How many confirmed cases of cyclosporiasis will there be in the United States in 2026?

TL;DR

My median forecast is 14,600 confirmed U.S. cyclosporiasis cases in 2026, with an 80% interval of about 9,100 to 23,200 and a 19% chance above 20,000. The year is already a record-scale outbreak: CDC’s week-27 NNDSS table showed 3,713 U.S.-resident cases by July 11, 2026, and Michigan alone reported 5,002 outbreak-period cases by July 17, 2026 (CDC week 27 NNDSS, Michigan MDHHS). The July 17 lettuce recall cuts the future exposure tail, but reporting lag and the more than 5,100 CDC-pending cases keep the final count far above the recent 2,400-4,700 annual range (FDA recall, CDC surveillance).

Context

Cyclosporiasis is nationally notifiable, but the final resolving number will come from reconciled annual CDC/NNDSS data or a final CDC surveillance publication, not from the weekly figures available now. CDC says weekly NNDSS data are provisional and can change as states reconcile case notifications; annual data are the finalized series (CDC NNDSS data tables).

The 2026 season is not a normal seasonal rise. CDC’s disease-specific page, updated July 15, said that as of July 13 it had 1,645 confirmed domestically acquired cases, 34 states reporting cases, and more than 5,100 additional cases needing analysis to confirm whether they were domestically acquired; CDC also says its current surveillance updates do not include probable cases (CDC surveillance). FDA then linked a five-state Taco Bell cluster to shredded iceberg lettuce from Taylor Farms de Mexico, with 1,644 illnesses and last onset July 13, and Taylor Farms announced removal of iceberg lettuce sourced from central Mexico on July 17 (FDA outbreak investigation, FDA recall).

Evidence

The historical base rate is low compared with 2026, but the series is not perfectly comparable across time. The 1997-2008 row source is a CDC summary of laboratory-confirmed sporadic cases, not all outbreak-linked NNDSS cases; the 2011-2015 source reports confirmed plus probable cases, with 1,988 confirmed out of 2,207 total cases; 2024 and 2025 are still provisional year-end weekly NNDSS values, not finalized annual tables (CDC 1997-2008 MMWR, CDC 2011-2015 MMWR, CDC week 52 2025 NNDSS).

The live NNDSS signal is already far outside the recent history. The week ending July 11, 2026, showed 874 new current-week cases, 3,713 cumulative 2026 cases among U.S. residents excluding territories, and 1,473 for the comparable 2025 year-to-date column; the same table shows Michigan at 1,506 and Ohio at 437, versus 25 and 16 respectively in the 2025 comparison column (CDC week 27 NNDSS). By itself, the 3,713 week-27 count is already 94% of the provisional full-year 2025 value of 3,945, before the usual July-August reporting wave (CDC week 27 NNDSS, CDC week 52 2025 NNDSS).

State data show why the federal weekly table is behind the actual outbreak. Michigan reported 5,002 outbreak-period cases as of July 17, with 102 hospitalizations indicated as of July 16, and its notes say the count covers all cyclosporiasis cases reported during the outbreak investigation period (Michigan MDHHS). Indiana reported 327 cases since May 1 as of July 16 and explicitly labels them all reported cases, while West Virginia reported 139 outbreak-associated cases and 15 hospitalizations as of July 17 (Indiana Department of Health, West Virginia Department of Health). These are not one-for-one confirmed NNDSS cases, but they are strong evidence that the CDC’s 1,645 confirmed-domestic headline is a processed subset, not the final national count.

The source-control evidence lowers the upper tail but does not remove it. FDA reported that the Taco Bell cluster had 1,644 illnesses in Indiana, Kentucky, Michigan, Ohio, and West Virginia, with illness onsets from May 13 through July 13, and that a shredded iceberg lettuce sample supplied by Taylor Farms de Mexico tested positive for Cyclospora (FDA outbreak investigation). Taylor Farms’ recall announcement says shredded iceberg product was distributed June 29 through July 16 to 27 states, with listed best-if-used-by dates running into early August for some products (FDA recall).

I used three numerical models. A historical NNDSS multiplier model says the current week-27 count should finish well above 10,000 even with a lower-than-usual multiplier, because 2026 is already more front-loaded than normal. A state-backlog model starts from Michigan’s 5,002 reported cases and the CDC’s more than 5,100 cases needing analysis, then applies a confirmed-case share below one and a reporting-tail multiplier above one. A source-control model caps the tail because the main lettuce source was removed, but still adds late-July and August diagnoses from exposures before and shortly after the recall. I combined these as a scenario mixture: 15% low-containment centered near 9,000, 55% main centered near 14,000, 25% severe centered near 19,000, and 5% extreme centered near 28,000. The resulting median is 14,600, the mean is 15,700, the 10th percentile is 9,100, the 90th percentile is 23,200, and the probability above 20,000 is 19%.

What's non-obvious

The easy mistake is to treat 1,645 as the national year-to-date total. It is not. CDC labels that figure confirmed domestically acquired cases since May 1, while the week-27 NNDSS total already shows 3,713 U.S.-resident cases and Michigan’s state system had 5,002 reported outbreak-period cases by July 17 (CDC surveillance, CDC week 27 NNDSS, Michigan MDHHS).

The second trap is to treat the recall as an immediate stop in the case curve. It stops many new exposures, but people exposed before July 17 can develop symptoms later, seek testing later, and enter NNDSS later; the recalled product distribution window ran through July 16 and some listed use-by dates ran into early August (FDA recall). This is why my distribution puts most mass in the low-to-mid teens rather than around 5,000-8,000.

Limitations

The largest uncertainty is the confirmed-only endpoint. CDC’s current surveillance page says its 2026 updates report laboratory-confirmed cases and exclude probable cases, but NNDSS annual cyclosporiasis tables often present one case-count column rather than a confirmed/probable split; for 2011-2015, CDC’s surveillance summary showed 90.1% confirmed and 9.9% probable cases (CDC surveillance, CDC 2011-2015 MMWR). If the resolving source publishes a strict confirmed-only count, state reported totals will overstate the endpoint; if it uses the annual NNDSS row without a split, the realized value could sit above my strict confirmed median.

The second uncertainty is the state-to-federal conversion rate. Michigan and Indiana label their counts as all reported cases during the investigation period, while FDA warns that state data may include probable and confirmed cases and initial reports not yet sent to CDC (Michigan MDHHS, Indiana Department of Health, FDA outbreak investigation). The third uncertainty is outbreak scope: FDA says the investigation is ongoing and additional brands, restaurants, retailers, or distribution channels may be identified, while the recall and import screening make a long uncontrolled tail less likely (FDA outbreak investigation).