混合效应meta分析及其应用

Mixed Effects Meta-Analysis and its Application

目的介绍混合效应meta分析模型的基本理论,并通过实例详细阐述其应用。方法Francesco等人提出将线性混合效应模型与meta分析方法结合,可用于合并复杂的多参数相关效应。本研究对该方法做详细的阐述,并采用该方法结合二阶段分析策略探究日均气温对英格兰与威尔士10个地区人群死亡率的影响。结果英格兰与威尔士10个地区日均气温与人群死亡的综合暴露—反应关系呈U形,且气温的影响存在地区异质性(Cochran Q=99.6513、I 2=63.9%、P<0.0001)。相对于最适宜温度,第1百分位数气温(低温)与第99百分位数(高温)的相对危险度分别为1.807(95%CI:1.605~2.035)与1.201(95%CI:1.150~1.254)。结论混合效应meta分析可用于合并不同研究的多个复杂相关参数,可广泛应用于多地区暴露与健康关系的流行病学研究以及临床试验中多中心纵向数据分析等。

Objective To introduce the basic theory of the mixed effects meta-analysis model and illustrate its application using an example.Methods Francesco proposed a method to synthesize complex estimated results across studies by casting the meta-analytical problem as a linear mixed-effects model.This study elaborated on the method and applied this model combined with standard two-stage analysis to explore the impact of daily mean temperature on the number of daily deaths in ten regions of England and Wales.Results The mixed-effects meta-analysis revealed a U-shaped exposure-response relationship and found spatial heterogeneity(Cochran Q=99.6513、I 2=63.9%、P<0.0001).Besides,relative to minimum-mortality temperature,the relative risk of the 1st percentile temperature(low temperature)and the 99th percentile(high temperature)was 1.807(95%CI:1.605~2.035)and 1.201(95%CI:1.150~1.254),respectively.Conclusion Mixed effects meta-analysis extends the use of meta-analysis,summarizing multiple correlated parameters across different studies.The approach can be used for the study of the exposure-response associations from multi-centers,individual studies in longitudinal clinical trials at multiple sites,etc.