Riley San Miguel said her son, Kru, was barely a month old when he started crying all the time, not wanting to eat. When he developed a fever, he was quickly admitted to an intensive care unit. His spinal fluid was infected with bacteria, and it was spreading to his brain. The doctors believed it had probably come from his infant formula.
“Initially, doctors were worried Kru wasn’t going to make it,” said Ms. San Miguel, 23, who lives in Sonora, Texas. The baby underwent the first of several operations on his brain to control escalating seizures and swelling, and when he could no longer breathe on his own, they put him on a ventilator for 21 days. Eventually, most of the right side of his brain had to be removed.
Her son, she learned, had been exposed to Cronobacter sakazakii, a deadly bacterium that is often linked to babies who consume powdered formula and can cause serious brain damage, developmental disabilities and death. Health officials analyzed two open cans of formula from Ms. San Miguel’s home; one tested negative and the other did not have enough product left to test.
Kru fell ill in August 2019, more than two years before similar infections in four babies — two of them fatal — triggered a voluntary recall at an Abbott Nutrition plant in Sturgis, Mich., that manufactured formula consumed by the latest sickened babies.
The recall helped lead the way to a nationwide shortage of infant formula that has left parents struggling to feed their children; the Biden administration has responded by working with Abbott to restart production and flying in additional supplies from Europe.
As with Kru’s illness, no link was established between infections in the four recent sick babies and production at any Abbott Nutrition plant. But while most of the attention has been focused on fixing the dire supply shortfall, regulators are confronting deeper issues of safety that persist in the manufacturing of powdered infant formula.
There is no comprehensive mechanism for detecting or investigating Cronobacter infections, and only one state, Minnesota, requires doctors and laboratories to report cases to the authorities. Manufacturers are left to conduct tests themselves on a relatively small sample of their production, with less than 1 percent of the Abbott plant’s total output required to be tested. The combination of blind spots makes it difficult to trace infections to their source and could leave families at risk of future shutdowns and infant fatalities.
The Sturgis plant has had troubles going back to 2010, when beetle parts in formula triggered another recall. In recent years, inspection reports show, federal inspectors found standing water in manufacturing areas, a fan loaded with dust and debris positioned near a formula drying area and sanitary zones that were not cordoned off.
In October, a whistle-blower complaint from a former Sturgis quality assurance worker landed with the Food and Drug Administration. He claimed that, in 2019, the plant did not destroy an entire batch of formula where problematic micro-organisms had been found because “senior management was under significant pressure to meet its ‘numbers’”; and that the company was “sanitizing” files before providing them to F.D.A. auditors, failing to disclose details about micro-organisms it found.
Abbott has said that no Cronobacter samples from the four sick babies matched ones found at the Sturgis plant. An opened container of formula consumed by one of the babies had tested positive for two strains of Cronobacter, the company said, but those strains also did not match any that were found at the plant, and one matched a sample found on a water bottle that had been used to mix the formula.
“The formula from this plant did not cause these infant illnesses,” the company said.
Abbott has said that the former employee who filed the whistle-blower complaint was fired for violating its food safety policies and that he has made “evolving, new and escalating” claims since then. The company said it expected to be able to address all of the F.D.A.’s concerns by the time it reopens the Sturgis plant within the next two weeks. A consent decree the company signed with the government provides a series of new safeguards, including hiring a qualified expert to oversee improvements at the plant and notifying the F.D.A. if it finds any contamination. Samples of Cronobacter and salmonella found must be stored for three years.
Read More on the Baby Formula Shortage
“We are confident that we can continue to produce safe, high-quality infant formula at all of our facilities as we have been doing for millions of babies around the world for decades,” Abbott said.
The lack of testing and public reporting makes it difficult to know how many people are infected with Cronobacter. One study estimated about 18 cases per year in U.S. infants, according to Dr. Julia Haston, an official with the federal Centers for Disease Control and Prevention. Outcomes are usually severe, she said, and up to 40 percent of infants who go on to get bacterial meningitis, a frequent complication, die from the infection.
Cronobacter is common in dirt and water. That means it is not an unusual bug to find in homes — much less in food processing plants. But there are only a few ways it is likely to get into an infant’s body, researchers say.
“It’s not falling from the ceiling,” said Dr. Janine Jason, a specialist in infectious diseases who worked at the C.D.C. for 23 years. “Where we know it grows well is powdered formula. It’s one of the few organisms that can survive in powdered milk or food.”
One C.D.C. study said powdered formula “has been identified as the transmission vehicle in nearly all Cronobacter infections for which a source was found.”
The relatively low rate of testing makes it difficult to provide a positive link between illnesses and contamination at a particular plant. The fact that there was not a genetic match between the bacteria found in a sick infant and bacteria found at a plant might be because the contamination occurred in the home, in a batch that was not tested or in a portion of a sample that was not analyzed.
“It’s like reaching into a big box of Legos, pulling out a few and saying that proves there are no blue Legos in the box,” said Samuel Geisler, a lawyer who is representing families in a lawsuit against Abbott.
The Abbott factory sits on the outskirts of Sturgis and towers over the rest of the town, its hulking white structure visible far before the city limits.
The plant employs hundreds of people, and many more in the city of 11,000 are closely connected to the plant. To some, it was their first secure job; to others, it was the lifelong employer of a parent or close friend.
“We take great pride that products made in Sturgis can be found all over the world and are extremely happy to know they will be producing infant formula again soon,” said the city’s mayor, Jeff Mullins.
The plant has had a history of regulatory issues with the F.D.A. The 2010 recall over beetle contamination came as an F.D.A. inspection noted the company’s three-year battle with warehouse and flour beetles.
