Frank Yiannas has been making an argument for the better part of a year that deserves more attention than it has received. In a STAT First Opinion piece last November, written after a Listeria outbreak in prepackaged pasta meals killed six people, he called for an independent National Foodborne Outbreak Investigation Board. He brought it back in July in Food Safety News, in a piece asking whether Cyclospora is a seasonal nuisance or a preventable problem. His diagnosis is short and hard to argue with. Foodborne outbreaks in this country are investigated by the same agencies that are supposed to prevent them. The work is split across FDA, USDA, CDC and fifty state health departments. And when an investigation goes nowhere, no single agency is accountable for the fact that it went nowhere.
I have known Frank since his Disney days, through his ten years at Walmart and through his time at FDA, and I guest lecture in his food safety culture class at Michigan State every year. I want his board built. I want to argue with one part of how he sells it.
The model he reaches for is the National Transportation Safety Board, and the comparison he draws is a real one: aviation accidents per mile flown have collapsed over forty years while foodborne illness rates have not moved at all, and GAO found that FDA and USDA missed their own illness reduction goals, one of them by a wide margin. It is a good comparison. It is also where most readers stop, because within a few seconds anybody who works in this industry starts listing the ways a lettuce field is not a runway. There is no wreckage. There is no flight recorder. There is no single moment. Roughly half of all multistate investigations never name a food at all.
Set the airplanes aside. The better question is smaller, and you can answer it out of one file. What happens to a finding after somebody writes it down?
In August 2013, a team of FDA investigators spent eight days inside a Taylor Farms de Mexico processing facility at Doctor Mora, in Guanajuato, working alongside COFEPRIS, SENASICA, CESAVEG, Guanajuato state health officials and CDC. They were there because 631 people in twenty-five states had come down with cyclosporiasis. They covered the plant and five ranches. They collected roughly 835 samples. Nothing came back positive, though one recycled wash water sample could not be read at all, because the water was too turbid for the filters and the method simply failed. FDA has since said so in its own method literature and rebuilt the test because of it.
What they did find, they found by reading the company’s own paperwork. A ranch entered the assessment only because FDA reviewed Taylor Farms de Mexico’s irrigation water records and saw a fecal coliform result off drip irrigation tape. It had triggered no corrective action, because it sat just under the firm’s own single sample action level. The investigators tabulated prescriptions written for harvest crews and found a drug effective against Cyclospora dispensed during the outbreak window. Then they wrote a recommendation. The second one in the report was that the firm determine whether Cyclospora is a reasonably likely hazard in the Guanajuato growing region, and if it is, re-evaluate the wash step.
That is careful work. It was done in a foreign country in eight days by people who had every reason to come home with less. When the plant reopened that August with FDA’s concurrence, the company committed to a sampling program covering both product and water.
And then nothing happened to any of it. The recommendation had no addressee. Nothing set a clock on it. No database tracked whether it was ever answered, and no public record shows whether the sampling program ran for a month or a decade or at all. FDA was back inside that plant once in the thirteen years that followed, in 2019, and because that visit found nothing objectionable it produced no public document whatsoever. Nobody was in a position to ask the 2013 question again, because nobody had been assigned.
2026. Same region. Same facility. Same parasite. Same commodity. Thirteen years later FDA and CDC now attribute 9,481 illnesses, 398 hospitalizations and two deaths across seventeen states to iceberg lettuce out of that plant, and my own fifty state floor for cyclosporiasis this season stands at 29,912 people. FDA did not report an onsite inspection at Doctor Mora until August 13, twenty-seven days after the recall.
That is the argument for Frank’s board, and there is not an airplane anywhere in it.
One piece of the airplane comparison is worth keeping after all, and it is the clock. A Go Team is in the air within hours, because everybody involved understands that a scene degrades. Taylor Farms recalled this lettuce on July 17. FDA did not report an onsite inspection and sampling at the plant in Guanajuato until August 13, twenty-seven days later, and its announcement does not say when that inspection began or whether anyone is there now. Taylor Farms says it suspended all production at that facility on July 18 and brought in outside experts to review it, which means that whatever an inspection turns up now, it is turning it up in a plant that has been shut for nearly a month. The 2013 visit was slower still. The last illness onset in that outbreak was July 2 and the team arrived on August 12, and FDA’s own explanation for why almost every sample came back clean was that more than five weeks had gone by. Nobody would defend a six week wait to walk a crash site. And food is the harder case here, not the easier one. A fuselage sits in a field until somebody photographs it. Shredded iceberg lettuce has about two weeks of shelf life, which means that by the time anyone arrives, the people who got sick first have already eaten the evidence.
That delay was not an investigator’s choice either. An American inspector cannot simply walk into a plant in Guanajuato. The 2013 assessment was carried out jointly with COFEPRIS, SENASICA, CESAVEG and Guanajuato state health officials, which is to say it had to be negotiated between two governments while people were still getting sick, and the same thing had to be negotiated again this summer. Standing authority, with access arranged in advance instead of assembled in the middle of an emergency, is exactly the sort of thing a country builds once and then uses every time. We have never built it.
