NIGMS Data on the timeline from BA to PhD to Asst Prof to R01

The previous director of NIGMS, Jeremy Berg, continues his established interest in transmitting NIH career and grant award data to you, Dear Reader. This is, in essence, a guest post. I received the following email…

Hi DM: I was reading your recent post and remembered that I did some analysis at NIGMS that I presented at our advisory council and on other occasions, but it predated the Feedback Loop.

The study was to look at newly funded assistant professors or equivalent ranks from 2004-2006 (360 individuals) and to examine the times (in years) between when they received their BS/BA degrees, when they received their doctorates, when the started in their assistant professor positions, and when they got their R01s (manually from their biosketches). The results are attached. The median time from BS/BA to funding was 15 years. The average age of award was estimated to be 38 years.

Let me know if you have any questions. Feel free to post this if you want.

Best, Jeremy

I most certainly do want. Take a gander at this, folks (click to embiggen).

Now obviously the 360 individuals are but a tiny slice of the NIH-wide pool of PIs. And the 2004-06 window is just one point along the long trendline that we were just discussing. Still, it reflected the situation in one IC right at the time the NIH was getting exercised over the 42 years-to-R01 graph. So it hits the right note.

As a completely untethered personal opinion, I think graduate studies that last 6.7 years are far too long. If I was running the zoo I’d like to see that median back at 5.0 and be much, much peakier. That long-ish tail extending to 10 years and beyond is ridiculous. The post-doctoral training interval, I have less problem with. Five years doesn’t seem too horrible. Although I suppose that could be shaved back a little bit too.

The time to first award after appointment just puts a histogram on the problem that we already know about. Again, were I the Boss of Science, I’d want to get major funding in the hands of good people faster. If you work backwards from the population that eventually won an award, i.e. are “deserving” in some sense, wouldn’t you rather they had the $$ as early in their independent career as possible?

It would be really fascinating to know if the ESI hoopla has shifted this distribution back to the left by any significant amount.

Now thinking about these data some more….wow, take heart o ye of dismal training experience! I was thinking about the prediction for success/failure for a half second until I realized these data only capture those who were eventually successful at landing an R01 from the NIH. Look at those 10-18 (!) year grads. Look at the poor souls stuck for 10-15 years in postdoctoral hell. Sure, they are the exceptions to the distribution and no doubt the successful exceptions to the distribution of folks who got stuck for that length of time in graduate school or postdoctoral “training”. But it was possible for some to succeed at last. Wouldn’t you like to hear their stories? I know I would….

Updating the age of first NIH R01 award trends: Flatlining

Way back in 2004, a News of the Week bit in Science by Jocelyn Kaiser [PDF] included the now infamous graphic showing the average age of first R01 award from the NIH had risen to 41.9 years of age. This led to a lot of “Jesus-is-coming-look-busy” behavior out of the NIH, as reflected in youth-focused initiatives, including the creation of the Early Stage Investigator definition (no more than 10 years past the PHD award) out of the New Investigator (no prior major/R01-equivalent award) class.

We now have an update from the Rock Talk blog which continues the trend up through the 2011 FY that just ended in October.

My initial reaction was “That’s it????”. After all these measures and attention, all they’ve managed to do is to flatline the trend at 42 years for the past 7 years?

My slightly more considered reaction is “Damn, I hate being right.”

I’ve maintained in both direct and indirect ways on this blog and elsewhere that the NIH approach to fixing the problem of younger investigators fairing poorly at the NIH trough is too little and too indirect. The major input to the system has always been the study section. Peer review. Conducted by “peers”. Except, oopsie, the “peers” are diverse in sex, geographical region, job category, ethnicity/race, University/Institution type….but most certainly not diverse where it counts for the young. The NIH choose to correct the bias against younger investigators by using Programmatic Priority decision making to pick up awards out of the order of funding (which doesn’t compromise science, btw). While I am generally in support of the system by which Program staff of NIH ICs decide to fund grants mostly, but not entirely, in the order determined by peer review I was not a fan of this as the only mechanism to fix the problem of young investigator hosage at the study section level.

