He who is accused of sorcery should never be acquitted, unless the malice of the prosecutor be clearer than the sun; for it is so difficult to bring full proof of this secret crime, that out of a million witches not one would be convicted if the usual course were followed!
- 17th century French legal authority
While looking for some references to help make sense of the (tortured?) reasoning responsible for Messick v. Novartis Pharmaceuticals,Corp. (a recent U.S. Ninth Circuit opinion arising out of a California intravenous bisphosphonate/osteonecrosis-of-the-jaw (ONJ) case) we came across the quote above in the very enjoyable Does Your Model Weigh the Same as a Duck? Though the aim of that paper is to expose two "particularly pernicious" fallacies of logic infecting drug research methodology, the point it makes when referring to the pre-Enlightenment era's view of the appropriate evidentiary burden in the trial of witches applies equally to modern courts that lower standards of proof in toxic tort cases lest, they fear, all present day witches (i.e. chemicals/pharmaceuticals) go unburnt.
A strong suspicion that the usual course, i.e. skeptical gatekeeping, won't be followed in Messick arises early on when the court chooses to demonstrate the strength of its argument (that the trial court erred when it found the opinions of plaintiff's causation expert to be irrelevant and unreliable) by fighting not one but three causal straw men. As usual there's Certainty, which advances the quixotic claim that plaintiff needs to prove causation essentially by deduction ( failing to recall Hume, D., A Treatise of Human Nature: "all knowledge degenerates into probability"). Then there's How, which stands for an argument nobody makes - e.g. we can't reasonably infer that aspirin reduces the risk of heart disease until it is proven how it does so. Last there's Sole, which defends the defenseless argument that a putative cause must also be the sole (i.e. only and sufficient) cause of plaintiff's injury. Unsurprisingly, each straw man is dispatched in a paragraph or less.
What is surprising is the length to which the court goes to save the plaintiff from her own expert, Dr. Richard Jackson. Jackson admitted that the fact that bisphosphonates are a cause of ONJ doesn't mean that it was the cause of her ONJ. He even admitted that the plaintiff had multiple risk factors for ONJ and that he could not determine "which of those particular risk factors is causing [the ONJ]." You would think that such equivocal testimony would put an end to plaintiff's quest to prove causation and that's exactly what the District Court below had held; but the Ninth Circuit thought otherwise.
The appellate court held that while plaintiff's causation expert "never explicitly stated that Messick's bisphosphonate use caused her [ONJ]", Dr. Jackson had analogized plaintiff's use of bisphosphonates to "the oxygen necessary to start a fire." Also, he had said that "[bisphosphonate use] was at least a substantial factor in her development of [ONJ]." Finally he was prepared to opine, based on his "extensive clinical experience", that "a patient without cancer or exposure to radiation in the mouth area would not develop ONJ lasting for years (as had plaintiff) without IV bisphosphonate treatments". Somehow, that's enough for a plaintiff to get to the jury on causation.
The problem with the assertion that bisphosphonates are to ONJ as oxygen is to fire is that a quick PubMed search reveals numerous cases of ONJ in cancer patients decades before bisphosphonates were ever marketed to them. ONJ has been attributed to bacteria, dental work, radiation and cancer all by itself. Perhaps, given that Dr. Jackson diagnosed plaintiff with bisphosphonate-related ONJ (or BRONJ), he's really saying: "plaintiff has BRONJ; therefore she has ONJ related to bisphosphonates". But that would just be begging the question and presumably not persuasive to the court. Either way, how an argument that's either demonstrably false or a logical fallacy can support plaintiff's causal claim escapes us.
Next, what should we make of the court's reliance on Dr. Jackson's "it's at least a substantial factor" opinion? Apparently what the court is saying is that while (1) there are multiple causes of ONJ including bisphosphonates; and (2) plaintiff had cancer, and perhaps other risk factors, known to cause ONJ; even though (3) her expert can't say which one did it; because (4) he's prepared to testify "that Messick's bisphosphonate use was a substantial factor"; (5) such testimony satisfies California's substantial factor standard and is admissible. However, the only way that (5) follows from (1 - 4) is if proof of "but for", or counterfactual, causation is not an element of California's substantial factor causation test and "maybe" causes are good enough. Yet California's substantial factor standard actually "subsumes the 'but for' test". We're again left scratching our heads.
