The Warning and the Empty Donor-Milk Program

RaiLee Mar was born on New Year’s Day 2014, twelve weeks early and weighing roughly three pounds. An emergency cesarean section delivered her in Summersville, West Virginia. Medical personnel then airlifted her to the neonatal intensive care unit at Charleston Area Medical Center. For four days, nutrition reached her through her veins. On the fifth, doctors began feeding her through a tube. Over the next nine days, she received seventy-two feedings of her mother’s milk.

Then Ericka Mar could no longer produce enough usable milk. Over-pumping had left blood in what remained.

Another mother in the hospital offered hers. The hospital refused because the milk was untested and unpasteurized and could carry hepatitis, HIV, or another infection into a critically ill infant. Charleston Area Medical Center stocked neither banked donor milk nor a human-milk product capable of replacing Ericka’s dwindling supply. RaiLee still had to be fed.

Her doctors ordered Similac Special Care 24, Abbott Laboratories’ cow’s-milk-based formula for premature infants. The label carried five words: USE AS DIRECTED BY A DOCTOR. RaiLee received three feedings, each half her mother’s milk and half formula. Doctors diagnosed necrotizing enterocolitis. She died the next day.

On July 24, 2026, the Seventh Circuit affirmed summary judgment for Abbott without resolving the formula’s safety, its role in RaiLee’s disease, or the adequacy of the five-word label. The panel decided the case on a narrower ground: Ericka Mar had not produced enough evidence that her proposed warning—human milk carries a lower risk of NEC than formula—would have changed what RaiLee’s doctors fed her before she developed NEC.

Five Words on the Bottle

West Virginia law governed the claim. A manufacturer may be liable when it could reasonably foresee that a product would be unreasonably dangerous without a warning and when the missing warning caused the injury. Cause-in-fact turns on a practical question: would an adequate warning have made a difference?

The district court rejected Abbott’s broad effort to end the case at the duty stage. In 2007, West Virginia’s highest court rejected the learned-intermediary doctrine, which ordinarily permits a manufacturer to direct warnings to the prescribing physician rather than the patient. The legislature adopted the doctrine for prescription drugs and medical devices in 2016, after RaiLee’s death. Infant formula is regulated as food. The court therefore held that no learned-intermediary rule shielded Abbott from a duty to warn Ericka Mar.

The court also rejected Abbott’s proposed categorical bar against warning duties for comparative feeding risks. A jury could have been permitted to decide that Abbott owed a stronger account of the relationship among cow’s-milk formula, human milk, and NEC. Abbott left the duty stage without any ruling that its label was adequate. Its entire risk instruction consisted of five words directing doctors to control use; NEC and the comparative value of human milk went unmentioned.

Mar’s labeling expert proposed language that put the comparison on the package: human milk has a lower risk of NEC than formula. Mar also had to show that a better warning would have changed a doctor’s order, made donor milk available, or otherwise altered RaiLee’s feeding. At summary judgment, she needed evidence of that different course, not speculation about what someone might have done.

Mar therefore had to identify a realistic alternative before RaiLee developed NEC. That meant showing how a warning could have overcome the hospital’s lack of donor milk, its infection-control rules, the approvals needed to change its feeding system, and the doctors’ existing knowledge of the risk. The Seventh Circuit held that the evidence did not allow a jury to find that any of those things would have happened in time.

The Hospital’s Empty Donor-Milk Supply

Necrotizing enterocolitis attacks the intestines and falls with particular force on premature infants. The Seventh Circuit described abdominal swelling, vomiting, gas within the bowel wall, and death. It cited pediatric guidance reporting substantial reductions in NEC among premature infants fed human milk and a study finding that formula-fed premature infants developed NEC six to ten times as often as infants fed breast milk alone.

Later evidence reinforces the comparative point without resolving causation. A 2024 Cochrane review of twelve trials involving 2,296 infants found that donor human milk reduced NEC risk by about half when compared with formula, with high-certainty evidence for that outcome. In October 2024, the FDA, CDC, and NIH drew the same scientific boundary the litigation continues to confront: the evidence does not conclusively establish that preterm formula causes NEC, while the evidence that human milk protects against NEC is strong. The agencies also identified prematurity as the primary risk factor and acknowledged important gaps in the science.

The Seventh Circuit stayed inside those limits. Human milk may protect an immature intestine in ways formula does not. Formula may contribute to the disease. Several causes may operate together. The panel declined Abbott’s request for a broader ruling that Mar lacked evidence of but-for causation.

The clinical hierarchy was clearer. A premature infant should receive her mother’s milk first. Pasteurized donor milk comes next. Doctors turn to cow’s-milk-based formula when neither is available. RaiLee reached the third option because the first had run short and the second did not exist at her hospital.

Building a donor-milk program required specialized containers, labeling, transportation, trained personnel, storage, thawing and warming procedures, and approval from the hospital’s board and foundation. RaiLee’s treating neonatologist, Dr. Stefan Maxwell, described the undertaking as very expensive and measured in years. He believed an all-human-milk standard was probably impossible in the Appalachian region at the time. Charleston Area Medical Center established a donor-milk program five years after RaiLee died.

When counsel asked whether a stronger warning would have changed his decision to feed RaiLee formula, Dr. Maxwell answered: “Well, what else would we feed this baby?” Her mother’s supply was gone. Donor milk was unavailable. The infant needed nutrition then.

Dr. Maxwell also testified that he already knew formula feeding was associated with higher NEC rates, and another treating physician regularly read the human-milk literature. Yet that awareness had never yielded a donor-milk program. For the Seventh Circuit, the combination of informed physicians and an unavailable alternative strongly suggested that Abbott’s proposed warning would have left the feeding decision unchanged.

