To the Editor.
—We agree with Drs Sacristán and Obenchain1 that careful analysis of uncertainty is an essential element of sound pharmacoeconomic evaluations. However, their suggestion that a confidence interval (CI) should be reported for the cost-effectiveness (C/E) ratio of an intervention having a significant probability of a negative C/E ratio may prove problematic in applications.Most troubling is the fact that the magnitude of a negative C/E ratio—defined as the ratio of incremental cost (AC) to incremental effectiveness (△E)—conveys no useful information. For example, it is clear that an estimate of AC =-100, △E =100 is more favorable than an estimate of AC =-100, △E=50, which in turn is more favorable than an estimate of AC=-50, △E=50. However, this implies that a C/E ratio of-1 (-100/100) is preferable to a ratio of -2 (-100/50), and yet a ratio of-2 (-100/50) is preferable to a ratio of -1 (-50/50).
Stinnett AA, Mullahy J. The Negative Side of Cost-effectiveness Analysis. JAMA. 1997;277(24):1931-1932. doi:10.1001/jama.1997.03540480031025