On July 13, 2026, in an opinion by Senior Judge Guido Calebresi, the Second Circuit delivered its opinion. Given Judge Calebresi’s confusion at oral argument, between causation and association, and his conflation of precautionary action with causal assessment, informed readers knew the likely outcome of the 64 page opinion once they reached the authorship line. The appellate court’s opinion was a rebuke of the MDL Judge Cote for supposedly exceeding and abusing her discretion and rendering a “manifestly erroneous” ruling, as to three of the plaintiffs’ expert witnesses, Drs. Baccarelli, Hollander, and Pearson.[1] The Circuit upheld the lower court’s exclusion of two other plaintiffs’ expert witnesses, Drs. Cabrera and Louie. In reaching these conclusions, the Circuit opinion ran roughshod over the “highly deferential abuse of discretion standard” that applies to a district court’s decision to exclude expert testimony.[2]
The appellate court also ducked plaintiffs’ verkakte argument on separation of powers, by noting that the issue sub judice is “not political. It is not about positions taken by elected officials or political appointees.”[3] Inquiring readers may still wonder whether the brandishing of political opinions did not serve as a brush-back pitch that tilted the scales of analysis.
Judge Calebresi’s opinion acknowledged that there was a distinction between studies or reviews that claimed an association and those that put forward a causal claim,[4] and that he and his fellow judges were not deciding the general causation question. Instead, the court framed the issue as “how closely a trial court may scrutinize a qualified expert’s conclusions when that expert follows methodologies that are generally accepted in their field, and the standard of reliability required for the admission of expert testimony on issues that are the subject of ongoing scientific debate.”[5]
Although neither the district nor the circuit court was called upon to decide the causation issue, the evidentiary issue before them required an evidentiary assessment of whether an expert witness could reach the causal conclusion on the available evidentiary display. The appellate court improperly elided this key question, and declared that the trial court had abused its discretion in trying to answer the relevant question.
Although the appellate court block quoted the most recent version of Rule 702, its evidentiary analysis proceeds almost exclusively on the basis of case law that precedes the 2023 revision to Rule 702. Subsection (b) of Rule 702, which requires that “the testimony is based on sufficient facts or data,” is included in the block quote, but the appellate decision never addresses this specific prong of the legal standard to explain how a completely uncertain causal relationship can be allowed to masquerade as an expert witness opinion of causation. Baccarelli had very publicly acknowledged the insufficiency of the data to support a causal conclusion, two years after submitting an expert report in which he asserted causality. Given that the court was well aware of Bacarelli’s duplicity and his public acknowledgements of data insufficiency, its failure to address the admitted insufficiency of the facts and data to support the asserted claim was itself manifestly erroneous.
Judge Calebresi reviewed the preceding state of the art on the causal questions. His opinion acknowledged that scientists had been investigating a potential relationship between prenatal acetaminophen exposure and neurocognitive disorders for decades without any study or studies having established a causal relationship. The American Psychiatric Association, as recently as 2022, declared that precise causes of ADHD and ASD were unknown.[6]
Nor was the acetaminophen case an instance of the FDA’s having been asleep at the wheel, as the lawsuit industry so often claims. The court’s decision recounts how the FDA opened a Tracked Safety Issue for the acetaminophen issue in 2014, and had conducted periodic reviews up to, and including, one in 2023.[7] The FDA consistently and repeatedly rejected a causal interpretation of the available data. The appellate court also noted that the advocacy scientists’ self-styled consensus 2021 call for precautionary action, which incited the lawsuit industry, stopped far short of concluding a causal association: “it notably did not conclude that the available data allowed for an inference that a causal relationship exists.”[8] The circuit court’s opinion thus laid a strong foundation for a searching inquiry into the adequacy of the evidentiary base for Baccarelli’s putative conclusions. The opinion duly cited the Supreme Court for the propositions that “conclusions and methodology are not entirely distinct from one another,” and that “nothing in either Daubert or the Federal Rules of Evidence requires a district court to admit opinion evidence that is connected to existing data only by the ipse dixit of the expert.”[9] Previous cases within the Second Circuit recognized that “when an expert opinion is based on data, a methodology, or studies that are simply inadequate to support the conclusions reached, Daubert and Rule 702 mandate the exclusion of that unreliable opinion testimony.”[10] Judge Calebresi’s opinion evaded this Supreme Court precedent.
