Rule 702 Limits on “Gatekeeping”: Courts May Not Exclude Generally Accepted Epidemiologic Reasoning as “Unreliable” Because They Disagree with the Conclusion

I. Introduction

These tandem appeals arise from an MDL in the Southern District of New York concerning over-the-counter acetaminophen (Tylenol and generics). Plaintiffs—children and their parents/guardians—assert state-law failure-to-warn claims alleging that prenatal acetaminophen exposure can cause neurodevelopmental disorders: attention-deficit/hyperactivity disorder (“ADHD”) and autism spectrum disorder (“ASD”). Defendants include manufacturers, retailers, and pharmacies involved in producing and selling acetaminophen products.

The litigation turned on general causation—whether acetaminophen is capable of causing ADHD/ASD in the population—proved primarily through epidemiology. The district court excluded all plaintiffs’ general-causation experts under Federal Rule of Evidence 702 and Daubert v. Merrell Dow Pharms., Inc., then granted summary judgment. The Second Circuit largely reversed: it held the district court exceeded its gatekeeping role as to three experts, but properly excluded two others. It also rejected defendants’ argument that federal drug-labeling law preempted plaintiffs’ failure-to-warn claims.

II. Summary of the Opinion

• Expert admissibility (Rule 702 / Daubert): The Second Circuit held the district court abused its discretion by excluding the testimony of Drs. Baccarelli, Hollander, and Pearson, because they applied generally accepted scientific methods (including Bradford Hill reasoning) with the kind of discretion and debate typical of scientific practice. The court emphasized that Rule 702 demands reliability, not “correctness,” and that scientific disagreement does not render an opinion inadmissible.
• Proper exclusions: The district court did not abuse its discretion in excluding Dr. Cabrera (failure to weigh/synthesize Bradford Hill factors) and Dr. Louie (unsupported extrapolation regarding dose/duration thresholds; mischaracterization of study endpoints; “analytical gap” under Gen. Elec. Co. v. Joiner).
• Phippen remand: The court vacated and remanded Phippen without deciding whether Dr. Ness should be admitted, noting the district court may revisit her admissibility given the reinstatement of other expert evidence.
• Preemption: The Second Circuit agreed with the district court that federal law does not preempt the failure-to-warn claims. The required general OTC pregnancy warning in 21 C.F.R. § 201.63 does not forbid supplemental pregnancy-related warnings (e.g., about ADHD/ASD risks). “Impossibility pre-emption is a demanding defense” under Wyeth v. Levine, and defendants did not meet it.

III. Analysis

A. Precedents Cited

1. The Daubert trilogy and the Second Circuit’s reliability framework

• Daubert v. Merrell Dow Pharms., Inc.: The court reaffirmed that district courts act as gatekeepers to ensure expert testimony rests on a reliable foundation and fits the issues. Importantly, the inquiry focuses on “principles and methodology,” not mere acceptance of the expert’s bottom-line conclusion.
• Kumho Tire Co. v. Carmichael: The Second Circuit relied heavily on Kumho’s command that experts must employ “the same level of intellectual rigor” in court as in the field. Here, that principle cut against exclusion: the panel concluded the district court penalized experts for using modes of reasoning that appear in real-world epidemiology.
• Gen. Elec. Co. v. Joiner: The panel invoked Joiner’s caution that courts may exclude opinions connected to data only by an expert’s ipse dixit, and its “analytical gap” concept. This precedent supported affirmance of Louie’s exclusion (unexplained extrapolation from “>28 days” cohorts to a claimed 28-day threshold risk).
• Amorgianos v. Nat'l R.R. Passenger Corp. and Ruggiero v. Warner-Lambert Co.: These Second Circuit cases supplied the operational test: a rigorous examination of the facts relied upon, the method, and its application; and exclusion where studies or methodology are inadequate to support the conclusions. The panel applied this to distinguish (i) debatable scientific judgments (admissible) from (ii) methodological non-synthesis and unexplained extrapolation (inadmissible).

2. Bradford Hill in products litigation

• Sarkees v. E.I. Dupont De Nemours & Co. and In re Zoloft (Setraline Hydrochloride) Prods. Liab. Litig.: These cases were cited for the proposition that Bradford Hill reasoning is a generally accepted way to structure causal inference from observational evidence. The Second Circuit used them to frame the dispute: the question was not whether plaintiffs proved causation, but whether their experts used accepted methods reliably.
• Daniels-Feasel v. Forest Pharmaceuticals, Inc. and In re Mirena (No. II) (and related district-court Mirena decisions): These cases influenced the panel’s treatment of factor “weighting.” The Second Circuit endorsed the idea—prominent in Mirena/Zoloft—that Bradford Hill is a methodology only if the expert explains how factors are weighed/synthesized, which supported excluding Cabrera. But the panel refused to let “weighting” become a vehicle for courts to impose their own scientific definitions or settle scientific debates, which supported admitting Baccarelli/Hollander/Pearson.

