ByHeart made infant formula for about three and a half years. Production started on March 23, 2022. Everything it ever made was recalled in November 2025.

In that short life, parents told ByHeart about a lot of sick babies.

In one eleven-month stretch, from April 26, 2024 to March 14, 2025, ByHeart received 3,247 complaints. Two-thirds of them, 2,167, were about a baby’s health. That is about six or seven health complaints every day, from a company with roughly 1 percent of the U.S. formula market. (March 2025 FDA Inspection Report, p. 4.)

One of those complaints, on December 22, 2024, was a baby in the hospital with suspected botulism.

And through all of it, ByHeart’s co-founders, President Mia Funt and CEO Ron Belldegrun, kept doing what they did best. They marketed.

I wrote last week about the complaints ByHeart closed. This post puts those complaints side by side with what ByHeart and its founders were telling parents at the same time.

Salmonella. Campylobacter. E. coli.

Of the 2,167 health complaints in that window, ByHeart sent only 69 to BlendHouse Portland, the Oregon plant that filled every can and every stick pack. It called ten of those “sensitive.” The FDA investigator listed all ten. (March 2025 EIR, pp. 4–6.)

A daughter diagnosed with Salmonella.

A baby with a Salmonella infection that had entered the bloodstream.

Diaper rash and diarrhea, then Campylobacter in the stool.

Two more babies with stool positive for Salmonella.

A baby who got Salmonella after drinking two batches of ByHeart. ByHeart opened two investigations, one per batch.

Mucous, diarrhea and blood, with cultures confirming Salmonella.

Two weeks of diarrhea, positive for E. coli.

Diarrhea, a high fever, and blood and mucous in the stool, diagnosed as Salmonella.

Eight Salmonella files. One Campylobacter. One E. coli.

Every investigation ended the same way: no evidence the batch was contaminated, and all process controls effective.

Then ByHeart told the FDA it never received more than one sensitive complaint on any single batch. Of course it didn’t. It counted one batch at a time. When one sick baby drank two batches, that became two complaints of one each. A system built that way never puts two sick babies on the same page.

The Botulism Complaint

On December 22, 2024, a parent reported a hospitalized baby with suspected Clostridium botulinum. (November 2025 FDA Inspection Report, pp. 4, 7–8.)

ByHeart never got a lot number. It marked the complaint lost to follow-up. That is the entire record.

Nothing in the FDA’s files shows ByHeart called the hospital, called California’s Infant Botulism Treatment and Prevention Program, pulled retained samples and tested them for C. botulinum, or told the FDA.

Three months later, the FDA inspected Portland and reviewed ByHeart’s complaints for exactly the period that included December 22. The botulism complaint was not among the ten “sensitive” complaints ByHeart showed the investigator. The word botulism does not appear in that report.

When the FDA came back in November 2025 for the outbreak, Portland’s Director of Quality said the company had only ever received one complaint of a hospitalized baby with suspected botulism. One. As if that were reassuring.

Infant botulism is rare. Before ByHeart, no outbreak of it had ever been tied to infant formula. A parent calling a formula company to say the doctors think my baby has botulism is not a routine complaint. It is a fire alarm.

I have done this work for more than thirty years. When a parent reports a hospitalized baby with suspected botulism, a responsible formula company calls back until it reaches the family. It calls the hospital. It calls the health department and California’s infant botulism program. It pulls its retained samples from every lot that baby could have been fed and sends them to a lab that can find C. botulinum. It tells the FDA. And it puts that complaint in front of the next FDA investigator who walks in the door.

ByHeart wrote lost to follow-up.

It Started Before That

ByHeart’s trouble did not begin in 2024.

In August 2022, five months into production, a base mix made at ByHeart’s own plant tested positive for Cronobacter sakazakii. In October 2022, a finished batch tested positive. ByHeart first blamed the lab. The lab was right. (FDA Warning Letter.) In December 2022, ByHeart recalled five batches.

On August 30, 2023, the FDA sent ByHeart a warning letter for failing to have process controls covering every stage of production to keep microbes out of baby formula. (FDA Warning Letter.) That winter, FDA found a moldy water tank, a leaking roof and dead insects at the Reading, Pennsylvania plant. (Food Safety Magazine.)

Between July 31, 2023 and April 23, 2024, ByHeart logged three “serious” health complaints. All three babies were hospitalized. ByHeart called all three allergies. (April 2024 FDA Inspection Report, p. 4.)

The first baby in the botulism outbreak got sick on December 24, 2023. (FDA CORE file, FOIA 2026-4033.)

What Mia and Ron Were Saying

Now listen to the other side of the conversation, much of it in the founders’ own words, on video that we saved.

In June 2021, before the first can shipped, Belldegrun explained the his philosophy: “the more you control of the process, the more you control the quality of the formula.” At the March 2022 launch, Funt said ByHeart had “sourced every ingredient, and applied rigorous quality standards, from farm to formula.”

In August 2023, ByHeart hit Target shelves nationwide and announced it was the first U.S.-made formula to win the Clean Label Project’s Purity Award. The FDA warning letter came eight days later. ByHeart’s Target page kept telling parents about its “meticulous safety and quality standards every step of the way.”

On November 8, 2023, ten weeks after the warning letter, Funt told parents in a video that “the ingredients are tested before they even ever enter the facility.” None of that testing was for C. botulinum. Six weeks later, the first baby in the outbreak got sick.

On March 21, 2025, three months after the botulism complaint, and the same week FDA investigators were inside the Portland plant, Funt told parents ByHeart ran “over a thousand tests on our ingredients and throughout every step of the process.” Not one of those thousand tests was for C. botulinum.

On May 13, 2025, ByHeart launched the Anywhere Pack, single-serve stick packs for feeding on the go. Funt said “today’s parents aren’t putting their lives on pause when they have children.” Belldegrun touted the patented protein blend. To celebrate, ByHeart gave away a chance at one million airline miles to babies born between January and May 2025, and bragged that its retail footprint had grown more than 200 percent in a year.

In June 2025, Belldegrun wrote an op-ed urging formula makers to “build their own supply chains, and offer traceability,” and told parents that they “directly handpick[s] all of our ingredients.”

The influencer ads, as Kristin Beck wrote in Food Safety News, showed mother after mother mixing ByHeart with bottled or tap water in the park, in the car, on the sidewalk. Powdered formula is not sterile. Some of those ads were still running on Instagram after the recall.

On October 6, 2025, ByHeart rolled out a celebrity campaign with Lance Bass, Mayci Neely and Trae Young, under the banner “Certified-Clean & High Standards,” in nearly 12,000 stores. “We’re a company for parents, by parents,” Funt said.

By then, babies were already in the hospital. The main cluster of the outbreak began on August 9, 2025. And parents were calling about the two lots that would be recalled first. One said a pediatrician confirmed ByHeart caused her baby’s constipation. It went away when the baby came off ByHeart and came back when ByHeart was restarted. The FDA’s entry reads: “No root cause analysis was conducted.” Another reported a colicky, constipated baby with a swollen belly. Lost to follow-up. (November 2025 EIR, p. 4.)

Constipation is usually the first sign of infant botulism.

After the FDA Called

On November 7, 2025, the FDA told ByHeart that 13 babies with botulism had been fed its formula. The next day ByHeart recalled two lots and told the public: “No ByHeart product has tested positive for any contaminants.”

That was not true. ByHeart’s own finished product tested positive for Cronobacter sakazakii in 2022, and ByHeart recalled it.

Then Funt went on Instagram. “There is no reason to believe that infant formula can cause infant botulism,” she told parents. In a TikTok video she said “multiple regulatory bodies such as Codex have come to this conclusion.” The Codex Committee on Food Hygiene had in fact listed C. botulinum as a pathogen of concern in powdered infant formula. ByHeart later deleted the TikTok video.

On November 9, the founders signed a letter to families, “Mia & Ron, ByHeart Co-Founders,” saying “no U.S. or global infant formula company tests for Clostridium botulinum.” By then, the California Department of Public Health had already found C. botulinum in an opened can of ByHeart fed to a baby with botulism. The founders also told parents no one had found it in an unopened can. (McGraw Hill summary.) New York’s Wadsworth Center later did, in an unopened can from Arizona. (Marler Blog on the FDA file.)

On November 11, ByHeart recalled everything it had ever made. That same day, it posted on Instagram recommending Nara Organics formula to its customers, with a 20 percent discount code. Nara would have its own botulism outbreak in 2026, from the same milk supplier. ByHeart deleted that post too.

Then came refunds. ByHeart offered them only on purchases from August 1, 2025 on, and explained that it had acted “despite the initial notification to ByHeart on November 7, 2025.”

Initial notification. ByHeart was told about a hospitalized baby with suspected botulism on December 22, 2024. Parents complained on ByHeart’s Instagram about waiting weeks for even those refunds.

What It Adds Up To

The final count: 48 babies in 17 states, sick from December 24, 2023 to November 29, 2025. Every one hospitalized. Every one treated with BabyBIG. The FDA never found a root cause. (FDA CORE file.)

ByHeart never tested for C. botulinum before the outbreak. Its Portland food safety plan did not identify any biological hazard in the ingredients it received. (March 2025 EIR, pp. 18–19.) The U.S. Senate HELP Committee wrote to Belldegrun and Funt by name in December 2025. And now the two of them run MeridianMade, which has agreed to buy Perrigo’s infant formula business.

Here is the whole story in two columns.

On one side: thousands of health complaints. Babies with Salmonella in their blood. Campylobacter. E. coli. Three hospitalized babies called allergies. A botulism report closed as lost to follow-up.

