Annie Waldman at ProPublica published a piece this morning that everyone who regulates, sells or drinks raw milk should read. It is the story of Emily Marris, her 17-month-old daughter Brooklyn, and a 6-year-old girl named Victoria Galindo. Both girls drank raw goat milk from Moonlight Ridge, an unlicensed farm in Riverside County, California, owned by Courtney Baker, in the spring of 2025. Marris found the farm through a Facebook group run by her chiropractor, Chris Boman. Both ended up in intensive care at Rady Children’s Hospital in San Diego with hemolytic uremic syndrome (HUS). Brooklyn’s stool culture confirmed E. coliO157:H7.

I have represented children with HUS since 1993. The details in Waldman’s reporting are ones I have heard at far too many kitchen tables: vomiting that turns bilious, then dialysis lines, a ventilator, seizures, a heart struggling against blood gone acidic, and parents who do not know whether their child will live through the night. Brooklyn survived. She came home with a limp, night terrors and a nephrologist she will see every six months, likely for the rest of her life. Victoria also needs ongoing monitoring.

What makes this story different is what the State of California did, and did not do, with the evidence it had.

What did Courtney Baker know about who was drinking her milk?

Everything. Baker has since said it was all “pet milk,” but ProPublica describes a paper trail showing she knew from the first day that a baby was drinking it. When Marris picked up her first unlabeled jug in January 2025, still warm, she pointed to Brooklyn asleep in her car seat and told Baker about the toddler’s digestive problems. Once home, Marris texted that Brooklyn was gulping it from a bottle. Baker wrote back that she was glad, and later that day asked, “How’s her tummy handling the milk?” The next morning Marris reported back on the baby’s diapers.

Marris and her husband later visited the farm to see the herd and asked about safety. Baker, they recall, told them repeatedly that the herd and milk were “clean and tested.” For the next six months Baker sold milk to a mother she knew was putting it in her toddler’s bottle.

The Galindos’ story is the same. Baker had once tutored their son in math and was advertising raw goat milk in a local Facebook group. When they picked up their first jug on June 2, 2025, the same day Brooklyn first vomited, the Galindos say they told Baker it was for their six-year-old daughter. And when CDFA’s undercover investigator asked to buy a gallon, nobody at the farm said anything about pets.

Then came what Baker did after she knew children were sick. On June 5, Marris texted from the hospital that doctors suspected the milk. Baker answered that her tests had shown nothing abnormal, then conceded in the same exchange that she had not tested for E. coli and that the tests she could find seemed to be for meat. On June 14, after a second child was hospitalized, Marris asked her to hold off selling. Baker did not reply. The next day she answered a buyer on Facebook offering raw goat milk from a “clean tested herd.” When Marris publicly warned the group, Baker asked her to take the post down. According to the families, Baker never told the hospital or either family that another child had fallen ill.

The first time anyone heard the words “pet food” was June 20, when regulators showed up at the gate.

What did California regulators have?

According to ProPublica, investigators with the California Department of Food and Agriculture (CDFA) had all of that: two cooperating families, hospital records, Marris’ screenshots of months of text messages, and the undercover investigator’s exchange with the farm.

Selling raw milk for human consumption in California without a license can bring fines of up to $10,000 and up to a year in jail. A CDFA regional supervisor wrote to her team, “She clearly knew the milk was going to a child.”

What did California actually do?

On June 20, 2025, regulators handed Baker a cease-and-desist order and a notice of violation for operating a food facility without a permit. Neither carried a penalty. The next day they accepted a half-gallon jug of milk that Baker herself said she had collected the night before. Because California law does not allow investigators onto unlicensed property, they did not collect milk themselves or sample manure, soil or water, as they have done at licensed dairies.

That sample came back negative. Regulators then helped Baker get back into business, selling the same raw goat milk, still without a license, inspection or pathogen testing, as long as it was labeled and registered as pet food. CDFA closed its case on August 1, 2025, finding she had sold unapproved milk without a license, and imposed no penalty. No referral went to the Riverside County District Attorney. Baker has since moved her operation to Arizona.

Why is a negative test weeks later meaningless?

Because raw milk lasts days, not weeks. The milk that sickened Brooklyn and Victoria was long gone before anyone collected a sample, and E. coli O157:H7 is shed intermittently by animals that look perfectly healthy. A single half-gallon, gathered unsupervised by the person under investigation, after she had been told her milk was suspected in two HUS cases, tells you nothing about the milk those girls drank. I made the same point about Dungeness Valley Creamery this month: in three of that dairy’s four outbreaks, Washington never found the bacteria in the milk. Absence of a positive sample is not evidence of safety. Two unrelated children with the same rare strain after drinking the same farm’s milk is evidence.

What is the “pet food” loophole?

In California, an unlicensed farm can sell raw milk if it is labeled for animals. According to ProPublica, 13 other states allow raw milk to be sold only as pet food, even though everyone, including the regulators, knows people drink it. Outbreaks in Louisiana this year and Florida last year sickened children, at least one of whom developed HUS. No dairy was held responsible because the milk was “legally sold and clearly labeled.”

Baker told Waldman that everything she sold was pet milk and that what customers did with it was not her business. A label does not change what goes into a toddler’s bottle. When a seller knows the buyer is a mother feeding her baby, the label is not a loophole; it is a fig leaf.

What role did chiropractor Chris Boman play?

Chris Boman is a Murrieta chiropractor who specializes in pediatric and prenatal care. He is also one of three administrators of a private Facebook group of backyard farmers, and for years, Marris told ProPublica, he sold raw milk from his own cow out of his clinic’s waiting room. In January 2025, Boman posted a photo in the group of a refrigerator full of fresh raw milk. Marris, a patient who was struggling to wean her premature “miracle baby,” commented asking whether it would be good for Brooklyn. Group members urged raw goat milk instead and pointed her to Moonlight Ridge. That is how a 17-month-old ended up drinking Baker’s milk.

After Brooklyn and Victoria were hospitalized, Marris posted a warning in Boman’s group when she saw someone else advertising raw milk. She was suspended. When she complained, Boman replied that he was not a mediator, that people could read her story and decide for themselves whether to assume the risk, and asked whether she wanted anything else from the group.

Boman told Waldman that Marris was “drumming up a bunch of fear.” He said he had talked with Baker, was satisfied her standards were high and pointed to her negative tests, the same unsupervised sample gathered weeks after the illnesses. He also connected Baker with the Farm-to-Consumer Legal Defense Fund, a nonprofit that assists raw dairy sellers, through his own membership. The fund says it is not representing Baker in the lawsuit. Boman still believes in raw milk; he told Waldman he stopped selling it only because his cow went dry.

Think about that. A health care provider whose practice is built around pregnant women and children, selling raw milk in his waiting room, running a group where mothers are steered to unlicensed dairies, and then silencing the one mother whose child nearly died. Boman is not a defendant in the case, but his part in this story deserves as much attention as the farmer’s.

What should change?

First, when there is evidence a seller knew “pet” milk was going to people, and here the evidence was in writing, regulators should refer the case for prosecution. CDFA told ProPublica that a cease-and-desist gets product off the market faster than a criminal referral. Fine. Do both. Nothing prevents regulators from stopping sales today and sending the file to the district attorney tomorrow.

Second, California and every state with a pet-milk exception should require the same pathogen testing for “pet” raw milk that licensed dairies face, and should give investigators authority to sample animals and the environment on any farm linked to human illness, licensed or not. A lab result from a jar the suspect hands over at the gate should never close a case.

Third, the label should say what the risk is. I am about to file a citizen petition with FDA asking for a warning label on raw milk cheese. States should require one on raw milk, whatever species it is sold for.

Where does the case stand now?

I filed suit on behalf of the Marris and Galindo families against Courtney Baker and Moonlight Ridge in Riverside County Superior Court, alleging Baker negligently sold them contaminated raw milk presented as fit for human consumption. Baker has said the farm will fully defend the case. When the state walks away, the civil courts are what is left.

Is raw milk worth the risk for a child?

No. Raw milk has no proven nutritional advantage over pasteurized milk and carries far higher risk of foodborne illness; a Johns Hopkins analysis put it at more than 100 times. This year alone we have seen Raw Farm, Graziers raw milk cheese, and Campylobacter outbreaks in Idaho. Most of the children with HUS are five and under.

Marris now spends her evenings in Facebook groups like Boman’s, telling her story to other mothers. Sometimes she gets attacked for it; sometimes her posts get deleted. She is doing the job the State of California declined to do. Read Waldman’s full story. Then drink pasteurized milk.

In the last twenty-four hours, two people have called my office. Both have culture-confirmed E. coli. Both have been told they are part of a multistate E. coli outbreak linked to a food that has not been named. When each of them went back through what they ate before they got sick, the same item came up: a sandwich wrap, meat and vegetables, bought at the same store. The same chain, that is. The two of them live in different states.

Two phone calls are not an epidemiologic study, and I am not naming the store. I do not have what the agencies have. The two epidemiologists on my staff are working on the link between the two callers now. They have ordered the state and local health department records on both, along with records from CDC, FDA and FSIS, and there will be more to say once those come in. What I have today is two interviews that line up with a file I wrote about last week, and they line up well.

