Immunization Registry Protocol Mismatch, Franklin County, Pennsylvania , Field Report, October 14, 2026

Franklin County, Pennsylvania is currently reporting 1,136 measles cases since January, with 157 new cases in the last seven days. Six people have died. Two of them were infants, below the age of vaccination eligibility. The others were old enough to have been vaccinated and were not. The outbreak is still expanding.

The first signal was not the case count. It was the registry.

The Franklin County immunization registry is running a decision protocol written for a coverage environment where MMR uptake holds above 95%. This is not that environment. County-level MMR coverage has fallen below the threshold required for population-level protection, a condition the registry's protocol does not recognize because the protocol was not written for this version of the county. The registry is logging cases against a baseline that assumes most people received the shot. The result is a system that reads as managing an outbreak it is actually failing to see.

The servers are lying, not through deception but through category error. The protocol queries the population and returns figures consistent with high-coverage assumptions. Those figures are accurate for a different operational environment. Cross-referencing the registry's response architecture against current coverage data reveals two query responses timestamped from conditions that have not existed in this county for at least three years. The system is not broken in the sense of producing errors. It is broken in the sense of producing confidence.

In four of the six nearest accessible versions of this region, the MMR coverage rate in Franklin County held. The outbreak presented similarly in those versions, the same virus, the same initial case clusters, the same age distribution in early reports. Containment followed within weeks because the surrounding population was immune enough to interrupt transmission. In this version, the virus is finding susceptible individuals because they are present in sufficient density to sustain spread. The registry's protocol, imported from a coverage environment where that density did not exist, has no handling instruction for the condition it is currently running inside.

This is a documentation problem with fatal tolerances.

The registry's threshold logic was set during a period of higher uptake. At some point, coverage declined. The protocol was not recalibrated to reflect the new baseline. The gap between what the protocol expects and what the population provides has been widening for years. A case count that should have triggered escalation to outbreak-level response was instead processed as a manageable cluster, because the protocol's definition of manageable was written for a population that no longer exists in these numbers.

The system is not reporting failure. It is reporting a condition it was designed to handle. The condition it was designed to handle is not the condition it is in.

In Timeline 7, the equivalent registry flagged the coverage decline in 2023 and recalibrated thresholds accordingly. The outbreak that emerged the following spring was contained in six weeks. The difference was not the virus, not the initial case count, not the public health response capacity. The difference was whether the registry knew it was running in a low-coverage environment before the outbreak began rather than after the sixth death.

This version of the registry does not know what it is running in.

The two infants who died had not yet reached vaccination age. Their safety depended on the population around them being immune enough to prevent exposure. That population was not immune enough. The registry did not flag the insufficiency because its protocol was not calibrated to this coverage rate. The infants were never entered into the registry's risk model as exposed. They were simply not there, in the way that a system will report no errors if it is not checking the right variables.

The protocol mismatch is not a discrete fault that can be patched in isolation. The registry's escalation thresholds, its reporting cadences, its case-clustering logic, and its outbreak-declaration triggers are all downstream of the same coverage assumption. Recalibrating the protocol means establishing what the actual coverage baseline is, county by county and age cohort by age cohort, and writing threshold logic that reflects that baseline rather than a historical one. It means teaching the system that two datasets with the same label can describe different populations depending on when the underlying survey was conducted. A registry that cannot distinguish between a 95% coverage environment and a 78% coverage environment is not a registry. It is a ledger.

In two versions of this county that register as immediately adjacent to the current one, the protocol recalibration happened before 2024. The coverage rates are similar. The difference is whether the system running surveillance on those rates was calibrated to see them clearly. One small policy decision, a budget line that either funded the registry update or didn't, a staff position that either stayed filled or went vacant, and the outbreak in those versions resolved differently. Not because the public health infrastructure was dramatically superior. Because it knew what it was looking at.

The case count here is 1,136. The trend line is not flattening.

The vaccine is available. It was available in every accessible version of this situation. The supply chain is not the constraint. The registry is not the constraint in the sense of limiting access. The constraint is the gap between who has been vaccinated and who needs to be, and the additional gap between what the registry believes about that first number and what is actually true. Both gaps are measurable. Both are currently being measured inaccurately.

A field note, logged from the position of the report: "The system reads normal. That is the finding. , G., Franklin County perimeter, Timeline 3"

Advisory: the vaccine exists. Check vaccination records. If the record system advising you is the same one described above, verify the record directly with a provider, not through the registry portal. End of bulletin.