Kularatna Sanjeewa, Whitty Jennifer A, Johnson Newell W, Jayasinghe Ruwan, Scuffham Paul A
Centre for Applied Health Economics, School of Medicine, Logan Campus, Griffith University, University Drive, Meadowbrook, Brisbane, QLD, 4131, Australia,
Qual Life Res. 2015 Jul;24(7):1785-93. doi: 10.1007/s11136-014-0906-2. Epub 2014 Dec 28.
The aim of this study was to derive an algorithm to estimate utility values for the EQ-5D-3L health states based on the preferences of a population sample from Sri Lanka.
The time trade-off method was used to directly value 198 EQ-5D-3L health states in a general population sample (n = 780) from Sri Lanka. Stratified cluster sampling with random selection within clusters was used to select the sample from four districts. Each participant valued 15 health states via face-to-face interviews. The best fit model was selected using consistency, parsimony, and goodness of fit. Based on logical inconsistency, numerous sub-samples were also used for model specification. For each model, the numbers of illogical orderings in the resulting value set were also examined.
Generalised least squares with random effects were found to be the best specification. The sub-sample consisting of participants with less than seven logical inconsistent observations produced no illogical ordering in the final value set and is considered the preferred model. Compared to value sets in other countries, a high disutility is associated with level 3 deficits in the mobility dimension. More than 50 % of health states in the Sri Lankan value set are deemed worse than death health states.
Sri Lankan utility values for EQ-5D-3L states deviate markedly from existing values for upper middle and high-income countries. It is important to have country-specific utility values to conduct cost-utility analysis.
本研究旨在基于斯里兰卡人群样本的偏好，推导一种用于估计EQ-5D-3L健康状态效用值的算法。
采用时间权衡法直接评估来自斯里兰卡的一般人群样本（n = 780）中198种EQ-5D-3L健康状态。通过在聚类中随机选择进行分层聚类抽样，从四个地区选取样本。每位参与者通过面对面访谈评估15种健康状态。使用一致性、简约性和拟合优度来选择最佳拟合模型。基于逻辑不一致性，还使用了大量子样本进行模型设定。对于每个模型，还检查了所得值集中不合逻辑排序的数量。
发现具有随机效应的广义最小二乘法是最佳设定。由逻辑不一致观察少于7次的参与者组成的子样本在最终值集中未产生不合逻辑的排序，被视为首选模型。与其他国家的值集相比，移动维度的3级缺陷具有较高的负效用。斯里兰卡值集中超过50%的健康状态被认为比死亡健康状态更差。
EQ-5D-3L状态的斯里兰卡效用值与中上收入和高收入国家的现有值明显不同。拥有特定国家的效用值对于进行成本效用分析很重要。