What the three federal pages say today

FDA’s CORE table, stamped September 25, carries reference #1421: E. coli O157:H7, 25 sick, product not yet identified, traceback initiated, no inspection, no sampling. The row carries USDA’s reference number, 2026-04, and it is the only active row on FDA’s table with one. FSIS’s outbreak table lists 2026-04 as a September investigation, product Unknown, with a note that FSIS and FDA are investigating it together. It is the only active investigation FSIS has posted all year. CDC’s weekly page, last updated September 23, counts six active multistate E. coli investigations and names none of them. No states, no onset dates, no hospitalizations, no ages. I went through all three pages this morning.

What being told you are part of an outbreak means

When a health department tells someone, they are part of a multistate outbreak, it is not the patient’s guess, and it is not the health department’s either. The patient’s stool culture grew E. coli. A public health laboratory sequenced the bacteria’s genome and uploaded the result to PulseNet, CDC’s national database of DNA fingerprints. The sequence matched the cluster. Then came the questionnaire: everything eaten in the week before illness, every store, every restaurant. My two callers answered that questionnaire in different states, and gave the same answer, before either of them talked to me.

CDC’s own timing rule matters here. Its outbreak pages say it usually takes three to four weeks to determine whether a sick person is part of an outbreak. A count posted on September 23 is made of people who got sick weeks earlier. The people whose sequences are matching this week are the next update.

Why a wrap explains the two reference numbers

Last week I laid out how FDA and USDA divide the food supply: meat and poultry to FSIS, everything else to FDA, a percentage formula for foods that contain both, and sandwiches as the standing joke. The joke is more specific than most people know. FSIS’s own directive says closed-face sandwiches, a meat or poultry filling between two slices of bread, a biscuit or a bun, are FDA’s, a decision the agencies trace to the 1930s. Wraps are different. When wrap makers argued that a wrap is close enough to a closed-face sandwich to be FDA’s too, FSIS disagreed: wraps were new, the old sandwich decision did not cover them, and a wrap with meat is a meat product, like a burrito. The two agencies put that in the Federal Register in November 2005, alongside a proposal, then under consideration, to move closed-face sandwiches to FSIS. The move never happened. Twenty-one years later the sandwich is still FDA’s, and FSIS’s stated position on a wrap with meat is that it is a meat product.

Now put a meat-and-vegetable wrap on the traceback desk. The turkey or roast beef or chicken inside it came from an FSIS-inspected plant. The lettuce, the tomato, the onion and the tortilla are FDA’s. If the wrap was assembled in the store’s own kitchen, the store is a retail operation under state or county inspection, and the traceback runs to its suppliers on both sides of the line. If it came in sealed from a manufacturer, that plant belongs to one agency or the other, but its ingredients still came from both. Either way, one sick person’s wrap produces two traceback questions, one to a meat supplier and one to a produce supplier, and neither agency can close its file until its half is ruled out. That is a file with two numbers on it. It is what #1421 looks like on paper.

The last time both agencies had a sandwich file

The last joint file before this one, #1305, carried USDA 2025-06, and FDA’s own closing summary of it is titled Deli Sandwich (suspect). That was the Fresh & Ready Foods Listeria outbreak: ready-to-eat sandwiches and snack items, turkey and cheese, ham and cheese, beef cheesesteak, sold at convenience stores, hospitals, hotels, airports and airlines in Arizona, California, Nevada and Washington. Ten people sick, all ten hospitalized, one dead. Here is the part worth remembering. FDA and CDC investigated that cluster in 2024 and could not identify a source; the specimens dated from December 2023 to September 2024. The answer came in April 2025, when FDA inspectors on a routine visit found Listeria in the San Fernando plant and the sequence matched the outbreak strain. The public heard the company’s name in May 2025, and even then, the summary says the sandwich was never confirmed as the vehicle; it remained the suspect. A joint file is no guarantee of an answer, and a routine inspection, not the traceback, is what found the strain in the last one.

What an announcement without a food looks like

The agencies do not need a brand name to say something. On October 28, 2020, FDA announced two E. coliO157:H7 outbreaks, 23 sick in one and 21 sick with one death in the other and said in so many words that it did not know what food was making people sick or whether that food was even FDA-regulated. It called the announcement early communication, and it promised a new web resource for early updates on active investigations. The clusters were posted as Unknown Source 1, 2 and 3, with the states listed. The table FDA keeps today grew out of that promise. It posts the count and withholds the states.

What I am asking for is what I asked for last week, plus one item. Say which states. Say the onset dates. Say whether the suspect food is a mixed product, and which ingredient each agency is tracing. And if a sandwich wrap is on the traceback desk, say so now, because a wrap sold this afternoon will take three to four weeks to show up in PulseNet. In October 2024, FDA, FSIS and CDC announced the McDonald’s Quarter Pounder outbreak while the onions and the patties were still being sorted out, and McDonald’s pulled both the same day. The public was told what was in play. That is the standard, and it was met two years ago.

To the people who are calling

The two people who called did the right things first. They went to a doctor, they were tested, and they went back through their food history item by item. If you have been told you are part of this outbreak, do the same, and keep the receipt, the loyalty-card record and any packaging that is left; write down the date and the store. Interviews are how outbreaks get solved. The percentages that put McDonald’s onions and patties side by side in 2024 came from them.

I have represented people sickened by E. coli since 1993. My hamburger cases went away because FSIS declared O157:H7 an adulterant in ground beef in September 1994; since about 2000, most of my E. coli cases have been lettuce. A meat-and-vegetable wrap is both halves of that history in one package. Last week I asked which side of the jurisdictional line this outbreak was on. Two phone calls suggest the answer is both, and that the reason no food has been named may be that the food is a sandwich. Twenty-five was the count on September 23. It is a floor. The agencies know what they are tracing. The rest of us are finding out one phone call at a time, and my epidemiologists and I will report what the records show.

Earlier today I wrote about the Mezzetta peperoncini recall, prompted by a consumer’s video of what appears to be a dead mouse in her jar. It made me wonder how often an animal in food produces an official recall, one on an FDA, USDA or state record rather than only in a viral post. The answer, looking back two decades: not often, and when it does, the details tend to surface weeks later in FDA’s Enforcement Report rather than in any notice a shopper would see.

The rule is simple. Under 21 U.S.C. § 342(a)(3), a food is adulterated if it consists in whole or in part of any filthy, putrid or decomposed substance; under § 342(a)(4), it is adulterated if it was prepared, packed or held under insanitary conditions whereby it may have become contaminated with filth. FDA’s Defect Levels Handbook tolerates a few insect fragments or rodent hairs in some commodities. There is no allowance for a carcass, and 21 CFR 117.35(c) says pests must not be allowed in any area of a food plant. Here are the recalls that met that bar, newest first, with what set each off.

The list

1. September 26, 2026 — Mezzetta Golden Greek Peperoncini, Medium Heat, 32 oz. (mouse). One lot, SKU 10106328, Lot 720106, recalled ten days after a consumer posted TikTok videos of what she described as a dead baby mouse in her jar. FDA posted the company’s announcement under “pest contaminant.” It does not say “mouse,” where the jars went or how many there were. Not yet classified.

2. January 16, 2026 — Dr. Praeger’s “Simply Nature” Spinach Bites sold at Aldi (rodent hair). Dr. Praeger’s Sensible Foods recalled 7,894 boxes distributed in Maryland and Pennsylvania because the product had the potential for contamination with rodent hair. FDA classified it Class II on March 16, 2026. No press release was issued; the recall sat in the Enforcement Report two months until Fox Business reported it on March 18. How the hair was detected was never stated.

3. December 26, 2025 — Gold Star Distribution, Minneapolis (rodent and bird contamination in a warehouse). After an inspection, FDA determined the facility was operating under insanitary conditions, with rodent excreta, urine and bird droppings where food, pet food, drugs, devices and cosmetics were stored. The recall ran to 44 pages and nearly 2,000 products shipped to dozens of Minnesota stores plus one each in Fargo and Indianapolis. Classified Class II on January 22, 2026. Prompted by FDA, not a complaint, and not a first offense: FDA had warned the same distributor in October 2018 after its lab confirmed mouse excreta and rodent hairs in the food storage areas.

4. September 30, 2024 — PCC Community Markets Hungarian Mushroom Soup, Seattle (insect larvae). Puget Consumers Co-op recalled 105 tubs of soup made by Ladle & Leaf, sold in its stores and hot bars September 26 to 30. FDA Recall F-0056-2025, Class II: “possible presence of larvae.” PCC’s own notice to members said only that the soup “may contain foreign material.”

5. May 10, 2024 — Lundberg Family Farms Sustainable Wild Blend Gourmet Rice (object “of rodent origin”). Wehah Farms of Richvale, California recalled 4,600 cases, 27,600 one-pound bags, because the rice “may contain a foreign object that appears to be of rodent origin.” Sold in seven states from Maine to Oregon. FDA classified it Class II on July 17, 2024, two months after the recall began. What part of the rodent, and how it was found, was never disclosed.

6. March 27, 2024 — Wegmans frozen cut green beans, packed by Nortera Foods (object “of rodent origin”). Wegmans’ own notice said “pest contamination.” FDA Recall F-1484-2024, Class II, in the Enforcement Report, said the bags may contain a foreign object of rodent origin. Distribution was limited to New York and Pennsylvania. The rodent was in the database, not the store notice.

7. June to August 2023 — G.O. Corporation romaine and lettuce blends, Minneapolis (deer feces). FDA Recall F-1013-2023, Class II, June 16, 2023: 11,500 pounds of romaine and salad blends sold in four Upper Midwest states “may be contaminated with deer feces.” A second round followed on August 1, and on August 7 USDA’s FSIS issued a public health alert for Taher, Inc. salads and wraps built on the recalled lettuce. 

