OBJECTIVES : The composite TTO (cTTO) is based on a series of adaptive paired comparisons that iterate toward an indifference statement (similar to the child's game, chutes and ladders). The aim of this project (EQ20170260) was to incorporate time preferences into the estimation of cTTO-based weights. Specifically, the time attribute, remaining years of life, has a marginal decreasing impact on value (i.e., discounting). Most analyses of TTO responses, particularly EQ-5D studies, do not account for this effect, which motivated this analysis.

METHODS : Using EQ-VT data from Spain and the Netherlands, we tested the effect of time preferences on the 20-parameter model between countries (the Netherlands, Spain), response interpretations (point, interval), heteroskedasticity specification (constant, hetero) and censoring (none, -1).

RESULTS : The results show that incorporating time preferences, namely the power function, increases model fit, reduces the number of insignificant level differences (p-value<0.05), diminishes ceiling effects in the value set, extends the range, shrinks confidence intervals, and improves face validity of cTTO-based weights. However, the interval examination also revealed the cTTO iterative procedure fails under time preferences, potentially producing a large spike at "immediate death" and low frequency of responses between -0.5 and 0.5 (i.e. overlap). Basically, the iterative procedure is not monotonic in errors around "immediate death", a flaw not in the original TTO task.

CONCLUSIONS : Incorporating time preferences may reduce sample size requirements and common biases in the estimation of EQ-5D values. Recognizing the flaw in the cTTO task may be strategically challenging due to its widespread use as part of the EQ-VT. Because time preferences vary by respondent, an analytical solution to this spike/valley is impractical; therefore, some researchers may choose to ignore time preferences (e.g., interviewer effects). Nevertheless, this study demonstrates a simple, yet fundamental flaw that forecasts the eventual obsolesce of the cTTO task.