Given that the problems are obvious, the strength of the pushback NIH has received on the GSI is amazing. I discussed the issues informally with a range of NIH-funded investigators. I have been trying to understand the underlying causes of the pushback and thinking about rationale alternatives that reasonable people could accept.

The community could make a wiser decision about the alternative, if we knew more about the 1% of PI’s, who hold the 11% of NIH funding, since that seems like the place to start with finding some money to fund more PI’s. We might ask, what does their productivity per dollar look like? More information about the next 9%, who have 29% of the funding, would also be helpful.

What is the obvious course of action? The community should push very hard to limit the total research funds controlled by individual PI’s. It is obvious form the NIH statistics and personal experience that productivity and good mentoring fall off when research groups get too large. One confounding factor is that PI’s have different tipping points. For some PI’s a total of 6 students, postdocs or technicians is too many, while other PI’s can handle 12 or 15 people. My best guess is that groups of 6-12 are the most productive.

However, a limit on lab size does not translate simply into a number of NIH research grants because the cost can vary considerably. This is where the GSI fell short. Therefore, any policies aimed at spreading the money around more evenly must take into account factors including but not limited to

  • the cost of the research (yeast are inexpensive while animals, human subjects, recharges for time on some instruments is expensive)
  • research support from sources other than NIH (or other federal agencies) for the PI and members of the lab (including fellowships)
  • the dependence of the PI on NIH to pay their salary
  • administrative contributions of the PI to research and institution building by heading large research projects such as centers or program project grants
  • intellectual contributions of the PI to research by helping multiple laboratories with specialized methods (say bioinformatics or electron cryo-microscopy).

I think that we have an elephant in the room that gets in the way of making the obvious, rational decision to limit the total funding to an investigator. That elephant is the financial model of medical schools that depends on recovering faculty salaries from NIH research grants. Remarkably, it is very difficult to actually nail down this number for any medical school, but anecdotally most people believe that basic science faculty at research-intensive medical schools are expected to raise 50-90% of their personal salaries on (NIH) grants. The so-called “contributed effort penalty” is a strongly negative incentive for institutions to pay faculty salaries for doing research (the main activity of most scientists at medical schools). These rules make institutions pay indirect costs on any salary that they pay faculty to do research. The current accounting practices should be replaced with new guidelines that reward institutions for paying their faculty. Until we can get this fixed, medical schools will fiercely resist any cap on the number of grants or total funds per PI. The solution should be to give institutions some time (say one decade) to adjust their financial model, so that they pay their faculty some minimal fraction of their salaries (say 50% for starters). This could be done by adjusting fund raising priorities to direct philanthropy or other support toward endowing faculty salaries rather than other expenditures such as new buildings for a decade or so.

Policy options:

  • A limit lab size with a nuanced procedure that takes into account the factors that contribute to the variable costs of doing research and the capacity of the PI to manage their group. At the same time we should resist formulaic, one-size-fits-all approaches such as the GSI, which failed to take the variable costs into account and included support such as training grants that do not contribute to the size of a research group.
  • Use a variable percentile cut off for funding grants depending on the size of the group. NIGMS already does this at the staff and council level. Specifically, funders could use a lower percentile cut off for each incremental grant. For example, they might fund every lab with no other grant up to the 30th percentile. This will take care of junior investigators trying to break into the system and will keep more labs from closing. Then require a 20th percentile score for the second grant and the 10th percentile score for the third grant (obviously on a sliding scale rather than in big steps). Even then, NIH should take into account the other factors listed above.

The views and opinions expressed in this blog are the views of the author(s) and do not represent the official policy or position of ASCB.

Other GSI Opinions
Opinion on GSI by Sam Dundon
Opinion by Charles Easley
Opinion by Mark Peifer
Opinion by Jessica Polka
Opinion by Tom Pollard
Thomas Pollard

Sterling Professor of Molecular, Cellular, and Developmental Biology and Professor of Cell Biology and of Molecular Biophysics and Biochemistry, Yale School of Medicine