What Your doctor was Never Told about Psychiatric Drugs

For decades, doctors were trained on a version of psychiatric drug data that had been filtered, reframed, and in some cases ghostwritten. The full record tells a different story.

In the early 1990s, a drug called Paxil was being taught to American doctors as a breakthrough. Cleaner than older antidepressants. Safer. More precise. The science, they were told, was settled. It wasn't. And the evidence for that was sitting inside the FDA's own files -- available through the Freedom of Information Act -- for anyone who knew to look.

When a pharmaceutical company seeks approval for a psychiatric drug, it submits its complete clinical trial data to the FDA -- not summaries, not press releases, the raw results. These are documents the public almost never sees. In the late 1990s, psychologist Irving Kirsch used the Freedom of Information Act to obtain the FDA's complete trial records for the most widely prescribed antidepressants in America: Prozac, Paxil, Zoloft, Celexa, and Effexor.

What he found was not what doctors were being taught:

~50% of submitted antidepressant trials failed to show benefit over placebo

1.8 pts average drug–placebo difference on depression rating scale (3 pts = clinically perceptible)

75–80% of improvement in trials occurred in the placebo group, not the drug group

The FDA reviewers noticed. On Paxil, an FDA medical reviewer wrote that the drug showed minimal improvement over placebo. In a Prozac FDA statistical review, the clinical significance of the observed difference was questioned. On Celexa, a reviewer wrote that the effect size is small.

That language exists in the approval documents. It was not transmitted to medical schools.

"Many trials failed. Average benefits were small. Placebo did most of the work. Clinical relevance was questioned by FDA scientists themselves. And yet doctors were taught that antidepressants correct a chemical imbalance."

In 1998, GlaxoSmithKline completed a clinical trial testing Paxil in adolescents. The trial had a number: Study 329. The data were unambiguous -- the drug failed to outperform placebo on its primary measures, and adolescents on Paxil experienced significantly higher rates of suicidal thoughts and attempts.

When Study 329 was published in a major medical journal in 2001, it concluded that Paxil was "generally well tolerated and effective." To reach that conclusion, the authors changed the outcome measures after the trial ended, reclassified suicidal behavior as "emotional lability," and minimized adverse events. The paper was not written by the listed academic authors -- it was ghostwritten by a medical communications firm working for GlaxoSmithKline.

Internal company documents, released years later through litigation, showed GSK executives knew the trial had failed. One memo described the study as "insufficiently positive." Another warned that publishing the raw findings would be "commercially unacceptable."

For years after publication, Paxil was prescribed to adolescents based on Study 329. No warning accompanied those prescriptions. The study was not corrected by peer review or by regulators. It came to light through lawsuits -- and a reanalysis that reached the opposite conclusion.

Pfizer conducted multiple pediatric trials with Zoloft. Several failed to show efficacy. The published literature emphasized benefit. In the Forest Laboratories' adolescent trial with Celexa, the trial failed its pre-specified primary outcomes. The company reframed secondary measures and published the study as positive. Court records later confirmed Forest knew the trial had failed.

In a 2008 analysis published in the New England Journal of Medicine, researchers found that across antidepressant trials submitted to the FDA, nearly half were negative -- but over 90 percent of the published papers were positive. The evidence base that trained a generation of psychiatrists was not incomplete by accident. It had been curated.

By the early 2000s, psychiatric disability was rising -- not despite expanded drug treatment, but alongside it. According to Social Security Administration data, Americans receiving federal disability benefits for mental illness tripled between 1987 and 2007, from roughly 1.25 million to more than 4 million. That increase occurred during the greatest expansion of psychiatric prescribing in history.

Psychiatrist Martin Harrow at the University of Illinois followed schizophrenia patients for 20 years beginning in the late 1970s. His findings, published in peer-reviewed psychiatric journals, were consistent: patients who remained on antipsychotic medication long-term were far more likely to remain psychotic and disabled than patients who, carefully and under supervision, discontinued. The World Health Organization's large international schizophrenia studies found similar patterns -- patients in lower-income countries, where antipsychotics were used more sparingly, had substantially better long-term outcomes.

These findings did not change the curriculum.

This is not a claim that psychiatric drugs never help anyone. Short-term symptom relief is real and, for some people in acute crisis, important. The concern is narrower and more serious: what patients and doctors were told about how these drugs work, how effective they are, and what the long-term consequences look like was not based on complete evidence. It was derived from a curated version.

More than 60 million Americans are currently taking at least one psychiatric medication. Over 4 million children are prescribed psychiatric drugs. They deserve access to the full record.

5 comments

What Your doctor was Never Told about Psychiatric Drugs