Radical changes in NIH funded research careers

A recent post at Galactic Interactions proposes what seem to be substantial changes in the current NSF- and NASA-funded astronomy research career. I had a few observations in the comments trying to understand just what was being proposed and how this might apply to the NIH-funded biomedical research career. In some exasperation with my not-gettin’-it comments, Rob Knop suggests that I’m a hidebound traditionalist. I don’t think so. I’m becoming increasingly convinced that the way astronomy careers work and the way biomed research careers work these days are sufficiently different that the specific problems and solutions may not look very similar. However, Rob raises an excellent challenge which I might view as “okay complainers, don’t like the systems determining your career? how would you change it?”.

Fascinatingly, while thinking about my answer and creating the above text, the NIH launched a similar query ! (as usual, Agent Writedit is on the case.)

So, how would YHN Go Big?

[ open thread ]

I discussed elsewhere the proposition that the MERIT extension (a 5 yr noncompeting turned into a 10 yr noncompeting interval) should be applied to those who need it, namely that rare New Investigator who does manage to get a grant. I still think this is a great idea.

Create career awards (not fellowships because of the way institutions use this to screw fellows out of the usual employment benefits) for that category of doctoral research scientist who is happy to labor away in someone else’s lab without being a PI. These people already exist, in great numbers and often work through to retirement in nebulous job categories. Let’s recognize that these people are an essential fuel for the NIH engine. It can be on the 5 yr cycle so that productivity is assessed and individuals are accountable to produce. This will create a great deal of independence in these individuals so that they are not beholden to one PI. Think of the side bennies on scientific fraud!

Ditch ALL F32 fellowships in favor of the K99/R00 transition mechanisms. Remember, the “/” translates to “get a job offer” so this does not equal “free jobs for any lame postdoc”.

Cap the number of awards or absolute grant dollars of any given PI. Let’s get serious. There is no miracle of ever-expanding capability of one person, no matter how brilliant, to really do justice to six R01 equivalent projects with <10% effort each because of all the other obligations. The only way the BigWigs pull this off is because there is a team of highly competent scientists under them doing the actual work of running the project. Not to mention coming up with many (most?) of the really brilliant ideas. Why not let some of those people be PI to get appropriate credit for their work? What would be lost? Not to mention cost/benefit gains. These bigwigs are over NIH salary cap which is some $180,000 per year. Are you telling me two lowly assistant/associate professors for this cost isn’t a better value? The big project and all the alleged “synergistic value” and “economy of scale” values could still be easily obtained, the BigWig would just have to share the credit (and maybe change to a more collaborative style, another GoodThing).

Update 07/10/07:

One of the major problems with NIH grant review is the churning of revised applications. There is a huge bias for revised grants, check CRISP for new grants in any given year, gate by your favorite study section and look for the proportion of -01, -01A1 and -01A2 to verify this. This becomes more intense as the funding lines shrink and the revision process becomes a holding pattern for highly meritorious grants waiting to land on the runway. This costs a lot of money and opportunity as scientists prepare these applications and other scientists review them. And I should be clear that in a very large number of cases, the critique/revise process results in very little change to the conduct of the eventual science (even if the PI is induced to say the “right” thing in the application).

To address the specific case of low funding lines, I think the NIH ICs should continue to “intend to fund” up to their historical funding line (pick a number but it will be something 20-25%ile). Then, in a given round they should fund some fraction of the current funding line from the new grants and fund some additional fraction from the “holdovers” from the previous round up to the historical funding line. For argument let’s say split the difference. 5%ile for new proposals and 5%ile for holdovers from the prior round in a 10%ile round. Now the key is that there is going to be considerable overlap between the just-missed from one round and the just-fundable two rounds down the road (hard to turn a grant around in one round). Over the course of a year, there is going to be a big bulk of proposals funded that would have been funded anyway at the savings of one less revised application. I’d argue that there will be little loss of the “best of best new proposals” on a per round basis because most of the brand new ones have to get in line behind revised apps anyway.

The more radical solutions I have in mind would be independent of specific funding climate. The first approach would be to significantly alter the instructions to reviewers. CSR provides specific instructions, instructions general enough to be useless and also explicitly avoids instructing on many issues (some of which I think could use more instruction). I’m not really privy to the whys of all this, the generic poles of the debate are NIH interests versus maintaining the independence of peer review. The point is, they can and do issue instructions, like the 5 review criteria, when they want to. For this purpose, there could be a series of instructions stating that the unrevised app is to be prioritized, in fact some of the changes toward significance and reduced application length will move in this direction anyway. I’d like to see it firmed up. Second, ICs could start looking at score percentiles by revision status and simply allocate a majority of their funds in a given round to unrevised applications. Going by the abovementioned CRISP technique, most of the CSR study sections close to my area end up with about 10% of the funded grants getting there unrevised. I’d like to see this up around 60%.