The argument for diversifying the NIH grant portfolio

Abstract

The United States has been a leader in biomedical science for decades, in large part because of the strategy used by the National Institutes of Health (NIH) to invest its budgetary portfolio. They identified talented young scientists from each generation and gave them the resources they needed to initiate and maintain strong research programs. However, recently this investment has become less diversified, with a larger fraction of grant dollars in the hands of a smaller fraction of researchers. This threatens the future of our field, as many productive early and midcareer scientists are facing having to close their labs. NIH and others have studied this problem, gathering data that suggest that over a certain level of funding to an individual investigator, there are diminishing returns in scientific output. Here I review these data and examine the issues that led NIH to propose and then reverse a cap on funding to individual investigators, the Grant Support Index. I consider other proposed solutions, and call on all in the field to examine whether the status quo is acceptable, and if not, urge them to propose and advocate for concrete alternatives.

FIGURE 1:

NIH research portfolio investment is skewed toward a small fraction of the investigators funded. Source: Presentation by NIGMS Director Jon Lorsch to NIGMS Council, May 26, 2017 (Lorsch, 2017). These data reflect all research dollars, but analyses restricted to research project grants are similar: 1% of scientists get 11% of funds, 10% of scientists get 37% of funds, and 20% of scientists get 52% of funds.

FIGURE 2:

Evidence supporting diminishing returns at the upper end of the funding spectrum. Association of the annual weighted Relative Citation Ratio (RCR) among all papers linked to a scientist’s grants with annual Grant Support Index (GSI). Owing to skewed distributions, both RCR and GSI values are natural log-transformed; the numbers inside the axes represent the raw, nontransformed values. An annual GSI value of 7 corresponds to approximately one R01 grant, while annual GSI values of 14 and 21 correspond to two and three R01 grants. Note the decreasing slope of the regression curve as annual GSI increases. Source: Presentation by NIGMS Director Jon Lorsch to NIGMS Council, May 26, 2017 (Lorsch, 2017).

FIGURE 3:

Evidence supporting diminishing returns at the upper end of the funding spectrum—non–log-transformed data. Source: Presentation by NIGMS Director Jon Lorsch to NIGMS Council, May 26, 2017 (Lorsch, 2017).

FIGURE 4:

The vast majority of the most highly cited investigators did not exceed the GSI funding cap. Top 50 most highly cited investigators plotted vs. their level of NIH support. An annual GSI value of 7 corresponds to approximately one R01 grant, while annual GSI values of 14 and 21 correspond to two and three R01 grants. More than half had the equivalent of two R01s or fewer while only seven of 50 exceeded the GSI cap (red line). Source: Presentation by NIGMS Director Jon Lorsch to NIGMS Council, May 26, 2017 (Lorsch, 2017).

FIGURE 5:

The fraction of NIH awardees who are midcareer investigators is declining. Source: Figure 4, Charette et al., 2016.