OBJECTIVES: To estimate the cost per ischemic event (death, nonfatal MI, and/or subsequent revascularization) avoided over six months and to estimate the precision of the cost-effectiveness ratio (CER). METHODS: The sampling frame consisted of patients undergoing a percutaneous revascularization procedure at the University of Maryland Medical System to treat a high-risk coronary lesion according to the American College of Cardiology/American Heart Association morphology classification system. Sixty-two patients treated with abciximab were matched with 62 patients who were not based on gender, diabetes, hyperlipidemia and coronary stenting. Three statistical methods were used to estimate the precision of the CER: Taylor series; Fieller’s Theorem; and confidence ellipsoids. RESULTS: Abciximab-treated patients experienced a 41% reduction in the rate of ischemic events at six months (16.1% vs. 27.4%, p=0.128). The mean total cost of inpatient treatment for abciximab-treated patients was about $2,460 higher than for patients not treated with abciximab. The point estimate for the CER was $21,789 per ischemic event avoided at six months. Overall, confidence intervals were relatively wide due to the non-significant difference in the rate of ischemic events. Using the Taylor series, the 90% CI ranged from $2,097 to $41,480. A 90% CI could not be calculated using Fieller’s theorem. However, the 80% CI ranged from $10,497 to $91,679. Confidence ellipses provide an intuitive approach for visualizing the variability of differences in costs and effects simultaneously. CONCLUSIONS: Although institution-specific analyses often lack power for hypothesis testing, the use of statistical methods can provide decision makers with a range of CERs that are consistent with the data collected.