It is also not a complaint about the people who investigate outbreaks, and I want to be clear about that, because this argument gets misused. The men and women at FDA’s CORE network, the epidemiologists at CDC, and the state and county investigators who make the phone calls are not the problem here. The 2013 assessment is the proof. They went, they looked, they read the records nobody made them read, and they wrote down the right question. What does not exist in food safety is anything sitting downstream of that work. The investigating already happens. The follow through was never built.
Nor did any of them choose the conditions they are presently working under. No epidemiologist voted to switch off federal surveillance reporting for six of the eight pathogens FoodNet was built to track, Cyclospora among them, ten months before the first illness in this outbreak. Nobody in a state health department pushed the food traceability rule out to 2028, or replaced pre-harvest agricultural water testing with an annual self-assessment, or arranged for five of more than 168,000 Taylor Farms shipments from Mexico to be physically examined at the border in the first half of this year. Nobody working a case file decided that roughly eleven billion dollars in public health grants, money that pays for disease surveillance, laboratory services and outbreak investigation in every state, should be terminated without notice in the spring of last year and then spend the next year tied up in federal court. A health department cannot staff a summer against a grant that may or may not exist. Every one of those decisions was made well above the people now being asked to explain a thirty thousand case outbreak with what is left.
Some of it was taken apart long before this administration or the last one, and the clearest example belongs to a program almost nobody remembers. In April 2001 the Agriculture Department launched the Microbiological Data Program, which pulled roughly fifteen thousand samples of fresh fruit and vegetables a year out of distribution centers in about eleven states, tested them for pathogens, and uploaded what it found to PulseNet. It was the only real produce surveillance this country has ever had. By one analysis it accounted for more than eighty percent of all public produce testing, and it triggered twenty-three recalls in 2010 and 2011 alone, fifteen of them involving people who were already sick. Among the things it sampled were bagged lettuce and cilantro. It cost four and a half million dollars a year. The produce industry lobbied against it for years, on the theory that a program which finds contamination causes recalls. It was zeroed out of the budget, the states were told to stop pulling samples on November 9, 2012, and it shut down at the end of that December, seven months before FDA investigators walked into the plant at Doctor Mora. I have been asking for it back for years.
The Microbiological Data Program would not have caught this parasite. It tested for bacteria. That is very nearly the point. We dismantled the produce surveillance we had, and we never built any for Cyclospora at all. The Agriculture Department research complex at Beltsville, whose laboratories do a meaningful share of the government’s Cyclospora work and coordinate it across agencies, is now being closed and its programs scattered. There is still no regulatory limit for this organism in agricultural water or on produce, and no validated kill step. FDA said in that same 2013 report that the sanitizers this industry relies on do not reliably kill it. A grower in flawless compliance with every rule on the books is doing nothing whatsoever about Cyclospora.
Frank knows this from the inside better than anyone arguing about it from the outside. He ran the internal version of this fix himself. The Foodborne Outbreak Response Improvement Plan he released as Deputy Commissioner in December 2021 was aimed at exactly this, at traceback, at root cause investigation, at getting outbreak data out the door faster. He had the title and the authority and he did the work, and his conclusion since leaving is that fragmentation cannot be repaired by a plan written by one of the fragments. That is a serious person telling you the inside fix was not enough.
Here is what I would want the statute to carry, and it is mostly machinery rather than investigation. What actually distinguishes the NTSB is not the team that flies to the crash site. It is the ninety-day clock that starts when a recommendation is issued, the permanent public database that tracks every recommendation until somebody answers it, and the annual list that names the agency that has not. Recommendations are not binding. They are simply impossible to lose. A board without those three things becomes the Chemical Safety Board, an agency with the same design and no leverage that has spent years fighting for its own budget while its recommendations pile up unanswered.
Three more things, and then I will get out of the way. There has to be a mandatory launch trigger, a case count or a death count or a duration, so that declining to investigate or hiding the data is not something an administration can accomplish by saying nothing. There has to be a way to reach the root cause when the root cause is in Guanajuato, because a board whose authority stops at the border cannot answer most of the questions worth asking. And whatever findings the board produces have to stay available to the families who were hurt. Aviation solved its version of that problem by making board reports inadmissible in civil cases and by keeping lawyers out of the investigation entirely. Copy it into food and you will produce the best root cause analysis this country has ever done on an outbreak, with the processor helping to produce it, and seal it away from the only forum where the people who got sick have ever gotten an answer.
Two things deserve saying on the other side of the ledger. FDA and CDC are aligned on the numbers in this outbreak right now, publishing the same figures on the same day, which has not been true all summer and is worth noticing when it is. And FDA now reports an onsite inspection and sampling at the plant, in coordination with Mexican officials.
I have written about Frank’s board twice before. I will keep writing about it. If somebody would rather argue with the idea than draft it, I would point out that a board of this kind takes regulatory authority away from precisely nobody, which is more than can be said for every single food agency reorganization proposed since the 1990s, all of which are dead.
One last thing, and it is not in any statute. Whatever this board turns out to be, build in an hour where the people doing the investigating sit down with the people who got sick. Not testimony. Not a hearing. I have watched that hour change how a person understands their own job for thirty-three years, and it would change how the report reads.
