过去我们曾用逐步回归、判别分析等多因素统计方法对胃癌病例对照资料进行了分析,但是对回顾性调查中的一些定性指标,如新鲜蔬菜及水果是否经常食用等,应用上述的统计方法不太灵敏;在分析配对调查资料时,常常损失较多有用的信息,特别是对各地区及不同样本,采用配对调查方法所计算的各因素的相对危险性指标,在需要相加 In the past, we used stepwise regression, discriminant analysis and other multi-factor statistical methods to analyze the case-control data of gastric cancer. However, for some qualitative indicators in retrospective surveys, such as whether fresh vegetables and fruits are regularly consumed, the above statistical methods are not applied. Too sensitive; in the analysis of paired survey data, often more useful information is lost, especially for each region and different samples, the relative risk indicators of each factor calculated by the paired survey method are needed to add