Deconstructing Milward (part I)

I could write a long and tedious post about Milward v. Acuity Specialty Products Group, Inc. and that’s why I’m having such a hard time writing a short and hopefully interesting blog post about it. The problem is that the opinion contains no formal arguments, eschews empiricism and denies the existence of objective facts. In other words, precisely because of the epistemological relativism it implicitly adopts, the attempt to blog about it has become a bit like trying to play croquet with a flamingo. But here goes – piecemeal fashion.

The trial court bifurcated the case into two phases. In the first phase the query was whether or not so-called "general causation" testimony by plaintiff’s experts was admissible under Daubert. If such testimony proved to be admissible a trial on the rest of the case, including so-called "specific causation", would follow. Let’s think about "general causation" and "specific causation".

The First Circuit’s opinion refers us to comment c(3),section 28, of the Restatement (Third) of Torts: Liability for Physical and Emotional Harm for the meaning of the supposed two types of causation in a toxic tort case; the former being "when a substance is capable of causing a disease" and the latter being when "when exposure to an agent caused a particular plaintiff’s disease". Everyone seems to think this is obvious but is it and, if it is, what should follow from it?

I readily admit that by 1995 I had added the general/specific causation idea to a CLE presentation I’d started to give on causal inference, A.B. Hill’s "criteria" and the inherent difficulty in proving that a particular plaintiff’s illness, sans aplastic anemia, asbestosis, etc, was caused by some agent established by frequentist methods to be a carcinogen. Now I’m not so sure. Here’s why.

We all agree that fired bullets generally cause holes in those they strike and in the toxin context that carbon monoxide (or cyanide, etc) generally causes chemical asphyxiation. But why do we so readily agree? Because everyone struck by a bullet or exposed to carbon monoxide will experience some degree of indentation or asphyxiation. Yet the vast majority of workers exposed to benzene or asbestos never develop any symptoms nor even any signs of prior exposure. Take the unibestos workers at Pittsburgh-Corning’s Tyler, TX facility. It’s true that 9% have developed mesothelioma but why haven’t the other 91%, after almost 50 years, developed mesothelioma? You can ask the same question about benzene or radium dial workers and you’ll get the same answer. You’ll also find that the dose response curves flatten out. What does it mean?

It means that amosite asbestos causes mesothelioma in people susceptible to getting mesothelioma from amosite asbestos and that benzene causes AML in people susceptible to getting AML from benzene. In essence it means that there is no "general causation" and that all causation is specific to the individual. That’s actually a good thing for plaintiffs because causation can then be established by means other than by epidemiology – by detecting biomarkers, asbestos bodies and the like.

But it also means that Milward has to prove that somebody, somewhere got the form of leukemia from which he suffered, APL, following benzene exposure. Milward’s expert could find no such case. The best he could do was to say that in some studies of benzene-exposed workers some cases of APL (a rare yet widely found and naturally occurring blood cancer) had been found. He basically admitted that such evidence would fail the inferential tests of strength and consistency and said instead that the fact that cases of APL were found bolstered his opinion by demonstrating that it was not manifestly false.

Thus my first beef (I’ll go page by page in the days to follow): asserting that a claim (here, that benzene causes APL) is probably true, or is more likely true, because it has not been proven false is in fact the logical fallacy known as argument from ignorance. Not pretty. Not pretty at all.