Frank Yiannas has been asking for a National Foodborne Outbreak Investigation Board, modeled on the National Transportation Safety Board, since he made the case in STAT last November. He came back to it in July in Food Safety News. I wrote about his idea in August, and again when De Ann Davis of Western Growerscalled it a strong idea if we follow through.
This time I do not want to argue the concept. I want to apply it. Three federal files from the past three weeks each ended, or stalled, with the central answer either missing or sitting somewhere the public was never pointed to. A standing board would have had to open a docket on each one and put something on the record. Here is what those dockets would hold today, and what they would still have to answer.
Docket one: romaine lettuce, FDA reference #1382
On September 10, FDA closed its investigation of a fourteen-person E. coli O157:H7 outbreak and named the food for the first time: romaine lettuce. The file had sat on FDA’s investigations table since July 1 with the food listed as not yet identified. CDC had declared the outbreak over on July 29, and by the time investigators identified the lettuce it was past its shelf life, as I wrote the day the file closed.
The closed row shows traceback initiated. The columns for on-site inspection and for sampling are empty. There was no recall and no advisory. No grower, processor or brand is named.
It is not the first romaine file to close that way. Reference #1280, the 2024 romaine E. coli O157:H7 outbreak, sits further down the same page: 89 sick, with traceback, inspection and sampling all initiated, and no firm named. Two romaine outbreaks, 103 people, and not one grower on the public record.
A board docket would have to answer four things. Where did traceback lead: to one farm, several, or none? Why was there no inspection or sampling on #1382? What is the probable cause, or if it is undetermined, what would have made it determinable? And what is the recommendation, to whom is it addressed, and has anyone answered it? FDA’s table page already promises that when a cause or contributing factors are identified that could inform future prevention, the agency will provide a summary. That promise has no clock on it.
Docket two: the 2026 Cyclospora season
CDC closed the iceberg lettuce outbreak on September 11 at 12,883 illnesses in 21 states, 570 hospitalizations and two deaths, linked to Taylor Farms de Mexico. Four days later CDC’s national case data page counted 19,883 laboratory-confirmed cases acquired in the United States with onsets between May 1 and August 31, against 1,180 for the same months of 2025, and said it would stop updating weekly. The two pages carry different dates, but the gap between them is 7,000 confirmed cases with no food named.
FDA’s table still lists four more Cyclospora files with the food “Not Yet Identified”: #1392 at 234 cases, #1385 at 38, #1398 at 22 and #1384 at 18. On the lettuce outbreak itself, CDC’s final update says FDA continues to investigate how the product became contaminated, and FDA’s table still marks that investigation active.
The docket questions follow directly. What made the other 7,000 people sick, and are those four open files one source or several? How did the parasite reach lettuce from a Guanajuato facility that FDA investigators assessed during the 2013 outbreak, and what became of the recommendation they wrote then, which I laid out in August? An NTSB-style recommendations database would have carried that 2013 recommendation as open for thirteen years. Nothing did.
Docket three: alfalfa sprout seed
FDA’s advisory on the Everything Sprouts outbreak counts 55 sick in 15 states and four hospitalized, infected with several strains of Shiga toxin-producing E. coli and with Salmonella Agona. On August 31 it said FDA had recommended that the common seed supplier recall the implicated lot, and that the supplier had not responded. On September 15 it said the supplier had agreed to recall. The seed went to growers and distributors in 16 states and Puerto Rico, and to Canada, Mexico, Tahiti and Venezuela. The advisory still does not name the supplier, the lot, or the additional grower FDA says used the same seed.
The name is public, just not on that page. FDA’s weekly Enforcement Report lists International Sprout Holdings, Inc., of Cookeville, Tennessee, recalling 43,799 pounds of alfalfa seed from lot SAL2-24JB, and puts the start of that recall at August 23, eight days before the advisory said the supplier had not responded. Phyllis Entis found it first, and I wrote about it on September 10; Food Safety News reported the same discrepancy. Two FDA records describe the same recall in ways that cannot both be complete.
FDA’s Enforcement Report data also lists six earlier Class I recalls by the same firm between 2014 and 2021: five of seed, for Salmonella, Listeria and possible E. coli, and one of mung bean sprouts for Listeria. A board docket would open with that history on page one. It would have to say how this lot became contaminated, why the recall record and the advisory disagree, which growers sprouted the seed, and whether anything recommended after the previous six recalls was ever done.
What the three have in common
None of this is a complaint about the people who work these cases. In each file investigators got somewhere: traceback reached romaine, traceback and genotyping reached a single processor in Mexico, and traceback reached one lot of seed. What is missing is everything downstream of that work: a required finding, an addressee for each recommendation, a clock, and a public record that stays open until somebody answers. That is the part Frank’s board would supply, and it is the part he tried to build from inside FDA with the Foodborne Outbreak Response Improvement Plan in 2021.
A board would not guarantee an answer in any of these. The NTSB sometimes concludes that a cause is undetermined. But it has to say so, explain why, and say what would have made the answer possible. “Undetermined, because no one inspected or sampled before the lettuce expired” is a finding. A blank column on a table is not.
There is one question I am not going to settle in a blog post. How a board’s findings relate to the claims of the people who were sickened is something any bill will have to answer, and the families should be at the table when it is written.
Until there is a board, somebody has to keep the docket. Every time a file closes without an answer, I intend to write down what a board would have had to put on the record. Consider this the first entry.