零膨胀的负二项模型在蛔虫与钩虫再感染模式研究中的应用

Application of zero-inflated negative binomial model to study the pattern of Ascaris lumbricoides and hookworm re-infection

目的应用零膨胀的负二项模型探索蛔虫与钩虫再感染的模式。方法在四川省宜宾市翠屏区邱场乡、江西省进贤县塘南乡和湖南省岳阳县三荷乡开展现场调查，调查内容包括：基线粪检与驱虫、一个月后复查、一年后复查及问卷调查。应用零膨胀的负二项模型拟合一年后个体虫卵计数数据，考察发生再感染以及再感染强度的影响因素。结果共计调查对象374人。治疗前蛔虫和钩虫感染率分别为38．08％（139／365）和39．17％（141／360），治疗后一年时蛔虫和钩虫再感染率分别为15．89％（58／365）和8．89％（32／360）。蛔虫模型结果显示：治疗前感染度高，则发生再感染的风险高（OR=I．94，95％CI：1.33～2．81），家用电器档次高、不喝生水都是保护因素，15-4岁组与45-64岁组发生再感染风险显著低于5～14岁组；治疗前感染度越高，治疗后再感染获得高感染度的风险越大（OR=1．56，95％CI：1．10-2．23），与5～14岁组相比，15～44岁组与65～70岁组再感染获得高感染度的风险小；不喝生水获得高感染度的风险小。钩虫模型结果显示：治疗前钩虫感染度越高，发生再感染的风险越高（OR=3．97，95％CI：2．07—7．60）。居住砖瓦房相对土坯房是再感染的保护因素；与5～14岁儿童相比，其他3个年龄组钩虫再感染获得高感染度的风险均较其高，其中45～64岁风险最高；接受中学教育较文盲或半文盲获得高感染度的风险低。结论零膨胀的负二项模型是拟合蛔虫和钩虫虫卵计数数据的最优模型，研究证实再感染过程中倾向性现象的存在，蛔虫和钩虫再感染的发生及强度与治疗前感染度、人口学特征、社会经济条件、环境及行为等因素相关。

Objective Use zero-inflated negative binomial model for modeling the pattern of Ascaris lumbricoides and hookworm re-infection. Methods Field survey was carried out in three villages： Qiuchang in Yibin City of Sichuan Province, Tangnan in Jinxian County of Jiangxi Province, and Sanhe in Yueyang County of Hunan Province. The survey consisted of baseline stool examination and treatment, re-examination after one month and one year, and questionnaire survey. Zero-inflated negative binomial model was applied to fit the egg count data collected after one year, meanwhile, the risk factors for the re-infection occurrence and intensity were evaluated. Results Totally 374 subjects were recruited in the survey after one year. Theprevalences of A scaris lumbricoides and hookworm at baseline 38.08% （139/365） and 39.17% （141/360）. The re-infection rates of Ascaris lumbricoides and hookworm one year after treatment were 15.89% （58/365） and 8.89%（32/360）, respectively. A. lumbricoides model showed： the higher infection intensity before treatment, the higher risk of re-infection oceurrence（OR=l.94,95%CI： 1.33-2.81 ）, high level of household electrical ap- pliances and not drinking un-boiled water were protective factors. The risk of re-infection in age group 15-44 years and 45-64 years was lower than that in group 5-14 years significantly. The higher infection intensity before treatment, the higher risk of high re-infection intensity after treatment （OR=1.56,95%CI： 1.10-2.23）. Compared to age group 5-14 years, 15-44 years and 65-70 years had low risk of getting high re-infection in- tensity. Not drinking unboiled water was also a protective factor for acquiring high intensity. Hookworm model showed： the higher infection intensity before treatment , the higher risk of re-infection occurrence （OR = 3.97,95%CI： 2.07-7.60）, living in brick house opposed to adobe house was protective factor for re-infection. Compared to age group 5-14 years, the other 3 age groups had higher risk of getting high re-infection intensity, with age group 45-64 the highest. People receiving high school education had lower risk of acquiring high re-infection intensity, relative to illiterate or semi-illiterate people. Conclusion Zero-inflated negative bino- mial model is the optimal statistical model for fitting Ascaris lumbricoides and hookworm egg count data. This study proved the existence of predisposition phenomenon in the process of re-infection. The occurrence and intensity of re-infection were related with infection intensity before treatment, demographic, socio-economie, environmental and behavioral factors.