I have spent the last month counting sick people. The fifty state health departments now report 26,755 cases of cyclosporiasis this season, against 249 nationally at this point last year, and I have written about almost nothing else since the middle of July. Today I want to write about something that worked, because I think it is the most useful thing I know, and because I can prove it out of my own billing records.
I do not try hamburger cases anymore. In the early 1990s ground beef was among the most dangerous things in the American diet. E. coli O157:H7 outbreaks tied to hamburger came several times a year, hundreds sick at a time, children in kidney failure. That is the practice I started with in 1993, and it is the practice I lost. That is because by the middle of the next decade the big ground beef outbreaks had become rare (no pun intended). That is not a complaint. It is a report on progress, and I would like more of it.
What happened with hamburger was not luck. On September 29, 1994, a month into the job, FSIS administrator Mike Taylor took the podium at the American Meat Institute convention in San Francisco and told the room that the agency considered raw ground beef contaminated with E. coli O157:H7 to be adulterated within the meaning of the Federal Meat Inspection Act, and that the government would begin testing for it. The Washington Post later described it as an oh-by-the-way moment several pages into a speech, delivered to an audience that had no warning it was coming. I have written about that day before. The industry sued, arguing that the agency had skipped the rulemaking administrative process and acted beyond its authority. The district court held otherwise — FSIS was entitled to interpret its own statute and to say what an adulterant is. No new law was passed. No money was appropriated. One official, Mike Taylor, decided to do something. He took to the podium and declared beef contaminated with E. coli O157:H7 to be adulterated.
Four things had to be true at once, and that is the part worth carrying forward. One, the toxic organism was named. Here: E. coli O157:H7. Two, a legal status was attached to that name, which meant a positive finding of the organism had a consequence (versus just filing the results away in some file cabinet.) Three, there was a required test, run by the government and then by industry, on a schedule nobody could opt out of. And four, there was verification that outlasted the news cycle: the Pathogen Reduction and HACCP rule in 1996, and beef trim brought inside the designation in 1999. Name, status, test and verification. Remove any one of the four and what you have is a press release.
The numbers moved, and they stayed moved. Across the original FoodNet sites, sporadic laboratory-confirmed O157 infections ran between 1.9 and 2.3 per 100,000 from 1996 through 1999. But by 2004 the rate had fallen below one per 100,000 — 42 percent under the 1996 to 1998 baseline, and the first time a Healthy People 2010 foodborne illness goal (an important metric) was met at all. CDC’s most recent long-run reportstill shows O157 down against the 2016 to 2018 reference period. Thirty years is long enough to tell the difference between a trend and a good year.
Then we did it a second time. The second time did not start with a willing administrator. It started with a petition. By the late 2000s the Shiga toxin-producing E. coli that were not O157 were sickening more Americans than O157 was, and no agency was required to look for them. I paid for the looking myself — roughly half a million dollars of my own money, with Dr. Mansour Samadpour at the Institute for Environmental Health, testing five thousand retail packages of ground beef for every E. coli that turned up. About two percent were contaminated. I briefed USDA scientists and took the findings to the National Meat Association and the American Meat Institute. Nothing happened. In 2009 my firm filed a citizen petitionasking FSIS to declare enterohemorrhagic non-O157 strains adulterants, and we filed it in the names of three people — June Dunning, who died of an O146:H21 infection in 2006, Megan Richards and Shiloh Johnson. The agency did not respond within the time the law gives it. I told them I would sue. They acknowledged receipt.
It took three more years, and then it worked the same way it had worked in 1994. In 2012 FSIS declared six more serogroups adulterants — O26, O45, O103, O111, O121 and O145 — and began routinely testing raw beef manufacturing trim for all of them, with positive trim barred from commerce and subject to recall. The industry predicted economic catastrophe, as it had the first time. But the system adapted, illnesses declined, and the sky did not fall. I keep that episode close for one reason. The 1994 declaration required an administrator willing to “walk to a podium” and declare change. (That is a rare event.) The 2012 declaration, however, required a lawyer, a microbiologist, three injured strangers and a threatened lawsuit. (The latter is not rare at all.) The door is not locked, and it does not only open from the inside.
