The Guardian reported on Aug. 31, 2026, that the US Centers for Disease Control and Prevention is not including two Pennsylvania measles deaths in its national count “while CDC reviews additional information regarding the circumstances and causes of death.” The agency’s note added that “available information does not establish whether measles caused or contributed to the deaths or whether the individuals died from other causes while infected with measles.” Pennsylvania reported the deaths in August, including at least one newborn, according to state reporting to the CDC. Three explanations account for the available evidence; the asymmetry between Pennsylvania’s cases and three deaths reported from Texas and New Mexico last year is the most diagnostic observation.
The procedural explanation — that the CDC is applying ordinary scientific review to two deaths where measles’s causal role is uncertain — predicts that comparable review would have been applied to the 2025 Texas and New Mexico cases. Former CDC Chief Medical Officer Debra Houry told The Guardian that those three deaths were added to the federal count “without the kind of pre-publication review Pennsylvania’s cases are now receiving.” Under the practice Houry described, “states notify CDC of reportable cases on things such as measles. CDC does not review or adjudicate these cases unless a state asked for additional assistance.” If the exclusion reflects standard protocol, the Texas and New Mexico cases should have been held under the same review. They were not. The procedural explanation is the weakest fit on the available evidence.
The political-pressure explanation — that HHS Secretary Robert F. Kennedy Jr.’s public statements shaped the agency’s handling — predicts that the exclusion would follow Kennedy’s intervention, not precede it. Kennedy said in a recent Fox News interview that the Pennsylvania deaths were “fabricated” and that “in 10 years, there have been three measles deaths in America.” The sequence the article documents is consistent with this prediction: the CDC’s measles landing page was updated twice on Sunday, first listing the deaths, then replacing them with an asterisk alongside an explanatory note. The source documents no comparable senior-official contestation of the 2025 Texas and New Mexico deaths — an asymmetry that strengthens the political-pressure reading relative to the procedural one. Houry told The Guardian that “the communication coming from CDC’s official channel on Friday was political and not focused on what should be done during an outbreak,” and warned that the new review step “will undermine that relationship” between states and the agency. The political-pressure hypothesis is the strongest fit on the available evidence; it does not require inference about motive beyond documented statements and the documented sequencing.
The scientific-uncertainty explanation — that the cases genuinely lack the clinical information needed to attribute cause of death — predicts that the CDC would specify what additional information it is seeking. The Guardian reports that “the CDC has not clarified what information it is seeking about the Pennsylvania cases or why earlier data was not acceptable.” The clinical picture the parents describe to The Atlantic is consistent with peripartum measles exposure: the mother was severely ill with measles and went into labor one month early, and the newborn was born with a pulse but was not breathing and could not be resuscitated. The scientific-uncertainty hypothesis is unfalsified by the procedural record but weakened by the agency’s silence on what it is reviewing.
The discriminators that would shift the ranking: if the CDC publishes the clinical or epidemiological data it is reviewing within a stated timeline, weight shifts toward scientific-uncertainty; if no such specification is published, the asymmetry with the 2025 cases becomes the dominant signal. The Guardian reports no such specification and no published review timeline.
The article does not report which hypothesis the agency endorses. An HHS spokesperson told The Guardian on X that the agency is “working to determine what actually occurred.” No criterion for resolving the review has been published.
The load-bearing element most exposed is the reporting relationship between state health departments and the CDC. Houry’s warning that the new review step “will undermine that relationship” identifies the structural risk precisely. State epidemiologists notify the CDC of reportable conditions under a long-standing protocol; introducing case-by-case federal adjudication of deaths that states have already reported converts a notification system into a review pipeline, and the conversion has no announced criteria. The asymmetry between the 2025 Texas/New Mexico inclusion and the 2026 Pennsylvania exclusion is the rare event the system was not built to handle.