Before and after that recall, the F.D.A. was pushing the industry to test its products for Cronobacter, spurred by a cluster of cases in a Tennessee neonatal intensive care unit that was linked to contaminated powdered formula from another company.
In 1996, the F.D.A. kicked off what would be an 18-year effort to institute tougher industry standards. When Cronobacter testing requirements were added in 2003, an industry group representing Abbott and other companies pushed back.
It argued that, while it made sense to test for the bacterium in formula meant for premature babies, it was unnecessary to do so in formula marketed toward healthy babies because they were not at risk. Abbott said the trade group’s position did not align with its own practices, which included voluntary testing for Cronobacter.
The F.D.A. held its ground, and in 2014 began requiring infant formula makers to take 30 samples, each the weight of about two nickels, from every lot of formula and test them for Cronobacter.
But the regulations did not define the size of a lot. Some factories have lot sizes of about 30,000 pounds. For Abbott, lots can reach 300,000 pounds, said Scott Donnelly, a former food product safety director who has testified as a witness for a plaintiff suing Abbott over problems with formula.
“The F.D.A. pretty much let industry — Abbott — define a ‘lot’ so as to minimize testing,” he said in an interview. “The end result is that the micro-testing for pathogens was inadequate.”
The F.D.A., armed with the new oversight authority, soon found problems in the Sturgis plant. While Abbott’s internal policy was to test more product than required, it had not, the F.DA. report shows. The company discovered Cronobacter in finished production batches that were ultimately destroyed in 2019 and the following year, F.D.A. records show. The Justice Department ultimately cited the contamination as justification to go to federal court this month to force reforms.
A September 2019 F.D.A. inspection noted that Abbott had received a complaint from a nurse practitioner about five babies who had consumed Similac Sensitive formula and were projectile vomiting for reasons that were not clear. A baby with confirmed Cronobacter was having seizures after consuming three types of Similac, the report said.
Ms. San Miguel’s son, Kru, was one of the babies who fell ill that year after being fed Similac, though his illness was not reported to the authorities at the time.
The company had discovered Cronobacter that year in its Similac Alimentum product for sensitive babies, records show, in the weeks before Kru fell ill drinking the same type of formula.
The 2019 F.D.A. inspection report showed that Abbott did not find a link between any of the Cronobacter complaints and its Sturgis manufacturing plant.
Navigating the Baby Formula Shortage in the U.S.
It was an inspection last September that uncovered serious process problems that would set in motion the most recent product recall.
On Sept. 16, the F.D.A. called the Sturgis plant to sort out Covid-19 procedures before the inspection. This was the plant’s first F.D.A. visit since the beginning of the coronavirus pandemic.
On the same day, officials in Minnesota reported to the C.D.C. that an infant had been hospitalized with a Cronobacter infection after consuming formula from the Sturgis plant.
When F.D.A. officials arrived in Sturgis, agency records show, they found water pooled on the floor in three areas, a situation that could create a growth environment for bacteria.
The F.D.A. also discovered that the plant had found Cronobacter in a finished powdered formula lot in June 2020. But the agency’s inspectors did not swab for the bacterium in or near production areas on this September visit, it confirmed.
The F.D.A. concluded in a report that Abbott had not maintained the building “in a clean and sanitary condition.”
Bill Marler, a Seattle lawyer who sues food companies on behalf of consumers, said he had “very seldom” seen such a blunt assessment. “I’ve been in a lot of dirty plants,” said Mr. Marler, who has not filed a case over the formula issue, “and I’ve seen a lot of inspection reports. And you never see that kind of language.”
By October, the C.D.C. found Cronobacter in the Minnesota infant’s open formula can, according to the agency. Yet five months passed before the F.D.A. took samples in the Sturgis plant that were compared with the bacteria from the formula, according to the C.D.C. (During some of those months, the F.D.A. again held back inspectors amid the Omicron surge.)
They were not a genetic match. That could have meant that the formula was contaminated from something in the baby’s home. Or it could have meant that formula in the Sturgis plant that might have been contaminated was not tested.
In January, a health department in Texas examined the only other sample that emerged from the four infant illnesses that resulted in the Sturgis recall. That baby’s formula had no bacteria and the genetic strain did not match the Minnesota baby’s.
Still, F.D.A. inspectors who returned at the end of the month found more standing water. This time they swabbed around the production area and found Cronobacter, agency records show. They discovered that the company’s own testing from Feb. 6 through 20 found the bacterium 20 times in the plant, including in critical areas, according to the inspection report.
By Feb. 17, the company issued a major voluntary recall and halted production. The F.D.A. is now seeking authority to require formula makers to do more testing and to ensure that details about the Cronobacter they find are entered into a national database.
Determining exactly how much should be tested is a difficult equation that requires regulators to make decisions about what an acceptable amount of risk is and at what cost, said Barbara Kowalcyk, the director of the Center for Foodborne Illness Research and Prevention at Ohio State University.
Meat and poultry are inspected far more regularly than infant formula, she said. “Do I think they need to re-look at the regulation? Absolutely.”
Ms. San Miguel’s son eventually spent about 80 days in a hospital. He has since had five brain operations, according to medical records, and though he is nearly 3, he speaks only a few words and has poor coordination on the left side of his body.
When Ms. San Miguel saw the news about other babies who had become ill with Cronobacter infections and the possible link to powdered formula, it reopened all her old anxiety.
No one ever told her, she said, that powdered formula, unlike its liquid equivalent, could not be considered sterile. The C.D.C. advises parents of newborns to use sterile liquid formula or to prepare powdered formula with very hot water, though no such explicit warning is required to be posted on formula labels.
“I was very angry for a very long time,” Ms. San Miguel said. “If someone told me about the risk I was taking, I never would’ve taken that risk.”
Julie Creswell and Madeleine Ngo contributed reporting.