In our recent post describing the best prescription medical product liability litigation decisions of 2023, no fewer than three of the opinions we discussed:  #4 In re Acetaminophen-ASD-ADHD Products Liability Litigation, ___ F. Supp.3d ___, 2023 WL 8711617 (S.D.N.Y. Dec. 18, 2023), #5 Onglyza Products Cases, 307 Cal. Rptr.3d 480 (Cal. App. 2023), and #16 Daniels-Feasel v. Forest Pharmaceuticals, Inc., 2023 WL 4837521 (2d Cir. July 28, 2023), involved courts rejecting attempts by plaintiff-side experts who had abused the so-called “Bradford Hill” analysis to claim medical causation where the science simply did not support a causal conclusion.  That’s not new.  Bradford-Hill abuse has figured in several of our previous “best of” posts.

That’s indeed unfortunate.  As another of our recent posts detailed, Sir Austin Bradford Hill was one of the most distinguished epidemiologists of the 20th Century.  He certainly would not have appreciated that his purported p-side acolytes have so abused his signature methodology to the point that it’s become a by-word on our side for flabby methodology that necessitates rigorous Rule 702 (or state equivalent) scrutiny.

As of December, 2023, moreover, the federal judiciary intentionally toughened up Rule 702.  One of the four substantive Rule 702 criteria was edited to emphasize that admissibility is contingent upon experts not only using valid methodologies, but requiring that they apply such methodologies reliably in the context of the specific case.  Amended Rule 702(d) now requires that “the expert’s opinion reflects a reliable application of the principles and methods to the facts of the case.”

We think that should matter with the Bradford-Hill criteria.  A major reason that this form of analysis has fallen into disrepute is that plaintiffs’ experts purport to apply its criteria in situations where causation, or even association, is highly questionable.  The Acetaminophen MDL was a classic example, where the FDA had independently looked at the science no fewer than six times (in 2014, 2015, 2016, 2017, and 2022, and 2023) and each time concluded that the science did not justify any warning.  In re Acetaminophen − ASD-ADHD Products Liability Litigation, 2023 WL 8711617, at *12 (S.D.N.Y. Dec. 18, 2023).

Minimal epidemiological support is simply not a situation where resort to causation analysis using the Bradford-Hill criteria is appropriate.  Who says so?  Sir Austin, that’s who.  In his original paper proposing this methodology, A. Bradford-Hill, The Environment & Disease:  Association or Causation, 58 Proc. Royal Soc. Med. 295 (1965), he expounded on when it was appropriate to invoke his nine factors as a method to extrapolate from association to causation.  The question Sir Austin sought to answer was “In what circumstances can we pass from this observed association to a verdict of causation?”  Id. at 295.

So, at minimum there must be an “observed association.”  But Sir Austin understood that this bare minimum was not enough.  Before one can even start an analysis with his nine criteria, the “association” must be more than merely “observed” – it must be “clear-cut”:

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?

Environment & Disease, 58 Proc. Royal Soc. Med. at 295 (emphasis added).

Thus, assessing the Bradford-Hill criteria in the terms of Rule 702, before any expert may “reliably apply” the Bradford-Hill methodology “to the facts of the case,” the proposed causal relationship must be “perfectly clear-cut and beyond what we would care to attribute to the play of chance.”  The Bradford-Hill article contains an example of such a “clear-cut” relationship:

From age 30 to age 60 the cardroom workers suffered over three times as much from respiratory causes of illness whereas from non-respiratory causes their experience was not different from that of the other workers. . . .  The evidence was so clear cut, the differences between the groups were mainly so large, the contrast between respiratory and non-respiratory causes of illness so specific, that no formal tests could really contribute anything of value to the argument.

Environment & Disease, 58 Proc. Royal Soc. Med. at 300 (emphasis added). 

That situation is when the author of the eponymous Bradford-Hill methodology thought his analysis was appropriate and would lead to reliable causation criteria.  His methodology is not appropriate to be used in the kind of muddled situations, such as Acetaminophen, where authors of relevant studies caution that “more work is needed to rule out confounding by indication and to assess generalizability.”  2023 WL 8711617, at *19 (citation and quotation marks omitted).

As Acetaminophen and many other Rule 702 decisions hold, it is not a reliable methodology:

to press conclusions that study authors are not willing to make.  This willingness creates an “analytical gap” between the conclusions reached by the authors and the conclusions [an expert] draws from their work.

Id. at *33.  That “analytical gap” language, of course, comes from General Electric Co. v. Joiner, 522 U.S. 136, 146 (1997).  Relying on published articles for propositions that their authors “were unwilling to say” was precisely what Joiner condemned.  Id. at 145.  Joiner “placed special emphasis on the fact that the authors of two of the epidemiological studies relied on by the plaintiff’s experts were unwilling to conclude that those studies established a causal link.”  Amorgianos v. National  Railroad Passenger Corp., 137 F. Supp.2d 147, 165 (E.D.N.Y. 2001), aff’d, 303 F.3d 256, 270 (2d Cir. 2002) (affirming “for the reasons stated” in the district court opinion).  Accord, e.g., McClain v. Metabolife International, Inc., 401 F.3d 1233, 1248 (11th Cir. 2005); United States v. USPlabs, LLC, 2019 WL 499101, at *9 (N.D. Tex. Feb. 8, 2019); Konrick v. Exxon Mobil Corp., 2016 WL 439361, at *6 (E.D. La. Feb. 4, 2016), aff’d, 670 F. Appx. 222 (5th Cir. 2016); Sparling v. Doyle, 2015 WL 4528759, at *21 (Mag. W.D. Tex. July 27, 2015), objections overruled, 2016 WL 236266 (W.D. Tex. Jan. 20, 2016); In re Heparin Products Liability Litigation, 803 F. Supp.2d 712, 743 (N.D. Ohio 2011), aff’d, 567 F. Appx. 359 (6th Cir. 2014); Cunningham v. Masterwear, Inc., 2007 WL 1164832, at *7 (S.D. Ind. April 19, 2007).

Some courts already refuse to allow experts to resort to Bradford-Hill analysis willy-nilly, without regard to its prerequisites.  E.g., In re Lipitor (Atorvastatin Calcium) Marketing, Sales Practices & Products Liability Litigation, 174 F. Supp.3d 911, 924 (D.S.C. 2016) (“it is well established that the Bradford Hill method used by epidemiologists does require that an association be established through studies with statistically significant results”) (footnote omitted).  We suggest that, particularly given the 2023 amendment to Rule 702, defense counsel take this approach to plaintiffs’ experts who purport to apply the Bradford-Hill criteria – as almost all of them do – to scientific facts that fall far short of the “clear cut” association that Sir Austin himself stated were necessary before it would be proper to use his nine-factor analysis.

In these situations, the plaintiffs’ experts should not even get to first base.  The Bradford-Hill methodology is not reliably applied to cases of questionable association.  Defendants should challenge the plaintiffs’ experts before they even begin their nine-factor guerrilla warfare.  Rather, we should be arguing that Bradford-Hill analysis is simply not applicable – and we should rely upon Sir Austin himself for that proposition.