8. July 10, 2023 — Trader Joe’s Unexpected Broccoli Cheddar Soup (insects). Winter Gardens Quality Foods of New Oxford, Pennsylvania recalled 10,889 cases sold in seven states because, in FDA’s words, the product “has insects in the frozen broccoli florets.” Trader Joe’s said its supplier alerted it. FDA classified it Class III.

9. February 18, 2022 — Family Dollar distribution center, West Memphis, Arkansas (rodent infestation). A consumer complaint sent FDA inspectors into the warehouse in January 2022, where they found live and dead rodents, feces, urine, gnawing, nesting, dead birds and bird droppings; fumigation turned up more than 1,100 dead rodents, and company records showed more than 2,300 caught the prior year. Family Dollar recalled food, pet food, supplements, drugs, devices and cosmetics sold at 404 stores in six states. Its plea agreement later admitted stores had reported rodents in deliveries since 2020. In February 2024 the company pleaded guilty to a misdemeanor and agreed to pay $41.675 million, the largest criminal penalty in a food safety case.

10. April 8, 2017 — Fresh Express Organic Marketside Spring Mix, sold at Walmart (dead bat). Two Florida consumers ate part of a salad before finding a decomposed bat in it. CDC’s rabies lab could not rule out rabies, and both were started on post-exposure treatment. Fresh Express recalled the production lot, code G089B19, from Walmart stores in the Southeast, with Florida health officials, FDA and CDC involved. It is the only produce-animal case I found that drew a federal public health response.

11. September 10, 2016 — MIH Marketing and Sales, Beebe, Arkansas (rodent activity in meat and poultry storage). FSIS announced a recall of about 662,049 pounds of meat and poultry sold at 17 Arkansas stores after an investigator on routine surveillance found it stored under insanitary conditions, including rodent activity; samples tested positive for alkaline phosphatase, evidence of possible fecal matter. Found by the agency, not a customer.

12. December 9, 2013 — Yauk’s Specialty Meats, Windsor, Colorado (rodent activity in production and storage). FSIS announced a recall of about 90,000 pounds of meat and poultry, expanded twice that week, after a food safety assessment found product made under insanitary conditions, including rodent activity in the production, storage and retail areas. Found by the agency.

13. February 12, 2009 — Peanut Corporation of America, Plainview, Texas (dead rodents, rodent excrement, bird feathers). The Texas Department of State Health Services ordered the recall of everything the plant had ever shipped after inspectors found the filth in a crawl space above production and an air handler pulling debris from it into the plant. The plant had operated unlicensed and uninspected since 2005. Texas acted on its own authority over an immediate and serious threat to human life or health, apart from the Salmonella findings that later sent PCA’s owner to prison.

What the list shows

  • Consumers find the animals; inspectors find the infestations. Every whole-animal case here started with a customer, usually holding a phone. Every warehouse case started with an inspector.
  • The animal hides in the database. Set aside Gold Star, where FDA wrote the notice. In the four other FDA cases since 2024 (Dr. Praeger’s, PCC, Lundberg, Nortera), the animal appeared only in the Enforcement Report, weeks after the recall began; the company or store notice, where there was one, said “pest” or “foreign material.” Mezzetta makes five: the mouse is on TikTok, not in the notice.
  • Most animals in produce never trigger a recall at all. A 2019 University of Illinois review in Science of the Total Environment counted 40 incidents from 2003 to 2018 of wild vertebrates found by customers in prepackaged produce: 21 amphibians, 9 reptiles, 7 mammals and 3 birds, and noted that they rarely draw a serious food-safety response. Two examples: in 2016 Western Family told a Utah woman who found a snake’s head in canned green beans it had “issued a recall to wholesalers”; I can find no FDA record of it. In 2010 Pictsweet’s insurer denied liability to a Houston mother who found a reptile head in frozen green beans, calling it a frog, not a snake. No recall.

None of these is an outbreak. Nobody died of a mouse in a jar. They matter because they show how the recall system talks to the public when the hazard is filth rather than a pathogen: late, in a database, in language that avoids the noun. Tell people what was in the food and where it went.

On Saturday, September 26, 2026, G. L. Mezzetta Inc. of American Canyon, California, announced a recall of one lot of Mezzetta Golden Greek Peperoncini, Medium Heat, in 32-ounce glass jars: SKU 10106328, Lot 720106. FDA posted the company’s announcement the same day under a headline that ends “Due to Presence of Pest Contaminant in the Jar.” The reason FDA lists is that the product “may contain foreign objects such as Pest contaminant in jar.” The company says it has received no reports of illness or injury.

The announcement never says what the pest was. The video that started this does. On September 16, a consumer posted a series of TikTok videos showing what she described as a dead baby mouse in a jar of Mezzetta peperoncini she had opened while making a Mississippi-style pot roast. She tagged the company. In a later video she said she had contacted the store where she bought the jar, an attorney and poison control. CBS News, Gray Television stations and others reported on the videos; CBS said Mezzetta did not respond to its request for comment.

Ten days after the first video, the recall notice went up on FDA’s website. The notice describes the company’s response under a bold heading: “We Moved Quickly.” I will leave the adverb to readers and stick to the dates. Video: September 16. Recall: September 26. The FDA page also carries a note that the company “updated the product information on September 26, 2026 in order to accurately reflect the recalled item.” The notice does not say what was wrong the first time.

What was recalled

  • Product: Mezzetta Golden Greek Peperoncini, Medium Heat, 32 oz. glass jar
  • SKU: 10106328. Lot: 720106
  • Company: G. L. Mezzetta Inc., American Canyon, California, a fourth-generation family companyfounded in San Francisco in 1935
  • Reason: “may contain foreign objects such as Pest contaminant in jar”
  • Illnesses or injuries reported: none, according to the company
  • Instructions to consumers: do not eat the product; contact CustomerSupport@Mezzetta.com “for assistance”
  • No other Mezzetta products or lots are included

What the notice does not say

FDA has told companies what a recall warning should contain. Its February 2019 guidance, Public Warning and Notification of Recalls Under 21 CFR Part 7, Subpart C, recommends images of the product; lot numbers and packaging details; the geographic areas and dates of distribution; a thorough description of the defect, the health hazard and the reason for the recall; the number and nature of complaints; instructions to consumers; and the company’s contact information. Measured against that list, the Mezzetta notice comes up short on nearly every count.

  • No photograph of the jar or the label. FDA keeps a Flickr album of recalled-product photos for exactly this purpose. This notice has nothing in it.
  • No states, no retailers, no dates of sale. The company says it has “detailed production and distribution records” and has “identified the customers that received product from the recalled lot.” It shares none of that with the people who bought the jars. Mezzetta is sold nationally; the notice does not say whether this lot went to one region or all fifty states.
  • No number of jars. One lot of a national 32-ounce item could be a few hundred cases or many thousands of jars. Consumers are not told which.
  • No description of the hazard. “Pest contaminant” is a category, not a description. “Mouse” is a word the company has so far not written, even though it says it has contacted the consumer and asked for the jar.
  • No packing location or date. The label on the 32-ounce Golden Greek Peperoncini reads “Product of Greece,” and the company says the peppers are imported from the Peloponnese. The notice talks about “our facilities” and “our facility,” about sanitation, pest control, an SQF audit and a HACCP plan, without ever saying whether Lot 720106 was packed in Napa County or in Greece and shipped here as finished goods. That is not a small point. The pest-control and sanitation records that matter are the ones for the plant where the jar was filled and capped.
  • No telephone number and no refund instruction. Consumers get an email address and the word “assistance.”

The notice also tells us, by omission, how the company learned of the problem. It did not learn from its SQF audit, its HACCP plan or its pest-control program. It learned from a customer with a phone. I do not write that to score a point. The company chose to put those programs in its recall notice as reassurance, and the sequence of events is what it is.

A mouse in a jar is adulteration, whether or not anyone gets sick

The Federal Food, Drug, and Cosmetic Act does not wait for an illness. Under 21 U.S.C. § 342(a)(3), a food is adulterated if it “consists in whole or in part of any filthy, putrid, or decomposed substance, or if it is otherwise unfit for food.” Under § 342(a)(4), it is adulterated if it “has been prepared, packed, or held under insanitary conditions whereby it may have become contaminated with filth.” FDA’s Current Good Manufacturing Practice regulation is blunter: “Pests must not be allowed in any area of a food plant,” and “[e]ffective measures must be taken to exclude pests from the manufacturing, processing, packing, and holding areas and to protect against the contamination of food on the premises by pests.” 21 CFR 117.35(c).

A rodent in a jar of acidified peppers raises the questions any investigator would ask. Was the vacuum seal intact when the consumer opened the jar? Was the animal in the jar before the brine and the cap, or did it get there afterward? What do the pest-control logs, the line-clearance records and the fill-and-cap records for Lot 720106 show? Were other jars from the same lot and the same fill window pulled and inspected? The company says it is reviewing “the relevant production, sanitation and food-safety records.” Good. It should tell the public what it finds.

