This presentation is part of: Poster Session - Utility Theory; Health Economics; Patient & Physician Preferences; Simulation; Technology Assessment
THE EFFECT OF AGE, RACE AND GENDER ON UTILITY VALUES FOR HYPOTHETICAL HEALTH STATES
Eve Wittenberg, MPP, PhD1, Nomita Divi, MS1, Elkan Halpern, PhD1, Sally S. Araki, PhD2, Lisa A. Prosser, PhD3, and Jane C. Weeks, MD4. (1) Massachusetts General Hospital, Institute for Technology Assessment, Boston, MA, (2) Stanford University, Center for Health Policy / Center for Primary Care and Outcomes Research, Stanford, CA, (3) Harvard Medical School and Harvard Pilgrim Health Care, Ambulatory Care and Prevention, Boston, MA, (4) Harvard School of Public Health, Health Policy and Management, Boston, MA
Purpose: Health states are assumed to have a unique quality of life value, expressed as a utility, consistent across people with the same level of experience with the state. This research examines this assumption by exploring the effect on utilities for hypothetical health states of the age, race and gender of the individuals from whom the values are elicited. Methods: Two parallel analyses were conducted: (1) a meta-analysis of the published literature (1976-2002), and (2) a pooled analysis of 4 primary data sets. For each analysis, a linear model was built by creating a rank-ordered outcome variable for utility and a rank-ordered predictor variable for health state severity, which assured a linear relationship within which to measure the effect of age, race and gender. The meta-analysis was conducted at the study level, and the pooled analysis at the individual level. Models also controlled for method of utility elicitation (meta-analysis) and study (pooled analysis). Health states were ranked through a modified Delphi approach with healthy individuals and physicians. An arbitrary subset of health states were included per study, 2 for the meta-analysis and 3 for the pooled analysis. Results: The meta-analysis yielded 328 studies, 9 of which reported community values for hypothetical health states in combination with either age, race or gender information on respondents. Gender and mean/median age were both non-significant in this model, and race was reported too infrequently to include (adjusted R2=0.60). In the pooled analysis (n=974), white respondents provided relatively higher ranked utilities for the same health state compared with non-whites, (p=.002); age and gender were non-significant (adjusted R2=0.40). Conclusion: Community utilities are recommended for use in societal perspective cost-effectiveness analyses (CEA), yet the composition of the sample from which such values are elicited is rarely reported. Individuals’ race may affect their valuation of health states, which implies that race must be considered in selecting samples to value states. More research is needed on variations in utilities due to individuals’ characteristics, and the validity of community values used in CEA.
