胃癌病因学研究中交互作用研究方法的比较

本文对原发性胃癌病例对照研究中筛选出的危险因素：不良饮食嗜好、常食咸菜、吸烟量、食用陈旧食品等运用三种不同的交互作用研究方法，即叉生分析，多因素ｌｏｇｉｓｔｉｃ回归模型中加入交互作用项和使用ＧＬＩＭ软件进行广义相对危险度分析等作了分析和比较；并指出了各自的优缺点。正确识别因素间联合作用的模式对于疾病的一级预防和公共卫生决策的制订具有重要意义。

Compression of the Interaction Effect Study Methods in the Etiological Study of Primary Gastric Cancer

The author used 3 different interaction effect mthods:interaction analysis ,adding interaction items to the multi variate logistic regression model, employing general relative risk models to study the risk factors of primary gastric cancer selected by conditional logistic regression model,pointed out their advantage and shortcomings. Correct recognition of the combined effects of the risk factors is of important significance for the primary prevention and making up the public health policies.