More important than Supreme Court precedent, the requirement of sufficient data is statutorily embodied in Rule 702(b), which mandates that “the testimony is based on sufficient facts or data.” Baccarelli’s failure to satisfy Rule 702(b) is manifested in Judge Calebresi’s own review of the state of the art, and his declaration that “[t]here is no doubt that any causal relationship between acetaminophen and ADHD and ASD is quite uncertain.”[11] The appellate decision never engaged with this specific requirement of Rule 702(b). Instead, the Circuit’s opinion cited decisions that predated the 2023 revision to Rule 702, including some notorious decisions that erroneously failed to apply Rule 702(b).[12] Given that the causal claim is quite uncertain, there must of necessity be serious questions about how any expert witness could arrive at a conclusion in support of the claim. These questions are largely unanswered in Judge Calebresi’s opinion.
Judge Calebresi focused principally on the opinions of Dr. Andrea Baccarelli, which was understandable at least in that Baccarelli had given the most comprehensive opinion of all the plaintiffs’ expert witnesses, and that he was the plaintiffs’ most credentialed witness.[13] Baccarelli proffered two epidemiologic approaches to get to his litigation conclusions, the Bradford Hill “criteria,” and the so-called navigation guide review.[14] The appellate court considered only the Bradford Hill analysis, which it ultimately found to be sufficient to satisfy Rule 702. The appellate opinion thus never considered the navigation guide approach, because it reached the same conclusion.[15] Presumably, the decision not to review the district court’s exclusion of the navigation guide opinion leaves the lower court’s exclusion in place. The decision not to review the navigation guide opinion allowed the appellate court to evade the full force of the fraudulent disconnect between Baccarelli’s for-litigation navigation guide review and his peer-reviewed, published version. The latter embraced a causal conclusion; the former explicitly declined to offer a causal claim. Perhaps Judge Calebresi was throwing a life line to plaintiffs and Baccarelli by helping them avoid the blatant contradiction. The problem for plaintiffs, however, is that the Bradford Hill analysis was similarly affected because Baccarelli spoke very publicly within hours of the Trump news conference (Sept. 22, 2025) to say that causation had not been established.[16] The scientific and legal community would have reasonably expected Baccarelli to note that, although the navigation guide could not reach a causal conclusion, he had some alternative methodological pathway to support his precautionary principle policy recommendations as causal conclusions. Baccarelli’s statement along with his silence on alternative approaches created a very strong negative inference that was real and palpable.
Judge Calebresi’s opinion thus focused on whether Baccarelli’s Bradford Hill analysis satisfied Rule 702(c) and (d), which requires that the proffered “testimony is the product of reliable principles and methods” and that the challenged expert reliably applied these reliable principles and methods to the facts in the case. The appellate court embarked on an attempt to parse the Bradford Hill factors, but the conclusion was already clear from the disconnect between Baccarelli’s litigation position and his statements in the published Prada review and his September 2025 pronouncements to the public. The Supreme Court, in one of its key decisions that led to the revision of Rule 702 in the year 2000, observed that the important goal of judicial gatekeeping was “to make certain that an expert, whether basing testimony upon professional studies or personal experience, employs in the courtroom the same level of intellectual rigor that characterizes the practice of an expert in the relevant field.”[17] Baccarelli had exhibited a deliberate disdain for the fact-finding objectives of the judicial process by proffering an opinion in court that he would not embrace or defend in either a scientific journal or in a public statement.
