前瞻研究中多因素人群归因危险度的Poisson回归模型估计

ESTIMATING THE POPULATION ATTRIBUTABLE RISK IN MULTIFACTORIAL PROSPECTIVE STUDY BY USING POISSON REGRESSION MODEL

本文提出多因素前瞻研究中利用Poisson回归发病率预测模型和相对危险度估计调整和综合人群归因危险度的方法,与Bruzzi等和Deubner等提出的多因素人群归因危险度估计方法进行了比较,强调在前瞻资料的人群归因危险度的估计中利用poisson回归模型考虑失访病例和随访时间效应,并能直接估计相对危险度的优势.应用所建立的方法对启东县肝病人群14年前瞻观察资料进行肝癌危险因素的人群归因危险度的估计。

This paper presents an approach for the calculation of the adjusted and summary PoPulation attributale risk(PAR)in multifactorial prospective study by ultilising Poisson regression incidence predictive model and the relative risk. The relationship of this approach to procedures proposed by Bruzzi et al and deubner et al is discussed. The authors emphasize on the benefits to be obtained from poisson regression model, which can ultilise persons lost to follow-up and consider person-years and directly estimate the relative risk in fitting prospective data. This approach is illustrated with data on four risk factors from fourteen years follow-up study of high risk population with hepatosis.