FDA has not yet classified this recall. When it does, the classification, the quantity and the distribution pattern will show up in FDA’s weekly Enforcement Report, which is where those facts usually surface: weeks after the announcement, when they no longer help anyone standing in a grocery aisle. FDA’s November 2020 guidanceon retail consignee lists reserves them mostly for foods that cannot be identified from their packaging; USDA’s Food Safety and Inspection Service posts retail distribution lists for its recalls as a matter of routine. A lot code on a jar should not be the reason consumers are left to guess which stores sold it.

What you should do

  • Check any 32-ounce jar of Mezzetta Golden Greek Peperoncini, Medium Heat, for SKU 10106328 and Lot 720106. If you have one, do not eat from it. Do not throw it out either. Keep it, photograph it and contact the company at CustomerSupport@Mezzetta.com.
  • If you find a foreign object in any food, keep the product and its packaging, photograph everything before you handle it further, write down where and when you bought it, and report it to FDA as well as to the company and the store.
  • If you ate from a jar in this lot and became ill, see a doctor and tell them what you ate.

Questions for Mezzetta and FDA

  1. Where and when was Lot 720106 packed, and how many jars did it contain?
  2. Which retailers, in which states, received it, and on what dates?
  3. Has the company examined the consumer’s jar, and was the seal intact when she opened it?
  4. What do the pest-control and sanitation records for the packing line show for the production window?
  5. Will FDA publish the retail consignee list and the recall classification now, rather than in an Enforcement Report weeks from now?

“For generations, families have trusted Mezzetta products on their tables,” the notice begins. Trust is earned with facts. Tell people what was in the jar, where the jars went and what the records show. That is what a recall notice is for.

Two federal pages carry the running account of what is making Americans sick right now: FDA’s Investigations of Foodborne Illness Outbreaks table and USDA’s Outbreak Investigations: Response table at the Food Safety and Inspection Service. I wrote them up on August 8, August 12 and September 2. Here is where they stand after FDA’s September 25 update.

FDA’s table carries seventeen investigations in its Active section. Add the case counts the agency publishes on the individual advisories and the table describes 14,197 illnesses, 680 hospitalizations where a number is given, and two deaths, both from Cyclospora in Michigan. Ten of the seventeen are ongoing; seven are files where the outbreak is over and FDA still calls the investigation active. Ten name no food, and those ten carry 645 sick people. Five are Cyclospora, and those five carry 13,195 of the 14,197, 93 percent of the table, nearly all of it the iceberg lettuce outbreak CDC closed on September 11. Nine of the seventeen show no on-site inspection.

USDA’s table lists four investigations for 2026. One is active. It is the same E. coli O157:H7 file FDA posted on September 23 with 25 sick and no food named.

Seven closed, seven opened

Since August 12, FDA has moved seven files to its closed list: frozen blueberries (17 sick, closed September 23), romaine lettuce (14, closed September 10 with no grower or processor named), the requesón Listeria outbreakthat killed one person (15 sick, closed September 16), the Ayco Farms cantaloupe file (70) and the moringa leaf powder file (131), whose advisory called FDA’s investigation complete on July 22, 63 days before the table said closed. Two small Cyclospora files from the August table, #1375 with eight sick and #1381 with two, are on neither list today.

Seven files took their places: Salmonella I 4,[5],12:i:- (43 sick, no food named); the alfalfa sprout file, which now carries multiple E. coli strains and Salmonella Agona (55 sick in 15 states); Salmonella Newport (116, no food named); Salmonella Bovismorbificans in Evergreen Fresh Sprouts broccoli sprouts (32 sick in six states); the joint O157 file (25); a Listeria file (40, no food named); and, on September 25, E. coli O26:H11 in Sierra Nevada Graziers raw milk cheese: 13 sick in nine states, eight hospitalized, three with HUS.

The total nearly doubled from 7,324 for one reason: the iceberg lettuce count went from 6,358 to 12,883 before CDC closed it. Take Cyclospora out and the other twelve files hold 1,002 people.

The table as it stands

Ref. #PathogenFood namedSickOutbreakInspected on site
1419E. coli O26:H11Raw milk cheese13Ongoing 
1421E. coli O157:H7 (USDA 2026-04)Not yet identified25Ongoing 
1417Listeria monocytogenesNot yet identified40Ongoing 
1411Salmonella BovismorbificansBroccoli sprouts32Ongoing 
1403Salmonella NewportNot yet identified116Ongoing 
1394E. coli (multiple) and Salmonella AgonaAlfalfa sprouts55OngoingYes
1402Salmonella I 4,[5],12:i:-Not yet identified43OngoingYes
1398CyclosporaNot yet identified22Ended 
1392CyclosporaNot yet identified234Ended 
1395Salmonella JavianaJalapeño peppers431OngoingYes
1379Listeria monocytogenesNot yet identified10OngoingYes
1390CyclosporaIceberg lettuce12,883EndedYes
1385CyclosporaNot yet identified38Ended 
1384CyclosporaNot yet identified18Ended 
1387Salmonella OranienburgNot yet identified99OngoingYes
1383Clostridium botulinumPowdered infant formula4EndedYes
1378Salmonella EnteritidisEggs134EndedYes

Counts for the See Advisory rows come from each advisory’s Case Counts block as of September 29. Hospitalizations where published: raw milk cheese 8, broccoli sprouts 3, alfalfa sprouts 4, jalapeños 57, iceberg lettuce 570, infant formula 4, eggs 34.

The newest files are the emptiest

The five newest rows, posted between August 26 and September 25, carry 226 sick people between them, and not one shows an on-site inspection.

Salmonella Newport, #1403, was posted August 26 with 84 sick and no food named. It is 116 today. Traceback and sampling are marked. No inspection, no advisory, no CDC notice, thirty-four days in.

The two files posted September 23 are the thinnest: E. coli O157:H7, 25 sick, carrying FSIS reference 2026-04, and Listeria monocytogenes, 40 sick. Traceback on each. No states, no ages, no hospitalizations, no onset dates. CDC has posted a notice for neither. I wrote about the joint file last week.

The broccoli sprout file, #1411, shows a recall, traceback and sampling. No inspection is marked at Evergreen Fresh Sprouts, the Idaho grower that recalled the bags on September 4.

The Graziers row, posted September 25, is the sparest of all. It shows a recall and nothing else: no traceback, no inspection, no sampling, on an outbreak in which nine of nine people interviewed ate the cheese, eight of thirteen were hospitalized, three developed HUS and CDC puts the median age at five. FDA says its investigation into the source of contamination is ongoing. FDA’s What’s New dates the recall September 25; its advisory says September 24.

The column records what has been initiated as of the stamp, and an inspection can begin after it. Eight rows show one; nothing posted after August 19 does.

Complete on one page, active on the other

FDA’s egg advisory was updated September 3. The final count is 134 sick, 34 hospitalized, 18 states, with illnesses from November 21, 2025, to August 12, 2026, a nine-month outbreak. FDA inspected Midwest Poultry Services in Texas, and three environmental samples matched the outbreak strain. Then this sentence: FDA’s outbreak investigation is complete. Twenty-two days later, at the September 25 stamp, the table still lists #1378 as Active.

The close also leaves a question FDA raised in July. On July 24 the agency wrote that Midwest Poultry does not account for all the illnesses in this outbreak. The September 3 update calls the company a likely source, confirms recalled eggs reached six states, Arkansas, Louisiana, Mississippi, New Mexico, Oklahoma and Texas, and lists sick people in eighteen. Twelve states have cases and no confirmed egg. Whether another source was ever found, the close does not say.

The lag is not consistent. The moringa file went 63 days between its advisory saying complete and the table saying closed; the blueberry file went two. The infant formula file at least explains itself: the outbreak investigation has ended, and the root-cause investigation into the Organic West Milk and Dairy Farmers of America supply chain it shares with the 2025 ByHeart outbreak continues. The iceberg lettuce advisory, untouched since September 11, says samples from the Mexican inspections are pending analysis. And the jalapeño advisory, 431 sick, 57 hospitalized, 32 states, still reads Ongoing on a page that has not moved since August 21, for an outbreak whose last illness began August 2.

USDA’s table

FSIS says it posts an investigation when there is compelling evidence that a meat, poultry or egg product is the source, and that it updates the table at least weekly. For 2026 it lists four. Three are closed: Listeria in retail-sliced headcheese in May, with a public health alert and an Illinois notice; E. coli O157:H7 in beef kofta served at The Kebab Shop in May, with a public health alert and a California notice; and Salmonella Enteritidis in chicken, listed as suspect, at the Broome County Correctional Facility in New York in June.

The fourth, 2026-04, is the first active row USDA has carried all year. Product: Unknown. Note: FSIS and FDA are currently investigating this outbreak. It is the one place the two federal tables point at each other, and neither says what the food is.

FSIS ref.PostedPathogenProductActive
2026-04SeptemberE. coli O157:H7Unknown (FDA #1421)Yes
2026-03JuneSalmonellaEnteritidisChicken (suspect), Broome County Correctional FacilityNo
2026-02MayE. coli O157:H7Beef kofta, The Kebab ShopNo
2026-01MayListeria monocytogenesRetail-sliced headcheeseNo

CDC’s count

CDC’s weekly page does not name investigations; it counts them. Its September 23 update listed 32 active multistate investigations: 20 Salmonella, six E. coli, five Listeria and one Campylobacter. FDA’s ten unnamed files are presumably among them; nobody outside the agencies can say which.

Marler Clark represents people sickened in the iceberg lettuce, Kebab Shop kofta and moringa outbreaks, the family of a child who developed HUS in the Graziers outbreak, and families in the Nara and ByHeart infant formula outbreaks.

