Is the R21 a dead end?

Aug 12 2014 Published by under [Education&Careers]

Long time readers may be aware that I've dabbled in the NIH game a bit in my wayward youth. In this effort I've had mixed success - one proposal landing 1%ile out of funding and two triaged. My last round of NIH reviews was particularly blunt in slapping me with some language to the effect of "You should learn to write for NIH before you resubmit", so I have concentrated on diversifying my NSF portfolio since last slinking away to lick my wounds.

In October I'm planning on submitting an R21 to support an off-shoot of a core project. I'm taking it in an unusual direction for me that aligns better with NIH and the R21 mechanism is, on paper just right for the proposal. That is because the R21 mechanism is geared towards limited support ($275k direct) for a short (2 year) and "risky" project. This mechanism was NIH's response to criticism that it only funds safe and proven projects that result in incremental science.

The idea, of course, is to get into the NIH system and generate data that would lead to an R01 proposal in a couple of years. I'm sure this is the intention of most people opting for an R21.


The numbers that Datahound churned up are a bit sobering with regard to future NIH success for R21 award winners. Although he's quick to point out the small sample size, it's very interesting that 72% of those with only R21 support in 2009 had no NIH money in 2013. Out of those who still had NIH money, 10% only had another R21. This suggests that the transition from R21 to R01 support is not all that common.

Why is that? There's several possible reasons. Certainly the smaller nature and shorter application (6 pages) of the R21 may be more attractive to PIs are smaller institutions. Perhaps the focus on the R21 takes away from landing the R01, stunting the progress of some labs. Maybe most R21 holders have no intention of taking the next step, but simply subsidize other funding with the R21. I don't know the answer and it would be damn near impossible to sort through all those (and other) options.

I'm sure the truth lays somewhere in between them all.

For those of you who have applied for an R21, why did you apply for that mechanism? Did/do you intend to shoot for an R01? If not, why not?

24 responses so far

  • DrugMonkey says:

    1) RFA and we didn't have preliminary data on a critical aspect

    2) a good idea...and we didn't have key preliminary data

    3) high risk idea

    4) straightforward developmental effort of a technology that we hope would lead to many R01 proposals.

  • proflikesubstance says:

    Were all funded?

  • dr24hours says:

    I have been Key Personnel on an R21 application. We applied to that mechanism because our methods were novel (my simulation combined with a clinical trial), and because the clinical trial itself had high risk in terms of effectiveness (we expected it to be effective for those who embraced it, but that not many would). Though not high risk in terms of danger to patients. We also chose it because 2 years was a great duration for our population.

    The PI was a junior faculty member. And I was junior at the time too. We scored at the 49th %ile, and the PI intended to resubmit, but never got around to it. I was pleased that the simulation aims were well-received. We got hit hard on the junior-ness, and ironically, the risk.

    Had it been awarded, yes, she was intending to do a larger R01 clinical trial that would no longer need a simulation aim.

  • drugmonkey says:

    A bit early to hit the sauce, isn't it PLS?

  • proflikesubstance says:

    I didn't mean "funded on first try", but were those ideas eventually funded under this or other mech?

  • Dr Becca says:

    I applied for an R21 before the paint had even dried in the new lab. There was a PA for research in my specialty area, and I had an idea for an "exploratory" project that seemed like a good fit for an R21 as opposed to an R01. It was funded with no preliminary data, only a demonstration that I had done all the techniques as a postdoc. However, 2 years goes by FAST. I submitted an R01 that was based off the R21 data but it was triaged, and I am now an unfunded person 🙁

  • boehninglab says:

    The R21 mechanism is a dead end, and many of the institutes know this (including NIGMS). In my experience having an R21 and submitting several others, they:

    1) may still require significant preliminary data. My last submission was dinged for this. It probably depends on the reviewers you get.
    2) If you do not publish several papers during that 2 year period on the project, it is nearly impossible to convert to an R01. I got severely dinged for this (in this end, the project produced 2 large papers several years after completion of the R21).

    If it is a side project and your lab doesn't need the funding, go for it. If you are junior faculty trying to get your first NIH grant, stay far, far, away.

  • @Mtomasson says:

    Why I applied for R21 mechanism: thought our preliminary data, all 2.5 years of it, were not enough to be considered solid for R01.

