Formula maker Abbott continues to firmly deny that its infant formulas sickened four babies, killing two. The denial is despite the same dangerous bacteria that sickened the infants—Cronobacter sakazakii—being found at the company’s formula factory in Sturgis, Michigan, which the Food and Drug Administration alleges was producing formula “under insanitary conditions.” And at least one container of Abbott’s formula tested positive for the same Cronobacter sakazakii strain found infecting one of the infants.
Still, Abbott argues that the link hasn’t been confirmed, and its formula isn’t to blame. In a lengthy Twitter thread on May 13, the company made the blunt assertion: “The formula from this plant did not cause these infant illnesses.”
But that is a brazen and misleading claim, according to the Food and Drug Administration. In a press briefing Monday evening, agency officials thoroughly dismantled Abbott’s defense.
The company’s unwavering denial will likely exacerbate frustration from US parents who are forced to navigate a dire shortage of infant and specialty formulas. The shortage is partly due to a recall of Abbott’s formulas and a shutdown of its Sturgis facility, which the FDA determined had numerous problems. Parents have seen empty shelves at store after store as they desperately tried to secure sustenance for their children, some of whom require specialized formulas due to metabolic conditions. Parents have faced purchasing limits, escalating prices, and scams in places where there is availability. Even if parents can obtain the formula, Abbott’s denials may raise safety questions.
Abbott’s defense is indeed questionable. In last week’s Twitter thread, the company reiterated that the link between its formulas and the four infant illnesses had not been confirmed—which is true. But, the company suggested that massive data gaps somehow support the company’s assertion that its formula is not the cause of illnesses.
Overall, the company argues that scant testing of its finished formula found no contamination before leaving its factory. The contamination found in the facility was present in “non-product contact areas.” Additionally, genetic sequencing of the strains found in the facility did not match strains found in two of the sick infants (no genetic information was available for the other two babies).
Opened formula containers from three of the four sick infants were tested, and only one tested positive for Cronobacter sakazakii. While the contamination in the one positive formula container matched the strain of C. sakazakii infecting the infant, it did not match any strains identified in the Sturgis facility. Lastly, Abbott notes that the four sick infants all consumed different types of formula produced in its Sturgis plant, and their illnesses occurred at different times in separate states. It’s unclear why that matters, but Abbott concluded that contamination at the plant did not cause the illnesses.
In the press briefing Monday, FDA officials all but called that reasoning nonsense. Most importantly, the lack of a genetic match is not proof that the formula is not the source of the infant’s bacterial infections.
C. sakazakii is not a reportable disease in this country, Susan Mayne, director of the FDA’s Center for Food Safety and Applied Nutrition, emphasized in the press briefing. That means when cases occur, outbreak investigations are not quickly initiated, and health officials don’t spring to collect bacterial isolates, begin genetic sequencing, and identify clinical clusters as they do for other concerning pathogens. As a result, the FDA and Centers for Disease Control and Prevention only have genetic sequences from two of the four sick infants. And overall, there are only about 238 genetic sequences of C. sakazakii strains in the CDC’s database, which is an extremely small number compared with other pathogens, such as E.coli, making genetic investigations difficult.
“Right from the get-go we were limited in our ability to determine with a causal link whether or not the consumption of the product from the Abbott Sturgis plant was linked to these four cases,” Mayne said.
Mayne also pointed out that the FDA isolated multiple strains of C. sakazakii from the environment inside the Sturgis plant when they were doing testing, which was after the cases were identified. “There certainly is the possibility that other strains that we didn’t detect at the time we were in the plant for the inspection certainly could have been in there.”
Frank Yiannas, the FDA’s deputy commissioner for Food Policy and Response, echoed the point, saying that the genetic data for C. sakazakii in this outbreak and overall is minimal. “It’s hard to read too much into that,” he said. He also highlighted that there was a diversity of strains at the plant—five different lineages—and noted there are examples in the scientific literature of multi-strain outbreaks over time from one source.
“The other thing we’ve heard emphasized quite a bit is that these products have been tested” and most tested negative for C. sakazakii, Yiannas said. But that also is not meaningful. Some of the end batches of formula are 400,000 to 500,000 pounds, but the end-product testing plans only involve a series of 30 samples that are 10 grams each, collectively less than a pound, Yiannas said. “The probability of detecting low levels of contamination through an end-product testing plan—it’s almost never going to happen,” he said. “Some statisticians calculate there’s a 97 percent chance that you won’t find low levels of contamination using that type of sampling plan.”
Overall, he said, “an over-reliance on end-product testing is not really the best way to assure food safety; it’s really about process control.”