In re Neurontin: First Circuit Issues Decision With Sensible View of Aggregate Evidence

The First Circuit has issued a critical decision both in its ultimate ruling and its reasoning. See In re: Neurontin Marketing & Sales Practices Litig., No. 11-1806 (1st Cir. Apr. 3, 2013) (slip opinion available here). In a decision written by Chief Judge Sandra Lynch, the unanimous three-judge panel reversed the district court’s grant of summary judgment and denial of class certification. See slip op. at 24-25.

In the underlying case, plaintiffs alleged that doctors prescribing the anti-seizure drug Neurontin were deceived by the defendant, pharmaceutical giant Pfizer, with respect to “off-label” uses for which Neurontin is not formally approved. The district court repeatedly rejected the plaintiffs’ use of widely-accepted statistical methods, including multiple regression, to prove causation. The First Circuit panel rejected the defendants’ argument that the plaintiffs’ use of aggregate evidence precluded class treatment. See slip op. at 20-22.

The plaintiffs’ expert had shown by regression analysis that essentially all Neurontin prescriptions for bipolar disorder were the result of Pfizer’s off-label marketing. In denying the plaintiffs’ second class certification motion, the district court adopted reasoning typically invoked by class action defendants, holding that the plaintiffs’ expert could not offer class-wide causation evidence because the regression analysis did not take account of doctors’ individual prescribing decisions, and only focused on Pfizer’s off-label marketing. See slip op. at 8.

While the second class certification motion was pending, Pfizer filed a summary judgment motion premised on the same causation argument, arguing that the doctor-by-doctor inquiry purportedly required to augment the regression analysis would be “unmanageable.” Slip op. at 13. The district court denied the second class certification motion and granted Pfizer’s summary judgment motion.

The First Circuit reversed both the district court’s entry of summary judgment and denial of class certification, with the centerpiece of the appellate ruling being the panel’s endorsement of the rigorous statistical methods that the district court had rejected. Recognizing that regression analysis is capable of distinguishing the relative causation effects among multiple independent variables, Judge Lynch’s decision noted that the plaintiffs had not relied exclusively on the regression analysis: “[I]n addition to the aggregate statistical evidence, . . . plaintiffs also presented circumstantial evidence that supported an inference of causation”, such as “documents showing that psychiatrists had almost never prescribed Neurontin for bipolar disorder until after Pfizer began its marketing campaign, at which point prescriptions jumped by 1700% in two years.” Slip op. at 20.

The first Circuit’s decision serves as essential guidance for plaintiffs’ counsel in complex litigation requiring sophisticated and rigorous scientific methods in order to confront facile arguments like those advanced by Pfizer, and adopted, repeatedly and emphatically, by the trial court.