Abstract: Background　In recent years，men who have sex with men （MSM） become a group with a high incidence of sexually transmitted diseases（STDs），especially HPV 16/18 infection incidence remains high.HPV infection will develop into STDs if it is not effectively controlled.So it is important to research the main influencing factors for high incidence of HPV 16/18 infection in MSM.Objective　To explore the risk of HPV 16/18 infection in MSM in Urumqi with latent class analysis （LCA），providing a scientific basis for the delivery of targeted interventions for this population.Methods　From March 2016 to August 2017，a survey was conducted in a snowball sample of 884 MSM in Urumqi.Participants' general demographic characteristics，sexual behavior characteristics and HPV infection-related factors were collected.HPV 16/18 genotypes were detected.Class of risk of HPV16/18 infection was identified by LCA.Results　Nine variables including sex，sexual intercourse pattern，commercial sexual behavior，use of drugs，history of sexually transmitted infections，condom use，anal intercourse，sexual orientation and circumcision were included.The AIC and BIC models with 3 potential categories are the lowest and the best models.According to the characteristics of sexual behavior and the related factors of HPV infection，the patients were divided into low risk group （520 cases），medium risk group （290 cases） and high risk group （74 cases）． The infection rates for these groups were 9.2%，14.5% and 24.3%，respectively，and high risk group showed much higher infection rate than low risk group（P<0.05）． Conclusion　LCA can be used to stratify the risk of HPV-16/18 infections in MSM.

Key words: Sexually transmitted diseases, Men who have sex with men, Human papillomavirus 16, Human papillomavirus 18, Latent class analysis

摘要： 背景　近年男男性行为人群（MSM）是性病发生率的高发人群，尤其是人乳头瘤病毒（HPV）16/18型感染率居高不下，而HPV感染若不有效控制会发展成性病，因此研究与MSM HPV16/18型感染高发相关的主要影响因素很重要。目的　采用潜在类别分析（LCA）探讨MSM的HPV16/18型感染风险，为实施针对性的干预措施提供科学依据。方法　于2016年3月—2017年8月在乌鲁木齐市采用滚雪球抽样法招募884例MSM，对其一般人口学特征、性行为特征、HPV感染相关因素进行调查，行基因分型检测HPV16型和HPV18型，并采用LCA对HPV16/18型感染风险进行分类预测。结果　共纳入了性行为对象性别、性交方式、有无同性商业性性行为、有无毒品使用、有无性传播感染史、肛交使用安全套、有无同性肛交、性取向、有无包皮环切术9个研究变量。潜在类别数为3的模型Akaike信息准则（AIC）和Baysian信息准则（BIC）最低，为最佳模型。根据性行为特征、HPV感染相关因素分为HPV感染低风险组（520例）、HPV感染中风险组（290例）、HPV感染高风险组（74例），感染率分别为9.2%、14.5%、24.3%，HPV感染高风险组的感染率高于HPV感染低风险组（P<0.05）。结论　LCA能用于区分MSM的HPV16/18型感染风险。

关键词: 性传播疾病, 男男性行为, 人乳头瘤病毒16, 人乳头瘤病毒18, 潜在类别分析