The National Institutes of Health has canceled about $15 million in Alzheimer’s research grants at Emory University and the University of Pittsburgh, and every one of the canceled projects was studying why Black Americans develop the disease at higher rates than everyone else. Black Americans are twice as likely to develop Alzheimer’s or dementia. An estimated 7.4 million people in the United States are living with one or the other right now.
The NIH is the federal agency that pays for most medical research in this country, running a portfolio of about $47 billion. When a scientist gets an NIH grant, it usually funds several years of work and it only gets awarded after other scientists review the proposal and agree the science is sound. That review process is the whole point. It is supposed to keep politics out of which questions get asked.
Here is what the canceled Alzheimer’s research was actually looking at. At Emory, clinical neuropsychologist Negar Fani was leading a five year study on how racial discrimination changes brain activity. Her team planned to recruit 220 Black volunteers who would log discrimination incidents and daily stress on a phone app, then get MRI scans while recalling those experiences. The specific target was white matter, which is the tissue that carries signals between different parts of your brain. Damage to it is strongly linked to cognitive decline. At Pittsburgh, Andrea Rosso was studying how Alzheimer’s risk changes for people living in divested, predominantly African American neighborhoods. Her colleague Ann Cohen was studying socially driven risk factors using surveys, brain imaging and blood samples, the kind of factors that used to be written off as individual choices and are now understood to be largely outside any patient’s control.
NIH Director Jay Bhattacharya signed the letters ending them. The reasoning given was that the projects “no longer effectuate NIH’s priorities” and that they lacked concrete, objective and measurable variables. The Emory notice went further. It argued that measuring racial discrimination is subjective, because two people can interpret the same experience differently, and that recruiting only Black participants “seriously limits possible comparisons and controls and does not seem scientifically justified.”
That last line is where this stops making sense. There is a federal law on the books requiring that NIH funded Alzheimer’s research study health disparities, meaning the differences in who gets sick and who dies. Cohen has pointed out that killing these grants runs directly against that requirement. And studying the population with the highest rate of a disease is not a quirk of these projects, it is how medical research has always worked. Nobody calls a prostate cancer study unjustified for not enrolling women.There is also the money. Cohen’s team had already spent $5.2 million of an expected $9.6 million when the termination came. That money is gone and the answers it was supposed to buy are gone with it. Both Pittsburgh teams had spent years building relationships in the neighborhoods they were recruiting from, which anyone familiar with the history of medical research in Black communities understands is the hardest part of the job. Tuskegee is not ancient history to the people being asked to give blood samples and sit in an MRI machine. Once you tell a community the study is over, you do not get to knock on the same door twice.
This pattern showed up early and it showed up crudely. In March 2025, a study at the University of California, Davis was terminated with roughly $15 million left of a $53 million award. The project ran across 28 clinical sites with about 1,700 participants who had mild cognitive problems, looking at what certain lesions visible on MRI scans actually mean. The termination notice said it “did not effectuate NIH priorities.” The lead investigator, Charles DeCarli, has said his read was that the study got killed because the word “diverse” appeared in its title. He was given a month to appeal and had to inform all 28 sites that the work had stopped.
The $15 million figure circulating right now also badly undercounts what has happened. Senator Tammy Baldwin flagged $65 million in halted funding across 14 of the 35 federally designated Alzheimer’s Disease Research Centers, some of which maintain banks of donated brains that make this research possible at all. Across the NIH as a whole, hundreds of grants worth more than $1 billion have been canceled. Courts have already weighed in once. A federal judge found an earlier round of mass terminations covering more than 1,700 grants to be illegal and arbitrary, and many of those grants were restored. NIH is now moving through a process it hopes will hold up better if challenged.
Meanwhile the White House Office of Management and Budget is advancing a regulation that would give political appointees direct authority to cancel grants that already passed scientific peer review. Senate appropriators approved a temporary spending bill that would block that rule. That fight is unresolved. What gets lost in the accounting is the science that will not exist. Vascular risk factors like diabetes and high blood pressure contribute to somewhere between 15 and 25 percent of dementia cases, and there is still no FDA approved treatment that targets them. Those are conditions that hit Black Americans harder and earlier. Figuring out how they drive dementia is one of the more promising paths to actually preventing cases rather than managing them after the fact, and it is precisely the kind of work now being classified as a priority mismatch.
The people in these studies were not statistics to the researchers running them. Fani’s participants were logging their worst days on a phone. Rosso’s and Cohen’s were giving blood and sitting for scans in cities where medical institutions have not always earned the benefit of the doubt. They did that because somebody told them it might help figure out why their mothers and grandmothers were getting sick. The letters ending that work said the variables were not measurable enough.
