【作者】 胡浩；

【导师】 姜宝法；

【作者基本信息】 山东大学 ， 流行病与卫生统计学， 2008， 硕士

【摘要】 研究背景由艾滋病病毒(HIV)感染引起的艾滋病(AIDS)是一种新发传染病,自1981年世界第1例被报道以来,它的流行与危害以及所造成的严重后果已日益引起全世界的广泛重视。目前我国艾滋病流行仍以共用注射器静脉吸毒为主要传播途径,而经性接触传播却呈逐年上升趋势。性传播的快速增长很大的原因是80年代以来的性产业的死灰复燃和暗娼的剧增。根据国家监测点的资料显示,2005年我国暗娼HIV感染率由1996年的0.02迅速上升至0.94%。针对暗娼开展艾滋病预防干预工作,已经成为控制艾滋病流行的关键性工作。与其他传染性疾病不同,行为因素在HIV感染的发生中起着决定性作用。要控制艾滋病在暗娼人群中的流行,主要是改变该人群的高危行为。这不仅需要了解这些行为发生的环境及其影响因素,还需要为这些行为的改变创造条件。这就迫切需要发展出基于科学理论、经过实践检验、有高度概括性的理论模型来帮助我们理解和促进艾滋病预防行为。但是迄今为止,我国很少有关于艾滋病行为干预理论的研究,基于某种理论的行为干预就更少。加强行为干预理论研究,探索适合国内特定高危人群的干预理论是目前亟待解决的问题。要研究艾滋病行为干预理论,应该从探讨其影响因素着手,目前用于影响因素分析的常用方法包括多重线性回归、logistic回归、主成份分析、路径分析、因子分析和综合评价等。但这些方法对行为问题研究的资料都有一定的局限性,表现在:(1)无法处理因变量多于一个的情况;(2)无法处理自变量之间的多重共线性;(3)无法对一些不可直接测量的变量进行处理;(4)没有考虑变量的测量误差,以及测量误差之间的关系。结构方程模型(Structural Equation Modeling,SEM)是因子分析(factoranalysis)与路径分析(path analysis)两种重要统计技术的结合体,它能将一般直接测量的观察变量作为潜在变量的指标,并允许这两种变量存在测量误差。它不但研究观察变量,又可以研究不能直接测量的潜在变量,既可研究变量间的直接作用,又可研究变量问的间接作用。随着行为干预实践的需要和有关的行为理论研究不段深入,结构方程模型因其在帮助建立,尤其是验证行为理论上不可比拟的优势,已经成为行为问题理论研究中的热点。目的(一)利用结构方程模型为济南市女性性工作者的艾滋病预防行为建立因果模型,对模型进行拟合、修正和评价,得到最优模型,经验的检验IMB模型在中国FCSW人群的适用性;(二)根据最终模型,确定艾滋病预防行为的影响因素,分析信息、动机和行为技巧和行为因素之间的直接作用和间接作用,为济南市女性性工作艾滋病行为干预工作提供科学依据;(三)通过对结构方程模型的应用,探索应用潜变量分析方法研究行为问题的方法,丰富国内艾滋病行为干预的相关理论,同时为今后行为干预理论的研究提供方法上的参考;(四)在干预过程中探讨基于IMB模型的艾滋病行为干预的可行性和有效性,希望找到一个适合中国国情的关于FCSW人群艾滋病高危行为预防的有效模式。方法采用方便抽样法,从济南市城区娱乐场所内抽取一定数量的女性性工作者为研究对象,采用自编的量表主要对受试的艾滋病预防信息、动机、行为技巧和行为等进行调查,调查结束后进行现场干预。利用Lisrel软件对资料进行结构方程模型分析,评价模型拟合情况,分析IMB模型各组分的关系。结果本研究共调查来自28个娱乐场所158名女性性工作者,其中有效问卷156份。从人口学特征来看,该人群以18-25岁的年轻人为主,平均年龄23岁,文化程度低,其中初中及以下学历的占68%,外省特别是东北三省的外来人员比例较高(52%)。该人群在AIDS方面的知识不全面,尤其值得注意的是她们对AIDS的预防方法存在诸多错误认识。在艾滋病预防动机水平上,对危险性的感知、社会关系的对预防行为的支持程度较低,自身对使用安全套的态度存在偏见。在行为技巧水平上,服务小姐缺乏跟客人协商使用安全套的技巧,安全套的正确使用率较低。最近一个月中,每次都使用安全套的比例为32.05%(50/156),最近一次接客中,使用安全套的比例为66.03%(103/156)。根据IMB理论模型,本研究设定了一个初始模型,并依据内在适配度指标、标准化残差、修正指数、参数估计值的显著性等,不断对模型进行剪裁和修正,在剪裁和修正的过程中通过标准系数、卡方值、GFI、AGF、CFI、RMSEA等指标对模型进行评价,然后采用似然比检验对这一系列模型进行比较,得到最终模型。其拟合指标如下:GFI=0.884,NFI=0.956,CFI=0.977,RMSEA=0.0796,说明最终模型拟合尚可。根据最终模型的结果提示,艾滋病预防信息和动机均可通过行为技巧影响预防行为,信息不直接作用于预防行为,动机可以直接对预防行为产生影响。信息和动机对行为技巧的影响相当,然而,动机对预防行为的作用远远大于信息的作用。标准化的参数估计显示,信息对行为技巧的直接效果为0.503,对预防行为的全体效果为0.327;动机对行为技巧的直接效果为0.510,对预防行为的直接效果为0.369,对预防行为的全体效果为0.700;行为技巧对预防行为的直接效果为0.650。结论(一)济南市FCSW人群预防动机和行为技巧水平较低,安全套使用率较低。(二)结构方程模型分析显示,影响济南市女性性工作者预防行为的主要因素是动机和行为技巧,而信息的作用较弱。结果揭示了这些影响因素对行为的直接和间接效应,最终模型拟合较好。(三)结构方程模型将数理知识与专业有机结合,是处理行为问题资料很好的选择。SEM与传统统计方法比较,具有许多独特的优越性,但也有局限和不足之处,如本研究中建立各个潜在变量都完全是借鉴相关理论而主观定义的,所以下结论不能绝对化。(四)研究支持IMB模型的理论假设,仅仅依靠知识宣教不能有效的促进高危行为转变,需要从信息、动机、行为技巧三方面进行综合干预。更多还原

