基于因子分析和Logistic回归分析的女性商业性行为者感染HIV危险因素研究  

Risk factors of HIV infection among female sex workers based on factor analysis and logistic regression analysis

作　　者：梁旭[1] 邓树嵩 谭兰芬[1] 陈坚[1] LIANG Xu;DENG Shusong;TAN Lan fen;CHEN Jian(Guangxi Baise City Center for Disease Control and Prevention,Baise 533000,Guangxi,China;School of Public Health and Management,Youjiang Medical University for Nationalities,Baise 533099,Guangxi)

机构地区：[1]百色市疾病预防控制中心,广西百色533000 [2]右江民族医学院公共卫生与管理学院,广西百色533099

出　　处：《中国艾滋病性病》2019年第12期1218-1221,1228,共5页Chinese Journal of Aids & STD

摘　　要：目的研究女性商业性行为者(FSW)群体内常见的流行病学变量之间关系,探索变量之中是否有潜在的影响因子,且影响因子与群体艾滋病病毒(HIV)感染之间是否存在相关性。方法通过调查一个区域内在社区活动FSW个体社会人口学和经济学变量,应用因子分析法,并使用SPSS 18.0统计学软件中对变量进行KMO检验和Bartlett球形检验,判定是否适于因子分析,如果适合因子分析的采用主成分分析法,并进行最大方差旋转(varimax rotation),以特征根>1及方差累计贡献率大于50%作为入选公因子的标准,使用回归分析计算法计算公因子得分,并保存生成新的分析变量。将生成的新变量与其他控制变量使用二分类Logistic回归进行分析。结果根据KMO的值为0.617,Bartlett的观测值为2 316.141,相应的概率P=0.000,表明适合因子分析;从原始变量中提取2个公因子F1和F2,通过Logistic回归分析发现F1公因子[调整比值比(aOR)=3.869,95%可信区间(CI):2.353~6.362]是感染HIV的危险性因素。结论 FSW个体社会人口学和经济学变量之间存在较强的线性关系,F1公因子作为反映FSW群体内社会结构的变量与群体的HIV感染相关。通过因子分析方法可将流行病学调查中常用分析变量转化为反映社会结构的变量,从而避免得出疾病感染风险的错误结论。Objective To investigate the relationship between common epidemiological variables among female sex workers(FSW), and to explore potential influencing factors among variables and the correlation of influencing factors with HIV infection among FSW. Methods Factor analysis was conducted to investigate the social demography and economic variables of FSW, with variables subjected for KMO test and Bartlett’s spherical test by SPSS18.0, to determine their suitability for factor analysis. If they were suitable, principal component analysis and varimax rotation were carried out. Characteristic root>1 and cumulative variance contribution rate>50% were used as criteria for selection of common factors. The scores of common factors were calculated by regression analysis and calculation method, and new analysis variables were saved and generated, which were analyzed by binary logistic regression. Results KMO value was 0.617, Bartlett’s observation value was 2316.141, and the corresponding probability P=0.000, indicating that factor analysis was suitable. Two common factors F1 and F2 were extracted from the original variables. Logistic regression analysis showed that common factor F1(adjusted OR=3.869, 95% CI: 2.353-6.362) was a risk factor for HIV infection. Conclusion There is a strong linear relationship between social demography and economic variables in FSW. As a variable reflecting social structure in FSW, common factor F1 is related to HIV infection in FSW. Factor analysis can transform the common analysis variables in epidemiological investigation into variables reflecting social structure, so as to avoid drawing wrong conclusion on the risk of HIV infection.

关 键 词：女性商业性行为者 因子分析 回归分析 艾滋病病毒 

分 类 号：R512.91[医药卫生—内科学] R373.9[医药卫生—临床医学]