Time preference for health in cost-effectiveness analysis

Abstract

In program evaluation, should a predicted health status gain of 1 quality-adjusted life year (QALY) occurring 10 years from now be valued the same as a 1-QALY increase realizable 5 years from now? Or 1 year from now? If not, how should these future gains (or losses) be evaluated from a present-time perspective? Such questions arise frequently in cost-effectiveness analyses of disease prevention-health promotion programs. This report argues there are actually two distinct interpretations of time preference jointly relevant in many multiperiod program evaluations. 1) In ongoing programs where both present and future population cohorts are, in effect, vying for resources, decision makers must establish a relative social weighting of cohorts by specifying (now) the dollar worth of any unit QALY gain achievable in each. This is a problem of intergenerational equity in the resource allocation process. 2) Individuals, in any cohort, may possess a time preference for the sequence of events comprising their own multiperiod health outcomes. Current models, typically discounting future health gains to present value at some constant rate (r), can well accommodate the first interpretation but not (simultaneously) the second. In response, this report introduces a two-step evaluation procedure featuring the "scenario strategy," a holistic multiattribute preference approach to evaluating multiperiod health outcomes. It allows one to isolate statistically time preference effects at the individual or group level and to incorporate them naturally into the overall evaluation of multiperiod outcomes. A survey-based example and an appendix illustrate the main points.