In re Avandia Marketing — Third Circuit vacated class certification and ordered a new predominance analysis

Case
In re Avandia Marketing, Sales Practices and Products Liability Litigation
Court
U.S. Court of Appeals for the Third Circuit
Judge
Thomas L. Ambro (Bill Clinton, 2000)
Date Decided
July 21, 2026
Docket No.
25-2278
Topics
Class Actions; RICO; Pharmaceutical Fraud; Statistical Evidence
Source
Read the full opinion

Background

Third-party payors that reimbursed prescriptions for the diabetes drug Avandia alleged that manufacturer GlaxoSmithKline LLC misrepresented the drug’s cardiovascular benefits and concealed its risks. The payors claimed that the alleged fraud caused physicians to prescribe more Avandia—and caused the payors to reimburse more prescriptions for the higher-priced drug—than would otherwise have occurred. They pursued RICO and state-law claims on behalf of a proposed class.

The District Court excluded the payors’ principal causation expert, finding her multiple-regression analysis unreliable, but admitted one damages model. It nevertheless certified a class covering qualifying Avandia payors during the narrowed period from January 1, 2005, through August 14, 2007. The court concluded that common evidence could establish causation and that purchase records, reimbursement data, supporting documents, and affidavits provided a feasible way to identify class members. GSK obtained permission to appeal under Federal Rule of Civil Procedure 23(f).

The Court’s Holding

The Third Circuit held that the proposed class was ascertainable. Potential members could be identified through objective purchase and reimbursement records and could confirm their status with verifiable affidavits and documentation. The need to cross-reference individual records or distinguish end-payors from fully insured plans did not make the process administratively infeasible.

The court nevertheless vacated class certification because the existing record did not yet establish that common questions predominated on causation. Joining the First, Second, and Ninth Circuits, it held that plaintiffs in a pharmaceutical-fraud RICO class action may use statistical evidence to prove class-wide causation, but that evidence must support causation rather than merely correlation. Because the District Court had not adequately determined whether the remaining evidence isolated GSK’s alleged fraud as the cause of increased Avandia prescriptions, the Third Circuit remanded for further fact-finding under its clarified standard.

The court also held that an appellate court may limit the issues it reviews in a Rule 23(f) appeal, analogizing that discretionary authority to review under 28 U.S.C. § 1292(b).

Key Takeaways

  • Statistical evidence can establish class-wide causation in a pharmaceutical-fraud RICO case only if it supports a causal inference and addresses plausible competing explanations.
  • A proposed class may be ascertainable through a combination of objective transaction records, corroborating documents, and verifiable affidavits, even when records from multiple sources must be cross-referenced.
  • Because the payors’ principal regression analysis had been excluded and the remaining evidence had not yet been shown to establish causation, the predominance finding required further examination on remand.

Why It Matters

The decision provides a framework for using regression analysis and other statistical proof at the class-certification stage in pharmaceutical marketing cases. A temporal relationship between promotion, safety disclosures, and prescription volume is not enough by itself; plaintiffs must present reliable common evidence capable of distinguishing causation from correlation.

The ruling also confirms that ascertainability does not demand a single comprehensive database. Courts may permit a documented claims process using multiple record sources, so long as membership can be determined through objective and administratively feasible methods.

✉️ Get tomorrow’s cases before your first coffee
Daily Case Law is our free morning digest — the most substantive new decisions, filtered to your jurisdictions and topics, each linking back here for the full analysis.

Leave a Comment

Your email address will not be published. Required fields are marked *

Scroll to Top