As the prior NIH Director, The Great Zerhouni, noted [PDF] in 2007, immediately after implementing the exception funding (aka “pickups”) strategy, study sections started “punishing the young investigators with bad scores,” says Zerhouni.

This initiative fell smack into the truism that despite the best attempts (and firmest admonishments) of the SRO that the study section is to rank applications in a smooth distribution and not pay one iota of attention to the “F-word”, people can’t help but bench mark their scoring to the Fund/Don’t Fund dichotomy. Which means they operate at some level on an internal prediction or expectation* about where paylines and/or eventual success rates are going to fall. When the reviewers got wind that special considerations were being extended to ESI or NI applications, they re-calibrated their perception of the payline and scored accordingly. It was not so much that people were being intentionally punitive, they were simply expressing their existing bias under new parameters.

Some may have even thought they were being super clever because now they could score one more Established Investigator application within the perceived payline and still give the ESI application a likely-fundable score, all the while appearing to spread out the scores in his or her pile of assignments. Bonus!

There is another post up at Rock Talk that provides further information on the longitudinal age trends of the NIH investigator pool. It smells to me of more fake learned-helplessness excuse making along the lines of “what can we do? the demographics are relentless.”. Sally Rockey seems particularly fond of this animation of slides they’ve put up in numerous powerpoint files, but I’m unimpressed. The trendlines for PIs over 66 and under 36 are more interesting to me.

Depressing to see, isn’t it? As a comment on that post noted:

http://nexus.od.nih.gov/all/2011/05/13/update-on-myth-busting-number-of-grants-per-investigator/ includes the total number of PIs on grants in 1986 and 2009. In 1986, there were 16,532PIs of which around 16% were age 36 or younger. In 2009, there were 26,183 PIs of which around 3% were age 36 or younger. That means, despite the large increase in PIs, the raw number of young PIs plummeted from around 2645 to 785. This is depressing.

Indeed.

“So, smart guy, what is your solution?“, I hear you ask. Well, I’ll tell you.

Number one, fix the diversity on study section issue. Strive for representation of those who are as-yet-unfunded that approximates their application numbers. The competition of biases is the only tried and true solution.

Two, continue the pickups. Do so in a balancing way that does not merely focus on “Young versus EveryoneElse”. Get serious about re-balancing the aging juggernaut of Baby Boomers against the two scientific generations that slot in just below them**.

Three, ditch the current definition of the ESI as being 10 years out from PhD. This was arbitrary, unfair and asinine from the beginning. It should have been tagged to the first professorial level appointment and/or the first time proposing applications to the NIH as a PI (any Research mechanism, not just R01).

Four, go after study section behavior. Hard. Train and educate. Provide measures of treatment of young versus established investigators, break it down by age if necessary. Give feedback and hold the section’s feet to the fire to do better. Expect SROs to ditch reviewers who just can’t get on board. Yes, they know. They can do it.

Five, snap the Universities into line. This is hard politically but duck soup in a practical sense. NIH awards funding to whom it chooses, as we’ve already discussed above, re: pickups. There is no “deserve”. There is no “right” to funding. There is ONLY what serves the interests of the NIH. So they are perfectly well within their rights to say things like “Yeaaah, we’re not going to fund any Training Grants at an institution that has a crappy record*** on hard money hires for Assistant Professors”. Or “No supplements for you!”. Or “We’re not going to use our discretionary exception funding for any of your**** applications until you get better.” Or even “Dudes, we’re totally serious. No new awards at all until you show us more Assistant Professor hires, better age distribution and yes we will be examining the deals just like we do for the R00 phase of the K99/R00 mechanism to make sure you aren’t pulling shenanigans”.

In my view, we will know the NIH is not really serious about these demographic trends until and unless they take a stab at item Five.
__
*A lack of calibration on this issue can lead to some sad results. Such as when those who are new to the section in question (or perhaps NIH review in general) lament at the dinner afterwards that they thought they had recommended “a good score” for some particular proposal.

**Self-serving alert!

***Define criteria however you like. Come up with something that you think will address the goal. This is not the point, the point is for NIH to actually deploy it’s considerable power of the purse.

****I know you didn’t forget that the applicant and the awardee for the NIH grant is the University and not the designated PI, Dear Reader. Right?