Kill the R21!!!!

The NIH's R21 grant mechanism is called Exploratory/Developmental. It is limited* to $275K in direct costs, to be split across 2 years however the PI sees fit. It cannot be competitively renewed.

Sally Rockey, head of Extramural research at the NIH, has just Twitted, and then blogged, advice that applicants not try for the R21 because success rates are higher for R01s.

Ding, dong the R21 is dead.

This seems like a very bad move on the part of the NIH. But clearly, with a statement like this coming from the top applicants should pay heed. And stop writing R21s.

No biggie, right? The R01s are superior because the project can be proposed for 5 years, and competitively renewed after that. More importantly, there doesn't seem to be any easier review of R21 vs R01s. These "Exploratory" and/or "Developmental" proposals are being sent back for revision and beat up for lack of preliminary data. Scores are being (to appearances) benchmarked against R01 applications.

And almost unbelievably, Program appears to be passing these over when using their discretion about funding priority and portfolio balance.

In theory, peer review should love these things. They are low risk! And we know the demand for ever increasing amounts of preliminary data for R01 applications is mostly about scientific "risk" and "feasibility".

Program should likewise be pleased to take a low cost look at an idea before jumping in with 5 years of funding.

At about half the cost, the NIH could boost it's R-grant funding stats by picking up 2 R21s for the price if one R01.

I have no idea what the NIH is thinking on this, but they have this 180 degrees wrong. They should be encouraging the funding of even more R21s.

In fact, I think there should be a boost given to the funding of any new R01 proposals that follow from a reasonably successful R21 award. (Where successful need not only mean papers but include the generation of solid preliminary data, models, systems, etc.) Among other benefits this would reduce the current reality that PIs generate substantial preliminary data for their next proposal(s) on the dime of their current projects**.

For this to work, the NIH would have to get real about making study sections review the R21 properly...but they should do that anyway. Rockey's position should not be to pass along the bad news but to FIX THE PROBLEM, imnsho. If they made these unrevisable and banned preliminary data, that would work wonders.

Can anyone point to any decent reasons why the NIH would rather get rid of the R21?
__
*currently. Previously it was limited to $100K per year for up to 3 years. There may be the occasional use via RFA that has these older parameters.

*technically illegal, btw.

20 responses so far

  • Pinko Punko says:

    I'm a fan of the R21, but it is so unevenly used across institutes and for different purposes, that it can be frustrating, especially if one's science is right up the alley of NIGMS (no R21s).

    And because success rate is low doesn't mean that those being funded are not making a lot of hay with them.

  • Martini says:

    In theory R21 is a great program. However, NIH uses it for different things in different institutes. Moreover, NIGMS doesn't even have them at all.

    NIGMS has said that they don't use them because they get too many applicants using these as regular grants instead of for exploring something new. Instead NIGMS has a high risk IDEA grant program that I think is much better than the R21. The program directors that run IDEA are very strict in awarding only new research directions without much prelim data

  • Morgan Price says:

    If scientists are putting too much time into writing and reviewing grant applications then killing off the small grants (i.e. R21) is only sensible. *Does anyone have data on the cost-benefit of grant review?)

  • WhizBANG says:

    We desperately need a small grant program of some sort. There simply is no funding source to generate preliminary data for new ideas (NIDDK, the institute I get routed through, does not participate in the R03 program) that depart from ongoing projects in the lab. As federal funding has dried up, the foundation money and other smaller sources that used to go to smaller labs and to less developed projects now get sucked up for worthy science that should be funded at the R01 level in a perfect world.

    Here's an idea. Since the peer review process hasn't worked for this mechanism, let's try something different. How about doing a PLOS sort of review for technical (no miracles required) and ethical issues. All proposals that pass muster then get reviewed by the POs or some other group within each institute who have dedicated funds for the mechanism and can pick the ones that they like within their grant portfolio.

  • pablito says:

    The idea has been floated that the NIH can't support the current number of PI's. Getting rid of the R21 may help cull the herd of those who have used this mechanism (in part) because it has been difficult to obtain R01 funding (myself included).

  • Confounding says:

    Unfortunate - the purpose of an R21 always made sense to me, and given we're worried about struggling young investigators, killing yet another mechanism that should be more amenable to sparse preliminary data seems...flawed.

  • The only way to fix R21s is to review them in R21-only study sections with heavy-handed oversight by SROs and program staff to enforce the terms of R21 FOAs.

    And BTW, you're sort of misrepresenting what Rockey wrote:

    Think about applying for a R01 instead if you are ready.

    I interpret "if you are ready" as meaning "if you have sufficient preliminary data and feasibility support". Her point about R21s is correct, and one that has been hammered on for years by numerous NIH officials: R21s are *not* "junior R01s" for new investigators and should not be seen as such, even though to some extent--and due to bad study section and program staff behavior--they have been used as such.

    She is not trying to kill R21s; rather she is trying to nudge them back into their proper role.

