Modern Salem-Witch Trials: Acetaminophen Actions for Autism

Autism is a neurological disorder typically diagnosed in childhood. It affects social interaction, including communication and learning, often manifesting in repetitive behaviors. Because of the significant variability in symptomatic expression, it is called Autism Spectrum Disorder (ASD). No one knows precisely what causes it, and through its tortured history (the name was first coined in 1911, and the diagnosis reformulated in the 1960s), it has been a disease in search of a culprit.  

In 1998, Andrew Wakefield fraudulently scapegoated the measles, mumps, and rubella vaccine, falsely claiming it caused autism to bolster his own discoveries. This pronouncement also assuaged the consciences of nervous parents who believed they were responsible for their autistic children’s intellectual limitations. Then came the attack on thimerosal, a mercury-based vaccine-preservative, claiming it was autism’s cause. Thimerosal was removed from childhood vaccines in 2001, but autism diagnoses continued to sky-rocket, especially over the last decade. [1] Regardless, the anti-vaxxers kept stoking vaccine autism-hysteria, and autism mercury myths abounded.

With autism rates still climbing, a new culprit had to be found. Enter Tylenol.

Along with complaints against acetaminophen manufacturers alleging autism, plaintiffs’ lawyers added another juvenile condition in search of a cause: Attention Deficit Hyperactivity Disorder. This disorder manifests as difficulty paying attention, controlling impulsive behaviors, or excessive activity.

Currently, about one in 36 children are diagnosed with autism spectrum disorder (ASD), which now includes ADHD, up from one per 150 children in 2000, providing a treasure-trove for lawyers in search of clients.

In June 2022, the first of the barrage of cases involving Tylenol was filed, claiming negligence and product liability for failure to warn that these conditions are caused by maternal exposure during pregnancy. [2] Proving “fault” is not necessary in product liability actions, but proving causation is [3], and proving causation depends on sound medical and scientific evidence. In federal court, where these first cases were filed, the caliber and admissibility of this evidence is decided at a Daubert hearing. [4]

By December, when the Daubert hearing was convened, some 600 cases had been lodged and amalgamated into a Multi-District-Litigation (MDL) to address overarching issues – such as general causation. Another 100,000 cases awaited filing pending the Daubert decision.

“Many lawyers laughed when we said the Tylenol autism lawsuit could be the next big thing…. No one is laughing anymore. Now there is a class action lawsuit to push the Tylenol and autism litigation forward. The Tylenol autism lawsuit is real.”

- Miller and Zois, LLC, plaintiff’s attorneys

Throughout the early stages of the litigation, the plaintiffs’ bar was optimistic (perhaps even euphoric), both in terms of the judge assigned to the case, Judge Denise Cote, who they called “one of the most well-regarded judges in the Southern District of NY,” and an anticipated positive outcome in the Daubert hearing.

Presumably this was their sentiment – until, that is, Judge Cote ruled on the Daubert motion on December 18th.

“Judge Cote issued a 148-page ruling that rejected all five of the plaintiff’s expert witnesses on the issue of causation….”

The plaintiff’s lawyers admit the federal cases are now dead- although litigation will continue in State courts, primarily in  Pennsylvania, New Jersey, and California, states where the Daubert standard for admissibility of scientific evidence has been rejected in favor of the more malleable Frye standard.

Judge Cote’s reasoning echoes that of Judge Robin Rosenberg in the Zantac cases, which may signal a trend in the federal judiciary. Instead of allowing a “weight of the evidence” approach, where the plaintiffs’ attorneys built a case plucking one shred of positive evidence from a litany of negative data found in a host of studies (a practice once afforded broad judicial approval), Judge Cotes carefully reviewed the studies, both individually and holistically.

Most offensive to the court was “the cherry-picking” habitually used by plaintiffs’ experts who focused only on the positive findings and ignored findings detrimental to their claims, a theme that also governed Judge Rosenberg’s Zantac ruling.

“The unstructured approach adopted by the plaintiffs’ experts permitted cherry-picking, allowed a results-driven analysis, and obscured the complexities, inconsistencies, and weaknesses in the underlying data.” – Judge Denise Cote

• If a study’s authors voiced limitations, the court required the plaintiffs’ experts to adhere to them and rejected expert testimony exceeding study caveats.
• If the study was overwhelmingly negative, the judge disallowed reliance on a discordant positive note. 
• If the study produced statistically insignificant findings – reliance on the data was outlawed.
• If the studies were inconsistent, the judge asked for an explanation. Where none was offered, she struck the testimony.

Among other issues irking the judge was commingling results from separate autism and ADHD studies, which can produce positive results merely as a consequence of amalgamating the data compared to investigating the diseases individually (a practice called stratification). Another issue was the failure to account for confounding variables, and a third was a misplaced extrapolation from animal studies.

Animal Studies

Animal toxicity tests can assess or predict biological damage in animals with physiological systems comparable to humans. The main objection is that doses employed are often orders of magnitude higher than typical human exposures or that the route of exposures differs, which might not affect the metabolic mechanisms experienced in humans.  Still, the endpoints are comparable: a cancerous cell is a cancerous cell, for example, although legally, care is taken that the results “may only be admitted where the gap between what [they] reasonably imply and more definitive scientific proof of causality is not too great.”

Using animals to evaluate human cognitive disorders like autism, however, seems to me just too remote, even if animals are supposedly specially bred for this purpose. How do we exactly determine that a rodent is depressed?  In the context of vaccines, some studies claimed male mice were depressed because they didn’t spend enough time with other mice (although maybe they just didn’t like their smell – we don’t know). Same with hyperactivity. In evaluating ADHD, hyperactivity was measured by how many marbles the rodent buried. Is this relevant to humans?

