Public preferences are typically incorporated into cost-effectiveness analyses (CEA) on the basis of the average health state utilities of a sample of individuals drawn from the general public. The cost-effectiveness of a programme is then assessed on an 'all-or-nothing' basis: the programme is declared either cost-effective or not for all patients in clinically homogeneous sub-groups. However, this approach fails to recognize variability between individuals in their preferences. In this paper, we consider how diversity in the preferences of individuals can be handled within CEA when the public's preferences are considered appropriate for defining benefit, with the objective of increasing the efficiency of health care delivery. The concept of preference sub-group analysis is described and some of its implications are assessed. These include the methods that could be used to identify sub-groups from amongst public raters, the appropriate approach to eliciting preferences and the possible implications of preference sub-group analysis for clinical decision making.
Copyright 2001 John Wiley & Sons, Ltd.