Sample size and power for cost-effectiveness analysis (Part 2): the effect of maximum willingness to pay.

Sample size and power for cost-effectiveness analysis depend on assumptions about the difference in cost and effect, the standard deviations of cost and effect, the correlation of the difference in cost and effect, the α and β errors and maximum willingness to pay (W). The first seven of these parameters share much in common in their effect on sample size and power for cost-effectiveness analysis, including that each is associated with a single pattern of power. W, on the other hand, is unique in that, when plotted for positive values, we can potentially observe any of six patterns of power associated with positive values of W. In addition, as W approaches ∞, power need be neither monotonically increasing nor decreasing and it can be multimodal. In this article, the relationship between W and sample size and power is explained.