Seventeen open. Ten with no food. Nine with no inspection, including all five files posted since August 26. One file complete on its own advisory for twenty-six days and active on the table. And the only meat investigation of the year is a file whose product line reads Unknown. The pages are public. I will keep reading them.

This morning KFF Health News and CBS News published Stephanie Armour’s Food safety becomes an election issue as major outbreaks and recalls cause concern. Read it. It is the best short piece I have seen this year on how a lettuce outbreak turned into a campaign issue, and it does something most political coverage of food safety does not: it slows down and asks what the numbers actually mean.

Stephanie gets the big thing right. Recall counts are a lousy yardstick. Agencies do not publicize every recall, one contaminated ingredient can spawn a dozen of them, and the FDA’s “fewest recalls in a decade” chart turns out to fold in drugs and devices, which former CFSAN director Susan Mayne called out. The better measure, as the epidemiologists she quotes say, is how many people got sick. By that measure the 2026 Cyclospora season is the largest outbreak of the parasite ever recorded here, roughly 18,000 to 20,000 laboratory-confirmed cases against 1,180 in the same months of 2025.

She also reached back to 1993 and Jack in the Box, and to how an E. coli outbreak put food safety on a new President’s first Cabinet agenda. I was there for that one. I represented Brianne Kiner. It is good to see a reporter remember that the last time the public got this angry about food, the government actually changed how it did business.

So, thank you, Stephanie. And, in the spirit of a friendly nudge, here are a few threads I hope you and your KFF colleagues will pull on next.

Who is answering for this outbreak? Behind the “robust response” talking points, the public record is thin. As I laid out earlier this month, the grower answered questions on its own FAQ page, FDA issued guidance nobody is required to follow, CDC stopped weekly updates, and thirteen congressional letters sit unanswered with no hearing on the calendar. CBS’s own reporting found the farm linked to the outbreak had not been inspected since 2019. Staffing numbers matter, but so does the simple question: when was the last time anyone looked?

The traceability rule is not a footnote. Stephanie notes that FDA pushed compliance with the FSMA 204 traceability rule from January 2026 to July 2028. That is the rule built for exactly this kind of outbreak, where lettuce moves through processors and restaurant chains and investigators burn weeks reconstructing the supply chain. Every month of delay is a month the next trace-back runs slower than it needs to.

Salmonella in poultry. The withdrawal of the Biden-era framework gets a sentence in the piece. It deserves a story of its own. USDA still does not treat Salmonella in raw poultry as an adulterant, and until it does, the chicken in the grocery case can carry levels that make people sick and still be perfectly legal to sell.

Babies. The election framing is lettuce and burgers, but the outbreak that should keep regulators up at night is infant botulism tied to powdered formula. I have written about the recent Reagan-Udall report on botulism in infant formula, a roundtable that ByHeart and the International Dairy Foods Association helped convene and paid for. Parents deserve to know who is shaping the science on what goes into their children’s bottles.

Accountability that bites. Fines and settlements are a cost of doing business. I just got back from Oslo, where I talked about the criminalization of food safety failures. Until executives believe that poisoning customers can carry personal consequences, the incentives do not change.

Stephanie closed with Sandra Eskin of Stop Foodborne Illness making the point that matters most: if the government wants people to eat real food, “they have to be safe.” Farmers plowing under lettuce nobody will buy is what it looks like when that trust breaks. Food safety should not be a red or blue issue. It becomes one when nobody in charge will answer the questions.

Great work. Keep going. I am happy to help.

As of September 17, CDC counts 37 people with hepatitis A in four states linked to La Serranita-brand concha negra shell meat from Ecuador, and 30 of them have been hospitalized. In April, the New Jersey Department of Health issued a notice headed “Illegally Imported Concha Negra Blood Clams” and told county inspectors to arrange destruction of the product because it came from an unapproved source. FDA’s safety alert and CDC’s outbreak page use neither word. Readers asked me why. The answer says a lot about how shellfish imports are actually policed, and about who could be held accountable.

The short answer

Under the shellfish sanitation rules the states enforce, the product was illegal the moment a certified dealer took it in. Under federal import law, it is not that simple. Coming from a country that has no shellfish agreement with FDA does not, by itself, make an import unlawful. But federal seafood safety rules, shellfish labeling rules and the Lacey Act give prosecutors real tools against the people who moved this product, whether or not anyone ever calls the import “illegal.”

What federal law says about importing clams from Ecuador

Last week I wrote that a package of La Serranita concha negra should never have cleared a U.S. port. As a matter of shellfish program policy, that is right. As a matter of federal regulation, it needs a sharper edge.

FDA’s seafood HACCP rule gives importers two ways in. Under 21 CFR 123.12(a), an importer must either get the product from a country with an active memorandum of understanding or similar agreement with FDA, or have and follow written verification procedures showing the product was processed in compliance with the rule. Those procedures must include product specifications and affirmative steps, such as collecting the foreign processor’s HACCP records, obtaining a foreign government certificate, inspecting the foreign plant or testing the product. Ecuador has no molluscan shellfish agreement with FDA, so the second path is the only one open. If the importer cannot provide those assurances, the regulation says the product “will appear to be adulterated and will be denied entry.”

FDA’s own compliance program for molluscan shellfish is candid about the limits. It says imported raw shellfish should come from certified shippers in agreement countries listed on the Interstate Certified Shellfish Shippers List. It also says that origin in a country without an agreement is not, standing alone, enough to detain a shipment at the border. When FDA has no other evidence of adulteration or misbranding, its staff are told to refer the entry to other agencies and the state shellfish authority rather than hold it. The program even concedes the weakness in that handoff: a state cannot know that shipments from uncertified dealers have been released into commerce unless FDA tells it.

That is the most likely reason the federal pages avoid the word “illegal.” FDA would be describing as unlawful an entry that its own guidance says it could not stop on origin alone. It is also possible FDA is silent because an enforcement matter is open. From the outside, there is no way to tell which.

Why New Jersey can say it

The National Shellfish Sanitation Program is a federal–state cooperative. FDA administers it, but the states carry it out through their own laws by adopting the NSSP Model Ordinance. Under those laws, certified shellfish dealers may buy only from approved, certified sources. The FDA Food Code, which the states adopt in some form, likewise requires shellfish received in interstate commerce to come from sources on the Interstate Certified Shellfish Shippers List. A New Jersey dealer that bought, held and resold concha negra from an uncertified Ecuadorian packer broke New Jersey’s rules, whatever happened at the port. That is what the state notice means, and it is why New Jersey could order the product destroyed.

The federal hooks that do not depend on the word “illegal”

Seafood HACCP source controls. An importer or dealer that stores, repacks or holds clam meat is a “processor” under 21 CFR 123.3, which defines processing to include handling, storing, packing and holding. Under 21 CFR 123.28, processors of raw molluscan shellfish must build source controls into their HACCP plans. They may process only shellfish harvested from growing waters approved by a shellfish control authority, and they may accept shucked shellfish only in containers labeled with the packer’s certification number, keeping records that show it. Under 21 CFR 123.6(g), a processor that fails to meet those requirements renders its product adulterated under section 402(a)(4) of the Food, Drug, and Cosmetic Act. No positive virus test is needed.

Shellfish labeling. A separate Public Health Service Act regulation, 21 CFR 1240.60, requires every container of shucked shellfish to carry the name, address and certification number of the packer. Shellfish without that label are subject to seizure or refusal of entry, and destruction. The package description in New Jersey’s notice lists the brand, “Concha Negra,” “Shell Meat,” “Fresh Frozen” and the net weight. It does not list a certification number. I do not know whether the actual packages carried one, but that is the first thing I would want to see. Violating a Part 1240 regulation is a federal misdemeanor under 42 U.S.C. § 271, punishable by up to a year in jail.

The Food, Drug, and Cosmetic Act. Introducing adulterated food into interstate commerce, or adulterating food held for sale after interstate shipment, violates section 301 of the Act. A first offense is a misdemeanor that requires no proof of intent, and under the Supreme Court’s Park doctrine, the responsible corporate officers can be convicted along with the company. The same conduct becomes a felony, punishable by up to three years, if it is done with intent to defraud or mislead.

Customs fraud. If the entry papers misdescribed the product’s species, form or origin, or claimed it was cooked, 18 U.S.C. § 542 (entry by false statement) and 18 U.S.C. § 545 (smuggling, including importing merchandise contrary to law) come into play. FDA’s import alert for molluscan shellfish notes past shipments labeled as cooked that had only been heated enough to open the shells.

The Lacey Act. Most people think of the Lacey Act as a wildlife law, but it covers fish, including shellfish. It makes it a federal crime to import, export, transport, sell, receive or buy, in interstate or foreign commerce, fish that were taken, possessed, transported or sold in violation of any state law or regulation or any foreign law. FDA and NOAA’s National Marine Fisheries Service have a memorandum of understanding to cooperate in using it against illegal commerce in molluscan shellfish.

There are two routes here. The first is state law. If New Jersey is right that its dealer’s possession and sale of these clams violated New Jersey shellfish law, then shipping them on to a New York dealer is interstate commerce in shellfish sold in violation of state law. The second is foreign law. Ecuador regulates concha prieta, the same black clam, with a 4.5-centimeter minimum size for harvest, transport, possession, processing and sale, and it subjects exports to its rules on products that are prohibited from export or need prior authorization. If these clams were undersized or exported without authorization, that is a Lacey Act predicate. A knowing violation is a felony, punishable by up to five years, and the government must prove the defendant knew of the underlying illegality (Ninth Circuit model instruction). A defendant who should have known, in the exercise of due care, faces a misdemeanor.

