QALYs When Health Varies Over Time: An Analysis of Model Identification and Parameter Estimation from Time Trade-Off and Standard Gamble Scores

In the rst part of the paper, we consider various QALY (quality-adjusted life year) models in situations where health varies over time, and provide a theoretical analysis of model iden-tication and parameter estimation from time trade-o and standard gamble scores.We investigate deterministic and probabilistic models, and consider ve dierent families of discount factors in all.The second part of the paper contains a discussion of some recurrent themes from related literature.Among other things, we question the standard gamble method as a 'gold standard' in health preference measurement, re-examine the role of constantproportional trade-o, and discuss so-called double discounting of QALYs.We also argue that it is not a matter of choosing between time trade-o and standard gamble procedures, since both types of scores are generally needed in order to be able to disentagle risk aversion from discounting.More broadly, we nd that conclusions drawn from the analysis of models involving chronic health states only may not necesssarily apply to situations where health varies over time.One reason is that risk aversion and discounting may collapse to the same thing when all health states are chronic.