回归模型在Meta分析中的应用及Stata实现

Application and STATA implementation of regression model in Meta-analysis

Meta回归是循证医学研究中一种重要的统计方法,可以用于分析异质性过大的原因,进而发现与筛检异质性的来源;也可以用于预测因变量与自变量的变化关系。本研究主要介绍了Meta回归的数学模型以及如何运用Stata软件建立Meta回归模型,寻找合并效应异质性过大的原因,并通过亚组分析来降低Meta分析中的异质性。另外,还介绍了如何应用回归模型自身性质来评估预测因变量与自变量之间的变化关系,并对Stata软件分析结果进行了解读。通过案例分析让读者可以重复文中Meta结果,验证文献中最终结果的产生过程。掌握回归方法在Meta分析中的应用与实现对于处理Meta分析中的各种问题具有重要价值。

Meta Regression is an important statistic tool in evidence-based medicine research. It is always used for detecting the reasons of heterogeneity in Meta-analysis. Also, the method can be used to predict the relationship between dependent variables and independent variables. This study focuses on how to adapt the Stata software to create a Meta-regression model to find the cause of the heterogeneity of the combined effect and to reduce the heterogeneity of the Meta-analysis through subgroup analysis. At the same time, article pays attention to how to apply the regression model to evaluate and predict the relationship between the dependent variable and the independent variable, and gives the specifically interpretation of the Stata outcome. Through detail case studies, the reader may repeat and reproduce the Meta results in the text and verify the process of final result in the literature. In short, the realization and detail of regression methods in Meta-analysis is of great value for dealing with various problems in Meta-analysis.