Poisson和Cox回归模型在队列随访资料分析中的应用和对比

Ｐｏｉｓｓｏｎ与Ｃｏｘ回归模型是流行病学队列随访资料分析中常用的两类多变量分析方法。本文对有关这两类多变量回归模型的统计方法等问题进行了系统的回顾（相乘模型），并用一个实例的结果来说明两者的应用。从本文的结果和讨论来看，Ｐｏｉｓｓｏｎ和Ｃｏｘ回归模型均适合于队列随访资料的分析，但两者各有一些优势和不足。最后，笔者就目前两者的应用情况和相互比较提出了一些看法。此外，还讨论了其它形式的回归模型（相加模型）及在回归模型中如何引入外部对照率等。

APPLICATION AND COMPARISON OF POISSON AND COX'S REGRESSION MODELS IN EPI-DEMIOLOGICAL COHORT STUDIES

poiason and Cox's regression models are the mostcommon teconiques of multivariable analysis used in epi-demiological cohort studies.A short review of statisticalmethods for the above two regression models(multiplica-tive model ),was presented,and an example was reported to illustrate their application. Poisson and Cox's regressionmodels are all suitable to analyse the data of epidemiologi-cal cohort studies,and the results from Poisson regressionmodel are close to the that from Cox's regression model.But there are some advantages or disadvantages in eachmodel.Some practical considerations or comments in theuse of Poisson or Cox's regression model were given. Addi-tive or general regression model,and how to introduce ex-ternal control rates into these models were discussed.