Who could be charged

The Ecuadorian packer and exporter is the most culpable party and the hardest to reach.

The New Jersey dealer and importer is the most exposed. It carries the importer’s verification duties under 123.12, the processor’s source-control duties under 123.28, and responsibility for the interstate sale to New York. New York’s notice says the product went from “a New Jersey dealer/importer” to “a New York dealer.” Its responsible officers are exposed under Park.

The New York dealer faces state approved-source violations, and federal exposure if it held the product for sale without the required source controls.

A customs broker ordinarily is not the “importer” under the HACCP rule and would be exposed only if it knowingly filed false entry information.

Restaurants and retailers face state and local health code enforcement and product destruction. Prosecution is unlikely.

Will anyone be charged?

Federal criminal cases after foodborne illness outbreaks are uncommon, but they happen: Peanut Corporation of America, Jensen Farms, Blue Bell and Quality Egg. Several of those rested on misdemeanor counts that did not require proof that anyone meant to hurt anybody. What we do not know is whether FDA’s Office of Criminal Investigations or NOAA’s Office of Law Enforcement is looking at this outbreak. If they are, the silence makes sense. If they are not, the silence is the problem.

What FDA should tell the public now

FDA does not have to call the import “illegal” to tell us what matters. It should say whether the importer had the verification procedures 123.12 requires for this product, whether the packages carried a shellfish certification number, and whether the importer or the Ecuadorian packer is now on Import Alert 16-50. And it should name the importer and the dealers, so that a restaurant can check its invoices against something other than a brand name.

Three questions readers are asking

Were the clams illegally imported? Under state shellfish law, New Jersey says yes: they came from an unapproved source, and certified dealers may not handle them. Under federal law, origin in a country without an FDA shellfish agreement is not automatically illegal. The import was lawful only if the importer had the verification procedures the seafood HACCP rule requires, and the product had to meet federal source-control and labeling rules once it was here.

Could anyone go to jail? Potentially. Shipping adulterated food is a strict-liability federal misdemeanor that can reach corporate officers, and knowing or fraudulent conduct can be charged as a felony under the Food, Drug, and Cosmetic Act, the customs laws or the Lacey Act. No charges have been announced.

Is the product still dangerous? Yes. Freezing does not kill hepatitis A virus, and the alert on La Serranita-brand concha negra shell meat remains in effect. Throw it out. If you ate it within the last 14 days and are not vaccinated, ask your doctor or health department about post-exposure prophylaxis.

In October I intend to file a citizen petition asking the Food and Drug Administration to require a warning statement on the label of raw and lightly cooked sprouts — the same kind of statement FDA has required on unpasteurized juice since 1998. 

Before it goes to the docket I want it read by the people most likely to find the holes: sprout growers, seed suppliers, retailers, restaurant operators, state and local health officials, consumer advocates, and anyone who has been sick from sprouts. Send comments to bmarler@marlerclark.com by Friday, October 9. I will post a revised draft before I file. Here is a link to Petition – https://www.marlerblog.com/files/2026/09/Sprout-Warning-Label-Petition_MarlerClark_9-28-26.pdf

Why now

Two sprout outbreaks are open at the same time. Alfalfa sprouts grown in Minnesota have sickened 55 people in 15 states, four of them hospitalized. Broccoli sprouts grown in Idaho have sickened 32 people in six states, three hospitalized, by CDC’s September 24 count. Together that is 87 people in 19 states, seven in the hospital. The broccoli sprout recall was announced September 10 for bags delivered between August 24 and September 2; twenty-five of the thirty-two people were already sick by August 23. On September 24 FDA reported that broccoli sprouts collected from a Washington grocery store and from a sick person’s home both matched the outbreak strain. The organism was on the shelf, in a package that said nothing.

The alfalfa seed lot went to growers and distributors in sixteen states, Puerto Rico and four countries, and on September 23 an Ohio seed company recalled 950 pounds of it sold to home sprouters in forty-four states, in small bags that carry no lot code. As of September 24 FDA still had not named the supplier.

Twenty-eight years of the same advice, none of it on the package

On August 31, 1998, FDA told children, the elderly and people with weakened immune systems not to eat raw sprouts. It said it again on July 9, 1999, and extended the advice to mung bean and lightly cooked sprouts on October 2, 2002. CDC said the same thing in 1999. That same year FDA wrote into the Food Code that hospitals, nursing homes, day care centers and senior centers may not serve raw seed sprouts at all; every state but California has adopted an edition of the Food Code, and California has the same rule in its own statute. FDA’s current guide for older adults and immunocompromised people lists raw sprouts of any kind among the foods to avoid.

Every one of those statements was delivered by press release, talk paper, model code or booklet. A press release lasts one news cycle. A label is there every time the product is picked up, which is where the decision to eat it is actually made. I first asked FDA for this in March 2003. Since then the agency has counted 52 sprout outbreaks and more than 2,700 illnesses in this country between 1996 and 2020 alone, and has said for years that contaminated seed is the likely source of most of them.

What the petition asks

The petition asks FDA to amend 21 CFR 101.17, the section that already carries the warnings on untreated juice, iron supplements and shell eggs, by adding a paragraph for sprouts. The statement it proposes:

WARNING: Sprouts may contain harmful bacteria, such as Salmonella, E. coli and Listeria, that can cause serious illness, kidney failure or death. Sprouts are eaten raw and cannot be pasteurized, and the bacteria may be inside the seed, where washing cannot reach them. Children, the elderly, pregnant women, and persons with weakened immune systems should not eat raw or lightly cooked sprouts.

It would be boxed, with WARNING in bold capitals, in type no smaller than 8 point. Packages too small to carry it could use a placard at the point of sale. The same statement would appear on a sign or menu wherever sprouts are sold loose — deli counters, sandwich shops, salad bars — and on the listing wherever they are sold online. It would also go on seed sold at retail for home sprouting, because FDA considers seed for sprouting to be food and this year’s seed lot reached home kitchens in forty-four states. Sprouts treated by a validated process that reduces pathogens would be exempt, exactly as treated juice is exempt today. The rule would take effect sixty days after publication; FDA wrote the juice warning in seventy-five days in 1998, proposal to final rule, and urged juice makers to start using the statement the day it proposed it. The petition asks FDA to do the same for sprouts.

The petition does not ask FDA to ban sprouts. It leaves the product on the shelf and the decision with the shopper, who for the first time would have the same information FDA has been giving her, everywhere but on the package, since 1998.

What I want you to tell me

1. The words. The proposed statement runs sixty-two words. The juice warning runs thirty-three. FDA’s own focus groups in 1998 preferred short statements, and a reviewer of the draft made the same point. My answer was to keep three facts the juice warning does not need — sprouts cannot be pasteurized, the bacteria may be inside the seed, and the at-risk groups should not eat them — and cut everything else. The longer version I published on September 11 is now offered as consumer-education text, not label copy. Is sixty-two words still too many? Should Listeria stay? Should miscarriage be named?

2. Small packages. A four-ounce clamshell has little room. Is a point-of-sale placard the right answer, or should the label carry a shorter form?

3. Seed for home sprouting. This is the newest request in the draft and the one most likely to draw objection. If you think it belongs in a separate petition, say so.

4. Shoots and microgreens. FDA’s sprout rule excludes plants grown in soil and harvested above the soil line. The draft asks that the warning follow the product rather than the growing method. Growers of shoots and microgreens, I want to hear from you.

5. Restaurants and delis. A large share of sprout illnesses trace to sandwiches and restaurant meals, not retail packages. Is a sign or a menu line workable, and who enforces it?

6. Cost and treatment. Sprout growers: what would relabeling cost, and what pathogen-reduction process should qualify for the exemption?

7. Labels that did not work. FSIS gave up on labeling breaded stuffed chicken in 2024 and declared Salmonella an adulterant instead. The draft answers that those labels failed because people cooked a raw product as if it were cooked, and there is no cooking step here. If you think the comparison cuts the other way, tell me why.

8. Anything I got wrong. Every fact in the draft carries a footnote and every source is listed at the end with a link. If a number, a date or a citation is wrong, tell me which one.

How this works from here

Comments are open until October 9. I will read every one, post a revised draft, and then file the petition in FDA’s citizen petition docket. Once it is filed, anyone can submit a comment to FDA directly at regulations.gov, and FDA has 180 days under its own rules to respond. The draft has already had one round of outside review from people who know this ground well, and it is better for it. A second round from the people who grow, sell, serve, regulate and eat sprouts will make it better still.

Twenty-eight years is long enough for the package to say nothing. Help me get the words right.

On Friday, CDC and FDA announced that E. coli O26:H11 in Graziers raw milk cheese from Sierra Nevada Cheese Company of Willows, California has sickened 13 people in nine states, put eight in the hospital and left three children in kidney failure. More than half of the sick are five years old or younger. The first illness began July 7. FDA contacted the company on September 24, eleven weeks later, and the company recalled every Graziers raw milk cheese the same day. The cheese was aged at least 60 days, which is what federal law requires of a raw milk cheese, and the label said exactly that. It said nothing else. I wrote about the outbreak on Saturday.