Concave exposures dominate the structure. The system absorbs small, frequent gains in data quality from ordinary notification; it loses disproportionately from rare events in which a federal agency holds reported deaths off the national tally without stated criteria. Vaccination uptake is a downstream concave exposure — the system appears robust under steady-state reporting but loses disproportionately from a single news cycle in which the national death count is publicly disputed. The Guardian reports that “the US has recently fallen below the threshold to keep measles outbreaks in check — a benchmark that would protect vulnerable infants, pregnant people and others at particular risk.” The herd-immunity threshold the article references — coverage commonly cited near 95% for measles — has fallen below the level the article describes. The structure the system sits in is one in which vulnerable infants, pregnant people and others at particular risk are most exposed to that fall.
A structural exposure the system underweights is pregnant people who cannot be vaccinated during an outbreak. The Lancaster county family’s account — the mother severely ill with measles, the newborn who could not be resuscitated — illustrates that exposure rather than carrying the structural claim: a mother who “was concerned about rising measles cases” but “was unable to get vaccinated because she was pregnant” sits inside a population the notification system was built to make visible, and whose visibility the current review step compresses. Kennedy’s 2021 praise, reported by The Guardian, of “the Amish of Lancaster county for not vaccinating their children” sits in the same exposure profile.
The agency has not gained from the review the way it would have from a published adjudication. No convex element is visible in the article — no component of the structure that benefits from the volatility the exclusion introduces.
Via-negativa recommendations: stop ad hoc federal adjudication of state-reported measles deaths; restore the prior notification protocol that Houry described; publish the criteria for any future case-level review or document that none will be conducted. Additions of robustness: pre-publication state-federal data reconciliation on a fixed timeline; a documented review standard applied symmetrically across jurisdictions, including to the 2025 Texas and New Mexico cases if the new protocol is to be applied retroactively.
If the breakage happens — a future outbreak in which the national measles death count is publicly contested while cases climb — the component that yields first is the state-federal notification compact. State health departments, seeing reported deaths held off the federal tally without stated criteria, stop treating the CDC count as the authoritative record. Local reporters and clinicians begin tracking state dashboards instead. The federal count becomes one dataset among several rather than the canonical one. Vaccine uptake, already below the herd-immunity threshold the article documents, drops further when the public cannot verify the case and death counts that justify vaccination.
The load condition that exceeds the structural property is a death report publicly disputed by a senior federal official before the federal review concluded. The Pennsylvania case shows the load: a reported death, a public statement from the HHS Secretary questioning whether the death occurred, and a federal agency revising its public count within days. The structural property the system relied on — that reported deaths appear in the federal count on the timeline states provide — does not survive that load. The Guardian’s parallel observation about the same Secretary’s response to a 2026 cyclospora outbreak that sickened more than 17,000 — where Kennedy characterized the outbreak as “under control” rather than engaging with the underlying mortality — adds a second documented data point on the same pattern.
Leading indicators that would have shown the structure approaching failure, all visible on Aug. 31, 2026: (1) a federal agency’s public landing page updated twice in one weekend with conflicting content, which The Guardian documents; (2) a former senior agency official publicly characterizing the agency’s communication as “political,” which Houry did; (3) the agency declining to specify what additional information it requires, which the article reports; (4) an HHS Secretary publicly labeling state-reported deaths as “fabricated” before the federal review concluded. The fourth indicator is the load itself; the first three are the strain it places on the reporting compact.
The breakage narrative reads identically for any agency in which senior political appointees publicly dispute the factual basis of state-reported mortality. The Pennsylvania measles case is the specific instance; the structural fragility is the general pattern. The Lancaster county family’s account — a pregnant mother unable to vaccinate, a newborn who could not be resuscitated, a father fielding a voicemail from the HHS Secretary — is the human record the system was built to capture and publish. Whether that record now appears in the federal count depends on a review whose criteria the article does not describe.
Analytical techniques used in this piece
This analysis applies the methods below. Each links to a short, plain-English explainer you can read and reuse.
- Differential Diagnosis
- Lists the candidate explanations for a symptom and rules them out one by one.
- Fragility / Antifragility Audit
- Asks whether a system gains or loses from volatility, shocks, and disorder (Taleb).
- Pre-Mortem (Fragility)
- Imagines a system has already broken and traces the structural fragilities that let it.