There is an interesting new survival analysis for PIs that have been awarded at least one R01 posted at RockTalk.

Briefly, what you are looking at is as follows:

We chose three cohorts of first-time R01-equivalent awardees — those who received their first R01-equivalent award in 1989, 1997, or 2003….We used data on these three cohorts for a Kaplan-Meier analysis to look at rates of retention. …So we used this to analyzing the number of years between the first year of R01-equivalent funding, and the last time an individual receives any additional research project grant (RPG) funding – whether it be from the non-competing continuation of their 1st R01 or another RPG award.

The bulk of Sally Rockey’s analytical comments focus on the sharp dropout associated with the first (presumed) interval of support, i.e., 3-5 years after the very first R01-equivalent award. And her conclusion seems to be that this is a problem that needs fixing.

These data seem to support the concept that if there is an intervention needed in retaining scientists in research, it would need to come at the renewal stage of the first award, or as some call it the “second” award. Indeed, we are giving increased focus to this stage through some of our new award mechanisms, such as the National Cancer Institute’s Outstanding Investigator award, and will continue to seek ways of keeping our talent from leaking out of the pipeline.

I am not sure that I agree with this general conclusion from the data presented.

Remember, we are talking system-wide statistics here, not the fate of your five closest colleagues, your training mentor or yourself.

The extreme case would be that once anyone manages to land an R01-equivalent award as a PI, that the NIH should move heaven and earth to keep them funded for the duration of their career. That is an arguable position, but I think it is wrong.

It is wrong for two reasons, which are related to each other. The first reason is that if we re-adjust the system to keep everyone in once they have entered, this will sharply reduce the entrants. It will reduce the number of people who get a chance to prove themselves. This may seem fine and dandy once you have passed the first-R01 hurdle yourself, but this is mind bogglingly forgetful of the position one was in before this and mind bogglingly arrogant in assuming you would have been one of the lucky few.

From a system perspective, this cuts down on scientific diversity. It cuts down on the ability to try out a range of scientific ideas and approaches to see which ones stick. It substitutes the limitations of advance prediction for the virtues of empirical testing.

This would also increase stagnation and slow progress. It would. When you have sinecure funding, I’m sorry but the pressures are not as high to be creative, productive and to diversity your scientific thinking. Yes, from our current vantage point of the amount of time spent securing funds versus doing the science, this may look better than the usual. But take the longer view here. Our competitive system has its virtues in terms of clearing out the dead wood and encouraging better efforts from those who are actively participating.

I think the real question here is the appropriate balance. The desired survival rate.

And we should be very clear that we expect there to be some amount of PI dropout.

Personally, I think that having the major reduction in PIs after the first interval of funding makes a lot of sense. Better than at year 11 or 16, for example. A given PI has had a chance to try out her ideas. He has been given the opportunity to show what he can do. Again, on a system-wide basis, some of these individuals are going to fail so badly that they are never funded again. Do recognize that many will suffer intervals of no-R01 and then come back. Datahound addressed that in a post earlier this year. But many will disappear. Wouldn’t it make more sense to have it happen before the NIH has wasted another 5 year interval on them?

Now before I get too far down the path, I will recognize that this is only the start of the data analysis. We want to know a few more things about who is shelled out of the system, never to return, who is able to fight back in after a gap and who is able to sail along with continual funding. This will allow us to see what may be undesired effects or categories of PI/applications that we wish to specifically protect. Is this part of the perfect storm that hits women particularly hard? Human subjects research? Physician-scientist PIs? Ecology or sociology?

Nevertheless, when I look at these initial survival curves, I just don’t see the problem that seems so obvious to Dr. Rockey. I don’t see how this is the next problem that requires special fixes from the highest offices at NIH.