On the other: farm to formula, a thousand tests, handpicked ingredients, certified clean, a million airline miles, celebrities, and, once babies were in the hospital, “no reason to believe that infant formula can cause infant botulism.”

Complaint systems exist for one reason: to hear the warning before the outbreak. Parents were warning ByHeart. Mia Funt and Ron Belldegrun kept selling.

That is why I keep pushing for a rule requiring C. botulinum testing of powdered infant formula and its dairy ingredients. Parents should not have to depend on whether a company is listening.

Sources

FDA, Clostridium botulinum/Infant Formula/Nov 2025 – CARA ID 1350: https://www.fda.gov/media/192722/download

FDA, Establishment Inspection Report, BlendHouse Portland (Apr. 22–26, 2024): https://www.marlerblog.com/files/2026/11/2-2024-04-26-Portland-EIR-2024-04-22-to-04-26-FEI-3013670080.pdf

FDA, Establishment Inspection Report Amendment 1, BlendHouse Portland (Mar. 17–27, 2025): https://www.marlerblog.com/files/2026/11/4-2025-03-27-Portland-EIR-2025-03-17-to-03-27-Amend-1-FEI-3013670080.pdf

FDA, Establishment Inspection Report Amendment 1, BlendHouse Portland (Nov. 11, 2025 – Jan. 23, 2026): https://www.marlerblog.com/files/2026/11/1-final_applied_blendhouse_portland_or_3013670080_eir_11-11-25_-1-23-26_amendment_1_br_nj_vlo.508.pdf

FDA, Warning Letter to ByHeart, Inc., MARCS-CMS 653854 (Aug. 30, 2023): https://www.marlerblog.com/files/2025/11/Reading-Warning-ByHeart-Inc.-653854-08302023-FDA.pdf

ByHeart, Voluntary Recall of Five Batches (Dec. 2022), posted by FDA: https://www.fda.gov/safety/recalls-market-withdrawals-safety-alerts/byheart-issues-voluntary-recall-five-batches-its-infant-formula-because-possible-health-risk

FDA CORE records, FOIA Request No. 2026-4033 (redacted): https://www.marlerblog.com/files/2026/08/FOIA-2026-4033_Redacted.pdf

Marler Blog, Eleven Months Early: ByHeart’s Botulism Complaint (Nov. 20, 2026): https://www.marlerblog.com/case-news/eleven-months-early-byhearts-botulism-complaint-and-all-the-other-sick-baby-complaints-it-closed/

Marler Blog, FDA’s Own ByHeart Botulism File (Aug. 31, 2026): https://www.marlerblog.com/case-news/fdas-own-byheart-botulism-file-48-babies-17-strains-a-death-complaint-no-root-cause-and-six-company-names-blacked-out/

Food Safety Magazine, ByHeart Finds Widespread Contamination; FDA Publishes Inspection Reports: https://www.food-safety.com/articles/10905-byheart-finds-widespread-contamination-in-infant-formula-as-botulism-outbreak-grows-fda-publishes-inspection-reports

Nutrition Insight, ByHeart brings protein-enhanced infant formula to US market (Mar. 24, 2022): https://www.nutritioninsight.com/news/byheart-brings-protein-enhanced-infant-formula-from-scratch-to-us-market.html

Nutritional Outlook, ByHeart makes retail debut at Target (Aug. 22, 2023): https://nutritionaloutlook.com/view/byheart-infant-formula-makes-retail-debut-at-target-locations-nationwide

Target, ByHeart Whole Nutrition Infant Formula product page: https://www.target.com/p/-/A-87956594

Business Wire, ByHeart Launches Anywhere Pack (May 13, 2025): https://www.businesswire.com/news/home/20250513513492/en

Business Wire, ByHeart campaign with Lance Bass, Mayci Neely and Trae Young (Oct. 6, 2025): https://www.businesswire.com/news/home/20251006647323/en

Food Safety News, Years before outbreak, ByHeart’s marketing encouraged dangerous practices (Jan. 28, 2026): https://www.foodsafetynews.com/2026/01/years-before-outbreak-byhearts-marketing-encouraged-dangerous-practices/

ByHeart influencer reel, Instagram: https://www.instagram.com/reel/DN3cpuvQBFY/

ByHeart, Voluntary Recall of Two Batches (Nov. 8, 2025), via Oregon Dept. of Agriculture: https://apps.oregon.gov/oregon-newsroom/OR/ODA/Posts/Post/In-Response-to-a-Broader-FDA-Investigation-ByHeart-Initiates-a-Voluntary-Recall-of-Two-Batches-of-Infant-Formula

McGraw Hill, Infants Hospitalized after Baby Formula Recall (Dec. 2025): https://www.mheducation.com/highered/blog/2025/12/infants-hospitalized-after-baby-formula-recall

ABC News, CDC says 51 infants hospitalized in botulism outbreak: https://abcnews.go.com/family/story/cdc-2-infants-hospitalized-botulism-outbreak-byheart-expands-127417947

Consumer Reports, ByHeart recall expanded to all formula ever produced: https://consumerreports.org/babies-kids/baby-product-recalls/byheart-formula-recall-expanded-infant-botulism-a2906529472/

MedicineNet, FDA links ByHeart outbreak to milk powder ingredient: https://www.medicinenet.com/fda_byheart_baby_formula_outbreak_milk_powder/news.htm

U.S. Senate HELP Committee, Letter to ByHeart (Dec. 18, 2025): https://www.help.senate.gov/imo/media/doc/letter_to_byheart_on_infant_formula_recall.pdf

ByHeart and founders’ videos and posts: Belldegrun (June 2021; June 2025 op-ed and video); Funt (Mar. 2022 launch; Nov. 8, 2023; Mar. 21, 2025; Instagram and TikTok, Nov. 2025); “Mia & Ron” letter (Nov. 9, 2025); Nara Instagram post (Nov. 11, 2025): 

Boursorama/Reuters, Perrigo to sell infant formula business to MeridianMade: https://www.boursorama.com/bourse/actualites/perrigo-va-ceder-son-activite-de-lait-infantile-a-meridianmade-pour-un-montant-pouvant-atteindre-175-millions-de-dollars-a97ddb1ee25f905b63f75389a68db0f1

Lisa Fletcher and her colleagues Andrea Nejman and Nathan Aaron at Spotlight on America, the investigative unit of The National Desk, aired a piece on November 19 that every member of Congress should watch before December 11: “Cyclospora sickened 12,000. The FDA delayed a rule that could have limited the spread.” It ran on Sinclair stations across the country. I was interviewed for it, and I am grateful she took the time to get it right.

What Lisa reported

The summer’s outbreak, tied to shredded iceberg lettuce from Taylor Farms de Mexico, ended at 12,883 confirmed illnesses, 570 hospitalizations and two deaths in 21 states. It is the largest Cyclospora outbreak this country has recorded.

Lisa’s piece walks through how we got here. Congress told FDA in 2010, in the Food Safety Modernization Act, to write a rule requiring companies that handle high-risk foods to keep traceability records and hand them over within 24 hours when an outbreak investigation asks. FDA took until 2022 to finish the rule. Compliance was set for January 2026. Last year FDA pushed that back 30 months, citing industry concern about the timeline. The outbreak began five months after the date the rule was supposed to be in force.

She interviewed Dr. Barbara Kowalcyk of George Washington University’s Institute for Food Safety and Nutrition Security, who did not soften it. “The traceability system isn’t working,” she told Lisa. “It’s broken.” Dr. Kowalcyk’s point was that a producer should be able to give investigators the location of every facility it uses in minutes, and that when it cannot, FDA’s investigation stalls.

Lisa also reported something that deserves more attention than it has received. Infant formula is not on the list of foods the rule covers, even after a botulism outbreak that put more than 50 babies in the hospital. Her team asked FDA whether it would reconsider. At publication, FDA had not answered.

What I told her is what I have said all year. The technology exists. The objection is cost. And while the tracing dragged this summer, thousands more people ate the lettuce.

How the rule was extended, then defunded

Lisa’s piece covers FDA’s 30-month extension. The rest of the story is in Congress. I laid it out in an October Publisher’s Platform, and the short version goes like this.

The push began before the extension. In March 2024, Rep. Scott Franklin, a Florida Republican and vice chair of the House subcommittee that funds FDA, introduced H.R. 7563. Rep. Sanford Bishop of Georgia, the subcommittee’s senior Democrat, was the lead cosponsor, and Rep. Jimmy Panetta, a California Democrat, signed on. The bill would have relieved retailers, restaurants and warehouses of keeping and passing along lot codes, which are the heart of the rule. It never got a vote. Its language moved into the House agriculture appropriations bills written by the subcommittee’s chairman, Rep. Andy Harris of Maryland. This was a bipartisan effort.

On March 20, 2025, FDA announced the extension to July 20, 2028. Its reason was that companies ready for 2026 depended on supply chain partners who were not. The agency said it did not intend to change what the rule requires. The grocery trade associations applauded.

Then Congress took the decision out of FDA’s hands. On November 12, 2025, the bill that ended the government shutdown, H.R. 5371, carried Section 780. It bars FDA from spending appropriated money to administer or enforce the rule before July 20, 2028, and directs the agency to meet with industry on “flexibilities” for lot-level tracking. That is the difference between extended and defunded. An agency can reverse its own extension. It cannot spend money Congress has forbidden it to spend.

Section 780 came from the House. Senate appropriators had said only that enforcement must not slip past July 2028. The trade groups that asked for the delay are named in their own lobbying filings.