The final (and apparently to the court most compelling) causation argument was that Dr. Jackson, on the basis of things seen only by himself (i.e. his clinical experience), had ruled out leading alternate causes and so thereby reliably ruled in bisphosphonates. This is of course just ipse dixit making an appearance in its de rigueur guise as "differential diagnosis" - the court admitting as much when it writes "[m]edicine partakes of art as well as science ..." while pretending not to notice the impact of the evidence-based medicine revolution. We've taken the position in prior posts that believing differential diagnosis (a/k/a differential etiology a/k/a inference to the best explanation) to be akin to the scientific method and that it produces the sort of reliable scientific knowledge contemplated by Rule 702 is simply the sort of pre-scientific thinking common among those prone to being mesmerized by credentials and jargon. Instead of rehashing those arguments consider the case of Dr. Franz Mesmer and what is revealed when the scientific method is applied to the beliefs of doctors drawn from their clinical experience.
After having seen many patients Dr. Mesmer came up with a hunch about how the body worked and how good health could be restored to the sick. His hypothesis was called animal magnetism and it entailed that an invisible force ran through channels in the body which, when properly directed, could effect all manner of cures. Redirecting that force via mesmerization became wildly popular and Dr. Mesmer became quite famous. In what would become the first recorded "blind" experiment clinicians who practiced mesmerism - the art of redirecting the invisible forces to where they were needed - proved unable, when they did not know what it was they were mesmerizing, to distinguish a flask of water from a living thing and neither could they produce any cures. On the commission overseeing the experiment in 1784 was none other than one of the leading lights of the American Enlightenment and rebel against authority - Benjamin Franklin.
Two hundred and ten years later doctors were still seeing in their patients what their hypotheses predicted rather than what was actually occurring. A classic research paper demonstrating the phenomenon is The impact of blinding on the results of a randomized, placebo-controlled multiple sclerosis clinical trial. Investigators assessing the efficacy of a new treatment for multiple sclerosis (MS) in a flash of brilliance decided to "blind" some of the neurologists who would be clinically assessing patients undergoing one of three treatments while letting the rest of the neurologists involved in the effort know whether a particular patient was getting the new treatment, the old treatment or the sham treatment. While the blinded neurologists and even the patients who had correctly guessed their treatment assignments (a check for the placebo effect) saw no improvement over the old treatment, the unblinded neurologists not only saw a significant positive effect that wasn't there but they continued to see it for two years. Is there some workaround; a way to test after the fact for the distortion of the lens through which a clinician in the know observes his or her patients? You could try but it would appear to be a mug's game and furthermore, by the very nature the bias produced, beyond the ability of cross examination to uncover.
A clinician's art and a differential diagnosis derived from that art saved the day in Messick. Along the way to deciding that objective, verifiable evidence is not required to prove causation in such cases the court listed her sister circuits said to be of like mind and in the first footnote added that the Fifth Circuit was now alone in not having similarly lowered the gates. How the Fifth Circuit feels about being Daubert's last redoubt is unknown to us but we're pretty sure that a plaintiff would win on causation in a bisphosphonate-ONJ case before that court. That's because there are five years worth of objective and verifiable (and verified) evidence that (a) ONJ incidence in bisphosphonate-treated cancer patients is drastically and consistently increased; and (b) the likelihood that ONJ in a bisphosphonate-treated cancer patient was due to the treatment is slightly over 98%. See: 2014 AAOMS Position Paper on Medication-Related Osteonecrosis of the Jaw.
We know why plaintiffs' counsel don't want courts to embrace the sort of causal reasoning that would make a case like Messick easy for both general and specific causation. It's because the day a court holds that "a probability estimate of 98% obviously passes the 'more likely than not' test" is the prelude to doomsday in low dose asbestos/benzene/etc litigation when that same court holds "a probability estimate of 2% obviously does not'". What we can't understand is why so many courts refuse to enforce the test by demanding something more than the musings of experts. Witches or bewitchment are our two working hypotheses.