Birth defects are scary. We suspect most parents among our readers can remember their relief upon hearing the pronouncement “normal” during a sonogram and when the post-birth count of fingers and toes came up as expected. When there is a child born with birth defects, parents search for reasons. On possible explanation may be a drug the mother took while pregnant. For more than fifty years, the vast majority of people in this country who have hit or passed childbearing age have understood that some drugs are associated with (or, more conventionally, “cause”) birth defects. The FDA’s familiar regulatory framework for drugs dates to 1962, when it was revamped because of concern over birth defects with a drug approved in other countries. Patient information for drugs, prescription and over-the-counter, is read unevenly, but the risks of use by pregnant women tends to get more attention in the real world. Not surprisingly, there has been quite a bit of litigation involving allegations of birth defects with drugs for many years, some of which has been detailed in various posts here (not to mention Daubert itself). Plaintiffs and their contentions have inherent jury appeal. The child (regardless of age at trial) is faultless and juries are disinclined to hold anything the mother (or father) did against the child. Although birth defects occur in the general population without regard to drug usage, the tendency of jurors to put themselves in the shoes of the parents of the child with birth defects can impede rationality and shift burdens to defendant drug manufacturers—after all, every knows that some drugs can cause birth defects and women are supposed to avoid taking them while pregnant for a reason.
So, it takes a brave judge and strong record to get rid of the general causation opinions of a well-qualified epidemiologist across all cases in an MDL about birth defects with a drug. That is what came together in In re Zoloft (Sertraline Hydrochloride) Prods. Liab. Litig., MDL No. 2342, 2014 WL 2921648, __ F. Supp. 2d __ (E.D. Pa. June 27, 2014). The expert, Dr. Berard, holds a Ph.D. and is a professor in a relevant discipline—Canadian schools count here—and actually had done research outside of litigation on birth defects with this the relevant class of drugs, SSRI antidepressants. The lack of research outside of litigation can be a major Daubert attack on an expert, but Dr. Berard’s non-litigation work was inconsistent with her opinions here. In addition, her work for plaintiffs suing over birth defects with another SSRI produced inconsistent opinions, which leads to the whole petard hoisting thing. (We are not suggesting Dr. Berard actually injured herself with an explosive device. A petard, given the context of the expression, sounds to us like an article of French clothing or perhaps an uncomfortable chair.)
The Zoloft court held a Daubert hearing on Dr. Berard and other challenged experts—at least one other opinion on mechanism experts is coming—and clearly understood the applicable law, scientific principles for evaluating causal relationships, and the evidence presented. Dr. Berard’s opinions that Zoloft and all SSRIs “cause an increased risk” of a wide range of birth defects were evaluated in the context of birth defects (when all lumped together like Dr. Berard and plaintiff did) being relatively common in the general population, mostly idiopathic, and not subject to study through human randomized exposure (because of ethical issues). Id. at ** 2 & 4. Thus, observational studies will be the meat of any evaluation of causation and good science requires first looking for consistent statistically significant results across multiple studies. We welcomed several propositions accepted by the court, which have been contested in many pastDaubert battles:
- “A statistically significant result does not necessarily indicate a large increase in risk; it simply indicates that the increased risk found is unlikely to result from chance alone.”
- “Even when the confidence interval is narrow and the increased risk is statistically significant, teratologists will not draw firm conclusions from a single study . . .”
- “[T]he teratology community requires repeated, consistent, statistically significant human epidemiology findings, and studies which address suspected confounders and biases.”
Id. at *3. The court also explained that an opinion on causation “requires analysis beyond the identification of statistical correlations reported in published epidemiological studies.” Id. Thus, the court looked closely at the reliability of Dr. Berard’s methodology in reaching her opinions, an inquiry that plaintiffs sometimes cut off by contending that the defendant is asking the court to decide whether an opinion is correct.
Dr. Berard’s opinions faltered against these standards in large part because her methodology allowed her to base causation opinions on graphing non-significant odds ratios from selected published studies and trying to discern a trend from them. Id. at *4. Rather than just saying “that sounds like nonsense to me,” the court looked to see if this methodology had been accepted by other epidemiologists or teratologists, how it compared to the generally accepted methodology, and if Dr. Berard could justify departing from the mainstream approach. Id. at *5. Dr. Berard and plaintiffs did not carry their burden on these issues, so she was seen as arriving at her opinions “by ignoring the basic requirements of the relevant scientific community’s methodology.” Id. (quoting Wade-Greaux v. Whitehall Labs., Inc., 874 F. Supp. 1441, 1478 (D.V.I. 1994)). As the court said, “it is not for the courts to be the pioneers, forging new trails in scientific thinking, especially when that means departing from well-established research principles, such as the principle of statistical significance.” Id. at *5. The court also applied its common sense in rejecting the contentions underlying Dr. Berard’s trend methodology that multiple weak associations across studies can substitute for statistical significance and that statistical significance should not be required when studying rare events. Id. Nor could Dr. Berard justify why she did not do a meta-analysis or consider the results of meta-analyses that had been done, something the court said is a widely accepted approach (and is about the only place in the opinion where we take issue). Id. at *6.
Dr. Berard compounded these methodologic flaws by how she arrived at her opinions here. First, she chose to graph the studies she thought were most supportive of her opinion, making trends illusory and raising the possibility for the court that her approach “may be driven by her desire to confirm her a priorihypothesis that Zoloft is a teratogen, rather than by her desire to test the possibility that individual studies were underpowered to detect true associations.” Id. at *6 n. 25. That seems to us like a nice way of saying that statistics can be worse than damn lies. Next, Dr. Berard tried to opine there was a class effect based on presuming there is a class effect and then trying to brush aside different results for different drugs and different types of birth defects in different studies. Id. at **6-7. This included her own publications through 2012 saying that another SSRI may have unique teratogenic properties and Zoloft was to be considered first-line therapy for depression during pregnancy. Id. at *8. And, as a plaintiff expert in relatively recent litigation over another SSRI, she denied that SSRIs were teratogenic as a class. Id. at *8 n. 36. Then, Dr. Berard could not provide a cogent explanation for why she cherry picked some studies for her analysis and excluded others. “[R]ather than simply ignoring certain studies, the accepted scientific practice is for an expert to explain why she gives more weight to certain studies in forming her opinion, discussing methodology, power, and other key factors.” Id. at *9. Again, her publications and work as a plaintiff expert in litigation on another SSRI gave lie to her excuse for her shifting opinions on a class effect as she attempted to rely here on studies she had previously interpreted as not supporting a class effect. Id.
Last, the court ran through the Bradford Hill criteria to see if the leap from a trend of weak associations to causation was reliable. Predictably, with issues like different results as to different types of birth defects and the lack of information to support a biological mechanism that would explain the full range of birth defect lumped together for Dr. Berard’s opinion, she did not fare so well under old Sir Austin’s criteria. Id. at **10-12. Noting that the mainstream view of the same studies Dr. Berard considered is that there is “no conclusive evidence of an association between Zoloft and birth defects,” the court held that Dr. Berard’s contrary view was not “a mere professional difference of opinion; Dr. Berard’s opinions regarding Zoloft are only made possible by her departure from use of well-established epidemiological methods.” Id. at *12. Dr. Berard’s causation opinions are gone and, it would seem, so is the plaintiffs’ ability to prove general causation, something the court understood requires competent epidemiological expert evidence. What is left is a decision that should be useful for other Daubert challenges on general cause opinions.