多元贝叶斯回归模型框架下绘制成本-效果可接受曲线

Cost-effectiveness Acceptability Curve Drawn from Multivariate Bayesian Regression Model Framework

目的:探讨成本-效果可接受曲线(CEAC)采用贝叶斯方法的原因,以及如何在贝叶斯回归模型框架下绘制CEAC。方法:首先从原理上分析贝叶斯方法的优势,其次在R软件中生成模拟数据并在Openbugs软件中实现模型,绘制CEAC,比较不同先验信息下该曲线的差异。结果:在强先验信息下,参数后验分布的均数比弱先验信息下更准确、标准差更小。在一定范围内,强先验信息下增量-净效益大于0的概率大于弱先验信息下的概率。结论:贝叶斯方法在参数估计、区间估计、概率解释上具有优势。在贝叶斯方法下,CEAC可得到合理的解释。贝叶斯方法先验信息对参数估计与成本-效果分析决策甚为重要。在强先验信息下,参数估计的准确度、变异度更小,同时CEAC也更准确。

OBJECTIVE： To investigate the reasons for cost-effectiveness acceptability curve （CEAC） using Bayesian methods, and how to draw CEAC in the framework of the Bayesian regression model. METHODS： Firstly, the advantages of Bayesian approach were analyzed fi-om the principle. Secondly, R generated simulation data, Openbugs software implemented model and CEAC was drawn. The differences in the curves of different prior information were compared. RESULTS： Under strong prior information,the posterior distribution of the mean parameters was more accurate than weak prior infbrmation, and its standard deviation was smaller than weak prior intbrmation. The probability of incremental-net benefit which was greater than 0 under strong prior was greater than the probability of weak prior within limits. CONCLUSIONS： Bayesian method is characterized with the advantages of parameter estimation, interval estimation and probabilistic interpretation. By Bayesian approach, CEAC can provide a reasonable explanation.The Bayesian approach prior information is important for parameter estimation and cost-effectiveness analysis decision-making. Under strong prior information, the accuracy and variability of the parameter estimation become weak, while CEAC is more accurate.