For several weeks I have been drafting a citizen petition to FDA under 21 CFR 10.30 asking the agency to require a warning statement on cheese made from unpasteurized milk. This outbreak is the reason to stop drafting and file. Before I do, I am posting the draft here and asking for comments. The full draft petition runs 22 pages with 47 footnotes, every one of them to a public source. I want to hear from cheesemakers, retailers, public health people, lawyers, parents, and from the people who are going to disagree with me. I would rather hear it now than read it in the docket after I file. Here is complete draft: https://www.marlerblog.com/files/2026/09/Raw-Milk-Cheese-Warning-Label-Citizen-Petition_WDM_9_27_26.pdf

What the 60-day rule is

Since 1987 FDA has prohibited unpasteurized milk in interstate commerce. The regulation carves out one exception, in its own words “alternative procedures to pasteurization,” meaning cheese cured at least 60 days at not less than 35 degrees under the federal standards of identity. The 60-day rule dates from the 1940s. It is the entire legal basis on which raw milk crosses state lines in a form people eat, and nothing in federal law requires the label to say what that means.

What FDA’s own science says about the rule

This is not my science. It is the agency’s. The joint FDA/Health Canada risk assessment, finalized in 2015, found that a serving of soft-ripened cheese made from raw milk carries roughly 50 to 160 times the listeriosis risk of the same cheese made from pasteurized milk, and that for those cheeses 60 days of aging can increase the risk rather than reduce it, because the pH rises as the cheese ripens and Listeria grows.

The same day it finalized the assessment, FDA published a request for comments on the 60-day rule. In it the agency wrote that the aging period had been presumed to control pathogens, that the available data raised questions about the safety of raw milk cheese even when aged, and that peer-reviewed studies showed E. coli O157:H7 surviving 60 days in cheddar. It cited a 1997 federal advisory committee memorandum saying the literature confirms pathogens survive the aging process. It asked whether a performance standard should replace the rule. Seventy-five comments were filed. Eleven years later there is no proposed rule, no completed reevaluation and no explanation.

FDA’s 2014 to 2016 sampling of 1,606 aged raw milk cheeses found pathogens in 14, including Salmonella, Listeria and a Shiga toxin-producing E. coli, and found generic E. coli, the agency’s indicator of unsanitary processing, in 5.4 percent. And a Cornell study FDA funded, published in Nature Medicine last October, found that 60 days of aging does not eliminate viable H5N1 bird flu virus in raw milk cheese. FDA began sampling aged raw milk cheese for the virus in December 2024, posted interim results in March 2025, promised final results that spring, and has not posted them.

Four outbreaks in 31 months, all in cheese that followed the rule

Each cheese was aged as the law requires. February 2024, Raw Farm raw cheddar, E. coli O157:H7: 11 sick in five states, five hospitalized, two with hemolytic uremic syndrome (HUS). October 2025, Twin Sisters Creamery aged raw milk cheese in Washington and Oregon, E. coli O103 and O26: 10 sick, one hospitalized, in cheese the state noted was “aged at least 60 days, as required by law.” I represent a claimant in that outbreak, and the petition says so. March 2026, Raw Farm raw cheddar and raw milk, E. coli O157:H7: nine sick, three hospitalized, one with HUS, more than half under five. September 2026, Graziers, E. coli O26:H11: 13 sick, eight hospitalized, three with HUS, more than half five or younger. Together, 43 people sick, 17 hospitalized and at least six children with kidney failure. Before that window, in 2017, a soft raw milk cheese from Vulto Creamery in New York was linked to a Listeria outbreak that hospitalized all eight of its victims and killed two.

Even California stops at the bottle

California has required a boxed WARNING on raw milk products since 1991. Its definition of a raw milk product expressly excludes cheese cured at least 60 days. Washington’s warning applies to containers of raw milk, not to cheese. Three of the four producers above are in those two states, and under state law none of their cheese had to carry a warning. Under federal law none of it did. The state exemption rests on the same assumption FDA’s own science has undermined: that 60 days of aging makes the warning unnecessary.

What the petition asks FDA to do

First, add a prescribed statement to 21 CFR 101.17, the regulation that has carried the unpasteurized juice warning since 1998 and the shell egg statement since 2000, modeled on the juice words:

“WARNING: This product is made from unpasteurized (raw) milk and may contain harmful bacteria that can cause serious illness in children, older adults, pregnant women, and persons with weakened immune systems.” Capitalized, bold, boxed, on the package.

Second, reach the cheese that has no manufacturer’s label. A five-pound Graziers loaf cut and re-wrapped at a deli counter carries nothing. The petition asks that cut-to-order cheese may be covered by a placard at the point of sale, that imported cheese be treated the same as domestic, and that online sellers put the statement on the product page.

Third, exempt cheese made from pasteurized milk or from milk treated by a process validated to achieve an equivalent reduction in pathogens, and exempt cheese shipped for further processing, exactly as the juice rule does.

Fourth, finish the reevaluation of the 60-day rule that FDA announced in 2013 and took comments on in 2015. Within a year, publish either a proposed rule or a written explanation of why not.

Fifth, publish the raw milk cheese sampling data the agency already collects, at the sample level, including the 2025 bird flu results. And follow the juice model on timing: labels turn over in the ordinary course, with a placard permitted for the first year.

What it does not ask for

It does not ask FDA to ban raw milk cheese, to lengthen the aging period, or to restrict who may buy it. Adults who want cheese made from unpasteurized milk would remain free to buy it, with the same information the government already puts on a bottle of raw milk in California and on a carton of unpasteurized juice everywhere. It asks for a sentence.

Where I want your help

Tell me what is wrong with the draft, and be specific. Some questions I am weighing:

Is the wording right? FDA’s juice warning names children, the elderly and people with weakened immune systems. I added pregnant women, because of Listeria, and said “older adults.” Should the statement name the diseases, or is that too much for a label?

Is a counter placard workable for cut-to-order cheese, and who is responsible for it, the retailer or the cheesemaker?

What validated processes should qualify for the exemption, and who validates them?

Is one year of placards before labels must change too long for public health or too short for a small farmstead cheesemaker?

And if any fact in the draft is wrong, say so. Every count, date and quotation is footnoted, and I would rather fix it than file it.

Send comments to bmarler@marlerclark.com by [Friday, October 9]. I will read all of them, I will change the petition where you persuade me, and I may quote from comments in a follow-up post without names unless you tell me I can use yours. Once the petition is filed, FDA will assign it a docket number on regulations.gov, and anyone will be able to comment formally there. The agency is required to respond within 180 days.

Twenty-eight years ago FDA put a warning on unpasteurized juice after outbreaks that included a child’s death, and wrote that consumers were entitled to know a product they thought was healthful could carry pathogens. The parents in this outbreak were entitled to the same sentence. I want to make the case for it as well as it can be made, which is why I am asking you first.

Saturday, September 26, 2026

This is the twelfth Saturday column. Last Saturday’s edition ran under the headline FDA Confirmed the Sprout Seed Recall and Still Did Not Say Who Sold the Seed; its point was that the answers existed, just not on the page where people look. This week the pages filled in, and every one of them was written after the fact. A raw milk cheese outbreak that put eight people in the hospital, in a group where more than half are five or younger, was announced Friday, seventy-nine days after the first illness and thirty days after the last. The broccoli sprout recall that reached the shelf on September 4 could have reached seven of the 32 people now counted. The alfalfa seed lot turned up in retail bags in 44 states, with no lot code on the small ones. Two outbreaks closed with 1,270 sick people between them, three files opened with 78 more, and the person nominated to run FDA sat for her confirmation hearing Thursday and, by one account, could not name the office that investigates any of it. The recalls are arriving. The outbreaks are already over.

Thirteen people in nine states have E. coli O26:H11 from a raw milk cheese, eight are hospitalized, three have HUS, and more than half are five years old or younger. CDC and FDAposted the outbreak Friday. The cheese is Graziers grass-fed raw milk cheese from Sierra Nevada Cheese Company of Willows, California: medium cheddar, sharp cheddar, jalapeño jack and Monterey jack, in 8-ounce and 16-ounce retail packages and five-pound food-service loaves, sold nationwide and online. Illnesses began July 7 and the last reported one began August 26; the sick live in California, Nevada, Colorado, Georgia, Kentucky, Michigan, Oregon, Tennessee and Utah, ages one to 58 with a median age of five, and all nine people interviewed ate the cheese. FDA contacted the company Thursday and the company recalled every Graziers cheese the same day, with no lot codes, because the investigation has not narrowed to a lot or a date. That is seventy-nine days after the first illness. San Diego County says two of California’s three cases are San Diego children under five. The labels FDA posted say the cheese was aged at least sixty days, which is what the federal rule has asked of raw milk cheese for decades, and this is the fourth outbreak in thirty-one months in cheese that met it: Raw Farm in February 2024, Twin Sisters Creamery in October 2025, Raw Farm again in March 2026, and now Graziers, 43 sick, 17 hospitalized and six with HUS between them. The same company had its Bella Capra raw goat cheddar recalled and quarantined by the state for Listeria in August 2025. I wrote about it Friday night, including the citizen petition I have been working on to put a warning statement on cheese made from unpasteurized milk and to finish the reevaluation of the sixty-day rule that FDA announced in 2013.

