Washington

Overall control: the US National Institutes of Health may be given greater autonomy. Credit: NIH

The US National Institutes of Health (NIH) should be reorganized so that it can support high-risk, pathfinding research more effectively, says an eminent group of scientists and public-policy experts in what is likely to be an influential report.

The study was commissioned by Congress and carried out by the Institute of Medicine, part of the US National Academies (see Nature 418, 572; 2002). It says that the director of the NIH should be given a formal role in the process of changing the number of institutes and centres — a switch from the present arrangement in which Congress takes the lead.

The authors also recommend that Congress should transfer the power to hire and fire institute directors from the secretary of the Department of Health and Human Services to the NIH director. In addition, they propose that Congress should give the director an annual fund of between $100 million and $1 billion for a programme to support “high-risk, innovative projects”.

But the authors, who released their proposals on 29 July, stopped short of endorsing the sweeping changes that some observers had anticipated, such as the regrouping of the NIH's 27 institutes and centres into clusters based around body systems, such as the brain.

Some of the report's proposals are likely to prove controversial. The 21-member panel, chaired by Harold Shapiro, former president of Princeton University, recommends merging the National Human Genome Research Institute, which has finished its major research goal, with the National Institute of General Medical Sciences. It also suggests re-evaluating the special status accorded to the National Cancer Institute, which has an unusual degree of independence from the NIH director.

Many researchers, including former NIH director Harold Varmus, have been calling for greater consolidation, claiming that the current arrangement of the NIH makes it inflexible and causes disparities in research funding. But changes to the status of individual institutes are likely to be opposed by lobbying groups and research organizations linked to the areas of science involved. The report's authors say such difficulties mean that further mergers would be politically impractical.

“Our discussions, correspondence and meetings made it quite clear that there would be very little agreement among these communities on what the right way to organize NIH is,” the report's authors write, “and there would probably be dozens of conflicting ideas in play and few clear avenues for narrowing these down.”

The authors also address the ongoing effort by the Department of Health and Human Services to centralize or outsource various NIH functions, such as some aspects of grant review. This move won the authors praise from the community for attempting to defend the NIH against interference from Congress and the Bush administration. David Baltimore, president of the California Institute of Technology, describes such behaviour as “gutsy”. “They're taking on two big interests here — the administration and the Congress — and I think that's a very good thing,” he says.

The question now is what will become of the report. A 1984 Institute of Medicine study, which included similar proposals, was not implemented by Congress. But politicians are currently more interested in the NIH than they were then. The agency is being investigated by Congress over payments to its researchers, for example. Last month, two members of Congress told NIH director Elias Zerhouni that they were beginning an investigation of payments for lectures made to NIH executives by large centres that received NIH money.

On 10 July, the investigation was expanded after an NIH programme administrator told reporters that even though he was removed from his role in 1995, he has since been paid an annual salary of $100,000 while doing almost no work for the agency. This, together with the fact that Congress requested the report, makes it less likely that the study will be ignored, observers say.