The appellate court chose to ignore the obvious disconnect between Baccarelli’s for-litigation statements and his for-scientists communications. Instead of facing the disconnect, Judge Calebresi proceeded to attempt a defense of Baccarelli’s application of the Bradford Hill factors. In conducting this otiose exercise, the appellate court presented Baccarelli’s method largely as a consideration of nine factors or considerations, none of which is essential or dispositive.[18] The court’s consideration of Bradford Hill’s factors failed to mention that that Bradford Hill had specified a fundamental predicate to those nine factors for assessing whether an association was causal:
“Disregarding then any such problem in semantics we have this situation. 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?”[19]
Bradford Hill makes clear that the predicate analysis is a prerequisite for engaging with the nine factors. Clear-cut associations are ones that not reasonably attributable to random or systematic error or confounding. Even the precautionary International Agency for Research on Cancer uses a causation analysis that requires associations to be “clear-cut,” and free from bias and confounding before declaring the epidemiologic evidence sufficiently to support a causal association.[20]
Judge Calebresi mentions the predicate of eliminating bias and confounding as what seems a mere afterthought; the parties do not clearly discuss the predicate as a required showing before considering the nine factors.[21] Baccarelli himself, in his for-litigation report, acknowledges the predicate to the Bradford Hill factors, but then asserts that the predicate was satisfied, a rather clear ipse dixit.[22]
Essential to Baccarelli’s approach to the Bradford Hill analysis was his use of transdiagnostic categories. Basically, Baccarelli wanted to include studies that lumped clear diagnoses with isolated symptoms of conditions, and to mix and match conditions to create a larger body of evidence from which he could select studies to support his claim. The district court recognized the dangers of “lumping” together distinct diseases, with different diagnostic criteria, as well as isolated signs and symptoms.[23] The selectivity becomes an even greater threat to validity given that Baccarelli worked without a review protocol.
The appellate court tried to resuscitate Baccarelli’s approach by its own ipse dixit. In Baccarelli’s own words:
“different NDDs [neurodevelopmental disorders] often have shared/overlapping symptomology as well as shared biological pathways or causes . . . . Therefore, in assessing the association between acetaminophen exposure in utero and neurodevelopmental disorders like ASD and ADHD, it is important to consider not just the clinical diagnoses but also the neurodevelopmental outcomes, including the symptoms.”[24]
While Baccarelli may be correct that it is important to consider clinical symptoms in a review, nothing in his attempted justification supports lumping diseases with isolated symptoms to reach conclusions with respect to the disease outcome of interest. Cancer is, for instance, not a single disease, but there are well understood “hallmarks” of cancer that are common to cancers generally.[25] These common hallmarks, however, would not validate a causal inference for any particular cancer from a study that examined all cancers as a composite end point. Nor would any epidemiologist today think it was appropriate to conduct a study of lung cancer by lumping pathologic diagnoses with self-reported complaints of shortness of breath.
There is no lack of cautionary warnings about the use of composite end points in epidemiology. One popular clinical journal urges scientists and clinicians to “[b]eware composite endpoints.”[26] Standard texts across clinical fields warn against the casual “lumping” of end points, especially when the included end points are not of similar importance or known to have similar causes or cures.[27]
In trying to vindicate Baccarelli, the appellate court erroneously declared that outcomes considered together allow causal conclusions across lumped disease-symptoms composites. In a footnote, the court cited three studies in which scientists supposedly created and used similar broad “transdiagnostic” or composite outcomes. None of the three studies even remotely supported the court’s attempt.[28] A fourth example in the text was equally inapposite.
The first example was a review by a well-known scientist, John Ioannidis, who looked at air pollution as a potential cause of various, disparate mental illnesses.[29] Ioannidis did lump disparate diagnoses together, but because he identified substantial shortcomings in the data, including heterogeneity of outcomes, he refrained from any causal conclusions.
The court’s second example was a systematic review that addressed irritable bowel disease and isolated symptoms together. Once again, the authors refrained from any causal conclusions.[30]
The third example cited by the court is perplexing because it appears to look at only at a single end point – oral cancers, without the transgressive transdiagnostic approach under judicial scrutiny.[31] The only relevancy seems to be that the review purports to employ an approach based upon Bradford Hill. The paper is replete with errors, but in any event, it does not use reach any causal conclusions and it does not employ transdiagnostic outcome variables for human subjects.
