In the last post, I discussed an important article by Professors Davidson and Guzelian, on the legal implications of evidence-based medicine (EBM). Terence M. Davidson & Christopher P. Guzelian, “Evidence-based Medicine (EBM): The (Only) Means for Distinguishing Knowledge of Medical Causation from Expert Opinion in the Courtroom,” 47 Tort Trial & Ins. Practice L. J. 741 (2012) [cited as Davidson].
Their criticism of the deficiencies in current gatekeeping practice cries out for reform of much current judicial practice. Education of the judiciary in EBM would be helpful to both plaintiffs and defendants in civil cases, as well as prosecutors and defendants in criminal cases. I will leave for another day a discussion of whether the boundary between knowledge and “mere opinion” is so easily discernible.
Although the authors’ call for EBM in judicial decisions is timely and needed, I noted my dissent to their assessment of one defense expert witness’s specific causation opinion in the hormone therapy litigation. Davidson and Guzelian criticize one district judge for having admitted the challenged testimony of a defense expert witness, Dr. Blaustein, who opined that (1) estrogen + progesterone combination post-menopausal hormone replacement therapy (CHRT) has not been shown to cause breast cancer, and (2) there is no generally accepted method for determining a cause of a woman’s breast cancer. Cross v. Wyeth Pharms., Inc., CASE NO.: 8:06-cv-429-T-23AEP, 2011 U.S. Dist. LEXIS 89078; 2011 WL 3498305 (M.D. Fla. 2011) (Merryday, J.).
Here is how Davidson & Guzelian put the matter:
“Blaustein also opined that there is no generally accepted method for diagnosing specific medical causation—that is, whether a medical intervention or treatment, even if it is known to generally cause a result ‘X’ (e.g., breast cancer), caused a particular patient’s result ‘X’. Blaustein’s statement, accepted by the judge as befitting of presentation to the jury, is ‘simply false’.”
The claim that Blaustein’s statement is “simply false” is pretty strong for both general and specific causation, and partially explains my initial dissent. The prestige of the Women’s Health Initiative (WHI), a large, randomized, clinical meta-trial of CHRT, with its finding of an increased risk for breast cancer among women, certainly has made Dr. Blaustein’s opinion on general causation a distinctly minority view. There are, however, some careful authors who have challenged the findings of the WHI on grounds of internal and external validity. See, e.g., Samuel Shapiro, Richard Farmer, Helen Seaman, J. C. Stevenson, “Does hormone replacement therapy cause breast cancer? An application of causal principles to three studies: part 2. The Women’s Health Initiative: estrogen plus progestogen,” 37 J. Family Planning & Reproductive Health Care 165, 165 (2011) (“HRT with estrogen plus progestogen may or may not increase the risk of breast cancer, but the WHI did not establish that it does.”). In any event, given the state of the science, most defendants would hedge their position on general causation even if they stopped short of saying it was not established. Still, the validity concerns may cause us to have some doubts that the conclusions drawn from the WHI and other studies are truly “knowledge.”
Professor Guzelian has persuaded me that their indictment of Blaustein’s opinion on specific causation is correct, at least technically. There are, to be sure, a few genetic causes of breast cancer, such as the BrCa 1, and 2, genes, which can cause breast cancer, and which, if present in a particular woman, would constitute an adequate description of the cause of that woman’s cancer. Blaustein claimed that there is no generally accepted method for attributing an individual woman’s breast cancer to known causes, and Davidson and Guzelian are correct that Blaustein’s claim is, therefore, “simply false.”
Given how cursory and conclusory the trial court’s opinion is, Davidson and Guzelian’s derision is, however, relatively uncharitable. We can probably credit the plaintiff’s counsel with sufficient entrepreneurial savvy to have not pressed the claims of a woman who had a known genetic cause of breast cancer. And we can similarly credit the defense counsel with sufficient intelligence not to have overlooked the presence of such a gene if it were present.
With genetic causes taken out of the equation, Blaustein’s opinion seems fairly unremarkable. Even if the criticism of the trial court’s refusal to preclude Dr. Blaustein’s testimony on general causation were correct under Federal Rule of Evidence 702, Davidson and Guzelian have failed to make their case against the trial court, and Dr. Blaustein, on specific causation.
Here is what Davidson and Guzelian have to say about Blaustein’s specific causation opinion:
“EBM has documented and catalogued the best practices regarding how to diagnose whether generally applicable evidence-based conclusions hold for specific patients.51 Moreover, even if in a specific case it were plausible to assert that it is impossible to reach a specific causation conclusion for a particular patient’s condition using the scientific method, that is very different than saying that there is no scientific method for analyzing specific causation. According to EBM’s strictures, Blaustein’s proposed blanket denial of the possibility of specific causation should have been precluded from his testimony.52”
Davidson at 757. The footnotes in this passage are to the section of an article on evidence-based toxicology, which deals with attribution of an adverse drug reaction. Philip Guzelian et al., “Evidence-Based Toxicology: A Comprehensive Framework for Causation,” 24 Human & Experimental Toxicol. 161, 190-91 (Table 9) (2005) (presenting “an overview of evidence-based specific causation diagnostic criteria”). Putting aside the very substantial differences between cancer causation and the general run of adverse drug reactions, we can see that the proposed criteria for specific causation in the referenced article contain this extremely important criterion:
“No good alternative candidate (unexplained exacerbation or recurrence of underlying illness).”
Id. Specific causation in a breast cancer case falls outside this criterion because most cases of breast cancer (with the exception of the genetic cases mentioned above) have no identifiable risk factor. Dr. Blaustein’s opinion on specific causation — that there is no generally accepted method for attributing breast cancer to exogenous cause(s) — seems then exactly on point, even on the EBM criteria for specific causation urged by the authors.
