applicants send proposals to the NIH, they identify their race, ethnicity and gender, but this information is removed from the application before the materials are sent to review. Collins said it is possible that reviewers are able to guess the race or ethnicity of an applicant by looking at names or where they trained.
The NIH plans to take several steps to address both of the possibilities suggested by the study. "This is not one of those reports we will look at and put aside," Collins said.
To address the potential for what Collins called "insidious bias" in the review process, NIH will conduct different studies in which grants are reviewed based on the scientific merits alone -- without offering any information about the applicant's qualifications or background.
The NIH plans to start a program in which promising young researchers from underserved groups are invited to serve on more grant review panels.
"We aim to have 50 assigned to each of NIH's three rounds of grant review panels in the 2012 grant year, which begins October 1," Collins said.
He also has set up two high-level advisory groups to gather expert advice about possible causes and actions NIH can take. The NIH will also support more mentoring efforts at research institutions.
Editing by Michele Gershberg and Mohammad Zargham