It is not over. The House passed its fiscal 2027 agriculture bill on June 4 by 213 to 210 with the rider in it again. The Senate has not passed its version. The rider lives or dies in whatever funding bill passes before the stopgap expires on December 11.

Ten things, short run and long run

Traceability makes an outbreak smaller. It does not keep the parasite off the lettuce. Both matter, which is why this list has two halves. It draws on what I said FDA’s ten commitments left out and on a summer of writing about this outbreak.

Before the 2027 season

1. Congress: leave Section 780 out. It costs nothing, needs no rulemaking and can be done by December 11.

2. Industry: comply with the rule now. Nothing in Section 780 stops a grower, processor or buyer from keeping lot-level records today. Buyers should write it into purchase orders. Restaurant chains and grocers have more leverage over a supplier than FDA ever will.

3. Importers: do what the law already requires. The Foreign Supplier Verification rule obliges a U.S. importer to verify that its foreign supplier controls known or reasonably foreseeable hazards. Cyclospora on imported produce has been foreseeable since the 1990s. FDA should enforce that duty and publish what it finds.

4. Test the water for the parasite itself. Generic E. coli testing does not find Cyclospora. FDA has a validated method. Growing and processing water should be tested for the organism, starting in regions with an outbreak history, and the results should be public.

5. Verify sanitation where the lettuce is grown and cut. Humans are the only known host. Every oocyst came from a person. Toilets, handwashing, worker health and wastewater kept away from fields, canals and product should be checked by auditors who are not paid to pass the farm.

6. Publish the findings, and agree on data sharing now. FDA has promised an outbreak investigation report. It needs a date well before the season and it needs the water data. Taylor Farms should publish its own root cause analysis. And industry, FDA and CDC should have a standing agreement that shipping, sourcing and test records move within hours, not weeks.

For the long run

7. Count the sick again. CDC cut FoodNet’s required surveillance from eight pathogens to two in 2025, and Cyclospora was one of the six dropped. Put it back, and fund the state and local labs that find outbreaks in the first place.

8. Rebuild the tripwire. USDA’s Microbiological Data Program sampled produce and triggered recalls until it was defunded in 2012. Nothing replaced it.

9. Set enforceable water standards. The pre-harvest agricultural water rule relies on assessments, not numbers. Raw, high-risk produce needs numeric microbial standards and routine testing, with the same standard applied to imports.

10. Fund the science and keep the tools. There is no kill step for Cyclospora on ready-to-eat produce and no way yet to fingerprint it the way we do Salmonella. That is a research problem, and research needs money and the people who know how to do it. Meanwhile, stop treating every FSMA requirement as a burden to be removed.

Why I keep making lists

In 2002, after the ConAgra ground beef recall, I wrote an op-ed asking the beef industry to put me out of business. It took a while, but it largely did. E. coli cases from hamburger, once most of my practice, are now close to none. I would be glad to say the same about lettuce.

None of the ten is a lawsuit. Six of them can be done before the first head of lettuce is cut for next season. Reporting like Lisa’s is how the public learns that the tools were written, finished and then set aside. Thank you, Lisa, Andrea and Nathan.

This week the USDA’s Food Safety and Inspection Service (FSIS) added a line to its outbreak investigations table: investigation 2027-01, October 2026, E. coli O157:H7, “Ground Beef (suspect),” active. As Food Safety News reported, there is no case count, no list of states, no named product, and no recall or public health alert. I have asked FSIS to publish what it knows, and I still think it should.

Look at that entry another way, though. It is the first ground beef E. coli O157:H7 outbreak on the FSIS table since November 2024, when Wolverine Packing recalled 167,277 pounds after Minnesota tied 15 illnesses to its ground beef. Two years between hamburger outbreaks. When I started this work in 1993, that would have been unimaginable. It is worth remembering how we got here, who deserves the credit, and what the lesson is for the foods making people sick today.

Where did the progress against E. coli in hamburger start?

It started with the Jack in the Box outbreak of 1992 and 1993: more than 600 people sick and four children dead. President Clinton and his Secretary of Agriculture, Mike Espy, had just taken office. Espy’s USDA launched a public education campaign, required safe handling labels on raw meat and poultry, and in the spring of 1994 recruited Mike Taylor from FDA to run FSIS.

On September 29, 1994, six weeks into the job, Taylor stood before the American Meat Institute convention in San Francisco and said that FSIS considered raw ground beef contaminated with E. coli O157:H7 to be adulterated under the Federal Meat Inspection Act. By his own account, the room went silent. FSIS began testing ground beef on October 17, 1994. Industry groups sued to stop the testing and lost. Texas Food Industry Association v. Espy remains my favorite case.

On July 6, 1996, with parents of children who died in 1993 beside him, President Clinton used his weekly radio address to announce the HACCP and Pathogen Reduction rule, the first real overhaul of meat and poultry inspection in 90 years, finished under Secretary Dan Glickman and Taylor.

Clinton, Espy, Glickman and Taylor took a political risk on behalf of children they would never meet. They were told it could not be done and that it was unfair to hold a company responsible for a bacterium. They did it anyway, and they deserve to be remembered for it.

What did the beef industry and FSIS do?

The industry fought the policy in court and then, to its credit, went to work. After 1993, and especially after ConAgra’s 19.6-million-pound recall in 2002, beef companies adopted test-and-hold sampling, steam pasteurization, organic acid rinses and hot water washes. They dropped suppliers that could not control the pathogen. They began meeting every year to share safety data with competitors and put about $30 million into beef safety research. Buyers such as Costco and Jack in the Box wrote those interventions into their purchasing specifications. Food Safety News told that story on the outbreak’s twentieth anniversary.

FSIS kept testing, moved upstream to beef trim in 2007, and in 2011 declared six more Shiga toxin-producing E. coli to be adulterants, after a petition I filed. Career inspectors and scientists carried that work through administrations of both parties.

How much has E. coli in ground beef declined?

The numbers come from three places, and they agree.

FSIS testing. By the end of 2010, the share of ground beef samples positive for E. coli O157:H7 was down 72 percent from 2000. In 2017, FSIS found 7 positives in 9,761 samples from federal plants, or 0.07 percent. In fiscal year 2024, 27 of 11,563 ground beef samples were positive for any of the seven adulterant strains, or 0.23 percent.

CDC surveillance. CDC reported that E. coli O157 infections fell 42 percent from 1996 to 2004. By 2010 the rate was 0.9 illnesses per 100,000 people, meeting the national health objective of no more than one.

My own practice. From 1993 to 2002, about 90 percent of my firm’s revenue came from E. coli O157:H7 in beef. By 2003 it was about 5 percent. Today it is close to zero. In a 2002 op-ed written after the ConAgra recall, I asked the beef industry to put me out of business. In hamburger, it very nearly has. 

That is the best thing to happen in food safety in my lifetime.

What about this outbreak?

Progress is not perfection. Cattle still carry E. coli O157:H7, grinding still spreads it, and ground beef still needs to reach 160 degrees. A rare outbreak is a reason for more openness, not less. If FSIS has traced illnesses to a lot, it should name the establishment and the product. If it has not, it should say how many people are sick and where, so the public can judge the word “suspect” for itself. The system that drove this pathogen down worked because test results were public and consequences were real.

How do we apply the hamburger lesson to other foods?

The beef story has four parts, and none of them is unique to beef.

A clear legal line. E. coli O157:H7 in ground beef is illegal. Salmonella in raw chicken is not. FSIS denied my petition to treat outbreak strains of Salmonella as adulterants and in 2025 withdrew its own proposed Salmonella framework for poultry. CDC said in 2011 that the measures that reduced E. coli O157 needed to be applied more broadly to Salmonella. That is still true.

Government testing, with results the public can see. Companies began testing seriously once FSIS did. Verification sampling of leafy greens, sprouts and infant formula ingredients would have the same effect.

Accountability at the source. For beef that meant the slaughterhouse, not the kitchen. For romaine and spinach, it means the field, the irrigation water and the cattle operation next door.

Safety as a shared project. Beef companies decided that food safety was not a place to compete and shared what worked. Buyers enforced it through their specifications. Produce buyers have the same power.

Notice what this did not require. Taylor did not wait for a new statute. He applied a law already on the books to a pathogen that was killing children, a court agreed, and a president backed the rule that followed.

If investigation 2027-01 turns out to be ground beef, the answer is the system doing what it was built to do: find it, name it, recall it, fix it. Then do for chicken, lettuce and sprouts what was done for hamburger. 

I would be glad to be out of those businesses too.

Saturday, November 21, 2026

Last Saturday’s edition was about where the E. coli came from. This week (week 20) was about who knew what, and who got the blame. FDA’s inspection report for ByHeart’s Portland plant shows the company was told about a hospitalized baby with suspected botulism eleven months before the recall and closed the file. Nara Organics told two courts the parents did it. Santa Cruz County reopened Capitola Beach and pointed at birds. The Little Wicomico Oyster Company sued the state that closed its river. FDA posted closing reports on three Salmonella outbreaks that sickened 261 people and blacked out the firms, one of the foods and the states. And the biscuit maker at the top of a four-company Listeria recall chain still has not said a word.

