Yesterday the FDA told the press, and then its own advisory, that a sample of shredded iceberg lettuce supplied by Taylor Farms de Mexico had tested positive for Cyclospora. Today the agency says it was not a positive at all. Citing the complexity of detecting this parasite, FDA laboratory experts re-reviewed the result and concluded that what they were looking at “does not represent true amplification.” As of today, there are no confirmed positive product samples in this outbreak.

Taylor Farms will likely make something of this, and so will the trade associations that spent last week insisting the case against this lettuce rested on nothing more than patient interviews. So be precise about what changed today, which is very little, and about why nobody was ever going to find Cyclospora on the lettuce that made 1,644 people sick.

Start with the calendar. Illness onsets in this outbreak begin on May 13. The recall covers product distributed between June 29 and July 16. The sample FDA thought was positive was pulled at the border under targeted import surveillance, off product still crossing into the country this week, and Taylor Farms confirmed that lot was not part of the recall at all. Three different sets of lettuce. The leaves that sickened people in Indiana, Kentucky, Michigan, Ohio and West Virginia were eaten in May and June, and the incubation period, the diagnosis, the reporting and the interviews consumed the weeks after that. By the time anyone knew what to sample, the implicated lots had been eaten or thrown out. The victims ate the evidence. That is not a failure of diligence. It is what a seven-day incubation period does to a product with a shelf life measured in days.

What the outbreak rests on was complete before FDA collected a single sample. There are 1,644 laboratory-confirmed infections in people who reported eating at Taco Bell across five states, with 94 hospitalizations. Michigan pulled ingredient-level exposure detail from 190 of those cases and 90 percent reported eating iceberg lettuce. FDA’s traceback converged on one supplier, Taylor Farms de Mexico, and the company responded by pulling all of its central-Mexico iceberg from the American market, while Taco Bell removed the ingredient nationwide. None of that came from a test tube, and none of it moved today.

It could not have come from a test tube. Cyclospora cayetanensis cannot be grown. There is no culture system and no animal model, so nothing can be enriched before testing; detection in food is washing the produce and running PCR on whatever DNA comes out. Canadian government scientists who verified FDA’s method found leafy greens spiked with two hundred oocysts ninety-three percent of the time and greens spiked with ten oocysts thirty percent of the time, and contamination on a field-grown leaf is patchy rather than uniform. A clean sample is not exoneration; it is close to no information at all. The false positive is that same weakness running the other direction, and it is the more revealing half. Because the assay reads DNA and nothing else, there is no isolate to go back to and no plate to look at, so confirming a signal means re-reading the amplification curve and deciding whether you believe it. With Listeria or E. coli O157:H7 the organism is cultured from the patient and from the food, both genomes are sequenced, and PulseNet and GenomeTrakr set them side by side until the link is microbiology instead of inference. 

I know I have been beating this drum for a bit.

Which leaves exactly two things: what sick people remember eating, and whether the supply chain can be reconstructed backward from a restaurant to a field. That is the whole toolbox for this pathogen. And both halves of it are things this country has spent the last several years defunding and delaying.

Take the paper trial first. The Food Traceability Rule was finalized in November 2022 after the Center for Food Safety went to court to force FDA to issue it. Leafy greens sit on the Food Traceability List. The rule requires lot-level key data elements at each critical tracking event, and it requires that a company be able to hand the government an electronic record within twenty-four hours. Twenty-four hours. That is the difference between knowing on May 20 which fields fed which restaurants, and standing here on July 19 with FDA unable to say where the rest of the lettuce went. Because that is where we actually are. The company recall notice lists twenty-seven states of distribution against a five-state advisory, identifies its customers by brand codes rather than names, and omits the lot codes FDA’s own update says it contains. FDA’s advisory concedes the product may have reached additional states. CDC and FDA are both still describing the scope of this outbreak in terms of what the firm has chosen to disclose, five days after the supplier was named. That is not a mystery. That is a records problem, and Congress and FDA have between them made sure it stays one.

The original compliance date was January 20, 2026. In March 2025 FDA announced it would push that out thirty months, to July 20, 2028, and in November 2025 Congress made the delay binding in the continuing appropriations act. Industry asked for it. The grocery association called the rule unachievable given available technology; the grocers’ association called the timeline nearly impossible; a bill in the last Congress would have stripped the lot-code requirement from restaurants and retailers outright. The capability exists. What was delayed was the obligation to use it on behalf of the public rather than on behalf of inventory.