Let me be careful about how much we celebrate this advance completely. The win is specific. In that same CDC report, non-O157 Shiga toxin-producing E. coli held roughly stable, and nonserogrouped infections rose substantially, most of that a consequence of laboratories switching to culture-independent panels that find what culture missed. I still have beef cases, albeit few and far between. This spring nine Californians were infected with O157:H7 traced to beef kofta, five hospitalized, two of them children who went into hemolytic uremic syndrome. Ground beef is safer than it once was. But it is not entirely safe. And anyone who tells you that the four-step protocol described above ends the problem has never sat in a pediatric nephrology waiting room.
Be honest about what a designation did and did not do. It cleaned nothing by itself. What cleaned hamburger was a decade of work the beef industry paid for and eventually defended — steam vacuums, hot water and organic acid carcass interventions, testing regimes that went well past what the government required, and an industry council that declared food safety a non-competitive subject and meant it. The designation created the reason. The companies did the work. That is why I say the beef industry put me out of that business. The naming by the government of an adulterant without an industry that decides to solve the problem is meaningless. It only insures that lawsuits will continue. It is not a public health program. What it does is make the problem impossible to ignore, and that turns out to be the hard part.
Here is the part I did not see coming. The E. coli did not disappear. It moved. The leading vehicle for O157 in this country is no longer hamburger; it is lettuce. I am in the leafy greens business because the beef fix worked, and the pathogen went looking for a commodity where none of the four elements exist. That is the single most useful sentence I can offer anybody who wants to understand how American food safety actually behaves. Pressure applied in one place is real, and “the problem” relocates.
Which brings me to the law I asked for, and to the promise inside it that has not been kept. In the spring of 2007, I testified before the House Energy and Commerce Committee at a hearing titled A Diminished Capacity, and my pitch was the one I have been making ever since: modernize the system and deny me the opportunity to sue. I brought dozens of clients to Washington across those years to put faces on it. The Food Safety Modernization Act was signed on January 4, 2011, and it was the right law.
Fifteen years on, here is what actually got built. GAO is still telling FDA to build the product tracing system that section 204 required, and the traceability rule itself now sits parked until July 2028. FSMA’s produce rule set an actual number for irrigation water and the agency repealed it. FSMA set a foreign inspection schedule that would have reached 19,200 inspections a year, and GAO found FDA averaging 917from fiscal 2018 through 2023 with 432 investigators covering the whole world and this country too. And the finding that ought to embarrass everyone: GAO reported that FDA has no process for measuring whether its FSMA rules prevent illness at all. We passed a prevention statute and never built the instrument that would tell us if prevention happened. In plain English: Congress told FDA to be able to trace a food back through the supply chain in a matter of hours, to inspect foreign suppliers on a fixed schedule, and to prevent illness rather than count it after the fact. The tracing rule is parked, the foreign inspections are running at roughly one in twenty of the promised pace, and the agency cannot say whether any of it has prevented a single case.
Now think of the four prerequisites for change and the summer we are experiencing, starting with the parasite. Cyclospora has a name. It has legal status in American food law. FDA has had a validated method (a test) for detecting it in fresh produce, romaine included, since 2017, and a validated method for testing agricultural water since 2020. Nobody is required to run either one, so almost nobody does. Name, yes. Status, yes. Test, available and optional. Verification, none. Result: Twenty-six thousand people have been made ill. And the number is growing.
Why is there no mandatory testing of pre-harvest agricultural water? The number setting the standard was written, and then it was withdrawn. The 2015 rule set a geometric mean of no more than 126 colony-forming units of generic E. coli per 100 mL and required growers to test against it. The compliance dates were pushed back, enforcement discretion was announced before the first of them arrived, and in May 2024 the criterion and the testing requirement were replaced outright by an annual assessment in which the grower evaluates his own water and decides for himself what to do about it. I walked through that record this week in connection with the jalapeños out of Sinaloa. Nine years, three delays and a repeal, and not one farm ever had to hit the number.
