Recognizing Preference Diversity: The Use of Preference Sub-Groups in Cost-Effectiveness Analysis

Societal preferences are typically incorporated into cost-effectiveness analyses (CEA) on the basis of the average health state utilities of a sample of public raters. The cost-effectiveness of a program is then assessed on an >all-or-nothing= basis: the program is declared either cost-effective or not for all patients in clinically homogenous sub-groups. However, this approach fails to recognize variability between individuals in their preferences. In this conceptual paper, we consider how the preferences of the individual can be handled within a societal CEA. The concept of preference sub-group analysis is described and its implications assessed. Consider, for example, the choice between lumpectomy and mastectomy for a group of women with breast cancer who are homogeneous clinically but heterogeneous in preferences, and assume that lumpectomy is the more costly procedure. If the incremental cost-effectiveness ratio (ICER) for lumpectomy on the basis of the public=s average preferences is considered acceptable, it would seem unreasonable to refuse mastectomy to those patients who would prefer mastectomy, the cheaper treatment. If the ICER of lumpectomy is not considered acceptable, should it be refused to all patients regardless of the strength of preference of those wanting the procedure? We argue that, in this situation, it is important to explore if sub-groups of individuals, from amongst the sample of public raters, exist (characterized by age, marital status etc.) whose preferences are sufficiently different to the average to generate an acceptable ICER. We address some important practical and conceptual issues of this framework including the implications of the search for sub-groups for the design of valuation studies. The framework has important implications for the practice of CEA and, in particular, for the use of health state valuation systems.