Harris v. CSX Transportation
I have borrowed and modified the phrase “Intellectual Due Process” from earlier writers because of its obvious implications for the presentation, interpretation, synthesis, and evaluation of scientific evidence in court. See Scott Brewer, “Scientific Expert Testimony and Intellectual Due Process,” 107 Yale L. J. 1535 (1998). The major reason courts write opinions is to explain and justify their decisions to litigants, present and future, and to a wider audience of lawyers, scholars, and the general public. Judicial opinions involving scientific evidence, whether in legislation, regulation, or litigation must satisfy the societal need to explain and justify the acceptance and rejection of scientific claims. Despite a great deal of hand waving that law and science are somehow different, in the end, when courts describe their acceptance or rejection of scientific claims, they are addressing the same epistemic warrant that scientists themselves employ. Even a cursory review of the judicial output reveals an unsatisfactory state of affairs in which many courts mangle scientific and statistical evidence and inference. There is much that is needed to correct the problem.
One proposal would be to require that the parties file proposed findings of facts in connection with Rule 702 gatekeeping challenges. Courts should file detailed findings of facts that underlie their decisions to admit or to exclude expert witness opinion testimony. Another proposal would require courts to cite properly the scientific studies that they discuss in reaching a legal conclusion about sufficiency or admissibility. These are small steps, but ones that would help reduce the gross inaccuracies and the glib generalizations, while increasing the opportunity for public scrutiny and criticism.
We do not think anything is amiss with special courts for tax, patent, family law, national security, equity, or commercial matters. There is an even greater need for scientific skill, knowledge, and aptitude in a specialized science court. The time has come for special courts to hear cases involving scientific claims in health effects and other litigation.
* * * * * * *
A decision of the West Virginia Supreme Court, late last year, illustrates the need for substantial reform of how claiming based upon “scientific evidence” is permitted and evaluated in court. Mrs. Harris sued the railroad for the wrongful death of her husband, who died of multiple myeloma. Mr. Harris had been exposed, in his railroad workplace, to diesel exhaust, which Mrs. Harris claimed caused his cancer. See Harris v. CSX Transportation, Inc., 232 W.Va. 617, 753 S.E.2d 275 (2013). The trial court excluded Mrs. Harris’s expert witnesses. Harris v. CSX Transportation, Inc., No. 12-1135, 2012 WL 8899119 (Cir. Ct. Marshall Cty., W.Va. Aug. 21, 2012).
1. The West Virginia Supreme Court reversed the trial court’s exclusion of witnesses on the basis of an asymmetrical standard of review, which would allow de novo review of trial court decisions to exclude expert witness opinions, but which would privilege trial court decisions to admit opinions by limiting appellate review to abuse of discretion. This asymmetry was, of course, the same dodge that the Third and Eleventh Circuits had used to keep the “gates open,” regardless of validity or reliability concerns, and the same dodge that the Supreme Court shut down in General Electric v. Joiner. A single judge dissented in Harris, Justice Loughry, who took the majority to task for twisting facts and law to get to a desired result.
2. The Harris Court cited a federal court case for dicta that “Rule 702 reflects an attempt to liberalize the rules governing the admissibility of expert testimony.” See Harris, 753 S.E.2d at 279 (citing and quoting from Weisgram v. Marley Co., 169 F.3d 514, 523 (8th Cir.1999). Remarkably, the Harris Court omitted reference to the United States Supreme Court’s unanimous affirmance of Weisgram, which saw Justice Ginsburg write that “[s]ince Daubert, moreover, parties relying on expert evidence have had notice of the exacting standards of reliability such evidence must meet.” Weisgram v. Marley Co., 528 U.S. 440, 442 (2000). The Harris Court’s lack of scholarship is telling.