What’s the most conspicuous missing designation? Salmonella. The organism should be named as an adulterant in meat and poultry. In January 2020 my firm petitioned FSIS on behalf of Rick Schiller, Steven Romes, the Porter family, all Salmonella victims, Food and Water Watch, the Consumer Federation of America and Consumer Reports to declare thirty-one outbreak serotypes adulterants. FSIS docketed it and took comments. Six years on, the agency has not granted it, and its position remains what it has been for decades — that Salmonella is not an adulterant in raw product because ordinary cooking kills it. That is precisely the argument that was made about O157:H7 before September of 1994, by the same agency, under the same law. I handed them the list six years ago, and the current head of food safety went on camera this summer to say she is more convinced than ever that it should not be done. People do not eat food under laboratory conditions. That was the answer in 1994, and it is the answer now. And it is not only what the agency will not grant. In August 2024 FSIS proposed the Salmonella Framework for Raw Poultry Products, which would have set enforceable final product standards for chicken and turkey. It drew 7,089 comments, and on April 25, 2025 the agency withdrew it. In its withdrawal notice FSIS listed its own legal authority to set those standards first among the issues that generated the most comment.
Mike Taylor, the man who named E. coli an adulterant in 1994, is trying to do the same thing today by naming Salmonella adulterant. Mike Taylor is now board member emeritus at Stop Foodborne Illness, and he has written publicly that our poultry safety regulation is not working, and that the ground beef rules were effective because there was a scientific basis for declaring that strain an adulterant. He put it more bluntly in March of 2025, five weeks before the framework was withdrawn. Writing in Food Safety News, he set the two commodities side by side: dangerous E. coli is legally prohibited in ground beef, there is no USDA limit on Salmonella in chicken, and contaminated chicken ships every day with the USDA mark on the label. He called the claim that USDA lacks the legal authority to act the most absurd of the industry’s arguments, for the plain reason that USDA used that same authority decades ago on ground beef. He also priced the objection: at the top of USDA’s own compliance-cost estimate, the cost to the industry is less than a tenth of a cent per pound. And he named the part that ought to sting. The food safety scientists who run the programs at Perdue, Tyson, Wayne Farms and Butterball signed a 2021 letter to Secretary Vilsack asking for enforceable standards and then sat silent at the December 2024 public meeting while the National Chicken Council and the Meat Institute made the no-authority argument for them. That is not a plaintiffs’ lawyer’s theory about what the law permits. It is the architect of the program describing the “load-bearing wall”.
One more small, but very big thing: baby formula. Babies were hospitalized this year with infant botulism after being fed powdered formula. In July, FDA wrote to the formula industry and named the organism: Clostridium botulinum, on its own letterhead no less. But it required no test for it. What the industry runs instead is a surrogate count, a general measure of sulfite-reducing clostridia that has stood in for the actual organism since an expert opinion in 2014, and that nobody has been made to revisit since. Element one is finished. FDA has already named the organism. Elements two, three and four are simply absent. H.R. 7867 would supply them. I have bought five hundred and thirty-five t-shirts, one for every member of Congress, and they arrived last week. Every one of them says the same thing, which is to test infant formula for botulism. It is not a sophisticated argument. (Nothing about the 1994 argument was sophisticated either.)
I am hopeful, and I want to be precise about why. Nothing in this argument for better food safety requires inventing anything. Every element that fixed hamburger already exists as an available action under laws already on the books. FSIS did not need a statute in 1994; it needed an administrator willing to say a sentence and then defend it in court. In 2012 it did not even need that — it needed a petition somebody was prepared to litigate. FDA does not need a new law to require a Cyclospora method it validated nine years ago, or to put a number back on irrigation water, or to build the traceability system Congress ordered in 2011. I listed fifteen of these last week, and four of them require no rulemaking, no appropriation and no act of Congress at all.
I have been asking the food industry to put me out of business for thirty years, and I said it under oath in a hearing room in 2007. With E. coli in hamburger it has actually happened, in one commodity, against one family of organisms. I would like to see it happen again and I do not think the obstacle is knowledge, or science, or money. We know the names of these organisms. We know how to test for them. We know exactly what a consequence looks like, because we attached one to hamburger in a single speech in 1994 and never took it off. What is missing is somebody willing to walk to the podium, like Mike Taylor did and name the organism— or, failing that, somebody willing to file another petition.
