Reviewing Grants For Priority Alignment
It's not the kind of grant review you're thinking of, and I don't know if it's good or not
The purpose of this post is to try to bring a teeny bit of clarity to the roiling landscape of biomedical research, healthcare, and public health in the U.S. This isn’t about outrage or blame or anything like that. I’m just hoping to highlight some things that I think are important to know about in the present moment, offer my thoughts on why they’re important to know about, and offer some ideas on what people and institutions do with the information. I’ll focus on things related to the NIH and the Dept of Health and Human Services, though if there are goings-on in other sectors that are likely to have a big impact on biomedical research and public health, I’ll go there.
SUPER DUPER IMPORTANT CAVEAT: Though I am a former federal employee and will look at things through that lens, I am not currently a federal employee. The golden rule is always this: TALK TO YOUR POINT OF CONTACT – your program officer, contracting official, whomever is appropriate. Talk to them early and as often as it takes. And if you’re not sure who that is, press pause right now and go do some digging. When appropriate here, I’ll offer thoughts on who the right POC might be for a specific thing. But there’s no substitute for up-to-date information coming straight from the people tasked with implementing policies, EOs, random tweets, whatever.
This entry in the series is a bit different, as we’ll be talking about an internal NIH policy memo that I have not personally read and that you most likely have not read and cannot yet read, either (unless you’re NIH staff). The memo, titled “Reviewing Grants for Priority Alignment”, *has* been obtained by journalists, and the reporting on the memo has generated a fair bit of anxiety amongst investigators. So, in response to requests that I think and talk about this one, let’s get to it.
Here’s the high-level summary:
· The memo is 7 pages of guidance intended to be used by NIH Program Officers.
· The guidance is intended to help Program Officers work with investigators to negotiate/renegotiate grants to bring them into alignment with priorities expressed by Jayanta Bhattacharya and the Trump administration.
· The guidance makes mention of a “computational text analysis tool” that scans grant applications for what I call no-no words and phrases.
· The guidance applies to new AND EXISTING grants.
· Implementation of this guidance is likely to result in a lot of extra work for Program Officers and investigators.
The first critically important thing to understand is that the guidance is meant to be used and implemented BY NIH PROGRAM OFFICERS. This is not guidance for scientists serving on NIH review panels, study sections, advisory committees, or any other role. This is not guidance intended for implementation at the peer review stage. While I can imagine some kind of additional guidance for peer reviewers coming along at some point, this is not that.
Before we dive into what the implications of this might mean, let’s address the likely reasons for this guidance existing at all and why people are paying attention to it now. As far as I can tell, the important bit about this new guidance isn’t the “what” – e.g., it’s not describing a bunch of new no-no words and phrases. The important bit is that this guidance talks a lot more about the “how”.
One of the major legal vulnerabilities in the mass termination of NIH grants, contracts, and cooperative agreements that did not align with administration priorities earlier in 2025 was the lack of process. You may have heard the phrase “arbitrary and capricious” used to describe what happened. Apart from being a fairly accurate description of what happened, that phrase has a lot of legal heft. According to the Administrative Procedure Act (APA), Federal agencies are not permitted to act in a manner that is arbitrary and capricious. Agencies have to have a process that can be communicated to others that explains how the agency is doing stuff. There’s a lot more in the APA, but that’s the key bit. You wanna kill grants for specific reasons? You gotta have a process that can be described and evaluated. (The APA also talks about public commentary on proposed rules, policies, and processes, but that’s for later discussion.)
A major contributor to the successful legal challenges to the grant terminations earlier in the year was the higgledy-piggledy nature of how they happened. No process meant no defensible explanation for the terminations. So, over the last 4-6 months, the NIH has developed a variety of documents and statements – and now, a guidance memo – that collectively seem to represent something of a more describable process for how grants and such may be brought into alignment with administration priorities. My guess is that the intent is to backstop actions with process and thus resolve the vulnerability to legal challenge on the basis of an “arbitrary and capricious” argument.
Lest you think that this will at least bring shared understanding to how determinations about what flies and what doesn’t will be made, and how things determined to be not kosher will be resolved and negotiated, let’s do a quick thought experiment. Have you ever seen a policy, SOP, guidelines statement, or really anything in writing that was impossible to misunderstand or misinterpret? Me neither. It won’t be possible for this guidance to be uniformly understood and applied across NIH ICs with perfect consistency. Some of this is inevitable for any policy, and that’s actually a good thing. As we’ve said before, policies are born of intent and live or die in the implementation. Having some flexibility and variability in implementation allows for the nuances to emerge and to be managed. Overly rigid policies rarely achieve their intended effect.