ByHeart was told about a hospitalized baby with suspected botulism on December 22, 2024, eleven months before it recalled anything, and closed the complaint as lost to follow-up. It is in FDA’s newly released inspection report for ByHeart’s Portland, Oregon plant, which filled the cans and stick packs parents bought. FDA’s CORE inspection, opened over reports of Clostridium botulinum in infants, ran from November 11, 2025 to January 23, 2026, reviewed the full batch records for the two lots recalled first, made April 23 and May 6, 2025, pulled six ten-can samples for its Arkansas laboratory, results redacted, and issued no Form 483. When the inspection opened, the plant’s Director of Quality told FDA the company had only ever received one complaint of a hospitalization with suspected botulism: CN-005563, December 22, 2024, no lot number obtained, no further investigation recorded. It is not among the ten sensitive complaints FDA reviewed at the same plant in March 2025, every one of them Salmonella, Campylobacter or E. coli; the word botulism is not in that report. The release testing FDA reviewed covered indicator organisms, Salmonella, Cronobacter, nutrients and heavy metals, and the environmental program looked for Salmonella,Cronobacter, Enterobacteriaceae, yeast and mold; neither looked for the organism that put 48 babies in the hospital in 17 states. The complaints logged for the recalled lots include a pediatrician-confirmed report of constipation that stopped when the formula was withdrawn and returned when it was resumed, closed with no root cause analysis. A complaint system that routes thousands of health complaints through headquarters and closes the serious ones one batch at a time is not built to see an outbreak. It did not see this one; FDA did. And watch what is happening to ByHeart itself. The brand is being quietly retired: the word that it is discontinued and will not relaunch came October 29 not from ByHeart but from a spokesperson for MeridianMade, the new contract manufacturer run by the same brother and sister, Ron Belldegrun and Mia Funt, who ran ByHeart. That new company is buying Perrigo’s store-brand infant formula business, a $380 million-a-year operation, for $40 million at closing and up to $135 million more through 2030, expected to close in the first half of 2027. Perrigo’s own release describes the buyer as an American contract manufacturing platform dedicated to infant nutrition with spray drying, dry blending and packaging capabilities, which is what ByHeart’s Allerton and Portland plants did; Reuters connected the two companies only through the people running them. The company whose formula put 48 babies in the hospital is being folded away under one name while its founders, under another, pick up a formula business many times its size. Nobody has said whether MeridianMade is running ByHeart’s plants, what becomes of ByHeart’s obligations to those 48 families, or whether the Perrigo plants will test for the organism ByHeart never tested for.

Nara Organics told two federal courts that the parents of babies paralyzed by its formula were the ones at fault. The company’s answers to the two complaints our firm filed, one in Washington and one in Pennsylvania, plead 61 and 59 affirmative defenses: that the parents misused or altered the formula, used it in contravention of warnings that did not exist, knowingly, voluntarily and willfully assumed the risk, were negligent and careless, failed to mitigate, failed to preserve evidence, and that the injuries were not foreseeable given the state of scientific knowledge. In the same filing Nara cross-claimed against Organic West Milk and Dairy Farmers of America, calling their milk powder the primary cause, the one defense in the stack that agrees with FDA: the June 2026 outbreak ran on the same milk, from the same supplier and the same Fallon, Nevada dryer, as ByHeart’s. One of the two infants went home from eleven days in the hospital on a feeding tube; the other, two months old, lost the ability to hold up his head or feed and came close to a breathing tube. Those are the parents Nara says assumed the risk.

Santa Cruz County reopened Capitola Beach on Thursday and said the preliminary findings point to birds. The county’s November 19 release puts the outbreak at nine children with E. coli O157:H7, three hospitalized and one with hemolytic uremic syndrome, up from two hospitalizations on November 11. CDC found the outbreak strain in water and sediment from Soquel Creek Lagoon and in sediment at the flume outlet by the beach, and the county says birds, not a sewer spill or other human source, with final results promised next week. I wrote Friday night about why that answer is incomplete: birds carry O157; ruminants produce it. The county’s own 2009 DNA fingerprinting study of Soquel Creek credited birds with 36 to 64 percent of the indicator E. coli and cows with zero, while staff noted cattle along Soquel San Jose Road. Marler Clark represents the family of the child who developed acute kidney failure, has filed a notice of claim with the City of Capitola and has records requests out to the city, the county and the state.

San Luis Obispo County confirmed a secondary case eight days after it closed the Brookshire Farms file. The county said November 14 that a person who never visited the farm was infected person to person at a San Luis Obispo preschool by one of the five confirmed October cases, developed HUS, was hospitalized and is recovering at home; the affected classrooms closed Monday. That makes four HUS cases and one death, the 18-month-old the county confirmed November 10, in an outbreak closed November 6 with no source found. I wrote about what a secondary case is and the Almquist case our firm tried on those facts. Will County found its baby goat on November 10, and Dallas County has said nothing about the State Fair of Texas petting zoo cases since October 28. The firm represents two Brookshire families, a culture-positive child in the Bengtson’s outbreak and three Dallas families, and has put all three operators on notice.

Ohio University’s count held at seventeen for the first time, and the health department still has no source. The Athens City-County Health Department’s November 18 and November 20 releases put the Athens-campus STEC cluster at 17 lab-confirmed students and four suspected cases that will likely stay suspected, onsets November 5 to 10, one hospitalized and released, no HUS and no exposure named. A six-day onset window in one student body is a point source, as I said at sixteen; the next update is 10 a.m. Monday, three days before students scatter for Thanksgiving. In Massachusetts, Tufts told its campus Thursday night of five confirmed illnesses suspected to be linked to the Carmichael Dining Center; the dining hall stayed open and nobody has said what kind of E. coli it is.

The oyster company sued the state while its river failed a fourth round of tests. Little Wicomico Oyster Company filed suit November 15 in Northumberland County Circuit Court against the Virginia Marine Resources Commission and the Marine Police for more than $9.1 million, over the October 9 closure of about 75 million seed oysters, 16 million of them now dead; a Lancaster County judge has let the company tend seed stock under Marine Police observation pending a December 11 hearing. I have six questions for that hearing. The same week VDH reported that all ten meat samples taken from the company’s cages November 9, from five leases, one more than before, were positive for Salmonella: four rounds in five weeks, zero clean, three clean rounds needed to reopen, and the fifth round collected November 16 still pending as of VDH’s November 20 update. CDC’s count stands at 28 sick in 13 states, five hospitalized, as of October 30. Marler Clark has two suits in this outbreak, in Denver and in Richmond, and no role in the company’s case against the state.

CDC and FDA closed the Graziers raw milk cheese outbreak at 18 sick in ten states, nine hospitalized and three children with kidney failure. The November 13 final update puts onsets from July 7 to September 20, four days before Sierra Nevada Cheese Company’s recall, with a median age of four; nothing in it addresses the 60-day aging claim on the label, and I said so. The firm has filed Tennant v. Sierra Nevada Cheese Co. in the Western District of Michigan for a child who developed HUS. In Texas, the Amarillo Area Public Health District warned Thursday of six cryptosporidiosis cases tied to raw milk, three of them in people who drank milk from Wethington Farms of Nazareth, sold from a parking lot off I-27: no recall, no listing among the state’s licensed raw-milk dairies, and a herd-share model Texas does not treat as a sale.

FDA’s table shrank to ten rows and the no-food ledger to 195. The November 18 update moved the Graziers file to ended, closed two Cyclospora files and adjusted two counts, Salmonella Enteritidis file #1425 from 35 to 33 and Javiana file #1424 from 70 to 73, the latter with inspection and sampling under way. Six rows still name no food, 195 sick, 183 of them in files still open. CDC’s dashboard moved from 26 to 27 active multistate investigations on November 18, adding three E. coli files and dropping two Campylobacter files, with a notice for none of them. And FSIS posted its first fiscal-2027 row: investigation 2027-01, E. coliO157:H7, ground beef (suspect), active, with no case count, no states, no recall and no public health alert, as Food Safety News reported, and less information than the FDA rows I complain about.

Three closing reports named a vegetable and blacked out everyone who grew, packed or shipped it. The abstracts FDA posted November 18 cover file #1402 (Salmonella I 4,[5],12:i:-, 43 sick, romaine), #1387 (Oranienburg, 99) and #1403 (Newport, 119): 261 people, 47 hospitalized. By my count about 65 redactions: firm names as trade secrets under exemption 4; the food in the Newport file, the commodity in the Oranienburg title and the states in the romaine file as deliberative under exemption 5. A vegetable is not a deliberation. The romaine abstract shows CDC named romaine the leading vehicle on August 7, FDA posted the file August 12 as not yet identified, inspected the processor from August 18 to September 1 and issued a Form 483, and told the public on November 4, 89 days later. I have FOIA requests in for that 483 and for the Oranienburg Form 4056. On Monday, Senators Murray, Shaheen and Baldwin asked HHS Inspector General Thomas March Bell to investigate how staffing and policy changes at FDA and CDC shaped the response to the 12,883-case Cyclospora outbreak. Their letter cites a first inspection of the Taylor Farms plant in Guanajuato a month after the July 17 recall and the first since 2019, a CDC Cyclospora laboratory cut from eleven people to three, and the lowest FDA inspection count since 2011. He should say yes.