Judge Rebecca Pallmeyer had stated the rule more starkly when she granted judgment in May 2025: Abbott could not be liable for failing to warn of a “better but unavailable alternative.”

When Knowledge Protects the Manufacturer

A product warning fixes the manufacturer’s own account of danger at the point of use. It may shape consent, procurement, transfers, hospital protocols, or escalation to administrators. It places the product’s risk inside the decision the product is asking someone to make.

Mar offered no testimony from a hospital administrator, no supplier prepared to deliver pasteurized milk before RaiLee became ill, and no transfer plan. Dr. Maxwell stood by the feeding decision. The court held that a jury could not infer from that record that a stronger label would have changed what RaiLee received.

The Three Proposed Alternatives

Mar offered three ways a different label could have altered her daughter’s feeding. The panel rejected each on the evidence before it.

First, Mar argued that a stronger warning could have prompted the hospital to create a donor-milk program. Dr. Maxwell said the hospital “would have tried,” perhaps sooner, had the danger been framed differently. The panel read that phrase against the rest of his testimony: building the program would have taken years, cost substantial money, required storage and transport systems, trained staff, and board approval, and the hospital already knew about the formula-NEC association. A willingness to try did not show that pasteurized milk would have reached RaiLee in time.

Second, Mar pointed to the milk another mother had offered. Charleston Area Medical Center prohibited informal sharing of untested, unpasteurized milk because it could transmit hepatitis, HIV, or another infection into a fragile infant. The policy addressed a danger different from NEC. Mar identified no evidence that stronger words on Abbott’s label would have caused the hospital to suspend its infection-control rule.

Third, Mar argued that a stronger warning might have led her to refuse formula and insist on using the milk she had been told to discard. The panel held that the theory failed at its threshold because she had never read Abbott’s existing label. The West Virginia authorities the court applied require a plaintiff pursuing this route to have read the warning that accompanied the product.

Mar relied on the NICU, just as Abbott’s label instructed. Because she never read the label, the court held that different wording could not have changed her own decision.

The Evidence Offered After Judgment

After judgment, Mar offered two witnesses who described what they said would have happened if Abbott had given a stronger warning.

RaiLee’s father, Anthony Mar, said that knowledge of the NEC risk would have led him to pursue donor milk or move his daughter to another hospital. Dr. Bar-Yam, a physician affiliated with a Boston hospital, said her milk bank could have sent donor milk to Charleston Area Medical Center within twenty-four hours.

Dr. Bar-Yam identified an existing supply and a delivery period measured in a day, rather than a local program that would take years to build.

Mar offered that testimony only after the district court had entered judgment. Rule 59 permits reconsideration for newly discovered evidence, not evidence a party knew about and could have presented earlier. Mar knew of both witnesses before summary judgment. The district court refused to reopen the case, and the Seventh Circuit found no abuse of discretion. Both courts also concluded that the testimony would not have changed the result.

The litigation therefore ended without testing at trial Dr. Bar-Yam’s assertion that a Boston milk bank could have supplied donor milk within twenty-four hours.

Mar was selected as one of four bellwethers from hundreds of consolidated cases. A bellwether is supposed to reveal the strength of recurring claims, but it remains one lawsuit built from one record. The district court recognized that limit. It said its ruling had “limited direct application” to other families, who may be able to prove that a physician, parent, administrator, milk bank, or hospital would have acted differently after a stronger warning.

Why the Design Claim Failed Too

The warning claim was one part of Mar’s case. The other—design defect—failed for a related reason, and she did not appeal its dismissal.

Her experts identified human milk and related products as safer feeding options. For design defect, however, the district court required evidence that Abbott could manufacture and distribute a feasible substitute at scale and cost. Prolacta’s ready-to-feed human-milk formula entered the market after RaiLee’s birth, while the fortifier discussed by the experts served only to add nutrients to milk already available. RaiLee’s central problem remained the loss of the milk itself.

Both claims turned on the same shortage. Design defect required a feasible substitute from Abbott; warning causation required a feasible response from the hospital. RaiLee needed immediate nutrition after her mother’s supply failed, hospital policy barred untested milk, and pasteurized donor milk was unavailable.

The Limits of Abbott’s Victory

Abbott won judgment without trial in the first federal bellwether, and the Seventh Circuit affirmed. The opinion also gives later courts a rule to apply. Plaintiffs must prove how a specific warning would have changed a specific feeding decision at a specific hospital before NEC developed; the general preference for human milk is insufficient by itself.

The ruling was substantial but narrow. The Seventh Circuit left the central scientific and warning questions unresolved. Its opinion neither declared Similac Special Care 24 safe nor approved the five-word label as an adequate NEC warning. It also declined to erase Abbott’s duty to warn or rule that cow’s-milk formula cannot cause NEC.

Abbott asked the panel to take that last step. The company argued that Mar lacked sufficient evidence that its formula was a but-for cause of RaiLee’s disease. The dispute turned in part on how exposure should be measured. Abbott counted every feeding from birth and characterized RaiLee’s diet as approximately 95 percent human milk. Mar focused on the three feedings after formula was introduced, each of them 50 percent formula. The district court said the answer could have broad effects across the multidistrict litigation and declined to decide it. The Seventh Circuit did the same.

The court recited evidence that human milk reduces NEC incidence and that formula-fed premature infants have experienced substantially higher NEC rates in published research. It also acknowledged that the mechanism and causal relationship remain uncertain. The warning claim therefore turned on whether different words would have changed the feeding decision.

The Record Before the Court

The treating physicians knew the association between formula and NEC, the hospital lacked donor milk, untested milk carried infection risks, Ericka Mar had not read the label, and the strongest evidence of an outside supply arrived only after judgment. On that record, the Seventh Circuit held that a stronger warning would not have changed RaiLee’s feeding before she became ill.