Feasibility and comparative assessment of incorporating immediate death in discrete choice experiments with duration on health utilities: attribute vs. alternative

To empirically assess the impact of immediate death as an alternative or an attribute on health utilities using discrete choice experiment with duration (DCE_TTO). Three DCE_TTO approaches were developed using the SF-6Dv2 to assess the impact of immediate death on health preferences in an online sample of the general adult population in Quebec, Canada. The first approach combined DCE_TTO with a probability of immediate death (POD) as an attribute, referred to as DCE_POD. The second approach combined DCE_TTO within a best-worst scaling with duration (BWS_TTO) including immediate death as an alternative, referred to as DCE_BWS. The third approach was a standard DCE_TTO model (used for a reference). Mixed logit models were used to analyze data. The models’ performance in estimating values was assessed based on parameters’ logical consistency, the significance of levels, and an analysis of the range of estimated values. A total of 369 participants were included for analysis. The number of parameters with inconsistent orders and non-significant coefficients was lower in DCE_POD compared to DCE_TTO and DCE_BWS. The dimensions of mental health and social functioning provided the largest and smallest decrements in all approaches, respectively. DCE_POD produced a wider range of values (-2.393 to 1) compared to DCE_TTO (-1.959 to 1) and DCE_BWS (-1.474 to 1). The findings demonstrated that when the probability of immediate death was designed as an attribute in DCE_POD, it decreased health utility values and produced a broader value set, while when immediate death was presented as an alternative in BWS_TTO, it increased health utility values and provided a narrower value set.