Incorporating feelings related to the uncertainty about future health in utility measurement

Abstract

The cost-effectiveness of health-care interventions is often evaluated using quality-adjusted life years (QALYs) as a measure of outcome. QALYs are valid representations of welfare only under the questionable assumption of additive independence of utility of health states across time periods. Therefore, some alternatives to QALYs have been proposed, among them healthy-year equivalents (HYEs) (Med. Decis. Making 1989; 9(2):142-149) and a semi-separable utility function (SSUF) (J. Health Econ. 2005; 24(1):33-54). This paper shows that HYEs using a single probability-equivalent standard gamble (SG) question measure the same health construct as the SSUF, which uses a series of probability-equivalent SG questions. The underlying assumption is that final health states are independent of initial health states. In contrast to the SSUF, however, HYEs are not bound by this assumption and also avoid propagating bias in the measurement of SG utilities from one question to the next. This paper also shows that both approaches can incorporate feelings related to the uncertainty about future health and capture them by using certainty equivalent questions.

A reply to Gandjour and Gafni

“…Because Gandjour's (2008) model is ill-defined and ambiguous, it is impossible to understand exactly what Gandjour and Gafni mean. Nevertheless we will try our best to interpret their writings as good as we can.…”

“…The first fundamental problem is that the model of Gandjour (2008), which underlies Gandjour and Gafni's (in press) analysis and is stated in their Eq. (1), is not well-defined.…”

“…In particular, it is unclear how the formula should be applied when computing probability weighted averages such as in expected utility or its generalizations. Gandjour (2008) claims that expected utility should not be used to compute probability weighted averages but in Eq. (4) of their comment Gandjour and Gafni (in press) do use expected utility to compute probability weighted averages.…”

The additive utility assumption of the QALY model revisited

“…The utility of the 3-period health profile is thus not only determined by expected mean future health, but also by anticipatory feelings about the uncertainty around the expected future health L (Gandjour, 2008). Furthermore, health states in the second and third period may also be influenced by preceding health states.…”

“…The utility of this profile can be described as an additive utility function described in Gandjour (2008):…”