"What If NIH Had Been 40% Smaller?"

“What If NIH Had Been 40% Smaller?”

That’s the question posed (and answered) by an all-star team of academic metascience researchers: Pierre Azoulay, Matt Clancy, Danielle Li, and Bhaven Sampat. (PS: Sampat recently moved to Johns Hopkins University and joined the Good Science Project as a Senior Fellow).

That is, what if the NIH budget had been 40% lower in the past (as the Trump administration is proposing now)?

To answer that question, they start out with detailed data on:

• The peer review scores given to all NIH grant applications between 1980 and 2007

• 557 drugs approved by the FDA between 2000 and 2023, and

• The patents linked to those drugs.

They then ask some straightforward descriptive questions:

First, if the NIH’s budget had been cut by 40% during those decades (as the current Trump administration is proposing), which specific grants would likely have been cut due to their peer review scores? They call these “at-risk grants.”

Second, if those at-risk grants had indeed disappeared, what would the impact have been on later drug development? They answer that question by looking for drug patents that specifically cite an “at-risk grant,” or that cite academic research that is linked to an “at-risk grant.”

The Results . . . .

As for direct links: Out of 557 drugs, 40 had patents that directly pointed to NIH funding, and 14 cited an “at-risk grant” in particular.

But that’s a wild underestimate of NIH’s impact. As the authors point out, more than 90% of patents in the Orange Book (a central source of information on new drugs) don’t cite government funding, but “surveys of laboratory managers indicate that public-sector science informs more than 40% of industrial R&D projects.”

So, the authors look at more indirect links. There, it turns out that 331 of the 557 drugs have a patent that mentions a published article that cites NIH funding. And 286 of the drugs (over half) are specifically tied to “at-risk” research. Their top example of this: Gleevec, a leukemia drug that has been called a miracle drug.

The authors then question their own results: patent filings can cite many research articles, and there’s no guarantee that any one citation was the crucial one without which the drug wouldn’t have been developed. So they look at a higher threshold: what if 25% or more of a patent’s citations are for “at-risk” research specifically?

The answer: 65 out of 557 drugs have patent filings that meet that threshold. So, 12% of drugs seem to depend substantially on NIH grants that would have been cut under a 40% reduction.

Worth Noting

The paper here is an underestimate of NIH’s impact in many ways:

First, the Orange Book (the main source of information on drug patents that are connected to government funding) doesn’t have information on “vaccines, gene and cell therapies, and other biologic drugs; diagnostic technologies and medical devices; as well as innovations in medical procedures, patient care practices, and surgical techniques.”

That’s huge! NIH funding almost certainly played a role in the development of many of these therapies. It’s a shame that we can’t track any of that.

Second, the data on NIH peer review ends in 2007, and it is possible that grants beyond 2007 had an effect on drugs approved through 2023.

Third, for obvious reasons, the authors couldn’t trace “second-order connections, when an NIH-funded scientific discovery enables additional research that in turn leads to drug development.”

Yet another huge limitation. When the NIH funds basic science (about half of the time, by their telling), there can easily be 2, 3, 4, . . . 100 steps between an original insight and the ultimate drug development years later. Think of the first drugs approved to use CRISPR for treatment of sickle cell anemia. Do those drugs’ patents cite Francisco Mojica’s 1992 research that led to CRISPR? No. Patent filings won’t cite the entire history of all research that was critical to the development of a given idea.

Fourth, the study doesn’t track or take into account the possible impact of the “likely many” scientists who wouldn’t have been able to do a PhD or a post-doc if NIH funding were cut by 40%.

Bottom Line

This is an impressive piece of research, if only for the work it took to match up NIH grants to publications, and then both grants and publications to later patents. It’s both timely and important: it shows that if NIH funding had been cut by 40% in the past, we would likely miss out on many important drugs and treatments. And it’s an underestimate in many ways: funding for basic biomedical science almost certainly has impacts far beyond what we can track through the usual sources.

A side note, but an important one: The NIH’s detailed data on peer review scores and unfunded grants is hardly ever available to anyone. And even in the rare cases where the NIH lets an outside researcher have access, NIH retains the right to veto publications (I know several academic researchers who have found this incredibly frustrating and have even had to cancel articles, although they all want to be anonymous for fear of annoying NIH). NIH should be much more willing to make this data available, without restriction, to metascience researchers who want to do studies like this.