Problems with Interval Estimates of the Incremental Cost—Effectiveness Ratio

The defining feature of a confidence interval is that it has a fixed minimum probability of covering the true value of the parameter being estimated, whatever the value of the parameter. The authors demonstrate by simulation that some recently proposed meth ods for interval estimation of the incremental cost—effectiveness ratio (ICER) either do not satisfy this definition or have other problems that limit their usefulness in applica tions. The problems are most prominent when the ICER is large and the true effec tiveness difference is small relative to its standard error. A modification of the percentile bootstrap confidence interval that involves a reordering of the sample space provides a partial solution of the problem. Key words: bootstrap; clinical trials; confidence inter vals ; cost—effectiveness ratios; Fieller's method; Taylor series. (Med Decis Making 1999;19:9-15)