    Did/do I intend to shoot for R01? Yes. It has been a year since my R21 ended, almost 4 years since the beginning of the project, and we are now ready to submit R01.

    Between the lines of my answers, I hope to point out that a major problem with this mechanism are differences of opinion/mechanics of what it is for. "No preliminary data needed," is completely loaded. I was surprised on a visit to the NIH to hear from POs that the intention of the R21 was to give ESTABLISHED investigators a way to branch out into new areas. I know many of us thought that it was a mechanism for junior people to get started.

    No preliminary data needed can be's written down after all...but my success, and from what I've seen on study section, a lot of preliminary data are needed to reach the "no preliminary data" threshold. I'm mean to say it's mostly a fib. Not to say it can't happen, just that study section reviewers have a hard time looking at a grant without data.

    I haven't thought this through as well as Drugmonkey, but in the back of my mind, this issue has made me wonder whether having many different mechanisms is actually counterproductive. I know the NIH means well by what they are doing...but.

  • proflikesubstance says:

    So, in practice, I wonder whether there isn't a self-fulfilling-prophecy thing going on here. Sounds like an R21, in isolation, is not enough funding to get to an R01. At least not at the end of the R21 because it takes time for the products to get out there. Is it possible that the reduced rates of future funding are simply because this mech is often being used to supplement other funds?

  • iGrrrl says:

    I think the reduced rates of future funding rest upon a number of things, but for R21s awarded to beginning investigators, one big problem (IMO/IME) is that when the R21 award lands, the investigator often takes a breath and relaxes. In order to keep funding going, the R01 needs to go in before the end of the first year, not 3/4 of the way through the second. Knowing what those R01 aims are likely to be, because you thought them through before the R21 went in, will help that application be more competitive.

    My usual advice is that no R21 or R03 should be submitted unless you already have sketched out the Aims of the subsequent R01. That helps you focus the R21/R03 on why you are not yet ready for the R01, and write the application to fill the holes. Having the longer-term goal of the R01 in mind helps keeps the work in the lab also focused on getting the needed preliminary data for all three aims, not just the one that required the proof of concept/removal of the barrier.

    Think about it. You put in the R21. If it is funded on the first round, it takes 9 months at least between submission and paid grant. You don't sit and wait during those 9 months. You work on the preliminary data for the subsequent R01. If you go in even 3/4 of the way through the first year of the R21, you've had 18 additional months to generate preliminary data for the R01, and if you already sketched out the Aims for that R01, you can be pretty focused during those 18 months. And if the R01 doesn't succeed on the first go, you can resubmit while you still have funding.

  • drugmonkey says:

    One of those funded and led, eventually to R01.

    One became two R21 apps but idea abandoned due to appearance of a RFA for a BigMech I couldn't compete for.

    One has the ideas on hold while I fudge around with some things.

    One is still under active pursuit in the lab.

  • drugmonkey says:

    I still say if the idea is right for r21, and a deadline is handy, then go for it. Make your case. Ignore the paylines (cause if we paid attention to the odds we'd never submit any grants)

  • proflikesubstance says:

    Indeed, in this particular case it seems like a good fit. It would be supplemental funding for an ongoing, NSF-funded project, but taking it in a new direction.

  • drugmonkey says:

    Sounds like an R21, in isolation, is not enough funding to get to an R01.

    My usual advice is that no R21 or R03 should be submitted unless you already have sketched out the Aims of the subsequent R01. That helps you focus the R21/R03 on why you are not yet ready for the R01, and write the application to fill the holes.

    I guess one aspect of this that may differ in my advice is that I see the new (especially) and not-so-new PIs' attempts to get R01 and R21 funding as independent. Particularly when you are a noob, your primary goal is to think about how to get to R01 funding. Going *through* an R21 as if this is a necessary step is too uncertain. Noob Asst Profs should pursue an R01 project* primarily and an R21 project only as a second option as time permits.

    The time line for an R21 just doesn't make it worth it to assume you will springboard off that to gain R01 funding *unless* you have additional major support in place and this is a sort of optional second or third program of research.

    *I am not that sure that the early stage process of data, design, argument, etc for an R01 is that similar to an R21. The are really different beasts in my view.