3. Meta-analysis appeared to play a role in the case, but the judicial decisions in Harris fail to describe the proffered evidence. The majority in Harris noted that one of plaintiff’s expert witnesses, Dr. Infante, relied upon a meta-analysis referred to as “Sonoda 2001.” Harris, 753 S.E.2d at 309. Neither the Court nor the dissent cited the published meta-analysis in a way that would help an interested reader in finding the paper. One could imagine the hue and cry if courts cited judicial cases or statutes by short-hand names without providing enough information to access the relied upon source. In this case, a PubMed search reveals the source so perhaps the error is harmless. Tomoko Sonoda, Yoshie Nagata, Mitsuru Mori, Tadao Ishida & Kohzoh Imai, “Meta-analysis of multiple myeloma and benzene exposure,” 11. J. Epidemiol. 249 (2001). Still, the time has come for courts to describe and report the scientific evidence with the same care and detail that they would use in a car collision case.
4. A quick read shows that the Sonoda meta-analysis supports the dissent’s assessment:
“‘Dr. Infante testified on direct examination that Sonoda 2001 considered 8 case-control studies specific to engine exhaust and stated it concluded that diesel and non-diesel engine exhaust causes multiple myeloma.’ Yet, as the trial court found, ‘[o]n cross examination Dr. Infante acknowledged that none of the 8 papers included in the Sonoda meta-analysis mention diesel exhaust’.”
Harris, 753 S.E.2d at 309. The dissent would have been considerably more powerful had it actually adverted to the language of Sonoda 2001:
“These results suggested that benzene exposure itself was not likely to be a risk factor of MM [multiple myeloma]. It is thought that several harmful chemical agents in engine exhaust, other than benzene, could be etiologically related to the risk of MM. Further case-control studies on MM are needed to obtain more information about detailed occupational exposure to toxic substances.”
Sonoda at 249 (2001) (emphasis added). Contrary to Infante’s asseveration, Sonoda and colleagues never concluded that diesel exhaust causes multiple myeloma. The state of scholarship and “intellectual due process” makes it impossible to tell whether or not Dr. Infante was telling the truth or the Harris Court badly misunderstood the record. Either way, something must give.
The dissent went on to note that Dr. Infante conducted his own meta-analysis, which included studies that did not mention diesel exhaust. Harris, 753 S.E.2d at 309. The railroad complained that some of the studies were small and had limited power, but that is exactly why a meta-analysis would be appropriate. The more disturbing complaints were that the meta-analysis left out important studies, and that it included irrelevant studies of benzene exposure and myeloma, which raised insuperable problems of external validity.
5. A half empty glass that is always full. According to the Harris Court, the West Virginia shadow of Rule 702 is a rule of “admissibility rather than exclusion.” Harris, 753 S.E.2d at 279 (citing and quoting from In re Flood Litig. Coal River Watershed, 222 W.Va. 574, 581, 668 S.E.2d 203, 210 (2008), which in turn quoted a federal case, Arcoren v. United States, 929 F.2d 1235, 1239 (8th Cir. 1991), decided before the Supreme Court decided Daubert.) This is just silly hand waving and blatant partisanship. A rule that sets out criteria or bases for admissibility also demarcates the inadmissible.
6. Cherry Picking. Dr. Infante was permitted by the Harris Court to aggregate data from studies that did not observe diesel exposure, while he failed to include, or he deliberately excluded data from, a large, powerful, exonerative study conducted by scientists from the National Cancer Institute, the International Agency for Research on Cancer (IARC), and the Karolinska Institute. See Paolo Boffetta, Mustafa Dosemeci, Gloria Gridley, Heather Bath, Tahere Moradi and Debra Silverman, “Occupational exposure to diesel engine emissions and risk of cancer in Swedish men and women,” 12 Cancer Causes Control 365 (2001). Dr. Infante inexplicably excluded this study, which found a risk ratio for men exposed to diesel exhaust that was below one, 0.98, with a very narrow 95% confidence interval, 0.92-1.05. Boffetta at 368, Table 2.
7. The West Virginia articulated an incohorent definition of “reliable,” designed to give itself the ability to reject gatekeeping completely. Citing its earlier decision in Flood, the Court offered its own ipse dixit:
“The assessment of whether scientifically-based expert testimony is “reliable,” as that term is used in [Daubert v. Merrell Dow Pharmaceuticals, Inc., 509 U.S. 579 (1993), and Wilt v. Buracker, 191 W.Va. 39, 443 S.E.2d 196 (1993)], does not mean an assessment of whether the testimony is persuasive, convincing, or well-founded. Rather, assessing ‘reliability’ is a shorthand term of art for assessing whether the testimony is to a reasonable degree based on the use of knowledge and procedures that have been arrived at using the methods of science — rather than being based on irrational and intuitive feelings, guesses, or speculation. If the former is the case, then the jury may (or may not, in its sole discretion) ‘rely upon’ the testimony. In re Flood Litig., 222 W.Va. at 582 n. 5, 668 S.E.2d at 211 n. 5.”
