Estimating a preference-based index from the Japanese SF-36
Estimating a preference-based index from the Japanese SF-36
复制标题
DOI：
10.1016/j.jclinepi.2009.01.022
复制
发表时间：
2009-12-01
影响因子：
7.2
通讯作者：
Kurokawa, Kiyoshi
中科院分区：
文献类型：
--
作者：
Brazier, John E.;Fukuhara, Shunichi;Kurokawa, Kiyoshi
目的：本研究的主要目的是从日本的SF-36中估计一个基于偏好的SF-6D指数，并将其与英国的结果进行比较。然后，使用标准赌博（SG），通过600名日本普通人群的代表性样本对SF-6D版本定义的249种健康状态进行评估。这些健康状态值使用经典的参数随机效应方法与个人水平的数据和普通最小二乘法（OLS）的平均健康状态值，以及一个新的非参数方法，使用贝叶斯估计方法建模。在日本数据上估计的所有参数模型在拟合优度较差，不一致性较多，更大的预测误差和偏差，以及预测中存在系统性偏差的证据。非参数模型在样本外预测中产生了实质性的改进。身体，角色和社会方面有相对较大的递减比疼痛和心理健康相比，在英国，结论：日本和英国之间的差异评估的SF-6D使用日本的估值数据集估计使用非参数贝叶斯技术在这篇文章中提出的。(C)2009 Elsevier Inc. All rights reserved.
Objective: The main objective of the study was to estimate a preference-bascd Short Form (SF)-6D index from the SF-36 for Japan and compare it with the UK results.Study Design and Setting: The SF-6D was translated into Japanese. Two hundred and forty-nine health states defined by this version of the SF-6D were then valued by a representative sample of 600 members of the Japanese general population using standard gamble (SG). These health-state values were modeled using classical parametric random-effect methods with individual-level data and ordinary least squares (OLS) on mean health-state values, together with a new nonparametric approach using Bayesian methods of estimation.Results: All parametric models estimated on Japanese data were found to perform less well than their UK counterparts in terms of poorer goodness of fit, more inconsistencies, larger prediction errors and bias, and evidence of systematic bias in the predictions. Nonparametric models produce a substantial improvement in out-of-sample predictions. The physical, role, and social dimensions have relatively larger decrements than pain and mental health compared with those in the United Kingdom.Conclusion: The differences between Japanese and UK valuations of the SF-6D make it important to use the Japanese valuation data set estimated using the nonparametric Bayesian technique presented in this article. (C) 2009 Elsevier Inc. All rights reserved.