Rosiglitazone: 42 trials, zero individual signals, one MA. #Shorts

Rosiglitazone (Avandia) was the best-selling diabetes drug in the world by 2006. Each of its 42 individual regulatory trials was powered to detect HbA1c change, not myocardial infarction. No single trial was big enough to detect a rare CV harm signal.

Steven Nissen pooled them all (NEJM 2007). 15,560 patients, 42 trials, MI as the endpoint.
- Pooled OR 1.43 (95% CI 1.03, 1.98). 43 percent more heart attacks.
- Pooled OR for CV death 1.64, just shy of significance.

The FDA imposed a black-box warning in 2007 and severely restricted use in 2010. Sales collapsed.

Meta-analysis lesson: TRIALS ARE POWERED FOR EFFICACY, NOT RARE HARMS. To detect a rare adverse event, you need:
- Many trials pooled (large total n)
- OR genuine post-marketing surveillance
- OR a dedicated CV outcomes trial

The Rosiglitazone case directly led to the FDA 2008 requirement that all new diabetes drugs run pre-approval CV outcome trials (CVOTs). Every modern SGLT2 inhibitor and GLP-1 agonist has CV-safety MA data because of this case.

Other echoes:
- Vioxx (Topol/Curfman pooled MAs detected the signal)
- COX-2 class broadly
- Antipsychotics in elderly with dementia

Sources:
- Nissen SE, Wolski K. NEJM 2007;356:2457-71
- Singh S et al. JAMA 2007;298:1189-95 (independent replication MA)
- Home PD et al. Lancet 2009;373:2125-35 (RECORD trial)
- FDA briefing document, July 2010

Methodological note: individual trial point estimates illustrative; Nissen 2007 pooled OR is exact published value.

Tools: mahmood726-cyber.github.io/allmeta/
App walkthroughs: at 786-MIII Meta-analysis

#MetaAnalysis #EvidenceReversal #Rosiglitazone #Avandia #Nissen #RareHarms #FDA #DrugSafety #SystematicReview #EBM

Видео Rosiglitazone: 42 trials, zero individual signals, one MA. #Shorts канала Evidence Reversal