  • JunionProf says:

    For me, the nice thing about the R21 is that it is not percentiled at NIAID. Awards are made strictly on the Impact Score. As of right now, I submit exclusively to a small study section and because only 10% (or less) of R01s are funded from this study section you can only realistically hope to have two R01s funded at any one time and even that is a very difficult proposition.

    Whereas if the study section gives the R21 a good enough score it will be funded. Period, end of story. Doesn't matter how many other R21 grants also got excellent scores. Thus far my success with the R21 has been mixed. I had an A0 score a 17 and get funded. I had another A0 score a 38 and I have resubmitted A1 and am awaiting reviews. For me, it is a worthwhile mechanism. However, if I didn't already have a R01 I would not be submitting many, if any, R21s. Getting the first R01 is key.

  • iGrrrl says:

    I totally agree that they are different beasts, DM. And I agree that new faculty should go for the R01 as fast as possible, and only use the R21 mechanism if they specifically need support for proof of principle/high-risk, high-payoff. It's a tool that can be used effectively in the right circumstances. Some I/Cs stopped supporting investigator-initiated R21s when they looked at the low rate of transition to R01, which I think was the result of misuse of the tool.

  • crystaldoc says:

    Supposed to be different beasts, but ... In recent years I have seen a lot of R21 apps that are chopped down versions of R01s that didn't get funded, in response to limitation on resubmissions and shrinking pay lines. I wonder if that trend will reverse itself now that unfunded R01s can go back in as new. I have not checked up on success rates of these chopped R01s as R21s, but it seems like some score pretty well. Although not as high in "innovation" as the mechanism would seem to call for, reviewers may be impressed with the preliminary data and extensively worked grantsmanship; they tend to compare favorably versus genuinely early-stage efforts. From the PIs perspective, the rationale for doing this is/was that it is not that much work compared with writing a brand new grant, and some money is better than no money.

    I have put in R21s for new ideas with limited preliminary data, one of which was funded. I have also put in an R21 chopped from an unfunded R01 (prior to change in policy re resubmission as new R01) which scored 15% and the resubmission is currently in review. I think it has a shot because of some pretty innovative aspects and it actually was pretty risky and light on prelim data to compete as an R01, so actually not bad for intent of R21 mechanism.

  • crystaldoc says:

    Sign of the times: by "light on preliminary data" for an R01, I mean something on which I have fewer than 10 publications, haha ...

  • proflikesubstance says:

    I would imagine that the "chopped R01" version is not what NIH had in mind, but I can see how people are trying their luck with that now. Not sure how one could trace the effect of the new resubmit policy, but it would surprise me if there is at least a change in the types of proposals being submitted as R21s, if not the number.

  • eeke says:

    My last R21 application was dinged for not enough preliminary data (even though one of the Aims was nearly finished). The only reason I used the R21 mech was that I felt we didn't have enough preliminary data to warrant an R01. It seems to depend on the study section that it goes to - maybe the PO's have some idea (or look up on Reporter) of which study sections actually follow the reviewer instructions. My current strategy is to get some additional preliminary data and submit a new R01 application.

  • AnotherAnon says:

    Thinking of going R21 with a project. Partly because I'm a new PI and my NIH contacts swear that new PIs have no chance at R01s but may do better with R03/21. I don't know. The pay lines are similarly low.

  • Dave says:

    A good discussion but, for me, the major issue is preliminary data, as many have alluded to here. How much preliminary data is expected from a new PI in an R01 app? I know that is a question with a lot of 'well it depends' answers, but I suspect this is what drives most noobs toward the R21 mech, even though in practice it sounds as though you need a fair bit of data to actually land an R01.

  • Dave says:

    ...meant R21 at the end there, obviously.

  • Mark P says:

    I would echo one of drugmonkey's comments which for me is the key

    "Particularly when you are a noob, your primary goal is to think about how to get to R01 funding. Going *through* an R21 as if this is a necessary step is too uncertain. Noob Asst Profs should pursue an R01 project* primarily and an R21 project only as a second option as time permits."

    A new PI needs to:

    1) Continue to be their own best postdoc and generate new data, to get papers out
    2) write ONE good R01

    Anything else leads to spending the whole day in your office which is the recipe for disaster

Leave a Reply