The recall ledger is four companies deep in biscuits, and the biscuit maker still has not spoken.Fresh & Ready Foods expanded its recall November 17 to Fresh To You biscuit sandwiches and Mmm! Mealsbiscuit-and-gravy trays made in Longview, Texas and sold through workplace vending and micro-markets in seven Southern states from October 15 to November 11, citing the expansion of C.H. Guenther & Son’s biscuit recall; FSIS updated its public health alert November 19 for additional products. Grand Strand Sandwich Company of Longs, South Carolina recalled four Lunch Box biscuit sandwiches November 19, distributed since March 2 to convenience stores, vending machines and prison canteens in the Carolinas and Virginia, after Dot Foods, not Guenther, told it on November 11 that the biscuits might be contaminated. Phyllis Entis at eFoodAlert also caught an FDA enforcement-report entry for Seattle’s Favorites LLC, three breakfast sandwiches pulled for Listeria with no public notice attached, and another for BWI Companies’ one-pound Broccoli Seeds Green Sprouting Calabrese, lot W4105, the seed lot FDA matched to the Evergreen sprout outbreak, now in its fourth retail recall. C.H. Guenther & Son has no notice on FDA’s recall page and no statement I can find. Elsewhere, ALDI pulled two Priano protein tortellonis in Alabama, Georgia and the Carolinas for Clostridium botulinum risk after a temperature failure, and Larbi Inc. recalled three lots of Herb To Body yellow dock root for Salmonella. Nothing from FSIS could be verified first-hand this week; the agency’s site would not load.

Three dates matter before the new Congress: December 6, December 11 and January 3. FDA’s docket on lot-level traceability closes December 6; the continuing resolution that carries Section 780, the rider that bars enforcement of the Food Traceability Rule before July 2028 and tells FDA to find flexibilities in lot codes, expires December 11; the new Congress sits January 3. I laid out four tracks last Saturday, from the docket comment to a short bill that puts the lot code and the date in statute. Who killed the MDP, the program that found Listeria before anyone got sick, is here. Project 2027 comments are due November 30, and the petitions, all of them, go to the dockets in December.

That’s the week — a botulism complaint closed eleven months before the recall; a formula company that pleaded the parents assumed the risk; a beach reopened on a theory about birds; a preschooler with kidney failure in a file the county had closed; seventeen students in Ohio and five in Massachusetts with no source; a river that has failed four rounds of tests and a company suing the state that closed it; 261 people in three FDA reports with the names blacked out; and a biscuit recall four companies deep with the first company silent. Next week: Athens on Monday, Santa Cruz County’s final lab results, Virginia’s fifth round, December 6 on the docket, and Thanksgiving. 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.

Capitola Beach reopened on Thursday, November 19, 2026, nine days after Santa Cruz County closed it because nine children had E. coli O157:H7 infections tied to the beach and Soquel Creek. Three of the children were hospitalized. One developed hemolytic uremic syndrome, the kidney failure that makes this bug so dangerous for kids.

The county now says its preliminary findings point to birds, not a sewage spill or another human source. I do not think that is a complete answer. Birds can carry E. coli O157. They do not make it. The animals that do are cattle and other ruminants, and the public has not been told whether anyone looked for them.

A disclosure first: my firm represents the family of one of the nine children, a child who developed acute kidney failure. We have filed a notice of claim with the City of Capitola and sent public records requests to the city, the county and the state.

What the county has said

The county’s November 19 press release, also reported by Lookout Santa Cruz, Good Times and KSBW, says this:

  • CDC detected the outbreak strain in water and sediment from Soquel Creek Lagoon and in sediment at the flume outlet on the beach.
  • The closure was lifted after a review of additional results from samples collected November 4 and a consultation with the California Department of Public Health.
  • Preliminary findings “suggest birds as the source associated with the outbreak,” and not sewage or another human source.
  • A health advisory remains. Health Officer Dr. Lisa Hernandez said that with waves, shifting sand and birds moving around, the bacteria could be anywhere on the beach.
  • Final laboratory results are expected next week and will be posted on the county’s Environmental Health website.

The release does not say how the county reached birds. It does not say that bird droppings were tested, which birds are meant, or whether anyone sampled livestock or deer upstream. It also does not say how the children are. On November 11 the health officer told KSBW there were two known hospitalizations, and the station reported that everyone had recovered. The count is now three hospitalized and one case of HUS.

The county adds that the risk from E. coli shed by birds exists at any beach where birds congregate (sounds more like a weak defense that facts). Birds congregate on every beach in California. Outbreaks like this one are rare. Something was different at Capitola.

Where E. coli O157 lives

The reservoir for E. coli O157:H7 is the gut of healthy ruminants. Cattle come first, followed by sheep, goats and deer. The animals are not sick, and they shed the bacteria in manure. That is the settled science, laid out in this review of the animal reservoir, and it is why I spent the 1990s in the hamburger business and the years since in lettuce grown downhill from cattle.

Birds are a different matter. A yearlong University of California, Davis study of wild birds on California farmland found E. coli O157:H7 in half a percent of birds or fewer. In controlled experiments, starlings housed with calves that were shedding O157 picked it up within a day and passed it to other calves just as fast. A bird is a vehicle. A gull or a duck carrying O157 got it somewhere, and that somewhere is usually manure.

Has a bird ever been blamed before?

In 33 years of E. coli cases, I have not had one traced to birds. I went looking and found three investigations where birds came up.

  1. Gloucestershire, England, 2002. Two young siblings and their mother fell ill on a beef cattle estate. Their isolates matched one from rook droppings collected at the cattle feed troughs, while the cattle samples tested negative. The authors called birds possible “intermediaries.”
  2. Battle Ground Lake, Washington, 1999. Children were sickened swimming, and investigators isolated E. coli O157:H7 from duck feces as well as from the lake.
  3. Utah, 2023. Thirteen children were infected through untreated municipal irrigation water. Seven were hospitalized and two developed HUS. CDC’s report says source tracking pointed to both birds and ruminants.

None of the three shows a bird as the place the bug began. Finding O157 in a bird tells you the bird was carrying it.

The closer parallel is in Northern California. In 2016, four children aged one to three were infected with E. coliO157, and the common exposure was a stream running past a city park playground, in a Marin County investigation. One developed HUS. E. coli O157 was in the stream four weeks running, and the closest match came from deer scat a mile and a half upstream. In Oregon in 2011, an O157 outbreak from strawberries was traced to deer in the fields.

What the Water Board’s own record shows

The county’s bird theory has a history. When the Central Coast Regional Water Quality Control Board wrote its bacteria cleanup plan for Soquel Creek, its project report included DNA fingerprinting of E. coli from five stations on the creek and Noble Gulch. Birds were the largest identified source at every one, from 36 to 64 percent. Cows were zero.

Three things about that table matter now.

  1. It counted ordinary indicator E. coli, the kind that shows feces are present. It did not look for O157. Knowing which animal leaves the most droppings in a creek does not tell you which animal left the pathogen.
  2. The second-largest category at several stations was wildlife, which the report defined as “raccoon, deer, and opossum.” It ran from 7 to 31 percent. Deer are ruminants, and they carry O157.
  3. Staff saw cattle along Soquel San Jose Road, upstream of the impaired stretch of the creek, and on Cunnison Lane near Noble Gulch, along with horses and corrals. The report says cattle might have been detected with more wet-season sampling, and that several data sets had too few isolates to be reliable.

The Board’s staff concluded at the time that farm animals were not a source of indicator bacteria in the creek or lagoon. That was nearly two decades ago, and it answered a different question. The question today is where one strain of O157 came from in the fall of 2026.

Why “birds” is a convenient answer

In the Water Board’s 2008 staff report, birds appear as an example of a natural, background source, and no responsible party is given a share of the cleanup for natural sources. Sewer leaks, storm drains and livestock belong to someone. Birds belong to no one.

Even on the county’s theory, this is not simply nature. The project report notes that birds gather on the lagoon’s sandbars, drawn by outdoor restaurant seating and by people who feed them. The lagoon itself is built every year, when the City of Capitola closes the creek mouth with a sandbar, as the Coastal Commission’s fact sheet explains. A warm, still pond full of birds, next to a beach full of toddlers, is a managed condition.

What the final report should answer

  1. Was E. coli O157:H7 cultured from bird droppings? From which species, in how many samples, and did whole genome sequencing match the children?
  2. Were cattle, goats, sheep, horses or deer upstream sampled, on Soquel San Jose Road, along Noble Gulch or in the Bates Creek drainage? If not, why not?
  3. Was the outbreak strain found at the upstream site near Nob Hill, above the lagoon? That bears on where it entered the creek.
  4. How does the outbreak strain compare with cattle and other animal isolates in the national sequence databases?
  5. Samples were collected November 4 and the beach closed November 10. When did the county have results, and when did it learn of the first sick child?
  6. When were the children exposed, when did they fall ill, and how are they now?

My view

I would bet that this strain began in the gut of a cow, a goat, a sheep or a deer somewhere in the Soquel Creek watershed, and that the lagoon and its birds did the rest. The way to find out is to publish the data, including what was tested and what was not.

If your child was at Capitola Beach or in Soquel Creek this fall and had bloody diarrhea, tell your doctor about the exposure and ask for a Shiga toxin test. My earlier post on what we knew when the beach closed has more for parents, and so does our page on E. coli.

On December 22, 2024, someone told ByHeart that a baby was in the hospital with suspected botulism.

Nothing in the FDA’s records shows that ByHeart recalled anything, called California’s Infant Botulism Treatment and Prevention Program, or did anything at all except mark the complaint “lost to follow-up.”

Nearly eleven months later, on November 8, 2025, ByHeart recalled two lots. Three days after that, it recalled everything it had ever made.

I know this because the FDA has now released three inspection reports for BlendHouse Portland, the Oregon plant that filled every can and stick pack of ByHeart Whole Nutrition: April 2024, March 2025, and the amended report from the November 2025 outbreak inspection. Read them in order and you watch a complaint system fail in slow motion.

“Only Ever Received One”

When the FDA walked into Portland on November 11, 2025, on what it called a “mission critical” outbreak inspection, the plant’s Director of Quality told investigators the company had “only ever received one complaint” where a parent reported a hospitalization and suspected C. botulinum. Complaint CN-005563. (Outbreak EIR, p. 4.)

