原发性肝癌影响因素的交互作用分析

摘要

目的 分析原发性肝癌影响因素间的交互作用.方法 采用配对病例对照研究,收集福州地区388例肝癌新发病例和388例对照,使用多因子降维法(MDR)分析因素间的交互作用.根据MDR模型结果,用多因素Logistic回归模型估计交互项及其他危险因素的危险度.结果 MDR最佳模型为乙型肝炎病毒(HBV)感染、吸烟和饮酒的交互作用,该模型的检验样本平衡检验准确度为82.73%,交叉验证一致性为13/15.多因素Logistic回归分析显示,调整了肿瘤史、吃霉变食物和喝茶,HBV感染主效应[OR=5.722,95%CI(1.809,18.418)]和交互作用[OR=5.646,95%CI(1.655,19.259)]有统计学意义,联合作用也有统计学意义(P＜0.05).结论 在原发性肝癌发生过程中,HBV感染是主要因素,吸烟和饮酒是辅助因素,它们协同作用.%Objective To explore the interaction between factors of primary liver cancer. Methods A matched case - control study was carried out in Fujian including 388 case with liver cancer and 388 controls. The interaction were analyzed with the MDR model. Base on the result of MDR model, an Logistic regression model was simulated to estimate the Ors of interaction factors and other risk factors. Results The interaction between HBV infection, smoking and drinking was discovered by MDR model with a statistically significant difference ( P <0. 0001 ). The model's testing balance accuracy was 82. 73% , and crossing validation consistency was 13/15. Multivariable Logistic regression showed that after adjusting cancer's history, moldy food and tea, the main effect of HBV infection [ OR =5. 722, 95% CI ( 1. 809, 18. 418 ) ] and the interaction [ OR =5. 646, 95% CI ( 1. 655, 19. 259 ) ] were significantly different. Conclusion The HBV infection was the major factor of primary liver cancer, and smoking and drinking were cofactor.