Black scientists are significantly less likely than white researchers to win grants from the National Institutes of Health, according to an audit released Thursday that confirmed disturbing suspicions inside the agency about a lingering bias against African Americans.

The analysis of data from more than 40,000 researchers who submitted more than 80,000 grant applications to NIH between 2000 and 2006 found that only about 16 percent of those from black applicants were approved, compared with about 29 percent of those from white scientists.

Even after the researchers accounted for other factors that could help explain the discrepancy, such as differences in scientists’ education and training, black applicants were still about 10 percentage points less likely than whites to get NIH funding, the researchers reported. About 27 percent of white applicants’ requests were successful, compared with only about 17 percent of blacks’.

Asians applying for money appeared to be slightly less likely than whites to get grants, but that gap disappeared when the researchers matched equally qualified white and Asian U.S. citizens. Hispanics were about as successful as whites.

The findings are troubling because they indicate that race remains a significant factor in who gets funding for research into diabetes, cancer, heart disease and other health problems from the premier funder of biomedical research, the researchers said.

“We have a very serious issue,” said Donna K. Ginther, director of the University of Kansas Center of Science, Technology and Economic Policy, who led the study published in the journal Science. “Science needs to reflect the diversity and power and potential of the population.”

NIH’s internal auditing had indicated that there might be a problem with bias in its scientific review process. The agency initiated and helped fund the study to investigate those concerns. Officials agreed the new findings were alarming and outlined steps the $31 billion agency will take to try to address the problem.

“This situation is not acceptable,” NIH Director Francis S. Collins told reporters during a telephone briefing. “This data is deeply troubling.”

For the study, Ginther and colleagues analyzed data collected by the NIH from all scientists with doctorates applying for the most common grants the agency hands out. The data included the applicants’ race and were combined with information garnered from other sources, including where researchers were educated, their training, whether they had applied for grants before and whether they served on NIH committees.

Only 1.4 percent of applications came from black scientists, even though they account for about 12 percent of the U.S. population, the researchers found. Moreover, the applications from black scientists tended to receive poorer scores than those from whites, resulting in bleaker chances of getting funded.

“Our research says, ‘If you hold everything else constant and the only thing different between these two investigators is the color of their skin, that person is less likely to get funded,’ ” Ginther said.

Ginther and her colleagues tried several methods to explain the discrepancy, including analyzing whether differences in the topics being proposed for study by blacks or the types of studies they hoped to conduct might be playing a role, but they did not identify any clear explanation. The researchers speculated, however, that several factors could be playing a role. Black scientists, for example, might not be as plugged into professional “peer-review” networks that judge scientific proposals as white researchers. They might also tend to work at institutions that offer less support.

“I don’t think it’s overt racism. I’m not thinking someone is going through the applications and saying: ‘Black, do not fund,’ ” Ginther said. “But it could be a matter of networks — that these investigators are not as well connected as others. Or it could be the resources of their home institutions in preparing the applications.”

NIH officials agreed and said they were taking steps to boost the number of black scientists on NIH committees that review grant proposals. Having served on such a committee appears to increase the chances of a researcher later getting a grant, the study found.

“It is a very valuable learning experience in terms of figuring out what works and what doesn’t work in your own application,” Collins said.

Collins said he has also asked two “high-level” NIH advisory groups to investigate: the NIH Diversity Task Force and the newly formed external Diversity in Biomedical Research Working Group, which will report back to him by June.

In addition, the agency planned to conduct more research to try to determine whether NIH reviewers are biased against blacks. Although an applicant’s race is removed before reviewers see applications, evaluators might be able to figure it out through a scientist’s name, where they work or simply because they know who they are.

For example, the NIH plans to conduct experiments in which all information that might indicate the race of the applicants, such as their names and where they work, are hidden to see whether that affects how applicants are evaluated. Another study might assess the ability of reviewers to infer the race of applicants. Reviewers might also receive sensitivity training.

“I would like not to believe that is intentional bias, but I can’t exclude, after talking to lots of colleagues, the possibility that even today, in 2011, in our society, there is still an unconscious, insidious form bias that subtly influences opinions of people,” Collins said. “That may be very disturbing for people in the scientific community to contemplate, but I think we have to think that’s one of the possibilities.”

NIH officials said they had shared their findings with other federal agencies, such as the National Science Foundation, to alert them to the possibility that a similar bias might be affecting their grant-making systems.