The complaint table gives the date: December 22, 2024. The parent was “lost to follow-up.” No lot number was ever obtained. That is the whole entry. (Outbreak EIR, pp. 7–8.)

Infant botulism is rare, roughly a hundred or so cases a year nationwide. Before 2025, no one had ever tied an outbreak of it to infant formula. A parent calling a formula company to say the doctors think my baby has botulism is not a colic complaint. It is a five-alarm fire.

And ByHeart’s own policy said so. Under the complaint procedure the FDA described, any health complaint involving hospitalization is a “serious” complaint, and ByHeart’s Chief Medical Officer is supposed to review every health complaint for “a potential for serious adverse health consequences.” (March 2025 EIR, p. 3.)

Three Months Later, the FDA Came Back and Never Heard About It

In March 2025, the FDA inspected the same Portland plant. The investigator reviewed ByHeart’s complaints from April 26, 2024 through March 14, 2025, a window that includes December 22, 2024.

Here is what ByHeart reported for that window: 3,247 complaints, 2,167 of them health-related. Of those 2,167, only 69 were forwarded to the Portland plant. Of those 69, ten were classified as “sensitive.” (March 2025 EIR, p. 4.)

The investigator then listed all ten, one by one. Every one was a bacterial illness. None was the December 22 botulism complaint. The word “botulism” does not appear anywhere in that 30-page report. (March 2025 EIR, pp. 4–7.)

The plant got “No Action Indicated.” (Food Safety News.)

So, one of two things happened. Either the December complaint never made it from ByHeart’s New York headquarters to the plant, or it made it and was not classified as sensitive. Either way, a hospitalized baby with suspected botulism did not make the list of complaints worth showing the FDA. Ten bacterial complaints did.

The Ten Sensitive Complaints

These are the complaints ByHeart did let the FDA see. Read them:

A daughter diagnosed with Salmonella. (24-003.)

An infant with “a salmonella infection that has entered the blood stream.” (24-011.)

A bad diaper rash and diarrhea, later diagnosed as Campylobacter. (24-035.)

An infant’s stool positive for Salmonella. (24-047.) Another. (24-059.)

An infant who “developed Salmonella poisoning” after drinking two batches. ByHeart opened two separate investigations, one per batch. (24-061 and 24-063.)

Mucous, diarrhea and blood, with cultures confirming Salmonella. (24-066.)

Two weeks of diarrhea, positive for E. coli. (24-080.)

Diarrhea, high fever, and blood and mucous in the stool, diagnosed as Salmonella. (24-0084.) (March 2025 EIR, pp. 4–6.)

That is eight Salmonella complaint files, one Campylobacter, one E. coli.

And every investigation ended the same way. “No evidence that the batch of infant formula was contaminated.” “All process controls were effective at managing any risks.” (March 2025 EIR, pp. 4–7.)

Here is the trick. ByHeart told the FDA it “did not receive more than one sensitive complaint for any batch.” (March 2025 EIR, p. 4.) Of course not. It counted batch by batch. When one sick baby drank two batches, that became two investigations of one complaint each. A system built that way will never see a pattern, because it never puts two sick babies on the same page.

Before That: Three “Serious” Complaints, All Called Allergies

Go back one more year. In April 2024, the FDA inspected Portland for the first time. Between July 31, 2023 and April 23, 2024, ByHeart reported just three “serious” health complaints. All three involved hospitalization. ByHeart called two of them milk protein allergies and one an allergy “not otherwise specified.” None of the complaints it sent the Portland plant in that period was health-related at all. (April 2024 EIR, pp. 3–4.)

Remember those dates. When CDC and FDA later looked backward, they found ByHeart botulism cases with onsets starting in December 2023, and California reported six more cases with onsets between November 2024 and June 2025. I have written about how this outbreak went on for years. Babies were getting sick during exactly the period when ByHeart’s complaint files showed nothing but allergies and, later, a botulism report closed as lost to follow-up.

What ByHeart Was Looking For

The March 2025 report also explains why ByHeart never found the organism. ByHeart’s food safety plan for Portland “did not identify any potential physical, chemical, or biological hazards” associated with the ingredients it received. (March 2025 EIR, p. 18.) The base powder, made with organic whole milk powder, was not treated as a hazard.

Its finished-product testing covered Cronobacter, Salmonella, Listeria, Staph aureus, E. coli and coliforms, Bacillus cereus, and aerobic plate count. Not C. botulinum. (March 2025 EIR, p. 19.) The outbreak report says the same thing about the two recalled lots: their release testing covered indicator organisms, Salmonella, and Cronobacter. (Outbreak EIR, p. 20.)

Even the Cronobacter protection did not work. ByHeart bought in-line equipment in 2022 “as an extra layer of protection against Cronobacter.” It never ran at its factory certification, and after troubleshooting from September 2023 to June 2025, ByHeart stopped counting it as a control. (Outbreak EIR, p. 20.)

The Constipation Complaints

Constipation is usually the first sign of infant botulism. The outbreak report lists what parents said about the two recalled lots:

A parent said ByHeart caused constipation, confirmed by a pediatrician. It went away when the baby came off ByHeart and came back when ByHeart was restarted. The FDA’s entry ends: “No root cause analysis was conducted.” (CN-007818.)

A colicky, constipated baby with a swollen belly. Lost to follow-up. (CN-007873.)

A ten-month-old constipated for four days. (CN-008241.) Another baby “getting constipated.” Three calls, lost to follow-up. (CN-008203.)

And an infant with no bowel movement for three weeks. ByHeart’s answer: it “did not receive similar complaints.” (193253.) (Outbreak EIR, pp. 4–8.)

Then the Botulism Complaints Came

After the recall, the complaints stopped being subtle. Two parents told the FDA their babies on ByHeart had infant botulism. (CN-008142, CN-008141.) A baby was hospitalized and tested for “GI bacteria including botulism” after four canisters from a recalled batch. (CN-008170.) And this, in full: “We are in the hospital with our daughter. She is being treated for botulism.” (CN-008192.) (Outbreak EIR, pp. 4–6.)

Many complaints the FDA forwarded were simply not in ByHeart’s system, or got “no response or follow up without a valid lot code.” For one family, Portland’s quality director “was not sure why” no one followed up, and ByHeart opened a deviation report to explain it. (Outbreak EIR, pp. 8–9.)

Who Was Responsible

The outbreak report names Ron Belldegrun, ByHeart’s CEO and owner, as the company’s “most responsible person.” (Outbreak EIR, pp. 10, 15.) The same executives keep appearing at close-out: the Chief Quality Officer, the head of supply chain, the VP of Quality, and the senior director of manufacturing sat in on March 2025 (March 2025 EIR, pp. 26–27), and again in January 2026, joined by the Chief Medical Officer. (Outbreak EIR, p. 27.) In March 2025 the FDA even reminded ByHeart that it was required to have a redundancy risk management plan. (March 2025 EIR, p. 27.)

What Should Have Happened

I have done this work for more than thirty years. When a parent reports a hospitalized baby with suspected botulism, a responsible formula company calls back until it reaches the family. It calls the hospital. It calls California’s Infant Botulism program, which talks to every treating doctor in the country. It pulls its retains and sends them to a lab that can find C. botulinum. And it tells the FDA, and it puts that complaint in front of the next FDA investigator who walks in the door.

ByHeart wrote “lost to follow-up.” Three months later it showed the FDA ten bacterial complaints, each closed one batch at a time. Nearly a year later, 48 babies in 17 states had been hospitalized and treated with BabyBIG.

Complaint systems exist for one reason: to hear the warning before the outbreak. ByHeart heard it. It just was not listening.

Sources

https://www.marlerblog.com/files/2026/11/1-final_applied_blendhouse_portland_or_3013670080_eir_11-11-25_-1-23-26_amendment_1_br_nj_vlo.508.pdf

https://www.marlerblog.com/files/2026/11/2-2024-04-26-Portland-EIR-2024-04-22-to-04-26-FEI-3013670080.pdf

https://www.marlerblog.com/files/2026/11/3-2026-01-23-Portland-EIR-outbreak-inspection-2025-11-11-to-2026-01-23-Amend-1-FEI-3013670080.pdf

https://www.marlerblog.com/files/2026/11/4-2025-03-27-Portland-EIR-2025-03-17-to-03-27-Amend-1-FEI-3013670080.pdf

Four of the country’s leading researchers on food and health published an editorial today in the American Journal of Public Health, “US Ultra processed Food Policy: Finding Common Ground.” The authors are James Krieger of the University of Washington and Healthy Food America, Kelly Brownell of Duke University, Ashley Gearhardt of the University of Michigan and Lindsey Smith-Taillie of the University of North Carolina. Their argument is that people who care about the American diet are losing time arguing with each other about ultra processed foods (UPFs) while the chance to do something about them is in front of us. They offer what they call “a pragmatic path forward.” I agree with them, and I want to say why.

I have spent more than 30 years on the part of the food problem that puts people in the hospital within days: E. coli, Salmonella, Listeria. Ultra processed food works on a longer clock. The policy fight, though, looks very familiar.

What does the editorial say about ultra processed foods and health?

The authors start with the size of the problem. The United States leads the world in UPF consumption, and more than half of the calories Americans eat come from these products. They describe UPFs as industrially produced foods built for profit: engineered to be hard to stop eating, to sit on a shelf for a long time and to be made from cheap ingredients.