Harris, 753 S.E.2d at 279-80. Surely, this is circular or vacuous or both. Opinions not “well-founded” will be ones that are based upon guesses or speculation. Opinions arrived at by the “methods of science” will be ones that have an epistemic warrant that will survive a claim that they are not “well-founded.”
8. The Harris Court evidenced its hostility to scientific evidence by dredging up one of its own decisions involving a multiple myeloma causation claim, State ex rel. Wiseman v. Henning, 212 W.Va. 128, 569 S.E.2d 204 (2002). Wiseman involved a specious claim that a traumatic rib injury caused multiple myeloma, a claim at odds with scientific method and observation:
“Some research has suggested that people in some jobs may have an increased risk of developing multiple myeloma because they are exposed to certain chemicals. But the International Agency for Research on Cancer (IARC) states that the evidence is limited overall. It has been suggested that people may have an increased risk if they work in the petrol or oil industry, farming, wood working, the leather industry, painting and decorating, hairdressing, rubber manufacturing or fire fighting. But there is no evidence to prove that any of these occupations carry an increased risk of myeloma.”
Cancer Research UK, “Myeloma risks and causes” (last visited May 28, 2014). Even the most non-progressive jurisdictions have generally eradicated specious claiming for trauma-induced cancers, but West Virginia has carved out a place second to none in its race to the bottom.
9. WOE. Not surprisingly, the Harris Court relied heavily on the First Circuit’s “weight of the evidence” end-run around the notion of epistemic warrant for scientific claims, citing Milward v. Acuity Specialty Products Group, Inc., 639 F.3d 11 (1st Cir.2011), cert. denied sub nom., U.S. Steel Corp. v. Milward, ___ U.S. ___, 2012 WL 33303 (2012). The Harris Court went on to conflate and confuse WOE with Bradford Hill, and cited a recent New York case that confidently saw through WOE hand waving, while ignoring its devasting critique of expert witnesses’ attempts to pass off WOE for scientific, epistemic warrant. Reeps ex rel. Reeps v. BMW of N. Am., LLC, No. 100725/08,
2013 WL 2362566, at *3, 2012 N.Y. Misc. LEXIS 5788; 2012 NY Slip Op 33030U (N.Y. Sup. Ct. May 10, 2013).
10. Link. Dr. Infante links a lot, even when his sources do not:
“Dr. Infante testified that the International Agency for Research on Cancer issued Technical Publication Number 42 in 2009, and that the publication stated that diesel exhaust exposures have been linked to multiple myeloma and leukemia.”
Harris, 753 S.E.2d at 294. The Harris Court neglected to give the title of the publication, which tells a different story. Identification of research needs to resolve the carcinogenicity of high-priority IARC carcinogens. The dissent was willing to go behind the conclusory and false characterization that Dr. Infante and plaintiff gave to this publication. Harris, 753 S.E.2d at 309. The trial court’s finding (and the dissent’s assertion) that the IARC Technical Publication 42 intended to express a research agenda, not to make a causation statement, seems unassailable. Furthermore, it appears to be precisely the sort of specious claim that a court should keep from a jury. The cited IARC source actually notes that the then current IARC classification of diesel exhaust was of inadequate evidence for human carcinogenicity, with a focus on lung cancer, and barely a mention of multiple myeloma.
11. The Benzene Connection. Plaintiffs’ expert witnesses, including Dr. Infante, argued that benzene was a component of diesel exhaust, and benzene caused multiple myeloma. This move ignored not only the lack of evidence to implicate benzene in the causation of multiple myeloma, but it also ignored the large quantitative differences between the benzene occupational exposure studies and the very small amounts of benzene in diesel exhaust. The Harris Court held that the trial court acted improperly by inquiring into and finding the following facts, which were “exclusively” for the jury:
- “There is substantially more benzene in cigarette smoke than diesel exhaust.
