Since 2005, a doctor in Brazil who suspects Creutzfeldt-Jakob disease has a legal duty to say so. The suspicion becomes a form, the form becomes a record in SINAN, the national notifiable diseases system, and the record joins a database built mainly to catch outbreaks: dengue, measles, snakebite, things that arrive in clusters. Prion disease does not arrive in clusters. It arrives one family at a time, usually with months of confusion first, and the question that Breno Barbosa, Maria Luiza Montenegro and Jose Eriton Gomes da Cunha put to that database in Frontiers in Neuroscience on 29 January 2026 is disarmingly simple: over 17 years, how many people made it onto the form at all?

The answer is 1,576. That is every suspected CJD notification recorded between 2005 and 2021, drawn from the Ministry of Health's own epidemiological bulletin published in 2022. This is a retrospective, descriptive study of existing public records, not a case-finding exercise, and the distinction matters: the authors went looking for the paperwork, not for patients. Of those 1,576 suspected cases, 547 (34.7 percent) were confirmed, most of them (65.6 percent) on laboratory criteria and 29.4 percent on clinical and epidemiological grounds, with the basis unstated in 4.9 percent. Set against the widely used figure of one to two cases per million people each year, the authors calculate that roughly 3,200 cases would have been expected in Brazil across that period. Notifications ran at about half the expected caseload; confirmations at about a sixth.

Where the notifications came from is as telling as how many there were. Sao Paulo accounted for 32.5 percent, Parana 12 percent and Minas Gerais 10 percent, with the Southeast, South and Northeast dominating. Prions are not choosier about Brazilian states than they are about anything else; what varies is who can be worked up. The paper is blunt about the bottlenecks: imaging and neuropathology are difficult to obtain in many units, initiatives to make RT-QuIC available in the country remain modest, and some diagnosed patients have had samples shipped abroad for biomarker analysis. An earlier Brazilian report on second-generation RT-QuIC, published in 2020, gives the shape of the problem in one line: of 603 suspected cases notified up to June 2018, 404 were left undiagnosed. Ninety percent of notified cases were urban dwellers, which says less about where prion disease occurs than about where a neurologist, an MRI scanner and a lumbar puncture can be found in the same building.

Now turn the telescope around. In the American Journal of Alzheimer's Disease and Other Dementias on 25 October 2025, a team at the Manuel Velasco Suarez National Institute of Neurology and Neurosurgery in Mexico City reported a cross-sectional study of their own emergency department. Among 19,003 patients seen across 2023 and 2024, 15 met criteria for probable CJD: an incidence of 0.55 per 1,000 emergency visits in 2023 and 1.01 per 1,000 in 2024. The apparent doubling was not statistically significant (relative risk 1.83, 95 percent confidence interval 0.63 to 5.35). Psychosis featured in 68 percent of cases and MRI was abnormal in all of them, most often with cortical ribboning. A companion conference abstract published in December 2025 counted 16 cases including one seen as an outpatient, and noted that only three of the patients came originally from Mexico City. The authors are careful to say their rate cannot be laid alongside population figures of roughly one to 1.5 per million per year, and they list their own constraint plainly: no neuropathological confirmation, and RT-QuIC and 14-3-3 testing that public institutions do not cover, leaving families to pay out of pocket.

Put the two papers side by side and the subject is not biology. It is the machinery of counting. Brazil has the legal duty and thin diagnostic reach, so cases fall out of the chain before they ever reach a laboratory. Mexico City has one referral hospital with the expertise to recognise rapidly progressive dementia, so cases pile up there at rates that look startling until you remember that a national institute is where the country's hardest neurological problems go to be solved. Both patterns are consistent with the same underlying incidence. Neither can prove it.

For a family, this is not an abstraction about registries. A confirmed diagnosis is what ends the round of misdiagnoses, what lets people stop searching and start planning, what makes genetic counselling possible where a familial form is in question, and what puts a name on a death certificate that will later be used to decide whether the state funds a reference laboratory. When a case is never confirmed, that family carries the uncertainty and the national count carries the gap. The Brazilian authors' recommendation is structural rather than clinical: a purpose-built CJD database instead of an outbreak system pressed into service, and consistency between official documents that currently do not agree with each other.