    Finally, I am an R21 awardee, and it has been a very successful project. I will be submitting an R01 built on the foundation of the outcome of the R21-funded research, and I plan on making a big fucken deal about the fact that it did, indeed, arise out of a successful R21. My expectation is that this will impress the study section quite a bit and give us a scoring bonus, in the same way that competitive renewals of productive R01s get a bonus.

  • Unfortunate - the purpose of an R21 always made sense to me, and given we're worried about struggling young investigators, killing yet another mechanism that should be more amenable to sparse preliminary data seems...flawed.

    No. What's flawed is using the R21 mechanism as a "junior R01". R21s are only "amenable to sparse preliminary data" as a side effect of being aimed at exploratory/developmental research. This is completely orthogonal to the reasons for new investigators having little preliminary data.

  • kelly suter says:

    The reason the R21 success rate is lower than R21 rates is because people mis-understand the mechanism. Similar issues come up with the R03 mechanism. The R03 is used in different ways by different institutes. Some use it for small, self-standing projects, some use it as a supplement for their K awardees, some use it for preliminary data generation.

    If one is not responsive to the mechanism, one is not going to fare well. I have had an R03 (my first R grant) then an R01, then an R21 and now a competing renewal on my first R01 and a new R01. I've been fairly successful and all of those grants except the competing renewal on my first R01 and the new R01 were before I had a tenure track position.

    My best advice to young grant writers is create a highly structured grant in terms of its organization. My second line of advice is focus on proof of concept in preliminary data as opposed to preview of outcome (this is absolutely needed for seasoned grant writers). By proof of concept I mean preliminary data to demonstrate every single method. Refer to earlier papers for methods you have done and focus your preliminary data section on methods you have not published with yet. Finally, you must have an outcomes and inferences section. You need to consider all outcomes, not just the ones you hope for and walk through your inferences for each outcome including when you would not be able to separate 2 inferences.

  • drugmonkey says:

    It is Exploratory AND/OR Developmental PP and the latter part is not at all inconsistent with why newbie PIs have limited Preliminary Data.

    What will you conclude if you fail to see a score bump or other sign of the panel respecting your R21 success?

  • It is Exploratory AND/OR Developmental PP and the latter part is not at all inconsistent with why newbie PIs have limited Preliminary Data.

    You are misinterpreting what NIH means by the term. Take a careful look at the R21 parent FOA:

    http://grants.nih.gov/grants/guide/pa-files/PA-10-069.html

    What will you conclude if you fail to see a score bump or other sign of the panel respecting your R21 success?

    FACTUAL ERRORS OF INCOMPETENT REVIEWERS!!11!!1!!

  • Finally, you must have an outcomes and inferences section. You need to consider all outcomes, not just the ones you hope for and walk through your inferences for each outcome including when you would not be able to separate 2 inferences.

    This is not true.

  • drugmonkey says:

    Hahahahah! ErRORrS!!!!!

  • Pinko Punko says:

    Thanks for being so helpful, CPP.

    I think Kelly is suggesting that in general, it is more useful than not to consider what models your experiments are and are not able to distinguish as opposed to the unseasoned first year graduate student of attempting to prove one model as opposed to testing hypotheses.

  • "Previously it was limited to $100K per year for up to 3 years. There may be the occasional use via RFA that has these older parameters.

    And the occasional one that bumps up the limit - I worked on a successful R21 in response to a specific RFA, and we got $500k over two years.

  • TeaHag says:

    I wonder if we aren't going to see the R21 format rolled in with other mechanisms to allow program to "suck it and see" before sending additional monies.

    I recently participated in the writing of an RFA which was requested as an R21/R33. The idea was to submit an R21 with self-proposed milestones for productivity/proof of concept that would be evaluated towards the end of the R21 phase of funding for extension into three years of R33 ($350K/pa).

    According to the RFA they projected funding ~ 50% of the awarded grants through to the end, which approximates arithmetic logic.

    I wonder how successful this mechanism is, seems tailor made to try funneling bright ideas to a quasi or even actual "translational" phase.

  • ProfessorX says:

    Very simple.....Reviewers don't know how to review them. Scores on these things are ridiculously out of whack as they are judged like R01s by lazy reviewers who refuse to read the review criteria for specific types of grants. If we could get them to show us the distribution of R21 scores and R01 scores, I wager it will prove my point. CPP is right. They have to reviewed in their own study sections to have a chance.

  • [...] of my interpretation of OER head Sally Rockey's comment about R21s having lower success rates as "Kill the R21", what should appear but a note from NHLBI (thanks to PiT for making me aware). The R21 grant [...]

  • drugmonkey says:

    Very simple.....Reviewers don't know how to review them.

    So why can't the NIH just enhance their reviewer instruction on this mechanism?

  • [...] little doubt. In fact, we (as a whole enterprise- scientists and Program staff are doubling down. Kill the R21! Circle the wagons on feasible projects where the "significance" and "likely impact" is obvious to [...]

Leave a Reply


8 − six =