Xie Feng, Pullenayegum Eleanor, Gaebel Kathryn, Oppe Mark, Krabbe Paul F M
Centre for Evaluation of Medicines, St Joseph's Healthcare, Hamilton, ON, Canada,
Eur J Health Econ. 2014 Apr;15(3):281-8. doi: 10.1007/s10198-013-0474-3. Epub 2013 Apr 4.
Choice-based methods have been used widely in assessing healthcare programs. This study compared the binary discrete choice experiment (DCE) and the multiprofile case of best-worst scaling (BWS) in eliciting preferences for the EQ-5D-5L. Forty-eight EQ-5D-5L health states were selected using a Bayesian efficient design and grouped into 24 pairs for the DCE tasks and 8 sets for the BWS tasks (each set has three health states). A total of 100 participants completed 12 pairs and 8 sets in a random order. A probit regression model and ranked order logistic regression model were used to estimate the latent utilities from the DCE and BWS, respectively. Both tasks were well understood by the majority of participants. The DCE tasks were relatively easier and took a shorter time to complete. The intraclass correlation coefficient (ICC) of the DCE was higher than that of the BWS. The variances associated with the latent utilities estimated from the DCE were larger than those from the BWS. The DCE is more feasible and reliable than the BWS in valuing the EQ-5D-5L. Future studies could focus on comparing the consistency and accuracy of these techniques in predicting the health utilities of the EQ-5D-5L.
基于选择的方法已广泛应用于医疗保健项目评估。本研究比较了二元离散选择实验（DCE）和最佳-最差尺度法（BWS）的多轮廓情况在引出对EQ-5D-5L的偏好方面的效果。使用贝叶斯有效设计选择了48种EQ-5D-5L健康状态，并将其分为24对用于DCE任务，8组用于BWS任务（每组有三种健康状态）。共有100名参与者以随机顺序完成了12对和8组任务。分别使用概率回归模型和排序逻辑回归模型从DCE和BWS中估计潜在效用。大多数参与者都很好地理解了这两项任务。DCE任务相对更容易，完成时间更短。DCE的组内相关系数（ICC）高于BWS。从DCE估计的与潜在效用相关的方差大于从BWS估计的方差。在评估EQ-5D-5L方面，DCE比BWS更可行、更可靠。未来的研究可以集中在比较这些技术在预测EQ-5D-5L的健康效用方面的一致性和准确性。
