It is not uncommon that terms are used without knowledge of their origin and that the origin is instructive about the meaning or proper application of the term. We offer two examples. Many lawyers who at least dabble in product liability litigation have heard the term “Bradford Hill criteria.” Was it named after two different people like Smoot and Hawley or Hatch and Waxman? Was it developed as a cheat sheet for lawyers and judges? No and no. Sir Austin Bradford Hill was already a famous (and knighted) British statistician when he delivered the President’s Address upon the inaugural meeting of the Occupational Medicine Section of the Royal Society for Medicine in January 1965. On that august occasion, he posed two lawyerly questions, “In what circumstances can we pass from this observed association to a verdict of causation? On what basis should we do so?” He then proceeded to identify nine factors, elements, or criteria for the process of evaluating such questions of general causation. Hence, the eponymous term Bradford Hill criteria. But note that this was not about evaluating whether there was any association between an exposure and an outcome each drawn from the proverbial hat. Instead, an association, typically established through observational research with statistically significant results, was the starting point. As SABH said, “Our observations reveal an association between two variables, perfectly clear-cut and beyond what we would care to attribute to the play of chance. What aspects of that association should we especially consider before deciding that the most likely interpretation of it is causation?”
Our second example is a “Daubert challenge.” Younger readers may have practiced with the Supreme Court’s Daubert decision as a long-established cornerstone of pretty much all litigation. Yet, we remember when it came out in 1993 and how its contours were still being shaped for years to follow. Before Daubert, the Frye standard had long reigned in federal court and many states. Frye came out of a racially charged murder conviction where the defendant’s exculpatory expert opinion based on the results of a lie detector test had been excluded. The affirmance by the D.C. Circuit set the new test, requiring that “the thing from which the deduction is made must be sufficiently established to have gained general acceptance in the particular field in which it belongs.” While Frye was seen as a high bar at the start, by the 1980s, a cadre of practitioners of the ancient art of junk science were routinely permitted to offer medical causation opinions at trial despite general causation being anything but generally accepted. They, and the lawyers who hired them, knew how to wrap their opinions in the cloak of general acceptance by saying their deductions on causation were drawn based on the application of “things” like the Bradford Hill criteria, Koch’s postulates, and differential diagnosis. Hearing the buzz words, many judges admitted the novel causation opinions, leaving the debunking of the junk up to cross-examination. One large litigation over a now-exonerated medication related to birth defects allegedly caused by maternal use of Bendectin. Daubert came out of this context and the absence in the then-applicable version of Fed. R. Evid. 702 of any mention of “general acceptance” or “reliability.” In 2000, after Daubert had already spawned many notable progeny, Fed. R. Evid. 702 and 703 were amended to reflect something of the reality from the caselaw. Thirty years later—an even one hundred years after Frye—Fed. R. Evid. 702 has been revised again, in large part to remind judges that the “gatekeeping” responsibility articulated in Daubert is also a thing. A Daubert challenge is really a Rule 702 challenge now.
We start with that diatribe not simply because we tend to include a bit of blather before diving into the decision we are discussing, but because the exclusion of plaintiffs’ general causation experts in an apparent MDL-ending Rule 702 decision in In re Acetaminophen ASD-ADHD Prods. Liab. Litig., ___ F. Supp. 3d ___, 2023 WL 8711617 (S.D.N.Y. Dec. 18, 2023), is a worthy heir to the history above. For one thing, it relates to alleged adverse effects on the offspring of women who used a certain medication while pregnant. For another, the drug at issue has been used very widely for many decades and it is on the World Health Organization’s List of Essential Medicines. The drug and the conditions at issue in the litigation, autism spectrum disorder (ASD) and attention deficit hyperactivity disorder (ADHD) have been widely studied. A causal relationship is not accepted and, before the litigation, a mere association probably was not accepted. Yet, there have been hundreds of cases, millions spent on advertising, and an MDL established. In addition, although we generally do not speak of judges, we note that the judge overseeing this MDL took the federal bench in 1994—a year after Daubert—and had a, shall we say, interesting history with issues of preemption and the FDA over the first year or so of the MDL. This decision is also the first MDL-wide decision of this scope applying the new version of Rule 702. And, in case the foreshadowing was insufficient, the court’s gatekeeping function, often utilizing the Bradford Hill criteria, was on full display.
The facts and posture are fairly simple. After rejecting preemption at the motion to dismiss stage, the court solicited and FDA refused to provide its input on the plaintiffs’ purportedly adequate warning as to the risk of ASD and ADHD with a range of acetaminophen-containing OTC products. The parties then proceeded to expert discovery on general causation, which is a sensible thing to do in an MDL like this if it gets that far and plaintiffs do not get do-overs when they lose (see here and here). The plaintiffs here put together a strong slate of experts, at least based on credentials, with an epidemiologist who will be taking over as Dean of the Harvard School of Public Health as their lead. 2023 WL 8711617, *2. Consistent with the Acetaminophen decision itself, we will focus on the challenge to his general causation opinion, which plaintiffs tried to buttress with subject matter experts trying to address issues like mechanism and biological plausibility. We will not go into all the ins-and-outs of the science discussed, but the court started with some basics. That included that usage of acetaminophen by pregnant women is very common, both ASD and ADHD are fairly prevalent in U.S. children, and the vast majority of cases of ASD and ADHD are considered to due to heritable genetics. Id. at *4. Combined with no known genes or biological pathways for these conditions, and relatively few recognized risk factors for them, these facts make it highly unlikely that an old, widely-used drug would cause either of these conditions, let alone both.
