The STAT News investigation reveals that last fall the Journal of the American Academy of Child & Adolescent Psychiatry (JAACAP) quietly issued an expression of concern for the seminal 2001 Keller et al. paper on paroxetine (Paxil) in adolescents. What the coverage only partially captures is how this 'cautionary note' was deliberately hard to locate—nested deep in the journal's site rather than prominently linked to the original abstract—exemplifying the half-measures that characterize corrections in pharmaceutical research.

Study 329 was a randomized controlled trial (RCT) enrolling 275 adolescents aged 12-18, funded by GlaxoSmithKline (GSK) with multiple authors holding financial conflicts of interest. The original publication claimed paroxetine was safe and effective. Yet a 2015 independent reanalysis published in The BMJ by Le Noury and colleagues, which restored access to the complete patient-level dataset, demonstrated the drug failed primary efficacy endpoints and produced a statistically significant increase in suicidal ideation and serious adverse events (number needed to harm approximately 1 in 10). This reanalysis stands as higher-quality evidence because it relied on raw data rather than the selectively reported outcomes in the 2001 paper, which had shifted endpoints after unblinding—a practice now recognized as outcome switching.

The original STAT coverage misses the broader pattern: GSK had internal documents showing the company knew of these risks yet marketed the study aggressively, leading to a record $3 billion settlement in 2012 for fraud related to Paxil and other drugs. Similar transparency failures appear in other high-impact trials, including those for rosiglitazone and rofecoxib, where initial positive RCTs were later contradicted by independent meta-analyses. In child psychiatry, the ripple effects are acute. Despite the FDA's 2004 black-box warning for suicidality in antidepressants for those under 24—partly triggered by Study 329 and its relatives—meta-analyses as recent as 2022 in The Lancet still cite the uncorrected Keller paper without sufficient qualification, influencing treatment algorithms and insurance coverage.

This case underscores chronic problems in research reliability. Journals resist full retractions fearing reputational damage; expressions of concern function as polite wrist-slaps that busy clinicians rarely notice. Observational studies following the black-box warning have shown conflicting outcomes: some report increased untreated depression, while others document reduced suicide attempts with improved monitoring. The core issue remains: when pivotal evidence is flawed yet foundational, the entire evidence base guiding SSRI prescriptions for young people becomes suspect.

True reform demands mandatory prospective trial registration, full data sharing, and prominent digital linking of all corrections. Until then, the buried Paxil note serves as a metaphor for how financial incentives and institutional inertia continue to compromise the safety of mental health care for adolescents.