R01 Distribution for 2014 by PI and IC

Jan 26 2015 Published by under Uncategorized

There has been much discussion (including recently e.g. here and here) about the distribution of the number of grants per PI. Here I examine the distribution of the R01-equivalent (R01 and R37 (MERIT) awards for Fiscal Year 2014.

New (Type 1), Competing Renewal (Type 2), and Non-Competing Continuation (Type 5) award information was downloaded from NIH RePORTER. This yielded 23526 awards. Awards were groups according to the PI. For awards with multiple PIs, only the corresponding PI was considered. The number of awards ranges from 1 to 6. This distribution is shown below:

R01 PI distribution-2

PIs with 1 R01 grant account for 77.8% of the awards whereas those with 6 awards account for 0.04% of the awards.

This can also be viewed in terms of the fraction of the dollars that are devoted to each group of PIs as shown below:

R01 grant amount pie chart

The fraction of the $9.9 B in this pool (not that this does not include supplements) going to investigators with single R01s is 60.0%. The average size of an R01 award does not vary significantly between the different groups of PIs so that PIs with 2 R01s have (on average) twice as much R01 support as those with 1. The percentage of R01 funds going to PIs with 2 awards is 27.7%. The percentage of R01 funds going to those with 6 awards is 0.17%.

I also examined the distribution of funds to PIs as a function of the NIH institutes and centers (ICs) that fund them. I have previously examined the distribution of R01 funds by IC. Here, PIs are classified according to the IC or ICs that fund them. If a PI has R01s from more than 1 IC, then they are classified as investigators of all ICs and all funds that this PI receives are counted for the involved ICs. This results in the double-counting of funds, but it is germane to issues such as the special scrutiny given to well-funded investigators.

The distributions of the number of PIs as a function of annual total cost R01 support for the four largest institutes are shown below:

4 IC PI Curves-1

 

Both the first and second peaks for NIGMS and NCI are shifted to smaller amounts relative to the corresponding peaks for NIAID and NHLBI. This may reflect policies of cutting budgets upon award and, in some cases, in out years.

The region of this graph for annual total R01 costs greater than $1 M is shown below:

4 IC Pi Curves-2

 

The curve for NIGMS is below those for the other three institutes, likely due to the NIGMS well-funded investigator policy (although some of the difference could be due to differences in the costs of the types of research supported by different institutes).

These data, together with some analysis of the publication patterns for these groups of investigators currently underway, may be useful as new policies regarding the distribution of funds across groups of investigators are proposed and discussed.

UPDATE

I presented graphs of the number of PIs as a function of funding level for just the four largest institutes. Below is a table that shows the number of PIs for each institute and center (recalling that a PI is associated with an IC if they have at least 1 R01 with that ICs; PIs with R01s from multiple ICs are associated with multiple ICs) as well as the amounts (in $ X1000) for the 10th, 50th, 90th, and 99th percentiles for each IC. Thus, for NIGMS, 10% of the PIs have total R01 funding less than or equal to $277K annual total costs while 99% of NIGMS PIs have less than or equal to $1604K in annual total R01 costs. The ICs with the lowest levels of the 99th percentile are NLM, FIC, NIDCR, NCCAM, and NIGMS. Those with the largest levels of the 99th percentile are NIDA, NICHD, NHGRI, NIAAA, and NIDCR.

IC Number of PIs 10th %ile 50th %ile 90th %ile 99th %ile
NIGMS 2939 277 348 845 1604
NCI 2675 283 394 1005 1979
NHLBI 1996 363 500 1207 2212
NIAID 1884 344 463 1138 2171
NINDS 1640 316 384 1036 1856
NIDDK 1633 309 426 1034 1889
NIMH 1212 346 568 1226 2231
NICHD 894 286 496 1106 2535
NEI 802 331 394 920 1810
NIDA 797 304 570 1318 2633
NIA 792 297 517 1223 2291
NIAMS 595 315 386 1052 2052
NIDCD 491 308 400 909 1394
NIEHS 356 310 458 1015 1897
NIAAA 339 297 507 1127 2354
NIDCR 318 364 474 1161 2333
NIBIB 308 315 482 1187 2251
NINR 162 370 511 946 1971
NHGRI 119 301 505 1200 2442
NCCAM 114 329 552 1166 1584
NIMDH 76 337 393 1013 1931
NLM 57 283 399 1041 1293
FIC 22 49 53 388 1393

13 responses so far

  • lurker says:

    Hi Datahound,

    In the P.I.E charts of the PI's with just 1 R01, can you break this down further by the portion that are likely modular (the ~$300-$350 sharp peak in 3rd graph) versus non-modular (everyone else)? The big slice of 1-R01 may mask several R01/R37 non-mods that can exceed 2 modular R01's. I've seen a current single R01's at ~$800K direct costs, and they may make up a bulk of the trailing tail inGraph #3. Or perhaps you could convert Graph #3 into a PIE chart, what fraction of the PIs are in the 1 mod funding level versus the 2 mod level versus the "well-funded"?

    Thanks,
    Lurker

  • drugmonkey says:

    I've seen a current single R01's at ~$800K direct costs,

    Yes! It is obvious to anyone that if we can just get rid of these greedy inefficient rat bastiges with $800K single R01s that there will be plenty of money to let the rest of us get our grants at a 30%ile score.

  • E Pratt says:

    I wonder what the target percentage is when people call for a more 'even' spread of funding across investigators. Should single R01 investigators constitute 70% of NIH funding? 80%? And how are investigators supposed to maintain a lab if there is a gap in funding? Especially as your earlier post showed that the likelihood of re-funding after 1 year is only 18%.

