Figure 1. Cost-effectiveness plane, thresholds, and confidence ellipses.

Figure 1 is a graphical illustration of how to analyze cost-effectiveness of a new intervention compared to an old intervention (“located” at the intercept). The center of the ellipses represents the difference in average costs and effects between the two treatments. The 2-dimensional cost-effectiveness plane can be dissected into regions of cost-effectiveness based on the socially acceptable cost of one quality-adjusted life year (QALY) or societal “willing-to-pay” (WTP) thresholds.[4] Here we use thresholds of $50,000 per QALY or $100,000 per QALY. Hence, the CE regions are “cost-effective at WTP of $50,000/QALY” (squares); “cost-effective at WTP between $50K–100K/QALY” (triangles); and not cost-effective using societal WTP of $100K/QALY.

In addition, we demonstrate confidence ellipses defined as confidence regions for true incremental cost-effectiveness, analogous to commonly used confidence intervals. The largest ellipse represents the 95% confidence region. In addition, the 90%, 80%, and 50% confidence regions are presented. Each percentage indicates the probability that the true incremental cost-effectiveness of the new treatment is located somewhere within that region.