Science1 distinct publisher3 min readPublished
Case definitions are boring by design, so that a measles death recorded in Pennsylvania means the same thing as one recorded in Texas. Two former CDC officials argue that the shared rulebook is what is now in dispute.
The Scientist · Science desk
Compiled by The ScientistSomething wrong?How this is made
Surveillance definitions are written slowly, by committee, with epidemiologists and agency reviewers moving them through the Council of State and Territorial Epidemiologists and the CDC [2]. That procedural sludge is the feature. A case definition is a measuring instrument, not a verdict on any one person's death, and its value comes from every jurisdiction pointing the same instrument at the same thing. A rule can be imprecise and still be useful, because the imprecision is shared and this year's count can therefore be set beside last year's. Reopen the rule during an outbreak and the comparison degrades even if the replacement rule is defensible on its own terms. That is the practical argument underneath the authors' warning that the standard process is now under threat [3], and underneath their claim that politically interpretable definitions cost you the ability to trust what is reported locally and nationally [4].
The Texas episode in their account shows where the determination actually sits. State officials investigated the cases, made the calls, and notified CDC, which then carried the deaths in national reporting [6]. What flowed the other way was support: money, vaccines, and staff who helped count cases, work through infection control, and advise on air handling [7].
On the epidemiology, those two deaths sit inside the expected range rather than outside it. At 1 to 3 deaths per 1,000 children infected [8], two deaths is roughly what you would expect among about 667 to 2,000 infected children, which is simply two divided by 0.003 and by 0.001 [16]. The thing that arithmetic cannot tell you is whether any individual child died of measles. That question is settled one file at a time, by a state investigation applying a written definition, which is precisely the machinery in contention.
The same op-ed describes pressure applied one layer down, at clinical guidance. HHS asked the authors to revise CDC's clinician information to include inhaled steroids and antibiotics for children with measles [12]; they declined, on the grounds that neither is a recommended routine treatment and neither has gold-standard evidence behind it [13], and they were fielding reports of vitamin A toxicity at the time [14]. Children's Health Defense, meanwhile, campaigned to pin the Texas deaths on the treating hospital rather than the virus [9], and the health secretary told a national television audience, falsely, that MMR kills more children than measles does and that its protection wanes [10].
The limits of this piece are worth stating plainly. It is an opinion column by two people who worked the measles response from inside CDC [15], and it lacks a figure for how many Pennsylvania deaths are disputed, a date for the federal objection, and evidence that any CSTE case definition has in fact been rewritten [17]. The mechanism they describe is credible, but a documented change to the instrument is a separate matter, and one that is not yet on the record.
Ranked by verification strength, evidence, and original report placement.
States have a standard process for notifying CDC about notifiable diseases using established case definitions; developing those definitions is usually a long, deliberative process involving epidemiologists, scientists, and agency review through the Council of State and Territorial Epidemiologists and the CDC.
Working off a common definition and process is what lets you know what the numbers mean and trust what is reported locally and nationally; if data or definitions are subject to political interpretation, that trust and the nation's ability to respond effectively are quickly lost.
When outbreaks occur, states may ask CDC for additional assistance, including highly specialized laboratory testing, help with contact tracing and data management, consultation on infection-control guidance, and countermeasures such as vaccines; on-the-ground help can come through an Epi-Aid supported by CDC's Epidemic Intelligence Service.
When Texas reported the nation's first measles death in a decade, followed quickly by the second, Texas followed the standard protocol: state officials investigated the cases, made the determinations, and notified CDC, which then included the deaths in its national reporting.
CDC provided Texas with financial support, vaccines, and personnel on the ground to assist with counting cases, infection control protocols, and advice on air handling systems.
On average, 1 to 3 in 1,000 children with measles may succumb to the infection and its complications.
Distinct publishers with included, body-backed reporting in this cluster.
1 article · September 4, 2026
Follow any of these and your For You feed starts watching them — no settings page required.
science
Resigned CDC officials urge Congress to put scientific review beyond political reach1 distinct publisher
leadership
Summer's 17,000 cyclosporiasis cases move produce verification onto the buyer's books1 distinct publisher
product
Enhanced Games' $62M quarter puts a price on buying legitimacy1 distinct publisher
science
An FDA panel wants unapproved peptides sold by prescription. That is deregulation, not wellness.1 distinct publisher
Evidence-backed comparisons of source perspectives and observed adoption signals. Read the methodology
Which Builder, Operator, and Investor concerns the observed source mix emphasized—not a truth score.
Evidence, demonstrated adoption, hype gap, incentives, and confidence are assessed independently, each on its own current evidence. How these are measured.
One column, two interested witnesses
Everything here rests on a single STAT opinion piece. Its strongest passages are firsthand: the authors were inside the measles response when Texas reported the first death in a decade, and their account of that chain of custody is specific and unchallenged. Its central passage is not — Pennsylvania happened after they resigned, and the withheld dashboard entry, the politicized post, and the internal correction memos arrive without a date, a document, or a word from CDC, HHS, or the state.
Nothing here to count
This story has no uptake to measure. The one thing that would function as adoption evidence — whether the national dashboard shows Pennsylvania's two deaths, and when — is asserted rather than shown, and no other state's reporting behavior is documented.
Framing runs ahead of the paper trail
The headline conflict is over a rulebook, but nobody in this account rewrites a rule: the case definitions are intact and the dispute, as described, is a delay and a social media post. STAT's authors also float the sharpest reading available — that the only difference from Texas was a blue governor — while conceding in the same breath that it is unclear what was different. That gap between the inference offered and the record produced is real, though modest, and the clinical claims about steroids, antibiotics, and vitamin A are stated with appropriate restraint.
Resigned in protest, writing about the people they left
The authors disclose the stake plainly: they quit CDC last summer over political interference and are now indicting the secretary, a named sitting director, and Children's Health Defense. That disclosure is to their credit and it does not neutralize the pull — a column vindicating the reason you resigned is not a neutral instrument, it ran on an opinion page rather than through a newsroom, and the officials it accuses were given no visible chance to answer.
Believable, unverified
The process description is the kind of thing that would be easy to falsify and nobody has, and the clinical statements match standard practice. But the specific allegation a reader will remember — a federal director sitting on a state's confirmed deaths — currently stands on one participant account, and until Pennsylvania, CDC, or the dashboard record speaks, confidence in it should stay well short of the confidence you can place in the mechanics around it.