In Cross, the trial court does not provide any insight into the basis for the plaintiffs’ challenge to Blaustein’s specific causation opinion, and I am aware of none. The trial court does not give us any particulars of the plaintiff’s use of CHRT or development of breast cancer, and there is no suggestion that she had an extraordinarily high risk (say > 10-fold increase, which I have never seen reported, in any event).
The plaintiffs did not attempt to infer causation from risk. Perhaps they thought better of it, or perhaps one of their testifying epidemiologists, Dr. Graham Colditz, refused to support such an inference:
“Knowledge that a factor is associated with increased risk of disease does not translate into the premise that a case of disease will be prevented if a specific individual eliminates exposure to that risk factor. Disease pathogenesis at the individual level is extremely complex. As Rose stated, a preventive measure that brings large benefits to the community offers little to each participating individual [3]. Accordingly, epidemiology must be harnessed to identify the population level strategies that will reduce the burden of illness.”
Graham A. Colditz, “From epidemiology to cancer prevention: implications for the 21st Century,” 18 Cancer Causes Control 117, 118 (2007). Professor Colditz’ view is hardly unique; there are many similar refusals to base an inference of specific causation upon an increased risk, whether or not that increased risk is quantified as a relative risk greater than two. The late David Freedman, who was the co-author of the chapters on statistics in all three editions of the Reference Manual on Scientific Evidence, was also a naysayer when it came to transmuting risk into cause:
“The scientific connection between specific causation and a relative risk of two is doubtful. *** Epidemiologic data cannot determine the probability of causation in any meaningful way because of individual differences.”
David Freedman & Philip Stark, “The Swine Flu Vaccine and Guillaine-Barré Syndrome: A Case Study in Relative Risk and Specific Causation,” 64 Law & Contemporary Problems 49, 61 (2001).
The plaintiffs’ challenge to Blaustein’s claim that there was no “generally accepted” method for specific causal attribution took the form of advancing their own “method”: differential diagnosis. At face value, the plaintiffs’ use of differential diagnosis to advance a claim of specific causation is “simply false.” There was no dispute about diagnosis, and no differential diagnosis at issue. Unfortunately, courts have permitted lawyers to corrupt the meaning of differential diagnosis and contend that it covers something akin to differential etiology.
Davidson and Guzelian, in their reference back to an earlier article on evidence-based toxicology, embrace differential etiology as a method of specific case attribution. I agree that there is really no dispute about the logically validity of such reasoning, generally. The logic of differential etiology is simple. If you can specify all the known causes of a disease, and eliminate all but one cause, then you have ruled in the specific cause. Logically, this is an iterative disjunctive syllogism, also known as the process of elimination. The syllogism requires an exhaustive statement of disjuncts, with the negation of all but one:
A v B v C v D
~A and ~B and ~ C.
Therefore, D.
In “The Adventure of the Beryl Coronet,” Sir Arthur Conan Doyle had his famous detective, Sherlock Holmes, articulate this method in ordinary English, with a bit more flair:
“It is an old maxim of mine that when you have excluded the impossible, whatever remains, however improbable, must be the truth.”
Arthur Conan Doyle, The Penguin Complete Sherlock Holmes 315 (Penguin 1981). The process of elimination was a mainstay of Holmes’ forensic thought:
“Eliminate all other factors, and the one which remains must be the truth.”
“The Sign of the Four,” chap. 1 (“The Science of Deduction”), in Arthur Conan Doyle, The Penguin Complete Sherlock Holmes at 92 (Penguin 1981).
The problem of course is, for a disease such as breast cancer, is that one of the disjuncts has been, and will remain, for some time: the proposition that this case is “idiopathic” or “sporadic.” The plaintiffs in Cross did not advance any plausible method for eliminating this disjunct. As a result, they can never arrive at a conclusion that CHRT was a cause of Ms. Cross’s breast cancer. The closest they can get to their desired conclusion with this “method,” once they have eliminated genetic causes, is a conclusion that:
The case is idiopathic OR the case resulted from CHRT.
This conclusion is not really a conclusion at all, but an indeterminate statement, which would be quite unhelpful to the trier in deciding the case. Furthermore, as the case is described by the trial court’s opinion, the plaintiffs did not even attempt a quantification of the probability of each of these two disjuncts. Thus, the plaintiffs failed to offer any substantial evidence that a jury could believe to find in their favor on specific causation. The trial court was correct to reject the challenge to Dr. Blaustein’s specific causation opinion, but the court should have granted the defendants’ challenge to the plaintiffs’ expert witnesses, who had no method at all on the crucial element of specific causation.
Interestingly, one of the plaintiffs’ better arguments against Dr. Blaustein was that he was, as a clinician, unqualified to opine on causation. Cross, *9. The trial court did not elaborate on the argument other than to point out that plaintiffs had emphasized that Dr. Blaustein relied upon his “unquantifiable and untested clinical experience.” Cross, *10. The courts have been remarkably resistant to the argument that physicians are generally unqualified to interpret scientific evidence of causation. Sadly, there is a good deal of empirical evidence to show that physicians are not particularly well trained in statistics or in interpreting clinical research. See, e.g., Donna Windish, Stephen Huot, and Michael Green, “Medicine Residents’ Understanding of the Biostatics and Results in the Medical Literature,” 298 J. Am. Med. Ass’n 1010, 1010 (2007) (“Most residents in this study lacked the knowledge in biostatistics to interpret many of the results in published clinical research.”).
