CN2: EFFECT OF INCLUDING (VERSUS EXCLUDING) FATES WORSE THAN DEATH ON UTILITY MEASUREMENT

Abstract

OBJECTIVE: To test impact of including (versus excluding) health states perceived to be worse than death on utility measurement using standard gamble (SG) and visual analogue scale (VAS) methods. METHODS: Face-to-face interviews were conducted in a convenience sample of women aged 22 to 50 years with no history of breast cancer or cancer requiring chemotherapy (n = 119). Data were collected between March 2000 and June 2000 at a Midwestern University in the US. Subjects were asked to rate their utility if faced with three hypothetical breast cancer health states: cure, treatment and recurrence. Utility weights were estimated using SG top-down titration and VAS methods scaled from zero (death) to one (perfect health). Linear transformation was used to allow for negative preference weights for health states worse than death. RESULTS: Unpaired t-test analysis showed statistically significant greater change in SG (n = 16) and VAS (n = 17) utilities for respondents perceiving cancer recurrence as worse than death than those perceiving death as least desirable health state (p < 0.05). Mean change in SG (VAS) utility allowing for health states worse than death for breast cancer cure versus treatment, treatment versus recurrence, and cure versus recurrence were 0.066(0.226), 0.335(0.339), and 0.401(0.564), respectively (n = 119). Excluding negative utilities from study resulted in statistically significant lower mean changes in utility weights (SG n = 103, VAS n = 102). CONCLUSION: Most studies typically measure health preferences excluding health states perceived as worse than death. Similar to Patrick et al (1994) results of this study indicate health preference elicitation methods can be successfully adapted to acquire negative utilities. Regardless of metric used, both SG and VAS yield higher changes in weight estimates when negative utilities are permitted. Addressing negative utilities in studies could significantly affect QALY estimates in economic analyses.