The fourth example advanced by the appellate court was more pertinent by subject matter – a review of acetaminophen and ADHD and ASD.[32] Although the review did evaluate end points relevant to the litigation, the authors evaluated constitutive symptoms of the two diagnostic categories as well. Notably, this review’s Bradford Hill analysis did not yield a causal conclusion.[33] All four citations originated in the plaintiffs’ appellate brief, which the appellate court credulously accepted and repeated. The existence of reviews that “considered” a mix of symptoms and diagnostic end points did not warrant Baccarelli’s methodology to conclude that a single exposure caused the entire grab bag of symptoms and diagnoses.
(to be continued)
自由西藏
[1][1] Rutledge v. Walgreen Co., Nos. 24-916-cv(L), 24-1121(Con), 24-2360(Con); 24-2594-cv, Slip Op. at 7, 27 (2nd Cir. July 13, 2026), available at https://scholar.google.com/scholar_case?case=13067399726798054350&q=rutledge+v.+walgreen&hl=en&as_sdt=2006.[cited as Slip op.]
[2] In re Mirena IUS Levonogorstrel-Related Prods. Liab. Litig., 982 F.3d 113, 122 (2d Cir. 2020).
[3] Slip op. at 6.
[4] Slip op. at 13.
[5] Slip op. at 6-7.
[6] Slip op. at 10, citing American Psychiatric Association, DIAGNOSTIC & STATISTICAL MANUAL OF MENTAL DISORDERS at 64 (ASD), 71 (ADHD) (5th ed., Text Revision, 2022).
[7] Slip op. at 10, citing Abraham et al., Functional Neurobehavioral Outcomes and Urogenital Outcomes Associated with Prenatal Acetaminophen Exposure, U.S. Food and Drug Administration (July 15, 2022), at 33; Abraham et al., Updated Literature Review of 18 Studies that Examine the Association between Acetaminophen Exposure During Pregnancy and Neurobehavioral or Urogenital Outcomes, U.S. Food and Drug Administration (March 10, 2023), at 17-18.
[8] Slip op. at 10, citing Ann Z. Bauer, et al., Paracetamol use during pregnancy — a call for precautionary action, 17 NATURE REV. ENDOCRINOL. 757 (2021).
[9] Slip op. at 26, citing General Electric Co. v. Joiner, 522 U.S. 136, 146 (1997).
[10] Slip op. at 26, citing Ruggiero v. Warner-Lambert Co., 424 F.3d 249, 255 (2d Cir. 2005).
[11] Slip op. at 35.
[12] See, e.g., Slip op. at 36, citing Milward v. Acuity Specialty Prods. Grp., Inc., 639 F.3d 11, 25 (1st Cir. 2011), cert. denied sub nom., U.S. Steel Corp. v. Milward, 565 U.S. 1111 (2012). The Milward decision has been critiqued in several places including in my review of the fourth edition of the Reference Manual. Nathan Schachtman, REVIEW: Expert Witness Testimony Admissibility in the Reference Manual on Scientific Evidence (4th ed.), Working Paper (Mar. 2026), DOI: 10.2139/ssrn.6480438 (citing other critiques as well).
[13] Id. at 28.
[14] Id. at 20.
[15] Id. at 50 n. 28.
[16] Andrea Baccarelli, Statement (Sept. 22, 2025), cited in Ashley Keller, Letter Brief on behalf of Plaintiffs-Appellants, In re Acetaminophen – ASD-ADHD Prods. Liab. Litig., Appeal Nos. 24-916(L), 24-1121 (con), 24-2360 (con), docket entry 226.1, at 74 (Sept. 24, 2025).
[17] Slip op. at 26, citing and quoting with obvious approval but with contumacious disobedience from Kumho Tire Co. v. Carmichael, 526 U.S. 137, 152 (1999).
