Friday, January 19, 2024

Experts, Standards of Review, and Meta-Evidence

Garner v. BNSF Railway Co., No. D082229 (D4d1 Jan. 4., 2024)

This is a wrongful death case where decedent’s family claims that exposure to diesel exhaust while working as a railroadman caused the non-Hodgkin’s lymphoma that lead to his death. In the run-up to trial, the Company moved in limine to exclude Plaintiff’s general causation experts for failing to satisfy the Sargon standard.

Plaintiffs put up three experts on this point. Generalizing a bit, each testified that diesel exhaust was an established cause of cancer. They pointed to, among other things, epidemiological studies that showed that the exposure could lead to three to four thousand excess cancers per million people, which is clearly significant. But that was cancer in general. None of the experts, however, pointed to any study that linked non-Hodgkins lymphoma, in particular, to diesel exhaust, or that suggested a dosing that could merit causation. But they testified that, more generally, the mutagenic way diesel exhaust acts on human issue merits a conclusion that it can cause cancers that were no just limited to one specific organ. 

The Company pointed out these apparent gaps in the analysis. But it didn’t provide evidence of its own that experts’ reasoning or methodologies were scientifically unsound.

After tentatively accepting the opinions of Plaintiff’s experts, the trial court ultimately excluded them. It found that the gap between the underlying epidemiological evidence and the ultimate conclusions as to general causation was too broad to be bridged by the experts’ opinions. And without the experts, there was no evidence of causation, and thus the case could not go to trial. Plaintiff appealed.

Before getting to the evidentiary issue, the Court of Appeal discusses the standard of review. It recognizes that rulings on evidence, and the admissibility of expert testimony in particular, are generally subject to an abuse of discretion. But, relying on a number of cases, the court says that when an in limine ruling precludes a whole cause of action, it is treated as a nonsuit and reviewed de novo.

But that’s not entirely right. No doubt, the court pulls in language from cases that say, literally, that a motion in limine ruling that excludes all evidence and therefore resolves a whole cause of action is essentially a nonsuit that gets reviewed de novo. But all the cases the court cites deal with circumstances where a trial court decides a legal issue, which then makes all evidence irrelevant and thereby dooms a cause of action. 

The legal issue could be the interpretation of a statute, a contract or case law. It could also be a determination that there’s just not enough evidence to make it to a juryitself a legal decision that basically a nonsuit. The cases cited all have these kinds of fact patterns. See Kinda v. Carpenter, 247 Cal. App. 4th 1268, 1285 (2016) (on in limine motion, trial court held that evidence was insufficient to get to the jury); McMillin Companies, LLC v. Am. Safety Indem. Co., 233 Cal. App. 4th 518, 529 (2015) (trial court excluded all contrary evidence on a duty issue based on a legal ruling that rendered the evidence irrelevant); Legendary Inv. Grp. No. 1, LLC v. Niemann, 224 Cal. App. 4th 1407, 1411 (2014) (similar); City of Livermore v. Baca, 205 Cal. App. 4th 1460, 1465 (2012) (exclusion of all evidence on the ground that the plaintiff's theory of liability was fatally defective); Dillingham-Ray Wilson v. City of L.A., 182 Cal. App. 4th 1396, 1401 (2010) (trial court interpreted contract as a matter of law and found that interpretation made evidence irrelevant and thus inadmissible); Fergus v. Songer, 150 Cal. App. 4th 552, 570 (2007) (trial court determined that, as a matter of law, attorney fee agreement was voidable and thus that evidence of damages based on it could not go to the jury). But the standard of review in these cases is de novo because the trial courts are fundamentally deciding questions of law, which always get reviewed de novo.

That, however, not the same thing as when a trial court makes an evidentiary ruling that makes certain evidence inadmissible, which then potentially has the consequence that the plaintiff can’t get to a jury. The ultimate significance of a pretrial ruling on a question of the admissibility of evidence should not change the standard of review. Notably, the court doesn’t cite any cases applying de novo review to the exclusion of a causation expert, even when that exclusion is potentially case dispositive. (This recent one, for instance, certainly didnt.) 

