The order certifying a class of third-party payors in the long-running Avandia MDL was the wrong result for the wrong reasons, and the Third Circuit has now vacated that order in an opinion that is well worth the read. In In Re: Avandia Marketing, Sales Practices and Products Liability Litigation, No. 25-2278, 2026 WL 2093904 (3d Cir. July 21, 2026) (to be published in F.4th), the Third Circuit held that the district court erred in ruling that common issues predominated over individual issues on the key issue that has undone the vast majority of pharma-related class actions—causation/reliance.
For anyone who hasn’t been following along, the FDA approved Avandia to treat Type II diabetes in 1999, and the MDL got going in 2007, after a widely publicized, but ultimately disproven meta-analysis purported to show an increased risk of heart attacks. That was almost 20 years ago. In that time, the FDA required a black-box warning and a REMS program, but then determined that neither was justified because there really was no increased risk. An MDL for a cardiac risk that never actually existed.
All that is left in the MDL is a third-party payor class action, where health plans allege that they paid for more Avandia prescriptions because the manufacturer fraudulently misrepresented Avandia’s cardiac risks—not because the medication increased cardiac risk (the original, bogus theory that the FDA repudiated), but because it supposedly did not reduce that risk. Causation is the Achilles heel for third-party payor class actions, and it should have been here, too. The district court, however, ruled that the plaintiff payors could prove causation on a class-wide basis and certified a nationwide class. We gave you our dim view of that order here, and the order came in as our fifth worst drug and device decision of 2025.
We knew where the Third Circuit was headed when we read the opinion’s first sentences:
We do not presume in law that x caused y merely because x happened first. The connection might be causal. But it might be coincidental. Or some z might be responsible for x and y alike. As statisticians emphasize, correlation alone does not prove causation.
In re Avandia, at *1. Correlation alone does not prove causation. We have heard it time and time again, and the Third Circuit made it the centerpiece of its opinion. The plaintiffs’ RICO causation theory was that the defendant’s alleged fraud caused physicians to rely on misrepresentations in prescribing decisions, which caused TPPs to pay for prescriptions they wouldn’t otherwise have reimbursed. The plaintiffs thus needed to show that physicians’ reliance could be proved on a class-wide basis.
On this score, the district court got it wrong. First, the district court ruled that evidence of “a common scheme to deceive” justified an inference of class-wide reliance. The Third Circuit, however, rejected that: “[O]ur cases do not support a broad, uniform rule that courts may infer class-wide reliance any time plaintiffs propose to prove the defendant committed fraud. . . . [And] we have also warned, in a case where the proof of the violation came from common evidence of a common scheme, that ‘reliance is nearly always an individualized question.’” Id. at *12. Moreover, “there is no fraud exception” in proving predominance, and a court should not “relax its certification analysis.” Id. Here, dozens of factors bear on a physician’s prescribing choices, and the Third Circuit faulted the district court for unjustifiably oversimplifying those decisions. The court pointedly noted that the district court’s “portrayal of prescribers’ decision-making process did not cite a single page of the several-thousand-page record” and did not represent “the ‘rigorous analysis’ required for class certification.” Id. at *13.
Second, district court found class-wide proof of causation from “statistical evidence,” namely the manufacturer’s internal studies showing that some of its marketing campaigns caused prescriptions to increase. Id. at *13-*14. But that was wrong too, because those internal studies did not purport to isolate the purportedly fraudulent messaging. Taken at face value, the internal studies could show the impact of marketing generally, but not the alleged fraud.
The plaintiffs therefore fell back on evidence that a decline in Avandia prescriptions correlated with the 2007 publication of the same disproven meta-analysis that started this whole MDL. So (the argument goes) if those same facts (even if false) had been disclosed earlier—say, 2005, the start of the class period—prescriptions would have dropped then, too. Voilà: causation.
The Third Circuit was having none of that:
[W]e are not aware that any court has permitted a putative class of TPPs to prove providers’ reliance by class-wide evidence without statistical evidence the defendant’s conduct caused the injuries. The Plans could have cleared the bar if they had introduced statistical evidence of causation, like a regression analysis. They tried. But the District Court struck that evidence after a Daubert hearing. Without it, the Plans have only statistical evidence of correlation and circumstantial evidence of causation. . . . That is not enough.
Id. at *15. There it is again—correlation is not causation. That is not to say that plaintiffs can never prove causation on a common basis, but the standard is rigorous. Following a long discussion of authorities in other circuits, the Third Circuit concluded as follows:
[W]e hold that TPPs in a pharmaceutical fraud RICO action may prove but-for causation with class-wide statistical evidence so long as that evidence is sufficiently rigorous to show causation, not just correlation. Statistical evidence has such rigor if, like a regression analysis, it can distinguish the causal significance of the variable at issue and justify the rejection of competing explanations.
Id. at *19. The payor plaintiffs here did not have this level of proof. The district court already excluded much of their statistical evidence as unreliable (which the plaintiffs wisely did not appeal), and what they had left fell well short of proving causation on a class-wide basis, or any basis. Importantly, the court vacated class certification and remanded rather than outright denying certification. The plaintiffs therefore may get another shot at developing new evidence—including discovery if the district court allows it—although we don’t understand why the Third Circuit has given the plaintiffs yet another chance after 19+ years of litigation that hasn’t proven anything.
So the saga continues, but the path back to class certification is considerably steeper now. Plaintiffs can use statistical evidence to prove class-wide causation, but it has to be reliable, admissible statistical evidence that isolates the effect of the fraud from everything else. Bare correlation—even striking temporal correlation—isn’t enough. A “common scheme to deceive” standing alone isn’t enough. And, after Avandia, “our marketing studies show marketing works” won’t cut it either. We will keep you posted.
Recent Comments