Map of life expectancy at birth from Global Education Project.

Thursday, July 20, 2006

Oh yeah . . .

Before I get to more of that brown shoe insurance stuff, I was meaning to write about the National Institutes of Health (NIH) peer review process. Assuming the U.S. still has enough resources 10 years from now to continue to fund health-related research, this does matter. This post will consist of exposition -- not the most exciting part of the movie. After a brief intermission, we'll start moving the plot.

There is a fairly bewildering tangle of institutes and funding mechanisms, which are served by a wide variety of review panels. Some panels only meet once, to consider one-time invitations for proposals called Requests for Applications. Other standing panels meet at intervals to consider proposals in one or another general field. Most of these are not associated with a single institute, but consider proposals directed at various institutes. Program Announcements, which are open invitations for proposals that may persist for 3 years or more, with quarterly deadlines, and some RFAs, may be issued by multiple institutes or single institutes.

Then, for RFAs and PAs alike, there are multiple funding mechanisms. For example, the R03 mechanism funds small scale, preliminary studies. The R21 mechanism is for slightly larger scale "exploratory and developmental" studies intended to lead directly to the big enchilada, the R01, which is funding for a large-scale research project of $500,000 a year or more and up to 5 years. There are mechanisms to fund research programs consisting of three or more linked studies with shared infrastructure; research center grants that focus even more on infrastructure development; grants for mentoring and professional development of investigators; and others.

All of these may be written in such a way as to shape the broad directions of research, with general shaping by Congress and more specific shaping by NIH staff, but for so-called extramural research - paid for by NIH but done independently - the applicants develop the specific research aims, questions or hypotheses, and study designs. Then they get a score from the peer review panel.

Typically, in one meeting, a panel may review some 40 proposals. Three panel members will be assigned as reviewers of each proposal, and each member will get about six proposals to review. If you do the math you'll see that there are about 20 members on a panel. All the panelists have access to all of the proposals, although only the assigned reviewers have access to the appendices, such as previously published research by the investigators, draft questionnaires, detailed protocols, etc., that may go with them. However, panelists are not required, or even particularly encouraged, to read proposals not assigned to them ahead of the meeting.

So, the assigned reviewers, ahead of the meeting, write critiques of their proposals. There used to be a lot of mail that flew around the country prior to the meeting, but now they post their essays on a special Internet site where the reviewers assigned to a given proposal can read each other's critiques. A federal travel agency gets everybody a flight and a hotel room, most likely in a strip mall wasteland in the Washington, D.C. suburbs, and the action begins.

[Curtain on Act One]

No comments: