Выбор редакции

Evaluation of Medical Technologies with Uncertain Benefits -- by Darius N. Lakdawalla, Charles E. Phelps

Cost-effectiveness analysis (CEA) remains the de-facto method of choice to evaluate and compare medical interventions. Standard approaches to CEA use the average (mean) outcomes from clinical effectiveness studies such as randomized controlled trials. This paper generalizes standard methods to include uncertainty in clinical outcomes and proposes a generalized version of the quality-adjusted life-year (QALY), referred to as a quality- and risk-adjusted life-year (QRALY). Our approach requires new information from clinical studies – not only means and variances of health outcomes, but also skewness. With that added parameter, this paper shows how Taylor Series expansions of expected utility can account for two distinct effects of uncertainty: the “insurance value” of reducing overall risks to health, and the “value of hope” produced by the presence of positively skewed outcomes. Simulations demonstrate that stochastic terms are particularly important when baseline disease severity is high, and mean treatment effects are low. They also demonstrate that the variance-based term has the greatest importance among the stochastic terms, although skewness- and kurtosis-based terms can be significant in some situations.