- Benzene is present only in trivial doses in diesel exhaust.
- The hypothesis that diesel exhaust causes multiple myeloma is confounded by the fact that cigarette smoking does not.”
The Harris majority further chastised the trial court for adverting to the ten or so studies that failed to find a statistically significant association between benzene exposure and multiple myeloma. Harris, 753 S.E.2d at 305-06. This inquiry directly calls into question, however, Dr. Infante’s methodology.
If these facts, found by the trial court, were reasonably established, then Dr. Infante’s argument was less than bogus, and a major underpinning for inclusion of benzene studies in his meta-analysis was refuted. These are precisely the sort of foundational facts that must be part of an inquiry into the methodological grounds of an expert witness’s opinion.
12. The Harris Court confused “proving causation” with “showing a methodology that provides an epistemic warrant for concluding.” Harris, 753 S.E.2d at 300. The Harris Court asserted that the trial court exceeded its gatekeeping function by inquiring into whether Mrs. Harris’s expert witnesses “proved” causation. Harris, 753 S.E.2d at 300. Speaking of “proof of” or “proving” causation is an affectation of lawyers, who refer to their evidence as their “proofs.” Epidemiologic articles and meta-analyses do not end with quod erat demonstrandum. Beyond the curious diction, there is a further issue in the majority’s suggestion that the trial court set the bar too high in declaring that the plaintiff failed to “prove” causation. Even if we were to accept the continuous nature of strength of evidence for a causal conclusion, Dr. Infante and the other plaintiff’s witnesses, would be fairly low on the curve, and their lowly position must of necessity speak to the merits of the defense motion to exclude under Rule 702.
13. Purely Matters for Jury. The Harris Court criticized the trial court for conducting a “mini-trial,” which set out to “resolve issues that were purely matters for jury consideration.” Harris, 753 S.E.2d at 305. In holding that the matters addressed in the pre-trial hearing were “exclusively grist for the jury and which had no relevancy to the limited role the trial court had under the facts of this case,” the Harris Court displayed a profound disregard for what facts would be relevant for a challenge to the plaintiff’s expert witnesses’ methodology. Many of the facts found by the trial court were directly relevant to “general acceptance,” validity (internal and external) of studies relied upon, and reliability of reasoning and inferences drawn. Aside from the lack of general acceptance and peer review of the plaintiff’s claimed causal relationship, the proffered testimony was filled with gaps and lacunae, which are very much at issue in methodological challenges to an opinion of causality.
* * * * * * *
The Harris case has taken its place next to Milward in the litigation industry’s arsenal of arguments for abandoning meaningful judicial supervision and gatekeeping of expert witness opinion testimony. See Andrew S. Lipton, “Proving Toxic Harm: Getting Past Slice and Dice Tactics,” 45 McGeorge L. Rev. 707, 731 (2014) (plaintiffs’ bar cheerleading for the Harris decision as “a lengthy and thoughtful analysis”, and for the Milward case as roadmap to evade meaningful judicial oversight). Not all was perfect with the trial court’s opinion. The defense seemed to have misled the court by asserting that “a difference between a case group and control group is not statistically significant then there is no difference at all.” See Respondent’s Brief at 5, Harris v. CSX Transportation, Inc., 2013 WL 4747999 (filed (Feb. 4, 2013) (citing App. 169, 228-230 (Shields) as having explained that the p-values greater than 0.05 do not support a causal association).
This is hardly true, and indeed, the lack of statistical significance does not lead to a claim that the null hypothesis of no association between exposure and outcome is correct. The defense, however, did not have a burden of showing the null to be correct; only that there was no reliable method deployed to reject the null in favor an alternative that the risk ratio for myeloma was raised among workers exposed to diesel exhaust.
Still, the trial court did seem to understand the importance of replication, in studies free of bias and confounding. Courts generally will have to do better at delineating what are “positive” and “negative” studies, with citations to the data and the papers, so that judicial opinions provide a satisfactory statement of reasons for judicial decisions.
