An expert witness is not supposed to pick a desired result and then reverse engineer inputs and methods that reach that result. As the Ninth Circuit observed 30 years ago, “[c]oming to a firm conclusion first and then doing research to support it is the antithesis of [the scientific] method.” Claar v. Burlington Northern R.R. Co., 29 F.3d 499, 502-03 (9th Cir. 1994). A recent opinion from the Southern District of Illinois offers a fine example of an expert with a results-driven approach and a court that called him out on it.
In re Paraquat Products Liability Litigation, 2024 WL 1659687 (S.D. Ill. Apr. 17, 2024), arises from a multidistrict litigation (“MDL”) in which the plaintiffs claim to have developed Parkinson’s disease as a result of exposure to an herbicide, paraquat. Four plaintiffs whose cases had been chosen for the MDL’s first trials offered a statistician (the parties disputed whether he also qualified as an epidemiologist) as their sole expert to establish general causation. He had a difficult task, as no peer-reviewed literature established a link between paraquat exposure and Parkinson’s disease. Indeed, when the court asked the plaintiffs to identify such literature, the plaintiffs could cite only a single opinion article. That article had been shared with another of the plaintiffs’ experts before it was published, leading the court to conclude in deciding a prior discovery dispute that there was reason to investigate “whether counsel for the MDL plaintiffs, their experts, or other third parties may have influenced the contents of the article for the benefit of one side in the MDL.” 2023 WL 8372819 (S.D. Ill. Dec. 4, 2023).
The plaintiffs’ expert tried to overcome the lack of evidence by conducting a meta-analysis that included just seven case-control studies – a small subset of the available literature – and concluded that paraquat exposure nearly tripled one’s risk of Parkinson’s disease. In justifying his use of just seven studies, the expert claimed, in part, that he had limited his analysis to occupational exposures. However, at deposition, he admitted that one of the seven studies – which accounted for 74% of the weight of his analysis – included residential exposures and thus did not meet his inclusion criteria. Ultimately, the expert conceded he had never reduced his criteria to writing and that his selection process was not objectively replicable.
Later, in a supposed “rebuttal” report, the expert offered what the court deemed “a methodological sea change” in his analysis and set out “much more granular and even previously undisclosed explanations of his study selection methodology.” The expert now claimed he had begun by considering 36 relevant studies, but he did not list them in his report and could not say at deposition how he had chosen them. He then narrowed the field by applying stated eligibility criteria, which were met by eight studies. He then disqualified one that had found “no evidence” of a link between paraquat and Parkinson’s disease, purporting to apply five “quality criteria.” But he admitted he had chosen those “quality criteria” only after reading the eight studies. The court observed that this “methodology” allowed the expert to arrive at a pre-determined result; quoting the expert’s own words from a deposition he had given in a previous litigation, the court commented that the expert’s methodology “may have allowed him to ‘cook the books.’”
The expert’s attempt to conduct a “weight of the evidence”/Bradford Hill assessment of causation built around his meta-analysis was similarly deficient. Neither his initial report nor his “rebuttal” report ever explained how he had actually conducted his assessment. Rather than disclosing how he weighed various criteria against one another, he merely “cite[d] favorable authority for each Bradford Hill factor before apparently concluding that it is satisfied for purpose of his analysis.” As the court noted, his lack of detail rendered his general causation opinion “virtually non-falsifiable” and thus contravened “one of the most basic requirements of the scientific method.”
In an impressively detailed 97-page opinion that cites myriad additional problems with the expert’s analysis, the court excluded the expert’s general causation opinion as the product of unreliable methodology. In so doing, the court cited the recent amendments to Federal Rule of Evidence 702 and confirmed that, “[m]indful of its role as the witness stand’s ‘vigorous gatekeeper,’ the Court will closely scrutinize the reliability of proffered expert testimony before permitting an expert to share her opinion with the jury.” The court certainly discharged that duty here, holding a four-day hearing at which the excluded expert himself testified and the parties offered “extensive” argument before issuing its lengthy and comprehensive Rule 702 opinion. And then, in a much shorter second opinion (2024 WL 1655500), the court granted the defendants’ motion for summary judgment in the four trial selection cases due to the lack of admissible expert testimony establishing causation.
With this opinion, the In re Paraquat MDL court follows the recent examples of In re Mirena IUS Levonorgestrel-Related Prod. Liab. Litig. (No. II), 41 F. Supp. 3d 213 (S.D.N.Y. 2018), In re Zantac (Ranitidine) Prod. Liab. Litig., 644 F. Supp. 3d 1075, 1270 (S.D. Fla. 2022), and In re Acetaminophen – ASD-ADHD Prod. Liab. Litig., — F. Supp. 3d —, 2023 WL 8711617 (S.D.N.Y. 2023) – which we discussed here, here, and here, and each of which was cited by In re Paraquat – in painstakingly evaluating the reliability of a general causation expert’s opinion before excluding that opinion and granting summary judgment to the defendants. A case in which the plaintiffs’ general causation evidence consists of unreliable expert opinion should not be tried to a jury, and it is encouraging to see MDL courts investing the time and effort it takes to faithfully serve the gatekeeping role Rule 702 envisions.