On the science, they point to a large body of observational research tying UPF consumption to obesity, cardiovascular and metabolic disease, dementia and mental health disorders, and to randomized trials linking UPFs to higher calorie intake and weight gain. Nearly all of that research rests on the Nova classification system. Taken together, they write, the evidence meets the Bradford Hill criteria for causation, which is the standard public health uses to decide when it is time to act.

They are also candid that not everyone agrees. A minority of scientists and advocates still question the UPF concept, how it is defined, and whether it should be the target of nutrition policy. The authors list the reasons fairly: some prefer a system built on nutrients, some have financial conflicts of interest, and some raise legitimate questions about biological mechanisms. Their concern is what the disagreement is costing. It is delaying policy and putting a rare opportunity at risk.

Why is this the moment for UPF policy?

The editorial lays out how much is moving at once. States have proposed and passed bills restricting UPFs in schools, removing certain ingredients from the food supply and requiring labels. FDA is working on a federal definition of ultra processed food. The Secretary of Health and Human Services is calling for action. Bills have been introduced in Congress. The 2025 Dietary Guidelines recommend avoiding highly processed foods. The San Francisco City Attorney has sued, alleging industry deception and fraud, and the authors expect more litigation to follow. Colombia and Brazil are moving ahead with their own policies. The authors describe the support as bipartisan, which in food policy does not happen often and does not last long.

What is the food industry doing about it?

According to the authors, the industry is opposing UPF policy with tactics borrowed from tobacco, alcohol and opioids. They describe an effort to cast doubt on the UPF concept in comments to FDA and through public relations, with sympathetic academics enlisted to help. They describe an industry preference for a nutrient-based system, which a company can satisfy by reformulating a product, for example by swapping non-sugar sweeteners for added sugar, without changing what the product is. They also describe a campaign called “Americans for Ingredients Transparency,” launched to preempt state and local action. The point of that strategy, they write, is to confine policy to the federal level, where the industry has the power to shape legislation and regulation to fit its interests.

What do the authors propose?

First, that advocates stop handing the industry its best argument. Disagreements among experts are normal, the authors write, and the same kind of disagreement surfaced over trans fats and sodium. Aired the way they have been, though, these differences look bigger than they are and let the industry claim the science is too unsettled to act on. Waiting for absolute scientific consensus would set a bar that few public health measures, the regulation of nicotine products among them, could have cleared.

Second, a definition that works in a statute. Krieger and Smith-Taillie chaired a Healthy Eating Research expert panel that recommended one: a product is ultra processed if it contains a cosmetic ingredient, such as a flavor, color or emulsifier, or an industrial ingredient not used in home kitchens, such as high-fructose corn syrup. A product that meets FDA’s criteria for a “healthy” claim would be exempt unless it contains a non-sugar sweetener.

Third, answers to the standing objections, set out in a table. More research is always welcome, but the authors say the current evidence is a sound basis for policy and rigorous enough to survive judicial scrutiny, and they note that policies on trans fats, tobacco and lead went forward despite scientific uncertainty. To those who would rather regulate nutrients, they answer that the two approaches work together: 90 percent of U.S. products high in saturated fat, added sugar or sodium are UPFs, while 39 percent of UPFs are not high in any of the three, a gap a nutrient-only rule leaves open for reformulation.

Fourth, five policies drawn from the panel report: an excise tax on selected UPFs such as sweetened beverages and processed meats; limits on UPFs bought and served in publicly funded settings, including schools, childcare and government facilities; publicly funded counter marketing modeled on the tobacco Truth campaign; guidance to limit UPFs in dietary guidelines; and a front-of-package warning label identifying a product as a UPF.

Why does a food safety lawyer care?

Because I have watched this argument before, and I have seen how it ends when government acts.

After the 1993 Jack in the Box outbreak, USDA’s Food Safety and Inspection Service declared E. coli O157:H7 an adulterant in ground beef and began testing for it. Meat and supermarket industry groups went to federal court in Texas to stop the testing and lost. In 2002, after the ConAgra ground beef recall, I wrote an op-ed asking the beef industry to put me out of business. Over time, with a clear rule and an industry that eventually paid attention, it largely did. Hamburger E. coli cases, which once were most of what my firm did, are now close to zero. I consider that the best thing to happen in food safety in my lifetime.

Nobody had every scientific question about O157 answered in 1994. What we had was enough evidence of harm, a definition that an inspector could apply and a rule with consequences. That is what the editorial’s authors are asking for here. Open questions about mechanisms are a reason to keep doing research. They are not a reason to wait.

I also pay attention when an industry’s first move is preemption. In my experience, a company confident in its products argues the merits. A company that wants to take the decision away from states and cities is telling you something about how it expects to do on the merits.

Where I come down

The authors end by saying the tools are in hand: strong evidence of harm, a definition that can be applied and a set of evidence-based policies. I agree. I would rather argue about the details of a rule than about whether to have one. Start with the places where children eat and with an honest label on the package. Let the research on mechanisms continue. And do not let the people who sell the products write the definition of what they are selling.

The beef industry showed that a food industry can change when the rules are clear and the consequences are real. I would like to see the same thing happen here.

https://www.marlerblog.com/files/2026/11/Krieger_AJPH_2026_US-UPF-Policy_Finding-Common-Ground.pdf

Little Wicomico Oyster Company, the Virginia grower whose raw oysters CDC says made 28 people sick in 13 states, has sued the Virginia Marine Resources Commission for more than $9.1 million. According to WRIC, which reviewed the court documents, the company says the Commission and the Virginia Marine Police kept it from tending about 75 million juvenile oysters in the Little Wicomico River after the Virginia Department of Health closed the river on October 9, and that about 16 million of them died.

A disclosure first. Marler Clark represents two people sickened in this outbreak in federal lawsuits against the company: a Colorado woman in Denver and, with Regan Zambri Long, a Maryland man hospitalized with sepsis, in Richmond. I have no part in the company’s case against the Commission, and I take no position on whether Virginia owes it money. My interest is in what happens to the oysters.

The short version

  • The outbreak: 28 sick, 13 states, five hospitalized, no deaths, as of CDC’s October 30 update. The investigation is open. Illnesses began as early as May 29.
  • The river: closed to oyster and clam harvest since October 9, indefinitely since November 8. Four rounds of oysters from the company’s cages have been tested. All four found Salmonella.
  • The lawsuit: filed Sunday, November 15, in Northumberland County Circuit Court. It asks for a permanent injunction protecting 58 million living seed oysters and payment for 16 million dead ones.
  • The court order: a Lancaster County Circuit Court judge has allowed the company to pull seed oysters from the closed river, work them at its plant, and put them back.
  • The open question: what those oysters will have to show before anyone eats one raw.

What the company has filed

WRIC reports that the complaint accuses the Commission and the Marine Police of unlawfully barring maintenance since about October 12, and alleges that officers threatened to arrest owner Andrew Cockrell if he took a bag or cage of seed oysters out of the water. The company’s lawyer, Tim Anderson, wrote on X that the damages claim is one for inverse condemnation. He has estimated a possible 30 percent loss, more than 20 million oysters, and posted on Monday that “these 480,000 are already dead.”

The husbandry is real work. Juvenile oysters are pulled, tumbled, broken apart, sized, rebagged and returned. 13News Now, which covered the hearing, reports that Mr. Cockrell testified the company has about 54,000 cages plus nursery bags, roughly three years of production, and that oysters were already dying in their bags.

The dates are not consistent in the reporting. 13News Now puts the hearing and the restraining order on November 6. WRIC dates the order November 12. Mr. Anderson wrote on November 13 that the Supreme Court of Virginia had rejected the state’s emergency appeal and dissolved a stay. I have not seen the orders. The next hearing is set for December 11 at 2 p.m. in Lancaster County.

As 13News Now describes the order, the company may clean, sort, resize and rebag its seed stock and return it to the river, and Marine Police may watch on the water or at the plant. The order does not reopen the river, allow sales from it, or establish that the oysters are safe to eat.

The numbers do not reconcile yet

  • 75 million oysters less 58 million living is 17 million, not 16 million.
  • Thirty percent of 75 million is 22.5 million.
  • The only loss described as already counted is 480,000.
  • $9.1 million for 16 million oysters is about 57 cents each, described as market value, for oysters that are not market size.

The complaint may explain all of this. The news reports do not.

The rule the judge relied on turns on six months

13News Now reports that the judge pointed to the Commission’s own relay regulation, 4VAC20-310-20, which says nothing in that chapter prohibits harvesting, transporting or handling seed-stock shellfish from condemned areas. The same chapter defines seed stock as shellfish “at least six months from marketing for human consumption.”

That definition is where the public health question sits. Mr. Cockrell testified, according to 13News Now, that the 75 million range from very small seed to oysters approaching market size. WRIC, describing the order, says the seed oysters are months from market size. How many months, for how many oysters, and by whose count?

Six months is a calendar, not a test. At the hearing, the Commission’s lawyer said no studies show whether seed oysters purge Salmonella in clean water, and a health department witness said the agency has no established purging rate for Salmonella and does not regulate seed stock the way it regulates market oysters. These oysters are not going to clean water in any event. The order sends them back to the same river.

What Virginia has found in that river

CollectedWhat Virginia sampledResult
October 13400 oysters from company cages, eight meat samples, four leasesEight of eight positive for Salmonella
October 21Company cages, four leasesThree of four leases positive
November 2Company cages, eight meat samplesThree of eight positive, two leases
November 9Company cages, ten meat samples, five leasesTen of ten positive
November 16Fifth roundNo results posted; page last updated November 17

Reopening takes three consecutive clean rounds, and there has not been one. Bird feces and a human sewage marker tested negative, and the health department has not identified the source. Its outbreak page, last updated November 17, says nothing about seed oysters or the court order. I laid out the third round on Food Poison Journal and CDC’s findings here.