The court next laid out the principles for interpreting causation from epidemiologic and animal studies. This recitation relied heavily on the Reference Manual on Scientific Evidence (3d ed. 2011), which might have been our third example above. Back when we started writing Daubert motions, the RMSE (as Acetaminophen calls it) was required reading. The first edition was published by the Federal Judicial Center months after Daubert was decided and it reflected a collaboration of prominent scientists, judges, and lawyers. The same is true for the two revisions since. Citing the RSME, the court stated, “It is important to note that epidemiology cannot prove causation; rather, causation is a judgment to be made by epidemiologists and others interpreting the epidemiologic data.” Id. at *7. Thus, the key question in Acetaminophen was “whether it is reliable to draw a causal inference from the associations that researchers have observed between prenatal acetaminophen exposure, ASD, and ADHD.” Id. This formulation reflects an understanding of what we noted above about the Bradford Hill criteria: There can be no proper causation analysis without predicate evidence from human studies, but a simple association reported in study does not supplant the need for an analysis. The court then proceeded to review the published studies. Boiled down, the vast majority of analyses in published studies did not find any increased risk of ASD or ADHD and those that did found modest increases, generally based on some combination of poor exposure ascertainment, inappropriate lumping of endpoints, and questionable subgroup analyses. (If we were to highlight an area where the court did not go far enough, then we might point out the need to identify when analyses were not predetermined looks at primary endpoints. In a paraphrase of the testimony of a difficult opposing epidemiology expert who passed away recently, “you have to say what you will do and then do what you say.” We have a strong suspicion that most of the increased risk calculations the plaintiffs’ experts liked were the result of post hoc slicing, dicing, and reassembling of the data (e.g., including all neurodevelopmental issues to get an increased risk in a study that was only designed to look at ASD).)
The next step was to identify the absence of any regulatory or professional body of note endorsing a causal relationship between acetaminophen use and ASD or ADHD. This included the FDA, which did not weigh in on plaintiffs’ proposed labeling but recently confirmed its view that “the limitations and inconsistent findings of current observational studies of APA and neurobehavioral and urogenital outcomes are unable to support a determination of causality.” Id. at *13. That FDA does not believe a true association requiring a warning exists is also obvious. Yet, three of plaintiff’s experts purported to have applied the Bradford Hill criteria to come to the opposite conclusion. “At no point have plaintiffs suggested that their experts could have satisfied their burden to offer reliable testimony of general causation without performing a reliable Bradford Hill analysis.” Id. at *22. This, along with the court’s emphasis on the proponent’s burden and the court’s gatekeeping function per the revision of Rule 702 (see id. at n.27), set the stage for the core analysis and this conclusion:
As explained infra, however, they have not used that literature to render discrete opinions regarding that exposure and the risk of ASD and the risk of ADHD. Instead, they have applied a “transdiagnostic” analysis that sweeps into their analyses (and conclusions) ASD, ADHD and other neurodevelopmental disorders. They have failed to show that their methodology in doing so is generally accepted by the scientific community. In any event, here, their analyses have not served to enlighten but to obfuscate the weakness of the evidence on which they purport to rely and the contradictions in the research. As performed by the plaintiffs’ experts, their transdiagnostic analysis has obscured instead of informing the inquiry on causation.
Id. at *16. For those who read many Daubert decisions, there is quite a bit packed into the above quote. Fundamentally, it reflects a rejection of the idea that a well-qualified expert can offer a novel general causation opinion just by citing some study findings and mouthing the magic words “Bradford Hill criteria.” Instead, the proponent of the expert evidence has to prove that each step of the expert’s analysis reflected a “reliable application of [reliable] principles and methods to the facts of the case.” Fed. R. Evid. 702(c)-(d).
