sertraline: the legal and safety record

first, the thing sertraline is not part of

There is a large, genuine scandal in American generic pharmaceuticals: a price-fixing investigation that has produced state attorney general complaints, Department of Justice criminal cases, deferred prosecution agreements and nine-figure settlements. Search for almost any generic drug and you will find pages gesturing at it.

Sertraline is not in it. This was checked three separate ways, and the strongest citation is the court's own list:

• The court-approved combined drug list for the state AG and end-payer cases runs to 58 pages and 212 molecules. The antidepressants on it are amitriptyline, clomipramine, nortriptyline, fluoxetine and trazodone. Sertraline is absent. Zoloft is absent. So are paroxetine, citalopram, escitalopram, venlafaxine, bupropion, mirtazapine and duloxetine [1].
• All three state AG complaints — 56, 524 and 606 pages — were downloaded and full-text searched. Zero occurrences of “sertraline” in any of them. The text extraction was validated by control searches that did hit on fluoxetine, nortriptyline, clomipramine and amitriptyline, so the zero is a real absence rather than a broken PDF [2].
• Zero occurrences in every DOJ criminal charging document and deferred prosecution agreement in that investigation.

A verified negative is more useful to you than an insinuation, and it is substantially harder to produce. If you find a page that implies otherwise, the list above is how to check it.

What we are not telling you, and why. Two manufacturers with sertraline connections do appear in that litigation as defendants — but the drugs alleged against them are other molecules entirely, and those claims are pending, unsettled and unadjudicated. Under this page's rule — settled or adjudicated only — they do not belong here, and we are not going to present an open allegation as though a court had decided it. There is also a consent-decree question involving a former sertraline application holder whose operative appendix was filed under seal; because we cannot read it, we cannot confirm the link, so we are not publishing it.

the birth-defect litigation: the courts ruled for the manufacturer

Between 2012 and 2017, roughly 315 claims were consolidated in federal court in Philadelphia alleging that Zoloft taken during pregnancy caused cardiac birth defects.

The court examined the plaintiffs' scientific evidence and found their expert witnesses' methods unreliable — first Dr Anick Bérard in 2014, then Dr Nicholas Jewell in 2016 — and entered summary judgment for Pfizer. The Third Circuit affirmed on 2 June 2017, and the docket was terminated that December [3].

No claim that Zoloft causes birth defects was ever decided in a plaintiff's favour in that litigation. That is the honest summary, and we would publish it identically if it had come out the other way.

the settled matter that does name Zoloft — and the distinction that matters

We got this wrong the first time, and the correction is instructive. Our earlier research concluded that the 2009 Pfizer resolution had nothing to do with Zoloft. That conclusion was drawn from the DOJ press release — and the press release enumerates only the four off-label drugs (Bextra, Geodon, Zyvox, Lyrica) and refers to the rest as “other, drugs.” Reading the operative documents instead of the announcement gives a different answer.

The civil Settlement Agreement names Zoloft twice. Preamble paragraph A defines it among the “Covered Drugs,” and the Covered Conduct paragraph F(5) states that from January 2001 through December 2004 Pfizer [4]:

“paid illegal remuneration… to health care professionals to induce them to promote and prescribe the drugs Aricept, Celebrex, Lipitor, Norvasc, Relpax, Viagra, Zithromax, Zoloft, and Zyrtec, in violation of the Federal Anti-Kickback Statute.”— United States v. Pfizer Inc., civil Settlement Agreement (2009), Covered Conduct ¶F(5) [4]

DOJ's own False Claims Act fact sheet names the same nine drugs [5].

The precision this requires, stated exactly:

• Zoloft was a kickback drug in that resolution. It was not one of the off-label promotion drugs. Conflating the two is the standard secondary-source error.
• Pfizer paid $1 billion civilly to resolve allegations. A civil False Claims Act settlement of this kind carries no admission of liability, and we are not going to write as though it did.
• The conduct period is January 2001 to December 2004 — more than twenty years ago, and before the approval changed hands.

We are showing our working here rather than quietly fixing the page, because “the press release and the settlement agreement say different things” is exactly the kind of gap this whole series exists to point at. A press release is a communication. The agreement is the record.

recalls: seven, all Class II, none Class I

FDA's enforcement database returns seven sertraline or Zoloft recall records. The headline is what is not among them [6]:

Read that table for what it is. Every entry is a manufacturing, labelling or packaging problem — a seal, a shelf-life date, a tablet thickness, a repackager's mislabel. None is a recall for the drug being unsafe as designed, and the most serious FDA category has never been used for it.

how many people take it — one payer, honestly labelled

This is Medicare Part D only. It is not the United States. We are not going to publish a national total, because no public source produces a defensible one: for 2024, Medicaid plus Part D comes to 77% of the only all-payer estimate available, which is not credible on its face given that commercially insured Americans outnumber Medicaid enrollees roughly two to one. We could not reconcile it, and that irreconcilability is the finding rather than something to average away.

The interesting thing is the ratio between those two charts. Part D patients went from 1.89M to 3.20M — a rise of about 69% — while claims rose about 46%. More people, each filling slightly fewer scripts per year. That is a more honest and more interesting story than a raw prescription count, and the distinct-patient figure that makes it visible is available in Part D and nowhere else [7].

Three things these numbers are not. They are not people starting the drug — a claim includes refills, and there is no new-versus-refill split anywhere in this dataset. They are not diagnoses. And they are not national. Any page that adds payers together to produce “US sertraline prescriptions” is constructing a figure the underlying sources cannot support.

two housekeeping facts most pages get wrong

• Zoloft is no longer Pfizer's. The approval is held by Viatris following the 2020 Upjohn spinoff. Zoloft was originally developed and marketed by Pfizer; writing that in the present tense is now factually wrong [8].
• FDA's own article URLs for this drug have rotted. The 2011 Drug Safety Communication on SSRIs and newborn pulmonary hypertension returned HTTP 200 through 16 February 2026 and 404s from 7 April 2026 onward. So do FDA's pages on antidepressant suicidality. We cite the current label PDF and Drugs@FDA, which are stable, and use archive permalinks for historical announcements. This is the same pattern we documented at what FDA approval actually means [8].