If the evidentiary question is one normally one which is reviewed for an abuse of discretion—a classic example of which is the admissibility of expert testimony, see Sargon Enterprises v. University of Southern California, 55 Cal.4th 747, 773 (2012)—it should be reviewed under that standard, consequences be damned. Then, whether whatever is left is enough to get to a jury is a legal and that question should get reviewed de novo. But reviewing discretionary decisions de novo just because a plaintiff might lose as a result puts a thumb on the scale in favor of a plaintiff because, especially in cases of causation, excluding expert testimony on causation issues is usually case dispositive, but admitting it is generally not. It is completely arbitrary and nonsensical for a standard of review to change based on whether the trial courts decision was to admit vs. to exclude. The standard of review should turn on the nature of the decision, not its consequence.

(FWIW, I just realized I digressed at length on this point in a post nearly eight years ago. While my memory is imperfect, my point stands nonetheless.)

In any event, moving on to the evidentiary question, the Court of Appeal drills down to the issue of the kind of inference a scientific expert can permissibly draw from underlying source material. As noted, these experts, relying on evidence of increased carcinogenicity in general, concluded that diesel exhaust could cause the specific kind of cancer that killed the decedent. They supported that conclusion with some more testimony explaining why that was a permissible inference, in their fields of expertise, to draw. Like that the kind of pathology caused by exposure to diesel exhaust—mutagenic damage to cellular level DNA—made it reasonable to conclude that the exposure could cause a bunch of different types of cancer. 

So the Court of Appeal concludes that in the absence of any evidence submitted by the Company establishing that that kind of conclusion was an inappropriate one for an epidemiologist to draw from the underlying data, the trial court abused its discretion in excluding the testimony because the inferential gap was too wide.

This is, admittedly, a hard epistemic task. As David Hume explained a couple hundred years ago in An Enquiry Concerning Human Understanding, you can never definitively prove the answer to a question of causation. All you can really do is observe a set of correlations that are significantly close from which an inference of causation can be drawn. So here, the Sargon question is how close does epidemiological statistical data—itself an observation of correlationneed to be to make a methodologically sound inference that a relationship is causal. And even more specifically, what kind of evidence (or law) validates the soundness of that inference.  

Some more extreme facts can illustrate the question. On one hand, say an expert relies on a study of a large population is exposed to some agent in a specific and measurable way. All or nearly all of them develop some rare disease that is almost never seen in the pubic at large. And there’s nothing else about the exposed group that otherwise meaningfully differs from the public. In that case, provided the methods in which the data were collected were sound, the gap between that data and the inference of a causal relationship between the exposure and the plaintiffs disease isnt very big at all.

On the opposite end, say the study of a small population correlates exposure a number of different outcomes, which also occur almost as frequently in the unexposed population, the sample was taken from a population that is atypical in some way, and the plaintiff suffers an outcome that is not among the ones measured in the sample. There, the gap between the data and a conclusion of causation is clearly too wide to stand. The fact that a couple of lactose intolerant people get a tummy aches from eating dairy does not justify an inference that ice cream causes pancreatic cancer.

Easy cases like these at the outer limits can likely be resolved as a matter of law or undisputed fact by well-informed judicial common sense. But there is obviously a wide field of grey area in between. And there, the question of “is this a reasonable inference to draw?” seems itself to be a factual question that could itself be a potential subject of expert opinion: Do professionals in the relevant field, employing appropriate methodologies, and outside of litigation, think that, data A rationally merits the more likely than not inference of a causal relationship?

Those situations can turn on a kind of meta-evidence. That is, evidence that, while potentially irrelevant to the merits, goes to a question of whether other evidence is even admissible. Here, that was, to some degree provided by Plaintiff. Its experts testified why it was reasonable to infer causation of non-Hodgkin’s lymphoma from general cancer data concerning exposures to diesel exhaust. On the other hand, the Company didnt put in anything to the contrary. So that resolves the appeal.

But what if the Company put in evidence that in the practices in field of epidemiology, the inference drawn by Plaintiff’s expert wasn’t a reasonable one to draw? In that case, it seems to me, under Evidence Code §§ 403 and 405(a), as a condition to admitting the testimony, the trial court would be obliged to decide whether, as a matter of fact, a preponderance of the evidence established that the inference was justifiable or the gap was to big. A trial court should not be able to engage in another round of meta-punting on the basis that that question too is a subject of debate within the field.

Reversed.




No comments:

Post a Comment

That's Not a Debate

Taylor v. Tesla , No. A168333 (D1d4 Aug. 8, 2024) Plaintiffs in this case are also members of a class in a race discrimination class action ...