3. Qualifications and the scope of expertise

• United States v. Tin Yat Chin: The panel cited this for how courts assess whether expertise matches the subject matter. Here, qualifications were not contested (“eminently qualified”).
• McCullock v. H.B. Fuller Co. and the district-court observation in In re Mirena IUD Prods. Liab. Litig.: These authorities supported the point that experts can rely on reviewing literature (not necessarily conducting their own studies) and still testify reliably.

4. Preemption doctrine in drug/warning cases

• Wyeth v. Levine: Central to the analysis. Wyeth supplies (i) the presumption against preemption in areas of traditional state police power and (ii) the principle that manufacturers retain responsibility for label adequacy. The Second Circuit treated the FDA’s general pregnancy warning requirement as non-exclusive—i.e., compatible with additional warnings.
• PLIVA, Inc. v. Mensing: Used for background on FDA approval pathways (NDA) and label control. The panel distinguished the monograph context and emphasized that OTC monographs do not necessarily entail preapproval of exact label text.
• Gibbons v. Bristol-Myers Squibb Co., Marentette v. Abbott Laboratories, and Medtronic, Inc. v. Lohr: These framed impossibility preemption and the presumption against “cavalier” displacement of state remedies.
• Safe Haven Home Care, Inc. v. United States Dep't of Health & Hum. Servs.: Cited to reinforce that unambiguous regulations control without deference to contrary agency interpretations—a key move when the panel refused to read the pregnancy-warning regulation as silently forbidding supplemental warnings.

5. Additional cited persuasive authority on risk ratios and plausibility

• Hardeman v. Monsanto Co.: Cited to rebut a rigid “risk ratio > 2.0” admissibility rule and to validate that smaller relative risks can still support causal inference when properly analyzed.
• Milward v. Acuity Specialty Prods. Grp., Inc.: Used for a pragmatic, science-consistent framing of “biological plausibility”: the mechanism need not be proven with certainty for plausibility to count within causal reasoning.

B. Legal Reasoning

1. The opinion’s central evidentiary rule: “Reliability” is not “correctness”

The Second Circuit anchored its reversal in the 2023 amendment to Rule 702 (committee note), emphasizing that courts should not exclude experts merely because the judge believes one side has the better interpretation of contested facts. The proponent must show the opinion is reliably reached—not that it is ultimately true.

2. Policing the boundary between gatekeeping and factfinding

The panel repeatedly faulted the district court for “substituting its own understanding” of epidemiological criteria and for treating scientific uncertainty as a disqualifier. The court reframed the gatekeeping role: a district judge should screen out “junk science,” but should not resolve open scientific disputes by declaring one side’s interpretation “unreliable” where it is within ordinary scientific discourse.

3. Admissible: transdiagnostic Bradford Hill and symptom endpoints

A key methodological dispute was Dr. Baccarelli’s “transdiagnostic” approach: one Bradford Hill analysis for ADHD, ASD, and related symptoms (symptom endpoints plus diagnostic endpoints). The Second Circuit held this was not per se unreliable. It stressed:

• Scientists (including FDA reviews and defendants’ internal reviews) examine broad neurodevelopmental outcomes together.
• Symptom-endpoint studies are used in the relevant literature; using them goes to weight, not admissibility, where the expert explains the choice.
• Concerns that combining outcomes “obscures limitations” are, in this context, for cross-examination and competing experts—not exclusion.

4. Admissible: flexible, literature-consistent use of Bradford Hill factors

The panel rejected what it viewed as judicially imposed “requirements” for Bradford Hill factors (e.g., demanding precise dosage data for dose-response; demanding known disease mechanisms for biological plausibility). It treated such demands as inconsistent with how observational epidemiology actually functions, particularly where randomized trials are unethical or impossible.

5. Inadmissible: failure to synthesize and explain “weighting” (Cabrera)

Cabrera’s Bradford Hill opinion was excluded not because Bradford Hill was invalid, but because he did not explain how the factors worked together—leaving the factfinder without a coherent method to evaluate what happens if one factor is weak or contested. This is the opinion’s clearest endorsement of the Mirena/Zoloft principle that Bradford Hill must be presented as an actual method, not a checklist leading to a conclusion.

