Turner EH, et al. Selective publication of antidepressant trials

One-paragraph summary

The investigators compared FDA judgments for 74 antidepressant trials with the published literature. Nearly all FDA-positive trials were published as positive, while most negative/questionable trials were unpublished or published with a positive impression; 94% of published trials appeared positive versus 51% by FDA assessment, and published effect size was 32% larger.

Key findings

• Reporting status tracked regulatory result.
• Selective publication changed both apparent success rate and effect magnitude.

Limitations

• Historical registration program; contemporary transparency rules differ.
• Does not estimate every form of bias or prove no drug effect.

Why it matters

It directly measured evidence distortion using a regulator's near-complete trial set.

Cited by wiki pages

• The efficacy debate
• Overview