174 F. Supp. 3d 911
D.S.C.2016Background
- MDL where plaintiffs allege Lipitor (atorvastatin) caused new-onset Type 2 diabetes; plaintiffs must prove general causation (agent can cause disease) then specific causation.
- Court summarizes accepted epidemiological two-step method: (1) demonstrate a statistically significant association (p-values/CIs), (2) evaluate causality using Bradford Hill factors (strength, consistency, dose-response, plausibility, etc.).
- Lipitor is prescribed at 10, 20, 40, 80 mg; evidence varies by dose: several studies show associations at higher doses (20/40/80 mg) but no study shows a statistically significant association at 10 mg (ASCOT, Cederberg, Navarese). Some mechanistic/metabolic studies find effects at mid/high doses but not 10 mg.
- Court previously required experts to opine by specific dose (CMO 49); parties reopened limited discovery for supplemental reports addressing dose-specific causation.
- Daubert/Rule 702 gatekeeping applied: court evaluates methodology, reliable application to facts, and sufficiency of data; rejects experts who cherry-pick, fail to use accepted epidemiologic methods, or lack requisite expertise.
Issues
| Issue | Plaintiff's Argument | Defendant's Argument | Held |
|---|---|---|---|
| Proper methodology to establish general causation | Use epidemiology/Bradford Hill or other reliable methods; some experts rely on literature review/mechanistic data | Daubert requires reliable scientific method; epidemiology is primary accepted method for chemical/drug causation | Court: Epidemiologic/Bradford Hill method is appropriate; expert may use other methods but must apply them reliably and with sufficient data |
| Dose-specific causation requirement | Plaintiffs initially treated risk as across all doses; later say some experts use class-based or mechanistic reasoning to generalize risk | Pfizer: experts must show sufficient data for each dose, especially because evidence differs by dose | Court: Dose matters; experts must support causation opinions at particular dosages; general/undifferentiated opinions are unreliable |
| Admissibility of Dr. Singh (epidemiologist) | Singh conducted systematic search, meta-analysis, applied Bradford Hill; opines statins (including atorvastatin) can cause diabetes and that 80 mg causes diabetes | Pfizer attacks sufficiency at lower doses and Singh’s handling of dose-response/conflicting studies | Held: Singh's opinion that 80 mg can cause diabetes ADMISSIBLE; Singh’s 10 mg opinion EXCLUDED for lack of a demonstrated association; by his testimony he cannot reliably opine on 20 mg/40 mg without relying on 10 mg, so those are effectively excluded |
| Admissibility of Drs. Quon, Roberts, Gale | Plaintiffs proffer mechanistic, literature-review, and clinical perspectives to support causation across doses | Pfizer contends these experts lack epidemiologic methodology, cherry-pick studies, fail to account for contrary evidence, or lack expertise to apply Bradford Hill | Held: Court EXCLUDES causation opinions of Dr. Quon, Dr. Roberts, and Dr. Gale under Rule 702 (methodological failures, cherry-picking, lack of dose-specific analysis) |
Key Cases Cited
- Daubert v. Merrell Dow Pharm., 509 U.S. 579 (gatekeeping requires reliability and relevance of expert testimony)
- Kumho Tire Co. v. Carmichael, 526 U.S. 137 (Daubert principles apply to all expert technical/scientific testimony)
- Gen. Elec. Co. v. Joiner, 522 U.S. 136 (courts may exclude expert opinions with too great an analytic gap from data)
- Westberry v. Gislaved Gummi AB, 178 F.3d 257 (4th Cir. guidance on expert evidence and liberal admissibility balanced with gatekeeping)
- Zellers v. NexTech Ne., LLC, 533 Fed.Appx. 192 (general/specific causation framework)
- Milward v. Acuity Specialty Prods. Grp., Inc., 639 F.3d 11 (scientists may reach differing causation judgments; methodology must still be sound)
- In re Fosamax Prods. Liab. Litig., 645 F.Supp.2d 164 (discusses use of Bradford Hill and epidemiological proof in drug cases)