Duloxetine: the legal and safety record

The record, which runs the other way

In the lead consumer case, Saavedra v. Eli Lilly (C.D. Cal., filed 2012), plaintiffs suing under four states’ consumer-protection laws alleged the label understated how often stopping the drug causes withdrawal — the court’s order records the label’s “greater than or equal to 1%” figure against the plaintiffs’ claim “that the risk of withdrawal symptoms following Cymbalta is in fact approximately 44%” [2]. The procedural record that followed is the cleanest defense run in this series: the JPML refused to centralize the cases — twice (2014, 2015) [1]; class certification was denied twice — December 2014, on the ground that plaintiffs had “no damages model at all” capable of classwide measurement, and again in July 2015 on predominance [2]; the case was then dismissed with prejudice at plaintiffs’ own request in October 2015, a tactic to manufacture an appeal that the Supreme Court’s Microsoft v. Baker decision later invalidated — so the Ninth Circuit dismissed the appeal for lack of jurisdiction and, in 2020, affirmed the refusal to reopen [2]. No court ever ruled on the merits of the withdrawal-rate claim, for either side.

What the courts did not decide

Here is the precision this page exists for: those outcomes turned on class-action mechanics and disclosure law — not on a scientific adjudication of discontinuation rates. No court ruled that duloxetine withdrawal is rare. The 1%-versus-44% dispute simply never got resolved in a courtroom, and the discontinuation literature since points firmly toward “common with abrupt stops” — evidence that stands on its own and lives in the stopping guide and on our duloxetine page. A drugmaker winning the lawsuit and a symptom being real are compatible facts; most coverage picks one. This page keeps both.

What is not in the record

• No DOJ case. Unlike most drugs in this series, duloxetine has no federal marketing resolution; Lilly’s 2009 plea was about Zyprexa, a different drug, and importing it here would be exactly the blur this series exists to prevent.
• No certified class, anywhere. Certification denied twice in the lead case, and the appeal never reached the merits [2].
• No Class I recall — sixty-three records, the most in this series, all Class II/III, within openFDA's ~2012-onward coverage window [3].

Common questions

Did the Cymbalta withdrawal lawsuits succeed?

Largely no, and the record is unusually clear about it: the Judicial Panel on Multidistrict Litigation refused to centralize the cases in 2014 and again in 2015; attempts at class certification in California, Massachusetts, Missouri and New York failed, with the Ninth Circuit affirming; and Eli Lilly won the individual cases that reached trial. Reported settlement talks resolved some remaining individual claims, terms undisclosed.

So is duloxetine withdrawal not real?

That is not what the litigation decided, and the distinction matters. The legal fights turned on class-action mechanics and on what the label disclosed — not on a scientific adjudication of how often discontinuation symptoms occur. The label reported figures around 1% in trials; plaintiffs cited analyses closer to 44–50%. Courts never resolved that gap, and the honest reading of the discontinuation literature is that abrupt SNRI stops commonly cause real symptoms — our stopping guide covers the evidence, which stands on its own regardless of who won in court.

What should I do if I take duloxetine and want to stop?

Exactly what the evidence — not the litigation — says: plan a gradual, prescriber-supervised taper. Duloxetine’s relatively short half-life is part of why abrupt stops are rough. Nothing about the defense verdicts changes the physiology.

Has duloxetine had serious recalls?

No Class I recall appears in the openFDA enforcement database (openFDA enforcement data is reliable only from roughly 2012 onward, so this is a statement about the covered window, not the drug’s whole history). Sixty-three records exist — the most of any drug in this series, largely nitrosamine-era and quality recalls — but all Class II or III.

Sources

Last verified 2026-09-02. Corrections change this date.