摘要
随机抽取启东肝癌高发区20个乡计84例存活肝癌病例为研究对象,每个病例匹配4名“健康”对照,分别组成1:1、1:2、1:3和1:4配比资料,对其进行条件Logistic回归分析。结果显示:各匹配比结果基本一致,即肝癌病因因素以HBsAg、抗-HBc和Pre-S_2蛋白较为重要。模型拟合均较好,说明资料的收集过程是随机的,有一定代表性。在各匹配比中,又以1:4配比模型中各因素的标准误最小,稳定性最强,因而所得结论是真实可靠的。
84 PHC patients were sampled randomly in 20 townships in Qidong county (a higher incidence area of PHC). Each patient was paired with 4 healthy persons as control and composed of 1:1, 1:2, 1:3, 1:4 matched materials respectively. Using conditional logistic regression model, we searched the practical value of 1:M matched case-control study method and analysed preliminarily the validity and reliability of these results. It showed that the results were consistent in all matched materials. The main agents of PHC were HBsAg, Anti-HBC and Pre-S2 protein, while tea seemed to be a protective factor.The models fitting were all good enough to demonstrate that the collection of materials is at random and representative. Compared with other models, the standardized error (SE) was the smallest among all factors in 1:I 4 matched model. So the stability of results obtained from 1: 4 model was the strongest, the conclusions were valid and credible.
关键词
肝肿瘤
回归分析
病因
配比病例
primary hepatocellular carcinoma
conditional logistic
regression
comparative study