The broccoli sprout count went to 32 in six states on Thursday, and 25 of those people were sick before the first recalled bag was delivered. CDC’s update adds ten people and two states, Minnesota and Oregon, to the Salmonella Bovismorbificans outbreak tied to Evergreen Fresh Sprouts of Moyie Springs, Idaho: Washington 24, Montana four, and one each in Idaho, Minnesota, Oregon and Utah, three hospitalized, onsets July 7 to September 3. Last Saturday the state’s page counted 27 across the outbreak and CDC’s counted 22; this week Washington counts 24 at home, in nine counties from Clark to Whatcom, and CDC counts 32 in all. Washington tested sprouts from a grocery store and from a sick person’s home, and both matched the outbreak strain. The recall, initiated September 4, covers 1,290 four-ounce bags delivered between August 24 and September 2; 25 of the 32 illnesses began on or before August 23, so the recall could have reached, at most, the last seven. Evergreen’s three outbreaks since 2011 now sum to 76 people. I ran the onset arithmetic Thursday.

The seed lot behind the alfalfa sprout outbreak was sold at retail in 44 states for seven months, and the small bags carry no lot code. Berlin Seeds of Millersburg, Ohio recalled 950 pounds of alfalfa sprouting seed Wednesday, lot SAL2-24JB, sold in its store and on its website from February 13 to August 23 to customers in every state but six. The lot code is printed only on the five-pound and fifty-pound bags; the quarter-, half- and one-pound bags a home sprouter buys have none. One illness has been reported to the company. It is the sixth sprout or seed recall since August 22, after Everything Sprouts, its expansion, the seed supplier, Evergreen and Todd’s Seeds, and FDA’s advisory, updated Thursday to add Berlin, still does not name the supplier, International Sprout Holdings of Cookeville, Tennessee, which has been in FDA’s own Enforcement Report since September 10, where Phyllis Entis found it. The outbreak itself has not moved since September 1: 55 sick in 15 states, four hospitalized, 46 with E. coli, seven with Salmonella Agona, two with both. Between the two sprout outbreaks, 87 people are sick, and the alfalfa advisory names neither the seed supplier nor the second grower it says received the seed. I wrote it up Wednesday night.

FDA’s table opened three files this week and closed two, and 65 of the new sick people have no food named. On Wednesday the CORE table added reference 1421, E. coli O157:H7, 25 sick, product not yet identified, carrying a USDA-FSIS reference number, 2026-04, which is what a joint investigation looks like on paper; and reference 1417, Listeria monocytogenes, 40 sick, product not yet identified. The SalmonellaNewport file opened August 26 went from 111 to 116 and FDA began sampling something it does not name. The rows with no food now number ten and carry 645 people, up from 575 last Saturday. On Friday the Graziers file went on as reference 1419. Frozen blueberries and moringa moved to the closed list. CDC’s own dashboard, dated Wednesday, shows 32 active multistate investigations, 20 Salmonella, six E. coli, five Listeria and one Campylobacter, down from 38 the week before. I wrote about the shared reference number Thursday; it is the most informative thing on any of the three agencies’ pages.

Two outbreaks closed with 1,270 sick between them, and the most useful fact in each came out on the last day. CDC declared the frozen blueberry E. coli O145:H28 outbreak over Monday at 17 sick and six hospitalized in Florida and Georgia. FDA’s closing note disclosed, for the first time on any agency page, that a Publix-collected sample of GreenWise frozen blueberries tested positive for the outbreak strain while every sample FDA collected was negative, and that FDA never inspected the Chilean packer in person; it did a remote assessment and put the firm on an import alert. Some of the recalled bags carry best-by dates in 2028. On Thursday CDC closed the 2026 backyard poultry Salmonella outbreaks at 1,253 sick, 256 hospitalized and two dead, in Georgia and Washington, across 46 states and Puerto Rico, seven serotypes, 22 percent of the sick under five. Eight hatcheries were linked to the strains. Not one is named. Both are on Food Poison Journal, the blueberries and the poultry.

The files that did not move are the ones where people are still waiting. Happy Hens: 23 sick, seven hospitalized, a recall on the state’s page since September 11 and none on the farm’s, and the egg test results San Diego County called pending on September 4 are still unpublished three weeks later. Bear River: 21 confirmed cases in Cache and Box Elder Counties, Utah, no source, no update since September 1. Hepatitis A in frozen blood clams: 37 sick, 30 hospitalized in four states, illness onsets into June on CDC’s page while FDA’s alertstill says they ended in February, and neither page says the clams came from a country with no shellfish agreement; I wrote that up Thursday. Woods Hill Table in Concord, Massachusetts reopened Tuesday with 14 confirmed Salmonella cases, nine employees and five customers, and no source; I wrote about it when it closed. Pet turtles hold at 20 sick in 12 states, half under five. And Cyclospora sits where CDC left it on September 15: 19,883 confirmed, 1,064 hospitalized, two dead, 12,883 of them in the closed iceberg lettuce outbreak and roughly 7,000 with no food named, with the next update whenever the agency decides.

The recall ledger, kept as every week by Phyllis Entis at eFoodAlert. FSIS announced Tuesday that Star Meat Delivery of North Carolina recalled 167,639 pounds of raw pork, beef and goat sold nationwide under the A&D Foods label, produced without federal inspection and carrying a counterfeit USDA mark; no illnesses are reported, and I wrote about the penalty for that on Food Poison Journal. FSIS added two Friday: 4,596 pounds of Sempio Korean ginseng chicken stew imported without reinspection, in eight states, and 2,320 pounds of Fontanini pork sausage links for hard plastic, in four. On the FDA side, Elegant Foods’ Pecorino Toscano wheels were recalled for Listeria and GreenTripe’s Xkaliber raw dog food for Salmonella, both in the Enforcement Report only, where Entis found them; Wakefern recalled Wholesome Pantry organic applesauce for patulin, Galil Importing recalled Lior ground cinnamon for lead in six states, Deano’s Pastarecalled ravioli for undeclared lobster in Massachusetts, and Mezzetta recalled a lot of Golden Greek peperoncini for what its notice calls a pest contaminant, after a video went around. In Canada, R.J. King Fisheries recalled frozen cooked lobster meat for Staphylococcus aureus. FDA also posted two warning letters Tuesday to raw pet food makers: Viva Raw, whose raw chicken dog food carried Salmonella Enteritidis and whose puppy food carried Listeria, and Raaw Energy, whose products carried Salmonella, Listeria and Campylobacter; that one is on Food Poison Journal too. The rest of the week’s ledger is in her September 20 to 22 and September 24 posts.

The nominee to run FDA sat for her confirmation hearing Thursday and, by Food Safety Magazine’s account, could not name the office that investigates outbreaks. Dr. Heidi Overton went before the Senate HELP Committee. Senator Patty Murray told her the Human Foods Program has about 1,700 employees, 13 percent fewer than at the start of 2025, and Senator Tim Kaine that FDA has lost 4,500 people, about a fifth of its workforce, and employs fewer federal food inspectors than in any year since 2005. Asked about mandatory GRAS notifications, she said the agency would need more resources to review them. Food Safety Magazine reported that when Murray asked about outbreak response, the nominee was not able to identify the CORE Network as the team that does it. Meanwhile the fifteenth congressional letter of the Cyclospora season turned up: 25 Democrats, 17 House members and eight senators including Schumer, led by Rosa DeLauro, wrote Secretary Rollins on September 17 asking USDA to stop closing the Beltsville research center that does the parasite work. Published answers to fifteen letters: none. On infant formula, the shirts have been on 77 committee desks for nearly two weeks with one reply, and Target argued in federal court that it had no duty to ask a question about the formula it sold during an outbreak; I answered that Tuesday. A federal judge in Minnesota entered a consent decree against a Minneapolis warehouse that stored infant formula among rodents after 26 state inspections; nobody was charged. I also wrote this week on what a food safety NTSB would have had to record this month, on Tom Gremillion’s deterrence piece, and on why E. colidoes not care how you vote, and I gave two talks at the Aquatiq Food Forum in Oslo on Wednesday and Thursday. And the firm is hiring an associate and a paralegal; the ad ran last Saturday.

That’s the week — a raw milk cheese recalled seventy-nine days after the first illness, a sprout recall that could have reached seven of 32, a seed lot in 44 states with no code on the small bags, 65 new sick people with no food named, two outbreaks closed at 1,270, an egg farm still waiting on its test results, and a nominee who could not name CORE. The recalls are arriving. The outbreaks are already over. Check back next Saturday.

About Bill Marler

William “Bill” Marler has spent more than thirty years as a food safety lawyer and advocate—work that began with the 1993 Jack in the Box E. coli outbreak and has never really stopped since. In the years that followed, he has represented victims in nearly every major foodborne illness outbreak in the United States. That case, and the movement it launched, is the subject of the book “Poisoned” and the Emmy Award–winning Netflix documentary of the same name. Bill’s work has been profiled in The New Yorker (“A Bug in the System”), the Seattle Times (“30 years after the deadly E. coli outbreak, a Seattle attorney still fights for food safety”), the Washington Post (“He helped make burgers safer. Now he’s fighting food poisoning again”), and many others.

Dozens of times a year, Bill speaks to industry, regulators, and universities across the United States, Canada, Europe, Africa, China, and Australia about a simple idea: outbreaks are preventable. He has testified before Congress on the Food Safety Modernization Act and teaches food safety at institutions including the Harvard T.H. Chan School of Public Health. He writes regularly about food litigation and food safety at Marler Blog, and in 2009 he founded Food Safety News, which he continues to publish.