疾病资料多元分析中交互作用指标及可信区间的估计

Estimation of Interaction of Two Factors in Epidemiology Data

目的旨在探讨疾病资料多元分析中两因子相互作用的三个评价指标。方法从理论上提出了两因子相互作用三个评价指标的构造，详细介绍了用直接法和多元回归方程系数法求解三个评价指标及可信区间的过程，最后用饮酒和吸烟在男性口腔癌发生过程中的交互作用的例子进行了验证。结果用直接法和多元回归方程系数法求解三个评价指标及可信区间的详细过程通过具体的实例进行了解答。结论在现实的疾病资料多元分析中，交互作用多采用在多元回归方程中加入两因子的乘积项的方法加以分析，本文提出的三个新的评价指标有助于更深地分析两因子的交互作用。

Abstract Objective This paper has proposed three new indices for measuring interaction of two factors in the analysis of epidemiological data. Methods The procedures for designing and calculating the three indices with confidence interval have been demonstrated through a real example of analysis of interaction of cigarette smoking and alcohol drinking among male oral cancer patients. Results RERI(Relative Excess Risk Index),AP(Attributable Proportion),S Index(Synergy Index),and their confidence intervals have been calculated and presented using data of smoking,alcohol drinking,and male oral cancer data. Conclusion The proposed indices would aid evaluation of interaction of factors in the multivariate analysis of epidemiological data. InteractionConfidence intervalsMultivariate analysisEpidemiology