In the case of autism: “Scientists … measure[d] the amount of time [rodents] spent exploring the cage with the ‘social peer’ compared to the time in the cage with the object,” and found “clear evidence” of impaired social behavior which they argued was relevant to humans. I’m not sure how many spouses spend equal time exploring their cages, apartments, or mansions – or even if we have baseline human measures of cage exploration to compare intra-species differences. Indeed, scientists admit the outcomes were highly variable and dispute the clinical relevance to humans.

After reviewing the proffered toxicologic testimony, Judge Cote’s struck the expert’s opinion, not because of the irrelevance of animal social disorders to humans, but because the studies were internally inconsistent, unreliable, heterogenetic, and there was too great an analytical gap between data and testimony.

There is a more fundamental problem. While rodents supposedly are reasonable models of human ASD, a recent study questioned this premise: The incidence of human ASD is three to five times higher in boys, a gender difference for which we have no explanation. [5] However, this study of ASD mediated by genetic mutation found no similar sex-based differential in mice. In evaluating objective markers, such as synaptic abnormalities, spine density, or neuronal signaling proteins between female and male mice, the researchers concluded the problem is with human diagnostics, leading to the underdiagnosis of ASD in females. I have another take: maybe mice are not a relevant model for human ASD.

Alternative Causes and Increased Risk

While we don’t know what actually causes autism, we do know certain risk factors increase the risk of the condition.  These include:

• A sibling with ASD
• Advanced parent age (either parent)
• Certain genetic patterns or conditions (e.g., Down syndrome or Fragile X syndrome)
• A very low birth weight and prematurity
• A mother whose pregnancies are less than one year apart or multiple.

However, studies relied on in the litigation didn’t address what impact, if any, these acknowledged risks have on acetaminophen-causality, a critical omission, per the judge.

“Because observational studies do not control for exposure to other risk factors for disease, their results must be interpreted with some caution.” - Judge Cote.

Other exposures in utero, such as valproic acid (an anti-epileptic medication), may be causal.  Post-natal exposures, such as infections, may also increase disease risk.  About 15% of ASD cases appear to be associated with a known genetic mutation, which may involve many genes making small contributions (the “polygenic effect”). Per the judge, over 70% of the variability in the population with ASD is due to genetic failures, not environmental factors, all potential causes or contributors that were not addressed in the evidence submitted.

The Importance of Addressing Confounders

“Confounding, … occurs when another causal factor (the confounder) confuses the relationship between the agent of interest and outcome of interest” - Judge Cote

A confounder is a variable associated with the outcome and another independent factor that caused the outcome. For example, because Coca-Cola consumption increased in tandem with polio rates, Coca-Cola was once believed to cause polio. In actuality, the correlation reflected the association of Coca-Cola with summer heat – we drink more Coke when it is hot -- and summer heat is the real culprit associated with polio. In the final untangled analysis, summer is causally associated with increased Coke consumption and polio; hence, measuring Coke consumption gives faulty conclusions.

Distinct from “feel-good” or recreational drugs, people use acetaminophen when they aren’t feeling well, often because of fever. The failure to account for the confounding effects of acetaminophen and the reason for its use is a significant omission identified by the judge. Acetaminophen exposures likely confound the effects of fever or the underlying disease for which the drug is taken. In the few studies where these variables were evaluated, it seemed that maternal fever correlated with increased childhood ASD and that acetaminophen might have had a protective effect.

Another confounding possibility might relate to a finding that nitrous oxide is correlated with ASD. Nitrous oxide is a common anesthetic used in dental procedures – for which Tylenol-type drugs may be taken to relieve post-procedural pain. (Alas, the finding comes from mice studies, so care must be taken in interpreting them).

Now that acetaminophen is litigation-proof, at least in federal court, what’s a parent or plaintiff’s lawyer to do?  Well, as my colleague and friend Susan Goldhaber has written, there are always the baby food cases.

As the plaintiffs’ attorneys in the Tylenol litigation write:

“If your child is under seven and used baby food from the usual suspect manufacturers, there is the possibility your family could have a claim in that litigation.”

[1] In the 2000s, the CDC expanded the diagnostic criteria for the disease, so an increase in prevalence around that time isn’t a useful rubric. The American Psychiatric Association changed the diagnostic criteria again in 2013.

[2] Springer v. Costco Wholesale Corp. (0:22-cv-1532), filed in the District of Minnesota on June 8th, 2022

[3] Proving causation is a two-part inquiry. The first is general causation: Can acetaminophen cause autism and ADHD in the general population? The second part of the inquiry is specific causation, i.e., did the drug cause the condition in this particular plaintiff? These specific instances are generally resolved case-by-case after an affirmative ruling on general causation in a Daubert hearing.

[4] In federal court (and 43 states), scientific evidence must satisfy the Daubert standard, and the judge is charged as the “gatekeeper” - barring evidence that is not reliable and relevant from being brought before a jury. In cases such as the acetaminophen-autism/ADHD litigation, scientific evidence takes the form of expert testimony based on epidemiological and toxicological studies. Epidemiological (or observational) studies are population investigations of human groups ostensibly similar to the plaintiffs. The results are enumerated as statistical likelihoods of increased risk of disease. Toxicological evidence derives from animal evidence, often using heavy doses of exposure, which is then extrapolated to humans. Neither data are precise, but epidemiological evidence is preferred for obvious reasons.

[5] Judge Cote notes the male-to-female differential as three times greater; the American Psychiatric Association notes it is four times higher, and other groups mention a five times higher differential. These figures signal that different standards are used in diagnosis, which casts further aspersions on research methodology.