Lipitor (Atorvastatin Calcium) Mktg. v. Pfizer, Inc.
892 F.3d 624
4th Cir.2018Background
- Over 3,000 women sued Pfizer alleging Lipitor (atorvastatin) caused new-onset diabetes; cases were centralized in an MDL in the District of South Carolina and four bellwether plaintiffs were selected.
- Plaintiffs relied principally on three experts: Dr. Nicholas Jewell (statistical reanalyses of clinical trials / NDA data), Dr. Sonal Singh (literature review, meta-analysis, Bradford Hill causation analysis by dose), and Dr. Elizabeth Murphy (specific-causation differential-diagnosis for a bellwether plaintiff).
- The district court excluded major portions of plaintiffs’ expert testimony under Rule 702/Daubert: Jewell’s analyses of ASCOT and NDA data, Singh’s opinions for 10/20/40 mg (permitting only 80 mg), and Murphy’s specific-causation opinion.
- After excluding experts and issuing show-cause orders, plaintiffs failed to produce alternative admissible evidence of specific causation; the court granted summary judgment for Pfizer across the MDL.
- Plaintiffs appealed, challenging the Daubert exclusions, the judge’s dose-specific requirement and reliance on statistical significance, the exclusion of differential-diagnosis evidence, and the MDL court’s decision to resolve summary judgment rather than remand cases.
Issues
| Issue | Plaintiff's Argument | Defendant's Argument | Held |
|---|---|---|---|
| Admissibility of Dr. Jewell’s statistical reanalyses (NDA, ASCOT) | Jewell’s reanalyses show a statistically significant association and should be admitted to prove general causation and notice. | Jewell cherry-picked methods, lacked clinical expertise to re-define diabetes endpoints, and applied unreliable/test-selective methods. | Exclusion affirmed: court found methodology results-driven, improperly substituted clinical adjudication, and unreliable under Daubert. |
| Requirement of dose‑specific causation and role of statistical significance (Dr. Singh) | Bradford Hill may be applied without a statistically significant association; not necessary to opine dose-by-dose. | Dose matters for pharmaceuticals; expert must show causation at the specific dose taken; weak/non-significant associations cannot reliably support Bradford Hill analysis. | Affirmed: court permissibly required dose-specific opinions; excluded Singh for 10/20/40 mg because associations lacked sufficient statistical support; 80 mg allowed. |
| Specific causation via differential diagnosis (Dr. Murphy) | Differential diagnosis is a standard, reliable method; Murphy reasonably ruled out alternatives and linked Lipitor to plaintiff’s diabetes. | Murphy failed to meaningfully rule out stronger alternative causes (age, BMI, family history); methodology not reliably applied to show Lipitor was a substantial contributing factor. | Affirmed exclusion: court found Murphy’s differential-diagnosis unreliable and conclusory as applied to the plaintiff. |
| Sufficiency of non-expert evidence / party admissions to avoid summary judgment; MDL disposition/remand | Plaintiffs: internal emails, label statements, NDA materials and website statements are admissions/strong evidence and can substitute for expert causation proof; MDL court should remand cases for individual adjudication. | Pfizer: those materials at best show association/notice, not causation; state substantive law typically requires expert proof for complex medical causation; MDL court may adjudicate common issues. | Affirmed: district court correctly ruled non‑expert materials insufficient to create a triable issue on causation; MDL court properly retained and resolved the cases via summary judgment. |
Key Cases Cited
- Daubert v. Merrell Dow Pharm., 509 U.S. 579 (gatekeeping role; reliability and relevance of expert testimony)
- Kumho Tire Co. v. Carmichael, 526 U.S. 137 (Daubert gatekeeping applies to non-scientific experts; focus on methodology and application)
- Gen. Elec. Co. v. Joiner, 522 U.S. 136 (abuse-of-discretion standard for excluding expert testimony)
- Westberry v. Gislaved Gummi AB, 178 F.3d 257 (4th Cir.) (need to show levels/doses of exposure and distinction between general and specific causation/differential diagnosis)
- Matrixx Initiatives, Inc. v. Siracusano, 563 U.S. 27 (statistical significance is not an absolute precondition to inferring causation; other evidence may be probative)
- In re Meridia Prods. Liab. Litig., 447 F.3d 861 (6th Cir.) (labels/admissions can in rare cases suffice but depend on the language and strength of the statement)
- In re Food Lion, Inc. Fair Labor Standards Act Effective Scheduling Litig., 73 F.3d 528 (4th Cir.) (MDL transferee courts have authority to resolve merits/summary judgment in centralized proceedings)