Questions I would want answered before December 11

  1. How many of the 75 million oysters are at least six months from market today, and who verifies it?
  2. Will oysters raised in this river be tested for Salmonella before sale, or does the six-month clock alone clear them?
  3. Will the three-clean-rounds standard that governs the river also govern this crop?
  4. WRIC reports the judge found no cross-contamination risk because the seed and market operations are in separate waterways. Is the plant where seed is tumbled the same plant that handles market oysters from the company’s other rivers, and what separates the two on shore?
  5. Where do the dead oysters and shell go?
  6. Will the health department finish a purging study before any of these oysters reach market size?

A court has agreed that these oysters need tending, and a dying crop is a real loss to a family business. I do not doubt either point. But the people in this outbreak began getting sick in May, months before the September 18 to October 8 harvest that was recalled. Who pays for the dead oysters is a question for the Commission and the court. What the living ones must show before someone eats them raw is a question for everyone else.

Yesterday I wrote about the dates in FDA’s closing reports on three Salmonella outbreaks. Today is about the black boxes.

The three Executive Incident Summary Abstracts cover investigations FDA closed on November 4: file #1402, Salmonella I 4,[5],12:i:- in romaine lettuce; file #1387, Salmonella Oranienburg; and file #1403, Salmonella Newport. Together they account for 261 illnesses and 47 hospitalizations. Coral Beach at Food Safety News has a story on each: romaine, Oranienburg and Newport.

Credit first. FDA does not have to post these abstracts at all, and a redacted report beats no report. But read them with a pencil and count.

What FDA withheld

FileSick / hosp.The foodThe firmWhere people got sick
#1402 Salmonella I 4,[5],12:i:-43 / 6Romaine, named at closureProcessor withheld under (b)(4)States withheld
#1387 SalmonellaOranienburg99 / 22Withheld under (b)(5)Packinghouse, grower and farms withheldStates withheld
#1403 SalmonellaNewport119 / 19Withheld under (b)(5)None identified. Growers’ location withheldStates withheld

By my count the romaine abstract carries 14 redaction marks, the Newport abstract 11 and the Oranienburg abstract about 40. That is roughly 65 marks across three short documents. In two of the three, the withheld item is the answer to the only question most people have, which is what the food was.

In no file does FDA say where the 261 people live.

Two exemptions, two different problems

FDA’s abstracts page says the reports are redacted to comply with disclosure laws, including the Trade Secrets Act and the Freedom of Information Act. The marks in the margins point to two FOIA exemptions, and they deserve separate treatment.

Exemption 4, marked (b)(4), covers trade secrets and confidential commercial information that a company gives the government. FDA uses it for firm names, and its long-held position is that supply chain relationships are confidential. I disagree with how far FDA takes that, but it is at least the right exemption to argue about.

Here is the argument. In the romaine file, four traceback legs converged on one supplier and processor of prepackaged, shredded romaine. FDA inspected that firm, issued a Form 483 on its sanitation preventive controls, corrective actions and verification, and concluded that the firm’s romaine was the confirmed outbreak vehicle. A customer list can be a commercial secret. The fact that a company’s lettuce made 43 people sick is not one.

FDA names firms all the time when there is a recall. It named a jalapeño importer, an egg producer and a lettuce processor this year. The difference in the romaine file is not the law. The lettuce was past its shelf life, so there was nothing left to recall, and with no recall there was no name. When the fuller summary of the 2024 romaine E. coli outbreak finally came out, the processor’s name was in it.

Exemption 5, marked (b)(5), is the bigger problem. It covers internal government memoranda that would be privileged in litigation. In practice that means the deliberative process: advice, recommendations and drafts exchanged before a decision is made. The Supreme Court said in EPA v. Mink in 1973 that purely factual material that can be separated from the deliberation is not protected. Since 2016 the statute has also required an agency to reasonably foresee harm before it withholds anything.

Now look at what is marked (b)(5) in these abstracts:

In the Newport abstract, every redaction is marked (b)(5), including the food in the document’s own title, the common ingredient among the sick, and the location of the growers.

In the Oranienburg abstract, the commodity in the title is marked (b)(5).

In the romaine abstract, the state agency that reported the cluster, the state where the subclusters were found and the party that ran the ingredient analysis are all marked (b)(5).

A vegetable is a fact. A state is a fact. Neither one is advice, and neither is a draft.

FDA knows how to print these facts. Its abstract for the jalapeño outbreak, posted last week, includes a table of cases state by state. CDC publishes a map of states for every outbreak it announces. FDA’s own investigations table names commodities as a matter of routine. Whatever harm FDA foresees from telling the public that a closed outbreak involved a particular crop, it has not said what that harm is.

No ongoing risk

FDA describes these abstracts as post-response documents. The outbreak is over and, FDA says, no longer poses an ongoing risk to the public. Take that at face value and the case for secrecy gets weaker, not stronger. If the danger has passed, naming the food cannot cause a panic and cannot interfere with an investigation that is closed.

The Oranienburg abstract also undercuts the premise. FDA traced six points of sale to one packinghouse. It inspected the farms outside the growing season, left a Form 4056 noting multiple sanitation issues, and recommended another inspection with sampling next season. Those farms will plant again. The cluster was flagged with a signal for Mexican-style restaurants in the Midwest, and those restaurants cannot ask a supplier one useful question about a commodity FDA will not name.

In the Newport file, FDA’s Division of Import Operations increased import screening for two growers. That tells us the suspect produce crosses a border. Which crop, and which country, is withheld.

What I am doing about it

First, I am asking FDA under FOIA for the Form 483 in the romaine file and the Form 4056 in the Oranienburg file. Neither has been made public in any form.

Second, the citizen petition I am drafting on outbreak disclosure will ask FDA to name the firm when it calls a product the confirmed vehicle, and to publish the facts in these abstracts without Exemption 5 marks: the food, the states and the dates.

Third, three questions for FDA, which I would be glad to print the answers to:

1. What deliberation is revealed by naming the food in files #1387 and #1403?

2. Why did the jalapeño abstract get a state-by-state table when these three did not?

3. Will FDA name the processor whose romaine it calls the confirmed vehicle in file #1402?

There are 218 people in the two files where no food was ever named. Nobody has told them publicly what made them sick. Forty-one of them were hospitalized. They are owed a noun.

I have been doing foodborne illness litigation for more than thirty years, and I have read a lot of “what not to eat” stories. Most are forgettable. The one Alice Callahan published in the New York Times this morning, “5 Foods That Food Safety Experts Avoid,” is not. It is short, it is accurate, and it explains why each food is risky rather than just telling people to be afraid of it. A tip of the pen to her.

Ms. Callahan, who covers nutrition and health for the Times and holds a Ph.D. in nutrition, asked several people who work in food safety what they personally stay away from. The five answers: raw sprouts, deli meat, packaged salads, raw shellfish and precut melon. Full disclosure: I am the one who gave her the bagged salad answer.

Why this list rings true to me

None of these foods is on the list because of a hunch. Each one is there because of how it is grown, processed or handled, and each one has an outbreak record behind it. I have spent my career in that record.

Sprouts. Seeds are germinated in warm, wet conditions that are just as good for bacteria as they are for sprouts. I first called on FDA to require a warning label on raw sprout packaging in March 2003, during a Salmonella outbreak traced to sprouts in Oregon, Washington and California. This summer brought two more sprout outbreaks, and I have drafted a citizen petition asking FDA for that same label.

Deli meat and melon. These are both Listeria stories. I represented the families of all 33 people who died in the 2011 Jensen Farms cantaloupe outbreak, and dozens of people sickened in the 2024 Boar’s Head outbreak. Listeria grows at refrigerator temperatures, which is why time in a deli case or a clamshell of cut fruit works against the consumer and not for them.

Packaged salads. I have been suing lettuce and leafy greens companies for nearly thirty years. I represented the families of the five people who died in the 2006 Dole spinach E. coli outbreak. As I told Ms. Callahan, when lettuce from many fields is cut, washed and mixed together, “If something goes wrong, it goes wrong in a big way.” I buy whole heads and wash them myself. That is not risk-free, but it is a smaller bet.

Raw shellfish. Oysters, clams and scallops filter the water they live in and concentrate whatever is in it, including norovirus and Vibrio. Cooking is the control step, and raw shellfish skip it.

The people she quoted

Ms. Callahan also chose her sources well. Barbara Kowalcyk of George Washington University is a friend and one of the most brilliant people working in food safety. Her son Kevin died in 2001 of an E. coli infection, and she has spent the years since making the system better for other families. Her point in the article is the right one: this is not about telling anyone what to eat. It is about understanding that some foods carry more risk than others, especially for young children, older adults, pregnant women and people with weakened immune systems.

The article also quotes Sandra Eskin of Stop Foodborne Illness, Ellen Shumaker of North Carolina State University and Matthew Moore of the University of Massachusetts Amherst. Readers get the science and the practical advice from people who know both.

What I would add

Consumers can lower their own risk with the choices in this article, and they should. But the burden should not sit with the person standing at the deli counter or the salad case. The companies that grow, process and sell these foods are the ones that can test, clean, keep product cold and trace it when something goes wrong. I have been asking the industry to put me out of business since 2002. The beef industry largely did. The foods on this list are where the work remains.

Read Ms. Callahan’s piece. Then take a look in your refrigerator.