Room: Phillips 231
Time: Mon 15:00 PM-16:30 PM

Eliciting valid utilities: challenges and strategies

Chair: Richard Willke

Session Description

Utilities are the underpinning of decision analysis, including cost-utility, cost-effectiveness and other decision analytic methods that depend upon Quality Adjusted Life Years (QALYs) as an outcome measure. Valid and effective utility elicitation is essential to these analytic methods, without which benefits can be inaccurately estimated and results can be erroneous. While the literature abounds with critiques of preference assessment methods and instruments, some fundamental issues remain unanswered in the most basic, direct utility elicitation methodologies. This session will discuss three aspects of utility elicitation: the quality of utility data from surveys, the use of proxies for individuals unable to answer for themselves, and the use of secondary data sources for utility estimation in situations where primary data are unavailable. Specifically, one presenter will discuss the frequency and handling of uninformative responses to utility surveys. Utility surveys often suffer from relatively high rates of confusing or uninformative data which can bias results and lead to inaccurate utility estimates. Bias and error becomes compounded when such estimates are used in cost-utility analyses, leading to biased decisions. This presentation will discuss the types of uninformative data that arise in utility surveys and approaches that have been proposed to deal with these data. Second, proxies are often needed to provide data in utility surveys when the individual of interest is unable or unwilling to respond directly, as in the case of children. There is often a natural person to provide proxy responses, though the use of proxies raises issues of conflicting motivations and incentives in survey responses. Parents are the natural proxies for children, though can confound their own preferences with those of the child in responding to conventional standard gamble or time trade off questions. This presentation will discuss the biases that may arise in using parents as proxies to value children’s health states, and how such biases affect results. In light of these and other issues in direct utility elicitation, alternative methods for assessing preferences are becoming more established in the field, including using secondary data sources for utility estimation. The final presenter will discuss approaches for using secondary data sources to estimate utility in the specific case of adult survivors of child maltreatment, where sample sizes can be maximized by using non-direct techniques. Secondary data presents unique challenges but also significant advantages, and correct approaches to using these data allows for substantial opportunities to accurately estimate utilities for populations and health states otherwise unmeasured. The presenters will focus on the methodological issues that arise in current approaches to utility measurement through preference surveys and strategies for dealing with the empiric obstacles that investigators face in estimating utilities. Each presenter will provide an overview of the issues involved in utility measurement and provide empiric data demonstrating their points. To allow maximum time for audience questions, the panel will serve as discussants for each other.

Key Terms None

Session Organizer: Eve Wittenberg (Brandeis University)

---

Presentations

1. Uninformative responses to utility surveys: a review of the issues and measurement challenges

   Presenter: Eve Wittenberg (Brandeis University)

   Discussant: Phaedra Corso (University of Georgia)

2. Use of proxies for valuing children’s health: potential biases and recommended approaches

   Presenter: Lisa Prosser (University of Michigan)

   Discussant: Eve Wittenberg (Brandeis University)

3. Estimating utilities for health outcomes associated with child maltreatment from secondary data sources

   Presenter: Phaedra Corso (University of Georgia)

   Discussant: Lisa Prosser (University of Michigan)