轨迹分析模型在男男性行为人群人乳头瘤病毒感染状态变化趋势研究中的应用

目的: 探索应用轨迹分析模型拟合HIV阴性男男性行为（MSM）人群肛周人乳头瘤病毒（HPV）感染状态变化趋势的可行性。方法: 2016年9月1日至2017年9月30日于乌鲁木齐市采用滚雪球法招募HIV阴性MSM者，以调查对象入组时间为基准，每6个月随访一次，采集肛管内脱落细胞并进行HPV DNA分型鉴定。纳入完成基线、6个月、12个月随访的研究对象，以感染不同型别HPV的累加数量为因变量，随访次数为自变量构建轨迹分析模型，分别探索将受试者分为一个、二个、三个及四个亚组时的HPV感染状态变化轨迹，并运用贝叶斯信息标准值（BIC）、贝叶斯因子对数值和平均验后分组概率（AvePP）评价模型拟合效果。结果: 共招募400名HIV阴性MSM者，其中187名MSM者纳入模型分析。结果发现，将HPV感染状态变化趋势按两组轨迹模型拟合效果最优。该模型中，第一亚组占54.5%（102/187），HPV感染状态变化曲线呈下降趋势；第二亚组占45.5%（85/187），HPV感染状态变化曲线呈上升趋势。结论: 应用轨迹分析模型能有效区分HIV阴性MSM人群HPV感染状态的变化趋势，有助于探寻HPV感染的高危人群。

Trajectory modeling for estimating the trend of human papillomavirus infection status among men who have sex with men

Objective: To investigate whether trajectory model can be used to explore the trend of anal human papillomavirus (HPV) infection status among HIV-negative men who have sex with men (MSM). Methods: HIV-negative MSM were recruited by using the "snowball" method from 1st September 2016 to 30th September 2017 in Urumqi. The subjects were followed-up every six months since enrollment. The cell samples in anal canal were collected and the 37-type HPV test kits were used for identification and classification of HPV infection at both baseline and follow-up visits. Taking the cumulative number of different types of HPV as the dependent variable and follow-up visits as the independent variable, the trajectory model was established for the study subjects who completed baseline, 6 months and 12 months follow-up. The model was used to simulate the trend of HPV infection status when the subjects were divided into 1, 2, 3 and 4 subgroups. Bayesian information criterion (BIC), log Bayes factor and average posterior probability (AvePP) were used to evaluate the fitting effect. Results: A total of 400 HIV-negative MSM were recruited at baseline and 187 subjects completed baseline and two follow-ups. The fitting effect attained best when the variation trend was divided into two subgroups. The first subgroup accounted for 54.5%(102/187) of the total, and the curve of change in HPV infection was decreasing; the second subgroup accounted for 45.5%(85/187) of the total, and the curve of change in HPV infection was increasing. Conclusion: Trajectory model can effectively distinguish the trend of HPV infection status in HIV-negative MSM to identify the high-risk group of HPV infection.