The 2026 cyclospora outbreak, explained

What is cyclospora?

Cyclospora cayetanensis is a microscopic parasite spread when food or water contaminated with faeces is swallowed. It does not pass directly from person to person, which is why outbreaks trace back to a product rather than spreading through a community. Symptoms usually begin about a week after exposure and are dominated by watery diarrhoea that can be prolonged and can relapse after apparently clearing; untreated, an infection can last from a few days to a month or more. It is treated with antibiotics.

The foods involved are almost always fresh produce eaten raw and hard to wash clean — past US outbreaks have been tied to imported raspberries, basil, cilantro, snow peas and, repeatedly, leafy greens. The parasite needs days to weeks in the environment after being shed before it becomes infectious, so contamination generally happens at a grower or processor, well before the food reaches a kitchen.

How big is the outbreak?

The CDC's outbreak count stood at 10,930 cases across 17 states as of August 20, 2026, with at least 454 hospitalisations and two deaths. It had been 9,481 in the same seventeen states on August 13, 6,358 in fifteen states on August 5, 1,947 in nine states for the twelve days before that, and 1,644 in five states in mid-July. Those steps have different causes, and the distinction matters. The jump on August 5 was not a jump in illness — the agency widened the definition of who belongs to the outbreak, which is explained below. The rises since are something else: the CDC attributes them to state and local health officials continuing to interview sick people and adding them under the definition already in force. No new state has been added since Maine and Massachusetts on August 13. Both deaths are in Michigan, in people with significant underlying health conditions, and both fell ill before the July 17 recall. For scale, the two largest multistate cyclosporiasis outbreaks the CDC documented in 2018 produced 250 and 511 laboratory-confirmed cases — 761 between them. This outbreak is more than fourteen times their combined total.

The outbreak figure is also only part of the season. Nationally, the CDC has recorded 17,180 laboratory-confirmed domestically acquired cases since May 1 — with 922 hospitalisations and cases in 48 states plus the District of Columbia — and says it is aware of a further 11,844 cases that are not laboratory-confirmed or that require further investigation, many of them reported by Michigan. That second figure has gone back up: this page carried 10,455 on August 13, so unconfirmed reports are now arriving faster than laboratories are clearing them. It is a queue, and its size measures the backlog rather than the number of extra people who are sick. A separate 1,805 confirmed cases were acquired abroad and are counted apart from all of the above. In a normal year, US cyclosporiasis runs at a fraction of this volume, which is why the agency moved to weekly surveillance updates for 2026.

Every one of these counts is a floor. Cyclosporiasis is diagnosed by a specific stool test that is not part of a routine workup, so infections are recorded only when a clinician thinks to order it, and mild cases that clear on their own are rarely captured at all.

Why do Michigan's numbers look so different from the CDC's?

Because they count different things. Michigan's health department reported 14,277 cyclosporiasis cases statewide as of August 20 — more than the CDC assigns to the entire multistate outbreak, but well short of the CDC's national count of 17,180 laboratory-confirmed domestic cases — and the gap is a matter of denominators, not disagreement. Michigan publishes every case reported in the state across the outbreak investigation period, whatever the suspected source; the CDC's outbreak count includes only laboratory-confirmed infections it has tied to the recalled Taylor Farms de Mexico lettuce, and the CDC notes that states may also be counting illnesses that do not meet its own reporting criteria. As the CDC itself puts it: "States may report higher case counts than CDC because they may include cases that do not meet CDC's formal reporting criteria."

The general rule, which The Fold learned by getting this wrong before correcting it: when a state and a federal count disagree, check whether they are counting the same population before calling it a disagreement. Michigan's own table now lists 73 reporting jurisdictions — 72 counties plus the city of Detroit, which the state counts separately — with the heaviest totals in the south-east: Wayne at 2,095 cases, Oakland at 1,538, Washtenaw at 1,105 and Ingham at 921.

What is the evidence against the lettuce?

Epidemiology, traceback and now genotyping — but still no laboratory match to the food. The FDA's traceback investigation converged on Taylor Farms de Mexico as the common supplier of shredded iceberg lettuce to the restaurants where sick people had eaten. On July 18 the agency announced a positive cyclospora test on a lettuce sample; on July 19 it withdrew the finding as a false positive that did "not represent true amplification." No confirmed positive product sample has been reported since. What has changed is on the human side: on August 20 the CDC said genotyping performed on samples from sick people shows that the parasites which caused these illnesses are genetically related, which it presents as additional evidence that they got sick from the same source. That ties the cases to one another. It does not tie them to the lettuce.

That is less unusual than it sounds. Foodborne outbreaks are frequently solved without recovering the pathogen from a sample, because suspect product is usually eaten or thrown away long before investigators reach it. And the case pattern points at the food independently: of the 14,271 Michigan cases classified by age, just 94 were in children under ten — consistent with a food adults eat routinely and small children largely do not, which is about the profile of salad greens.

Does the recall explain the outbreak?

More of it than it looked a week ago — but what changed was the definition, not the illness. Until August 5 a person counted as part of this outbreak only if they reported eating at Taco Bell. On that day the CDC widened it to anyone who reported eating at Taco Bell or eating the recalled Taylor Farms de Mexico iceberg lettuce, and the outbreak went from 1,947 cases in nine states to 6,358 in fifteen. The FDA, announcing the change, noted that many of the newly added illnesses began before the July 17 recall. Nobody became ill on August 5; the boundary moved. What has happened since is different: between August 5 and August 20 the count went from 6,358 in fifteen states to 10,930 in seventeen — Maine and Massachusetts joined on August 13 and no state has been added since — and the CDC attributes that to continued interviewing rather than another change of boundary.

Resist dividing one of these numbers by the other. The two figures that invite it — 10,930 in the outbreak and 17,180 confirmed nationally — sit on different pages with different cut-offs. The national page now carries data as of August 24; the outbreak count was last revised on August 20 and takes onsets through August 11. The national figure is currently the fresher of the two, which is the reverse of the position a fortnight ago. The Fold has published that ratio twice, at about 47 per cent on July 26 and about 29 per cent on July 31, and the second was low mainly because the outbreak count had not been revised since July 24 while the national count kept climbing. Both were computed on what each page said at the time; neither should be read as a stable measure of the recall's share.

What can be said without dividing anything: the outbreak has grown on every measure the CDC publishes for it, thousands of confirmed cases this season are still not assigned to it, and the CDC and FDA say they are investigating at least six other cyclospora clusters whose sources have not been confirmed. The FDA says it can take as long as six weeks to establish whether a sick person belongs to a cyclospora outbreak, so the assignment will go on changing.

What has been recalled?

On July 17, Taylor Farms de Mexico voluntarily removed all iceberg lettuce sourced from central Mexico from the US market. The recall notice reached past restaurants into retail, covering Marketside-brand product sold at Walmart — moving the exposure from a foodservice problem, where a diner has no way of knowing which supplier fed a given restaurant, to a retail one, where consumers could check a bag in their own refrigerator. That window has now closed: the CDC says the best-by dates for the recalled products have passed and they should no longer be available in stores or restaurants.