Axios Vitals

March 03, 2026
Happy Tuesday! Today's newsletter is 1,047 words, a 4-minute read.
1 big thing: FDA slammed on rare disease decisions
The FDA's rejection of several new drugs for rare diseases has alarmed patient advocates and pharmaceutical companies, some of which are publicly accusing the agency of inconsistent standards and unrealistic regulatory requirements.
Why it matters: Experimental rare disease therapies already face complicated logistical and financial hurdles, even when the path for approval is predictable.
The big picture: Emerging new technologies have given new hope to the millions of Americans who live with rare conditions. But they're also forcing the FDA to grapple with dicey regulatory decisions, particularly when a drug is designed to treat very small patient populations.
- Critics say a series of recent rejections fly in the face of the Trump administration's stated commitment to fighting rare diseases and finding innovative new approaches.
- "Something has changed in the last few months. I think of it as a new, toxic ideology that has crept into the FDA, and it's having tragic consequences," said Tim Hunt, CEO of the Alliance for Regenerative Medicine, which advocates for cell and gene therapies.
The other side: "Absolutely not the case. There's no new mindset, there's no methodological purity," said a senior FDA official in an interview.
- "We are looking for the smallest possible spark," the official added. "When we don't grant approval, it's usually because we don't even see the spark, let alone the fire, let alone the blaze."
Driving the news: UniQure, the sponsor of an investigational gene therapy for Huntington's disease, yesterday disclosed that the FDA views its existing data as insufficient, and that the agency has suggested the company perform an additional clinical trial.
- "It seems the FDA is leaning into their desire to pursue gold standard science," Leerink analysts wrote in an investor note.
- But what the agency is calling for "seems to be in direct contradiction to applying flexibility in rare diseases," the analysts added.
- The senior FDA official gave a different explanation: "We have not seen any data that persuades us that there's even a little effect."
2. Hospital billing shift found to drive up costs
Hospitals are increasingly billing health plans for more complex care than they actually provide, according to a new analysis of Blue Cross Blue Shield claims from 2022 to 2025.
Why it matters: The so-called coding intensity swells health care spending, and it corresponds with an increase in hospital use of AI to help document patient visits.
- While automated coding can boost productivity, newly billed diagnoses still have to reflect what's actually happening with a patient, according to the analysis from Blue Health Intelligence, which is affiliated with BCBS.
What they found: The top 10% percent of hospitals in the study sample drove most of the increases that were detected, with almost 60% of inpatient admissions that could be coded as complex done so at those facilities by the end of March 2025.
Case in point: Coding intensity contributed to about $22 million additional spending in maternity care over the study period, the analysis found.
- Admissions for postpartum anemia following sudden blood loss grew more than 8 percentage points among hospitals with high coding intensity growth.
- The condition is often treated with a transfusion. But the analysis found there was virtually no change in transfusion claim rates in the hospitals with the largest increases in postpartum anemia claims.
3. HHS moves to cut off Anthropic AI use
HHS is cutting off its workers' access to Anthropic's AI tools at the direction of President Trump after the company rebuffed the Pentagon's demand to lift safeguards on the use of its model, Claude.
Why it matters: It's the latest step in one of the most aggressive federal actions against a major American AI company.
Driving the news: The department yesterday instructed workers to stop using Anthropic and switch to another approved platform.
- "HHS-approved enterprise offerings, including ChatGPT Enterprise and Google Gemini, remain available for authorized mission-related use in accordance with Department policy and federal information security requirements," a department spokesperson said.
The FDA last summer launched a generative AI tool dubbed "Elsa" that's aimed at helping its reviewers, investigators and other employees streamline their work.
- The technology was built in Amazon Web Services' GovCloud environment and aimed at reviewing clinical protocols and speeding up scientific evaluations, as well as identifying targets for inspections.
- An FDA employee told the news outlet Notus that access to Claude was still possible through Elsa but that a banner on an internal FDA website said Elsa would be transitioning away from Anthropic models.
Trump on Friday blacklisted Anthropic, calling it a "supply-chain risk" after company officials expressed concern about the use of AI for mass domestic surveillance and the development of weapons that fire without human involvement.
4. AI tool for predicting preterm births gets OK
The FDA yesterday greenlit a cloud-based AI tool that can predict preterm births entirely from standard ultrasound images.
Why it matters: The technology could flag high-risk pregnancies and improve case management in "maternity care deserts" with limited obstetrics care that are often associated with a higher risk of premature births.
Driving the news: The FDA authorized the Delivery Date AI technology, which is trained on millions of de-identified ultrasound images and is compatible with most existing ultrasound machines.
- It was shown to predict days to delivery in a study of more than 5,700 patients with the University of Kentucky and retrained several times to achieve higher sensitivity.
- Robert Bunn, founder of developer Ultrasound AI, said the way the package delivers predictions as early as eight weeks into a pregnancy, even for first-time mothers, lets clinicians "shift from last-minute reactions to proactive preparation and intervention."
What's ahead: Since preterm birth is linked to multiple factors, there's more work ahead to integrate AI predictions into clinical practice and show reliability across diverse populations.
5. Catch up quick
👀 The Trump administration intends to nominate a new CDC director in the coming weeks, marking its first effort to permanently fill the role since last summer. (WSJ)
🍼 Consumer Reports found heavy metals in more than half of the infant formulas it tested, despite the FDA's pledge last year to tighten oversight. (Axios)
⚕️ Nearly half of all new colorectal cancer cases occur in adults under 65, signaling a major shift in the demographics of the disease. (NYT)
Thanks for reading Axios Vitals, and to editors Adriel Bettelheim and David Nather and copy editor Matt Piper. Please ask your friends and colleagues to sign up.
Sign up for Axios Vitals