The court walked through the opinions and the application of principles and methods to form them for each of plaintiffs’ five experts. As for the plaintiff’s epidemiologist, Dr. Baccarelli, he was not only well qualified but he had co-authored three studies on maternal acetaminophen use and various neurodevelopmental outcomes. This tied to one of his failings—shared by the other plaintiffs’ experts—in trying to rely on study data related to outcomes other than ASD or ADHD to fill in the gaps for the lack of study data supporting a relationship with ASD and ADHD. The plaintiffs called this a “transdiagnostic evaluation,” a gloss on the inappropriate lumping we discussed above. The court used plaintiffs’ term, but clearly saw the game being played. “After all, this litigation is brought to obtain recovery on behalf of those who have been diagnosed with ASD or ADHD, not on behalf of anyone with, for example, a deficit in communication or self-regulation.” Id. at *20. Yet Dr. Baccarelli never conducted separate and distinct analyses for ASD and ADHD. Dr. Baccarelli also cherry-picked when he would rely on a different endpoint, as his “assessment of a study’s use of non-ADHD, non-ASD endpoint seems to depend on whether the study’s result supports his ultimate opinion about a causal connection with prenatal exposure to acetaminophen.” Id. at *21. Thus, his “use of a transdiagnostic Bradford Hill analysis fail[ed] to meet the requirements for admissibility on the issue of causation for ASD or ADHD.” Id. at *23.
Nonetheless, the court proceeded to look at each of the nine criteria and how Dr. Baccarelli claimed to have applied them. The court found that he failed, or, rather, plaintiffs failed to show he reliably applied, seven of the nine criteria. Some of this may have been affected by the stubborn refusal to admit that more than one of the criteria weighed against causation, as consistency was clearly not met when most analyses in most studies found no increased risk and the strength of association was clearly quite low where all but one study found no statistically increased risk or an increased risk below 2.0. Id. at *23-29. He conceded that specificity was not met, but claimed temporality was met because by definition a child’s diagnosis of ASD or ADHD occurs after the maternal exposure (if any) while pregnant with the child. The court rightly saw this as sophistry, focusing on the absence of evidence or analysis that put the exposure in studies before the development of ASD or ADHD during pregnancy. Id. at *29. Baccarelli’s opinion on dose-response (or “biological gradient” per Bradford Hill or “biologic gradient” per Baccarelli) relied largely on another plaintiffs’ expert, whose opinion was itself excluded as unreliable, and otherwise“skate[d] over the complexities and limitations of the underlying literature.” Id. at *30. Baccarelli also tried to rely on the unreliable opinions of other plaintiffs’ experts in concluding that the biological plausibility (just “plausibility” per Bradford Hill) criterion was met. But “[s]cientists have at best developed hypotheses” about mechanisms for the development of either ASD or ADHD in utero. Id. Many courts have let the ability to articulate some hypothesis, no matter how unfounded, as satisfying this criterion, but Acetaminophen did not.
Plaintiffs carried their burden to show that Dr. Baccarelli’s coherence analysis was reliable. We will leave it at that, except to say pairing speculative mechanisms proposed for conditions with no known biological mechanisms does not sound very coherent. Dr. Baccarelli claimed that the analogy criterion was met because maternal use of a specific prescription drug has been associated with “adverse effects on neurodevelopment” per its label, but he could not explain why that was instructive about acetaminophen and ASD or ADHD. Id. at *31. The experiment criterion was largely punted given apparent agreement that human experimentation would be unethical and “the animal studies cannot bear the full weight of providing admissible evidence of causation in this case.” Id. (The two plaintiff experts who offered opinions based on cherry picking allegedly supportive animal study results and ignoring contrary studies were also excluded.)
The analysis of the other four plaintiffs’ experts is worth a read, but Dr. Baccarelli’s fate was clearly determinative of the admissibility of any general causation opinion. We wrap up with the court’s recitation of the excuses, er, arguments, offered by plaintiffs, all of which sought a lower bar for admissibility of a novel causation opinion:
At oral argument, the plaintiffs asked the Court to focus on the fact that Dr. Baccarelli is a preeminent epidemiologist, which he is. They ask that the Court ignore his published statements acknowledging the weakness in the literature, arguing that he has been correct to change his mind when rendering his opinion here. They stress the direction of the association evidence, ignoring those studies finding no association or a negative association. They argue that it is unnecessary to insist that a finding of association be statistically significant, arguing that a more flexible standard should be adopted. They contend that the limitations expressed by authors in their studies should be ignored as simply an overly conservative requirement that scientists impose on each other to get peer reviewed studies published. They suggest that the FDA’s surveillance of this issue since 2014 means little since the FDA was not vigilant in reviewing the risks associated with certain other drugs and that it has not performed a Bradford Hill analysis. These and more arguments like them do not relieve the Court of the obligation to scrutinize the methodology applied by Dr. Baccarelli to ensure that it is sufficiently rigorous to pass muster by the standards established by his discipline, Rule 702 and Daubert.
Id. at *35. To us, this reads like a summary of what plaintiffs often say about unfavorable science at trial. The only thing missing was accusing the defendants of buying the science. It also reminds us of the oft quoted scene from A Few Good Men where an attorney follows up on an overruled objection to excluding an expert opinion with a “strenuous” objection to the same opinion. The difference with Acetaminophen is that the plaintiff lawyers who created the litigation should have known from the start that they had very weak general causation evidence. Perhaps they expected the defendants to settle before that weakness was tested. While this litigation also would have been a good candidate for an early decisive preemption ruling in favor of the manufacturers, the court’s ruling rejecting what was clearly plaintiffs’ best shot should be equally decisive.