In In re: Onglyza and Kombiglyze XR Products Liability Litigation, MDL 2809, 2022 WL 43244 (E.D. Ken. Jan. 5, 2022), the MDL judge bifurcated discovery into two phases with general causation proceeding first. At the close of expert discovery, plaintiffs move to exclude three defense experts and defendants moved to dismiss one plaintiff expert. Plaintiffs went down swinging and defendants caught it right in the sweet spot.
The drugs at issue are in class that is generally used as a second-line treatment for diabetes. Plaintiffs allege that the drug causes heart failure and related cardiac conditions. At the request of the FDA, manufacturers conducted a cardiovascular outcomes study, a randomized controlled trial with multiple cardiac endpoints. The result of the trial was that there was no statistically significant difference between the drug and placebo for any primary endpoint (cardiac death, non-fatal heart attack, non-fatal stroke) or secondary endpoint except for hospitalization due to heart failure. Id. at *4. Following the cardiac trial, the authors of the study explained that when you test for multiple endpoints, the more endpoints tested, the greater the likelihood of at least one false positive where the statistical significance is due to chance. The authors cautioned that the outcome regarding hospitalization for heart failure warranted further investigation. Id. at *5. Further study was done in the form of observational studies and meta-analyses, none of which found an association between the drug and heart failure. Id. Trials of other drugs in the class also failed to show an association.
It is against that background that the court had to conduct its Daubert analysis. The court’s primary focus was on the Bradford Hill causation analysis. The Bradford Hill framework consists of nine factors used by epidemiologists to assess “whether an observed association may be causal.” Id. at *1. The court sets out and describes all nine factors. Id. at *2. We’ll focus on the analysis of each expert.
Defendants’ first expert was an endocrinologist – the type of doctor who treats diabetes. Plaintiffs argued that he was not qualified to opine on whether the drug causes heart failure because he is not a cardiologist. Id. at *7. But he does prescribe medications to diabetics, including diabetics with heart disease, and must understand the risks and benefits of those medications. Therefore, he was aware of the risks and benefits of and the scientific literature on the drug before becoming an expert in the litigation. “Thus, [the expert’s] testimony in this case is about a matter growing naturally and directly out of research [he has] conducted independent of the litigation.” Id. (citation omitted).
Defendants’ second expert was a cardiologist. Plaintiffs’ primary challenge to this expert was an unsupported claim that he did not write his expert report. They claimed it was “virtually identical” to another defense expert’s report. However, both experts testified to never seeing each other’s reports or spoke to one another. But they both are cardiologists who were asked to opine on the same issue – whether the drug can cause heart failure. So, there was nothing untoward about them both citing the same sources. Nor was it surprising that both reports used the same language when most of the phrases that plaintiffs point to are exact quotes from the relevant literature. The experts did not copy from each other, they quoted the relevant scientific literature, of which there is not a lot. Id. at *8.
Plaintiffs’ second attack on this expert is that he was not qualified to opine on how to correct multiple-endpoint studies to account for false positives. But the doctor had extensive research experience including with clinical trials and as the head of a research laboratory. Further, the expert explained his opinion on how the correction should be applied, a position supported by other scientists in the field. Id. at *9. If the jury were to accept his corrections, then the cardiac trial did not show any association and there is no need to conduct a Bradford Hill analysis. If, however, the jury discounted the corrections, defendants’ expert’s opinion contained a thorough analysis of why the association does not rise to the level of causation. Id. at *10. Nor could plaintiffs support an argument that defendants’ expert failed to consider any study that he should have in reaching his conclusions. Id. at *11.
The third defense expert challenged by plaintiffs was a pharmacoepidemiologist and based on his regular work and experience he was qualified to review the medical literature and opine on whether it established a causal connection between the drug and heart failure. Using the reverse of their first argument, plaintiffs took issue with this expert not being a cardiologist. But the court found he was not determining whether the patients in the studies had heart failure, rather the relationship between the heart failure diagnosed by others and the drug. Id. at *12. Next plaintiffs attacked the expert’s failure to undertake his own analysis of the raw data underlying the studies. That is not a requirement for a reliable Bradford Hill analysis. Nor was it an issue that defense counsel provided the expert with most of the literature where he also conducted his own searches to satisfy himself he had all relevant studies. Id. In fact, defendant’s expert reviewed plaintiffs’ experts’ reports and testified that he considered all the materials they had. Id. at *13.
The court then turned to plaintiffs’ expert – a cardiologist who testified that using the Bradford Hill methodology, it was his opinion that the drug more likely than not is capable of causing heart failure. Id. at *15. However, neither the expert nor plaintiffs’ counsel “have been able to identify any other expert who has reached this opinion.” Id. Having used the same methodology as defense experts but being the only scientist – inside or outside the litigation – to reach that conclusion, the court was “wary that the method ha[d] not been faithfully applied.” Plaintiffs’ expert had never published an article using the Bradford Hill analysis, nor written on it at all until this litigation. Id. Defendants argued that his analysis was flawed because he only considered the cardiac outcome trial and certain animal studies – no other human data. Indeed, plaintiff’s expert testified that the one finding of a statistically significant increase “should be interpreted as cause and effect unless there is compelling evidence to prove otherwise.” Id. That flies directly in the face of the Bradford Hill criteria that there should be consistency of association – “has the observed association been repeatedly observed by different persons, in different places, circumstances, and times.” Id. at *2. The answer here is no.
Moreover, the authors of the cardiac outcomes study themselves did not come to the same conclusion as plaintiff’s expert. They concluded the finding merited further investigation, not that it established causation. Also, relying on and drawing conclusions from limited data lacks “scientific rigor.” Id. at *16. While plaintiffs’ expert chose to ignore the observational studies, he could not identify any defects in those studies to warrant their exclusion from consideration. And certainly nothing to justify relying on pre-clinical animal studies over human data. Id. at *17. Finally, in an attempt to draw an analogy (the Bradford Hill criteria that asks whether exposure to analogous drugs causes the same outcome), plaintiff’s expert compared the drug at issue not to other drugs in the same class, but a different class of diabetes drugs known to worsen heart failure. “In other words, [plaintiff’s expert] cherry picked [the analogous drug] as a comparison only because they have been shown to cause the outcome at issue.” Id. at *18. Results-driven analysis is not permitted.
This is a jam-packed opinion with critical analyses of four experts – three who considered the totality of the evidence and correctly applied the Bradford Hill criteria and one who went out on a limb and found himself very much alone – and excluded.