The additional big challenge contributing to the variability for how this guidance will be applied has to do with the fact that it’s the PO who is placed at the tip of the spear. That could mean that individual PO differences could have outsized and unpredictable consequences. As a concrete example, I’ve had a friend ask if they should remove their health equity and structural racism work from their biosketch. The real answer is “I don’t know, and it’s kind of hard for anyone to know, especially since we haven’t actually seen the guidance.” Based on what we know, it is possible that a PO at one IC could use the text analysis tool (more on that in a moment), find those portions of the biosketch, and decide that though the investigator lists activities on their biosketch that would not align with current administration priorities, the work proposed in the grant application has nothing to do with those activities and so there is no issue. But a PO at a different IC could look at that same situation and determine that the investigator’s role needs to be changed, diminished, or eliminated to ensure that the grant funds will not be linked to any activities not aligned with current administration priorities.
There are, of course, things to do to help further clarify. The Golden Rule definitely applies here – TALK TO YOUR POINT OF CONTACT (in this case, the PO). The caveat is that the new guidance is going to mean a TON of extra work for POs (more on that in a moment as well), and POs are only human albeit truly wonderful humans in general. Individual ICs could establish implementation processes for the new guidance within their own IC that will help reduce confusion and variability, and I sincerely hope this is already happening.
Based on what I’ve read from journalists who have seen the guidance memo, there is discussion of a “computational text analysis tool” that is to be used by POs to scan grants for words and phrases that might indicate the presence of proposed activities that do not align with administration priorities. Based on some discussions with friends and former colleagues, it seems that this text analysis tool was developed in-house at NIH, perhaps at the National Institute for General Medical Sciences (NIGMS). It sounds like the tool might use AI-enabled approaches for text analysis, though that is only my best guess at this time. While it’s not unusual for there to be undisclosed details about each and every step of the funding decision-making process (indeed, some of those details may fall under one or more of the 9 exemptions from FOIA – see https://www.ecfr.gov/current/title-45/subtitle-A/subchapter-A/part-5/subpart-C if FOIA exemptions are a topic of interest to you), the possibility of using AI-enabled tools to inform decisions introduces the possibility of biases, limitations, and errors related to the training data for the AI algorithms, particularly if those algorithms make use of pretrained transformers (remember, that’s the PT part of ChatGPT).
The good news is that the guidance seems to require that the PO be the “human in the loop” when the text analysis tool finds no-no words or phrases. The bad news is that, as discussed above, there’s going to be variability in how POs respond to “hits” from the text analysis tool. The bigger bad news is that implementation of this guidance is going to create a TON of additional work for POs. More scrutiny of grants, more documentation of the results of that scrutiny, more discussions with investigators and institutions to respond to misalignments with administration priorities, more documentation justifying the actions taken by POs as a result of those discussions… if that sounds like a lot, it is.
And this is probably the right point to reiterate that all of this not only applies to new grants going forward, it applies to existing grants as well. So, if I’m understanding correctly, the kind of analysis and renegotiation being described in the guidance can/will be applied to currently active grants as well as newly submitted/newly funded ones.
What can we do?
1. Exhibit patience and grace with your Program Officers. They’re going to be doing the best that they can, and their jobs just got a whole bunch harder. Be patient. Plan ahead. Remember that they are your partner in your science, and however much you dislike these changes, they dislike them way, way more.
2. Expect inconsistencies. Saying, “But my PO at the National Institute of My Awesome Science said this should be fine!” is probably not going to get you the result you want, unless the result you want is to frustrate your PO.
3. Compare notes with as many colleagues as possible on an ongoing basis. This is another situation where no single investigator is going to have enough data to really figure out what’s happening at any given IC. But by comparing notes and experiences, it might be more possible to figure out what’s going on internally regarding how this guidance is being implemented.
4. Engage with IC leadership. Ask IC leadership team members how their IC is implementing this guidance. Share what you learn from them. This kind of engagement could happen at the level of individual investigators, single institutions, existing consortia, even professional societies engaging with the leadership teams of ICs.
Finally, it seems important to end by saying out loud that many of the things that do not align with current administration priorities – the scientific study of structural racism, health inequities, gender identities as distinct influences on health separable from sex as a biological variable, diversity of the scientific workforce, and many others – are in fact very much appropriate topics for specific attention and public funding through the NIH.
Further Reading:
https://www.statnews.com/2025/07/06/hhs-nih-grant-terminations-legal-guidance-internal-memo/
https://en.wikipedia.org/wiki/Administrative_Procedure_Act
https://www.ecfr.gov/current/title-45/subtitle-A/subchapter-A/part-5/subpart-C