Now the other half, the counting, which is worse. On July 1, 2025, CDC cut FoodNet from eight pathogens to two, keeping Salmonella and Shiga toxin-producing E. coli and dropping Cyclospora, along with Campylobacter, Listeria, Shigella, Vibrio and Yersinia. The reason given was funding. One year later this parasite produced what is on any honest reading the largest foodborne outbreak in the United States in decades, and the surveillance system built to see it coming had been switched off twelve months earlier. It gets more specific than that. States do roughly half of the food processing facility inspections in this country, ninety percent of the produce safety inspections and essentially all of the retail inspections, and in December 2024 four senators had to write to FDA objecting to cuts in the state and local funding that pays for it. The DOGE federal layoffs of 2025 then hit FDA’s foods program and CDC directly. None of this was a plot. It was a series of budget decisions, taken under two administrations, each one defensible in isolation, that together dismantled the only apparatus capable of telling Americans how many of them were poisoned this summer.

And it is bipartisan. In 2012 the Obama administration zeroed out USDA’s Microbiological Data Program, the modest five-million-dollar-a-year effort that tested high-risk produce and reported positives straight to FDA, CDC and the states, and that had triggered twenty-three produce recalls in its last two years alone. Dan Flynn credited the produce industry’s lobbyists with the kill at the time. The people who wanted less produce testing got less produce testing.

Look at what that costs in this outbreak. The official count is 1,644 in five states. Cyclosporiasis has been reported this year in thirty-four states, with more than 9,000 confirmed or probable cases nationally against roughly twenty-seven hundred in all of 2025, and Michigan alone reported more than five thousand confirmed cases as of Friday with a hundred and two hospitalizations. I do not know how many of those people ate this lettuce. Neither does FDA, neither does CDC, and neither does Taylor Farms. Nobody built the machine that would know. A confirmed case requires a physician who suspects the parasite, a laboratory that runs the right test, a state health department with staff to interview the patient, and a federal system willing to count the result. Cut any one of those and the illness still happens; it simply stops appearing in the number that recalls and enforcement and, eventually, juries are built on.

Which brings me back to where this started. FDA collected the sample, FDA ran the assay, FDA announced the positive, FDA graded its own result, FDA reversed itself, and FDA then edited its own public record.

Which brings me to the case Frank Yiannas has been making. Frank ran food policy and response at FDA under two administrations, and before that spent a decade running food safety at Walmart and Disney. When he left the agency in 2023 he warned that its fragmented structure was undermining its ability to protect the public, and in November 2025, after a Listeria outbreak in prepackaged pasta meals killed six people and the original investigation failed to find a root cause, leaving the company to keep operating and keep making people sick for the better part of a year, he laid out the remedy: an independent National Foodborne Outbreak Investigation Board, modeled on the NTSB, investigating across every food category, outside the agency silos and outside the political pressure, publishing what it finds so the lessons are banked instead of lost. His argument is short and hard to answer. Foodborne outbreaks in this country are investigated by the very regulators charged with preventing them, responsibility is scattered across FDA, USDA, CDC and the states, and no one is finally accountable for any of it. He renewed the call this month, on this outbreak, arguing that we have been treating Cyclospora like hurricane season when it is a preventable problem of sourcing, agricultural water and controls in the field rather than at the shopper’s sink.

Set the two systems side by side, because Frank does and the comparison is devastating. When an airplane goes down, an independent board with no stake in the outcome takes the wreckage, publishes a probable cause and issues recommendations in public, and accidents per mile flown have fallen to almost nothing. When a bag of lettuce goes bad, the agency that regulates the lettuce investigates the lettuce, and by CDC’s own FoodNet data the incidence of Salmonella,E. coli and Listeria has been flat or rising for twenty years. GAO has found that FDA and USDA missed their own illness-reduction targets, one of them badly. An NTSB does not let the airline write the accident report.

The ask is not complicated, and it is cheaper than what we are paying now. Fund FDA’s human foods program and the state and local inspection workforce that does most of its actual work. Fund CDC and FDA and restore FoodNet to the eight pathogens it was designed to watch, this one included. Fund state epidemiologists, who are the people who found this outbreak in the first place, in Michigan, by interviewing sick people one at a time. Put the Food Traceability Rule into force instead of moving it again, because when microbiology cannot deliver a match, the records are the evidence and everyone in the supply chain already knows it. And build Frank’s board, so that the next time an outbreak sickens thousands of Americans, somebody whose budget and reputation are not on the line gets to say what happened and why.