Simulated anginal frequency (episodes per week) for 1,000 hypothetical patients. If anginal frequency (f) is log normally distributed, with mean µ and median m, then log(f) is unimodal with mean log(m) and variance 2log(µ/m) (32). Given µ = 10 and m = 3, (as reported in the COURAGE trial [3,30]), 12.2% of patients are angina free (0 episodes per week) and 27.7% manifest more than daily angina (>7 episodes per week). In the COURAGE trial, 12.4% were angina free and 21.1% manifested New York Heart Association functional class III angina at baseline. The COURAGE investigators recently corrected the mean value from µ = 10 to µ = 6 (31). Because the mean of the log transformed distribution depends only on the median of the untransformed distribution, this correction has little effect on the simulation. Using µ = 6 and m = 3, 16.6% are angina-free and 20.3% manifest more than daily angina.

Imputed effect of drug-eluting stents in the COURAGE trial by which the risk ratio (RR) for drug-eluting stents (DES) versus optimal medical therapy (OMT) was imputed from the RR for bare-metal stents (BMS) versus OMT in the COURAGE trial multiplied by the RR for DES versus BMS from a meta-analysis of historical trials in the medical literature (22). PCI = percutaneous coronary intervention; PES = paclitaxal-eluting stent (Taxus, Boston Scientific, Natick, Massachusetts); SES = sirolimus eluting stent (Cypher, Cordis Corp., Miami Lakes, Florida). Squares represent mean RR, and horizontal rules represent 95% confidence interval (CI).