Michigan's July 28 surveillance update put the state's Cyclospora count at 9,680 cases — roughly 400 more than a day earlier — while the Centers for Disease Control and Prevention's parallel ledger documented 6,707 laboratory-confirmed cases nationally as of the same date, with about 11,500 more unconfirmed. [1][2][3]

Neither number is wrong, and neither number can touch the other. The state figure is a surveillance-wide count built on Michigan's definitions and reporting clocks. The federal figure is a confirmed-plus-unconfirmed structure built on different definitions and a different clock. [3] Adding them, subtracting them, or dividing one by the other would produce arithmetic, not epidemiology. The gap between the ledgers — flagged by the July 27 account that refused to merge Michigan's count with the federal categories, when the state figure stood at 9,253 — is widening, not closing.

Local television reports the state number straight, as it should: 9,680 is the count Michigan's health department published, and its movement is the fastest signal the public has. [1][2] But movement between state updates is not onset incidence. Of the roughly 400 cases added in a day, some fraction is new illness, and some is report delay, laboratory confirmation, backfill, or reclassification working through the pipeline. The update does not decompose its own movement, so nobody outside the department knows the fractions. [1]

The federal ledger's shape tells a different story about the same parasite. CDC's 6,707 confirmed cases are the subset that cleared laboratory confirmation nationally; the roughly 11,500 unconfirmed sit beside them in a category Michigan's headline count does not use. [3] A reader of the state ledger believes the outbreak is larger than a reader of the federal confirmed column does, and both are reading the same underlying reality through incompatible definitions. That definitional gap — not either total — is the measurable fact of this edition.

Two limits bind the numbers. The hospitalization figure circulating late in the week — 193 — carries a July 31 date, after this edition's cutoff; the last in-scope figure remains 160, and the update is excluded here. [1] And the evidence labels stay attached: "surveillance" describes Michigan's count, "confirmed" and "unconfirmed" describe CDC's columns, and none of the three converts into the others. [3]

The open questions are the ones the ledgers cannot yet answer. What do the Taylor Farms traceback and the Taco Bell exposure records establish or exclude as of July 29? When will Michigan and CDC publish compatible definitions, dates, and confirmation-status fields so the public can reconcile the two counts without a translator? [3] And beneath the definitional dispute sits the staffing question this paper has tracked through the FoodNet cuts: surveillance capacity decides what surveillance can know, and both ledgers are downstream of it.

On X, the outbreak was unobserved. The only pre-July-29 hit was a stale July 14 Fox News status, too old to carry a community frame. [1] The discourse this story deserves — patients, restaurant workers, produce suppliers — is happening off the indexed platforms, or not at all.

Michigan will update again. CDC will update again. Until the definitions meet, each update will widen the same gap this account refuses to close.

-- NORA WHITFIELD, Chicago