[18] Slip op. at 14 (drawing on the discussion of Bradford Hill’s presentation in the new edition of the Reference Manual’s epidemiology chapter ); Steve C. Gold, Michael D. Green, Jonathan Chevrier, & Brenda Eskenazi, Reference Guide on Epidemiology, in National Academies of Sciences, Engineering and Medicine & Federal Judicial Center, REFERENCE MANUAL ON SCIENTIFIC EVIDENCE 897, 974-985 (4th ed. 2025). Brenda Eskenazi is a member of the Collegium Ramazzini.
[19] Austin Bradford Hill, The Environment and Disease: Association or Causation?, 58 PROC. ROYAL SOC’Y MED. 295, 295 (1965) (emphasis added).
[20] IARC MONOGRAPHS ON THE IDENTIFICATION OF CARCINOGENIC HAZARDS TO HUMANS – PREAMBLE at 31, (2019) (requiring that evaluators must rule out chance, bias, and confounding “with reasonable confidence,” before finding sufficient evidence to support causality), at https://perma.cc/LJE4-K7HH.
[21] Slip op. at 15. See Frank C. Woodside, III & Allison G. Davis, The Bradford Hill Criteria: The Forgotten Predicate, 35 THOMAS JEFFERSON L. REV. 103 (2013). Judge Calebresi’s downplaying of the predicate for considering the Bradford Hill factors can be traced to the inversion of the analysis in the Reference Manual; see Green, et al., supra note 18.
[22] Andrea Baccarelli, Rule 26 Litigation Report – The Causal Relationship Between Prenatal Acetaminophen Use, Neurodevelopmental Disorders (NDD), Attention?Deficit/Hyperactivity Disorder (ADHD), and Autism Spectrum Disorder (ASD) (Oct. 10, 2023), Document 1261-1, at 23, in Case 1:22-md-03043-DLC, In re Acetaminophen – ASD-ADHD Prods. Liab. Litig. (S.D.N.Y. 2023).
[23] 707 F. Supp. 3d at 340 (noting that although ADHD and ASD share some signs or symptoms, “the diagnostic criteria of the two disorders are undeniably distinct.”
[24] Slip op. at 29, citing App’x 1778.
[25] Douglas Hanahan, Hallmarks of Cancer: New Dimensions, 12 CANCER DISCOVERY 31 (2022).
[26] Montori, et al., Users’ guide to detecting misleading claims in clinical research reports, 329 BRIT. MED. J. 1093 (2004).
[27] See Wolfgang Ahrens & Iris Pigeot, eds., HANDBOOK OF EPIDEMIOLOGY 1840 (2d ed. 2014) (47.5.8 Use of Composite Endpoints); Kip, et al., The problem with composite end points in cardiovascular studies, 51 J. AM. COLL. CARDIOL. 701 (2008).
[28] Slip op. at 31 n.10.
[29] John P.A. Ioannidis, Air Pollution as Cause of Mental Disease: Appraisal of the Evidence, 17 PLOS BIOL. 1 (2019) (evaluating the connection between “air pollution” and “mental disease,” including depression, bipolar disorder, schizophrenia, and personality disorder).
[30] Schoultz et al., Assessment of Causal Link Between Psychological Factors and Symptom Exacerbation in Inflammatory Bowel Disease: A Systematic Review Utilising Bradford Hill Criteria and Meta-Analysis of Prospective Cohort Studies, 9 SYSTEMATIC REVS. (2020).
[31] Raj et al., Reviewing the Oral Carcinogenic Potential of E-Cigarettes Using the Bradford Hill Criteria of Causation, 9 TRANSLATIONAL CANCER RESEARCH 3142 (2020).
[32] Slip op. at 31.
[33] See Alemany et al., Prenatal and Postnatal Exposure to Acetaminophen in Relation to Autism Spectrum and Attention-Deficit and Hyperactive Symptoms in Childhood, 36 EUR. J. EPIDEMIOL. 993, 1000 (2021).
