The valuation of health states worse than dead is challenging. Currently used time trade-off methods face problems in (1) detecting time-dependent preferences and (2) insensitivity toward severity for states worse than dead. The better-than-dead (BTD) method has the potential to detect time-dependent preferences. This study compares the BTD and composite time trade-off (cTTO) methods at the ordinal level.
An experiment was conducted in a convenience sample in which respondents (N = 200) valued the same set of 7 health states in the BTD method and cTTO method. Binary BTD responses were used, with response categories of better than dead and worse than dead. Ternary cTTO responses were used, with the additional equal-to-dead response category. Polychoric correlations were used to determine the agreement between these methods. Consistency and test–retest reliability were assessed within methods.
Overall agreement between the cTTO and BTD method equaled 77.1% and differed between health states and respondents. For both methods, there were few inconsistencies, and the test–retest reliability was comparable (88%). Health states were more often considered worse than dead in the BTD method (BTD: 54.7%, cTTO: 37.2%).
The high agreement between both methods and the comparable amount of inconsistencies and test–retest reliability suggest that the methods have similar measurement properties. The BTD method yielded higher frequencies of worse-than-dead responses while essentially asking respondents to make similar choices. This accounts for part of the disagreement between the methods. Several explanations are offered for this difference, yet more research is needed to explain this phenomenon.