健康效用值测量中的映射法及其相关模型概述

发布时间：2018-03-03 00:18

本文选题：映射法　切入点：健康效用值　出处：《中国药房》2017年29期 　论文类型：期刊论文

【摘要】：目的:为将非效用量表测量结果转化为健康效用值提供方法学参考。方法:查阅国内外文献,总结健康效用值测量中的映射法及其相关模型,并以生存质量量表测量结果转换成欧洲五维健康量表效用值为例,介绍各模型在概率映射中的应用。结果:映射法可通过建立非效用量表和效用量表之间的映射关系,进而得到健康效用值;其常用模型包括普通最小二乘法(OLS)模型、最小绝对离差法(CLAD)模型、Tobit模型、多元Logistic回归(MNL)模型、贝叶斯网络(BN)模型等。其中,OLS模型相对简单,预测效度较好,但会受到天花板效应的限制;Tobit模型不受天花板或地板效应的限制,当误差项满足方差齐性及正态性时,其预测结果优于OLS模型;CLAD模型可用于误差项非方差齐性不适用Tobit模型的情况;MNL模型先通过回归分析确定一种健康状态,再确定其效用值;BN模型预测效度较好,且不涉及计量经济学中的诸多假设和限制条件,但其构建过程受领域专家的影响较大。使用MNL或BN模式计算健康效用值的方法主要有蒙特卡洛模拟法、期望效用值法和最大可能概率法等。利用拟合优度、调整拟合优度、平均误差、均方误差、平均绝对误差等指标进行模型性能评价,可选出最优模型,进而计算健康效用值。结论:由于各映射模型各有优缺点,在临床研究中需要根据实际情况选择不同的映射模型。
[Abstract]:Objective: to provide a methodological reference for converting the non-utility measurement results into health utility values. Methods: the mapping method and its related models in the measurement of health utility values were summarized by consulting the literature at home and abroad. The application of each model in probabilistic mapping is introduced by taking the result of quality of life scale as an example. Results: mapping method can establish the mapping relationship between non-utility scale and utility scale. Then the health utility value is obtained. The commonly used models include the ordinary least square method (OLS) model, the minimum absolute deviation method (CLAD) model and the Tobit model, the multivariate Logistic regression model and the Bayesian network model, among which the OLS model is relatively simple. The prediction validity is good, but limited by the ceiling effect. The Tobit model is not restricted by the ceiling or floor effect. When the error term satisfies the homogeneity and normality of variance, The result of prediction is better than that of OLS model. The model can be used to determine a health state by regression analysis, and then to determine the validity of OLS model when the non-variance homogeneity of error term is not applicable to Tobit model. Moreover, it does not involve many assumptions and restrictions in econometrics, but its construction process is greatly influenced by domain experts. Monte Carlo simulation is the main method to calculate health utility value using MNL or BN model. The method of expected utility value and the method of maximum probability are used to evaluate the performance of the model, and the optimal model can be selected by using the indexes of goodness of fit, adjustment of goodness of fit, mean error, mean square error, mean absolute error, etc. Conclusion: because each mapping model has its own advantages and disadvantages, it is necessary to select different mapping models according to the actual situation in clinical research.
【作者单位】：中国药科大学国际医药商学院;
【分类号】：R95

【相似文献】