6. Inadmissible: analytical gap and mischaracterization (Louie)

Louie’s dose/duration opinion failed for classic Joiner reasons: he asserted a “clear answer” that 28 cumulative days increases risk two-fold, but his cited studies examined broad “>28 day” groups (28 to 280 days). The court treated his step from “group average risk among a wide exposure range” to “a threshold at day 28” as an unexplained extrapolation. He also misstated endpoints (citing ASD where studies assessed only ADHD or symptoms), reinforcing unreliability.

7. Preemption: the pregnancy warning is a floor, not a ceiling

On preemption, the panel treated the text as decisive: 21 C.F.R. § 201.63 mandates a general pregnancy warning verbatim, but does not prohibit an additional, specific warning about plausible risks. Because manufacturers could comply with both federal law and any state-law duty by keeping the mandated warning and adding supplemental language, impossibility preemption was not shown. The panel also noted that Congress’s OTC labeling preemption provision, 21 U.S.C. § 379r, contains a product-liability carveout that undercuts broad preemption arguments.

C. Impact

1. Practical effect in the acetaminophen MDL and similar mass torts

• More expert testimony reaches juries in scientifically contested fields: When experts apply methods commonly used in the literature (even amid controversy), exclusion becomes harder to justify.
• Trial courts must distinguish “debatable inference” from “methodological failure”: The opinion provides a blueprint: admit where the expert (i) identifies relevant literature, (ii) acknowledges limitations, and (iii) provides scientifically plausible reasons for interpreting the data; exclude where the expert (a) misstates evidence, (b) fails to synthesize Bradford Hill factors, or (c) leaps from data to conclusion without explanation.
• Encourages robust adversarial testing: The court’s repeated emphasis on cross-examination and competing experts signals that, in close scientific disputes, the courtroom should function more like a forum for competing scientific narratives than a judicial resolution of the scientific question.

2. Doctrinal effect: Rule 702’s “hard look” without outcome-driven gatekeeping

The decision is a notable Second Circuit statement on the post-2023 Rule 702 landscape: it confirms that judges must take a “hard look” at application, but it rejects “hard look” as a license to enforce the judge’s own epidemiology or to demand forms of proof observational science cannot usually supply.

3. Preemption effect for OTC monograph drugs

Defendants in OTC-warning cases may find it harder to convert “exact language” requirements for mandated warnings into a general bar on supplemental risk disclosures. The opinion reads the pregnancy warning regulation as compatible with additional warnings—placing significant weight on textual limits and on the demanding nature of impossibility preemption.

IV. Complex Concepts Simplified

• General causation vs. specific causation: General causation asks “Can this product cause this injury at all?” Specific causation asks “Did it cause this plaintiff’s injury?”
• Epidemiological association vs. causation: An association means two things occur together more than expected by chance; causation means one contributes to producing the other. Observational studies typically establish association; Bradford Hill reasoning helps evaluate whether the association is likely causal.
• Bradford Hill criteria: A set of commonly used considerations (consistency, strength, dose-response, plausibility, etc.) to assess whether an observed association is causal. They are not a checklist where every box must be checked.
• Cherry-picking (in expert analysis): Unreliable selection of only supportive studies while ignoring contrary ones. The court clarified that it is not cherry-picking to prefer some studies when the expert gives a coherent scientific explanation and addresses limitations.
• “Analytical gap” (Joiner): A logical leap between data and conclusion that the expert does not explain—making the opinion ipse dixit rather than reasoned inference.
• Impossibility preemption: State law is preempted only if it is impossible to comply with both state and federal law. If a manufacturer can satisfy federal requirements and still add a warning to satisfy state duties, preemption usually fails.

V. Conclusion

Rutledge v. Walgreen Co.; Phippen v. Walgreen Co. meaningfully tightens (and clarifies) the boundary between legitimate Rule 702 gatekeeping and judicial resolution of scientific disputes. The court held that when qualified experts apply generally accepted epidemiologic methodologies and give reasoned explanations—especially in an area of ongoing scientific debate—their opinions are ordinarily admissible even if contestable. At the same time, the decision confirms that courts may (and should) exclude experts who fail to synthesize Bradford Hill factors into an intelligible method or who jump from studies to conclusions without explaining the inferential steps. Finally, the opinion rejects broad preemption arguments based on the FDA’s general OTC pregnancy warning requirements, treating them as non-exclusive and compatible with supplemental risk warnings under state tort law.