    I wonder what these distributions look like for the NRC, which is often used as an 'ideal' example.

  • lurker says:

    @DM,

    For a prolific blogger and twitter-er, you have quite the number of hypocritical statements. On the one hand, you claim to be an advocate for the plight of noob investigators clawing for survival (yeah, I've read your many entries, no need to refresh my memory). On the other hand, you think more R37's are the answer to the funding crisis, but that does nothing for noobs, only gives a nice entitled cushion to the BSD class. And your sarcasm above seems to put you in the pocket of the BSDs, but your more righteous than that, aren't you?

    I'm not saying we take away the entire humongous R01 from the one BSD, who also happens to be HHMI to boot. I'm just wondering if said BSD is so awesomely amazing to deserve HHMI AND an R01 worth 2.5X of a modular. Wouldn't HHMI and one mod R01 be reasonable? If this "leader of the field" gave a real crap about how the funding crises is demoralizing new investigators, might he have more modesty not to take such a big slice of the finite cake? Better than writing an opinion piece in Genetics bemoaning the decline in funding of his favorite model organism.....

    No wonder Mbeisen wants to smack you...sometimes you are full of it.

  • potnia theron says:

    DH - I know you focus on the big IC's, but I would really like to know if the little ones (NIDCR, NIDCD for example) follow the same patterns, both in general and in specific. A future post?

  • PI in RI says:

    NIH's Project RePORT is an awesome resource for data mining. I wish there was some way to integrate (strip identifying information?) into this system ALL applications submitted, not just the funded ones. For one example, I'm interested to know what is the funding profile for the whole applicant pool for a given IC and/or study section. I would guess that the % of zero-R01 applicants (for an R01) exceeds the % of one-R01 in the total funded pool. Also would love to know new R01 success rate of PIs w/ zero vs. one, two, etc. R01s.

  • drugmonkey says:

    On the other hand, you think more R37's are the answer to the funding crisis, but that does nothing for noobs, only gives a nice entitled cushion to the BSD class.

    You are not reading me carefully. I suggest something like the R37 needs to be "greatly expanded" to address one aspect of the crisis, i.e. grant churn from those who are going to eventually get funded anyway. I am not suggesting it only be a bennie for the already-privileged class (which is why I am not pleased with MIRA). PECASE is the R37-lite for noob investigators and that is the other supporting example of existing mechanisms NIH could draw upon for both evidence of success and a way to do it. I am further arguing that if the NIH wants to stabilize the grant churn they need to do this for a LARGE number of investigators, including all PI demographics.

    And your sarcasm above seems to put you in the pocket of the BSDs, but your more righteous than that, aren't you?

    Why does my sarcasm about your ridiculous suggestion mean I am "in the pocket of the BSDs"? The Hound presented data that once again demonstrate the relative amounts of grant dollars that are available from different categories of "the overly funded". You go the usual complainant one further by limiting your definition of the overly funded to those who have R01s in excess of $800K per year. How many of those can there possibly be? And since NIH gets some 60,000 apps per year and DH has posted data showing that about 5,000 PIs enter and exit the ranks of the funded, how are we supposed to view giving the six dudes with $800K/yr R01s a haircut as anything remotely worthy of discussion?

  • lurker says:

    @DM

    Who are the ones getting the current R37's? BSD's only, from what I can tell. How would you realistically "expand" this to all demographics it its only going to be based on the standard roulette of study section review? This is the first I recall you detailing expanding R37-vis-a-vis a PECASE-like mechanism, which isn't clear to me how this is any different from your standard beauty prize pageant for noobs. Not really sustainable.

    If you say I'm not reading you carefully, same could be said about you in thinking I'm only gunning for one or six BSDs with a single R01 >$800K. What I was really thinking is that there needs to be a system-wide revolt, like a French Revolution or a Progressive Era/TR Trust Busting shaming of the NIH 1%'ters. Expose the BSDs with HHMI AND are grubbing for more than 2 mod R01's worth of NIH support. This is more than six dudes. Go look at the Y-axes of DH's most recent analyses of PIs with more than $1M support, tally up the IC's and you get >100, maybe even >200 such BSDs. That could be 300-600 mods that could help alot of noobs and midcareers keep the lights and 2 staff max running instead of shutting down for good.

    I would concur cutting these BSDs down to size won't save and prevent the churn of the 5000 PI's that enter and exit, and it shouldn't. But it could save a lucky 8%, some of whom are going to simply be collateral damage from a funding system that has gone sclerotic.

  • Comradde PhysioProffe says:

    Kill all those greedy fuckers with three or more grants and that'll totally double the number of poor fuckers with their single genius R01!! Oh, wait. It won't even move the needle significantly? How can that be???? It's those greedy fuckers' fault, right?

  • […] a recent post, I presented the distribution of the number of investigators who were PIs of from 1 to 6 R01 grants […]

  • […] comment to my previous post asked about the size distribution for R01s for investigators with a single R01 grant. This […]

  • Drugmonkey says:

    How is extending R37-like intervals of non-competitive renewal to more scientists not sustainable? Some 20%, maybe 40% depending how you look at it, of PIs are approximately continually funded for at least 10 years. How is it unsustainable to try to get these kind of numbers into the R37/PECAsE-like pool?

    (r37 *can* btw, be withdrawn for lack of productivity)

Leave a Reply