【Abstract】 AIDS that caused by HIV is a emerging infectious disease. Since the first case reported in 1981, Its spread and damage has become the most important public health and social problem all over the world. Currently, the main HIV transmission route is through injecting drug use, however, there is a emerging trend that the proportion of sexually transmitted HIV infections is increasing. The rapid growth of sexually transmission is mainly due to renascence of sex industry and the boom of FCSW from 1980s. Data from national sentinel surveillance shows that the prevalence rate among FCSWs increased from 0.02% to 0.94% between 1996 and 2005. Intervention works targeted to FCSWs is consided as a key to control the spread of AIDS.Unlike other infectious diseases, behavioral factor act as a crucial role when HIV infected. In order to control the spread among FCSWs, the first thing is to change their risk behaviors. We have to understand the situation and influencing factors of such risk behaviors; besides, we need to creat conditions and settings to promote the behavior modification. In this regard, it’s urgent to develop some theories which are scientific, general and experience confirmed to help us understand and promote the AIDS prevention behavior. Unfortunatelly, few studies focused on such field and intervention based on intervention theory even less. Enforcing intervention theory study and exploring sutiable theory for specific high risk population is on the top of agenda in China.Prevention and intervention should start from analyzing influencing factors. At present, statistical methods widely used in behavior research include multiple linear regression, principal component analysis, factor analysis and logistic regression etc. However, these methods have their limitations as follow when applying to data about behavioral problems: (1) can’t handle two and more dependent variables; (2)multi- collinearity between independent variables is unresolved; (3)not able of dealing with immeasurable variables; (4)measure error and their correlation is thoughtless.Structural equation modeling (SEM) is a combination of two important statistical techniques, i.e. factor analysis and path analysis. It treats observation variables as the indications of latent variables and permits the existence of measure error in both two variables. It’s capable to study the manifest variables as well as the latent variables, the direct effects as well as the indirect effects. With the requirement of intervention practice and the development of relevant behavior theories, SEM has became one of the hot research area due to its unique advantages.Objective:1. By means of SEM to develop a causal relationship model for AIDS prevention behaviors among FCSWs in Jinan city. After model estimation, modification and assessment, to come up with an optimum model to comfirm the applicability of IMB model.2. Based on the final model, to ascertain the influencing factors of AIDS prevention behavior. Analyzing direct and indirect effects of information, motivation and behavioral skills on prevention behavior and provide scientific reference for succedent intervention.3. To explore a reliable pattern to analyze behavior problems using latent variable analysis methods. To enrich relative intervention theories for combating AIDS and provide methodological referrence for behavior theory study.4. To probe into the feasibility and efficiency of IMB based intervention.Method:FCSWs are recruited from several entertainment venues in Jinan urban area using convenience sampling method. Self-compiled scale is used to collect relevant information required by IMB model.Using Lisrel 8.7 to perform SEM analysis. After model estimation, modification and assessment, to come up with an optimum model to study the relationship of each components of IMB model.Result:Total of 158 FCSWs were recruited from 28 entertainment venues, form which 156 valid questionnaires were collected. In terms of demographical character, Most of FCSWs in this research were young people aged between 18 and 25 and the average age is 23. Low educational background and spinsterhood were two primary characters. The people from other proviences accounted for a large contribution (52%) , especially from three Northeastern Provinces of China.Their knowledge about AIDS was incomprehensive, especially the knowledge about how to effectively prevent AIDS. In regard to prevention motivation, their perception of risk and social support were in low level; prejudice against condom use was prevalent. For behavioral’ skills, FCSWs were in short of the negotiation skills to persuade guests to use condoms. Their physical skills to use condom was also worrying with few of them could use it correctly. In the recent month, The rate of condom use was 32.05% (50/156) . The rate of condom use in the recent time was 66.03% (103/156) .Based on IMB model, we set a initial model at the beginning. Then according to Goodness of Fit Statistics, standard residual, modification index and estimation value of parameter, a series of pruning and modifying was conducted. In this process, standard coefficient, chi-square value, GFI、AGF、CFI、RMSEA etc. were used to assess the current model. When dealing with nested models, Likelihood ratio test was performed to determine the optimum model. Finally, we got the final model and its GFI、AGF、CFI、RMSEA were 0.884, 0.956, 0.977, 0.0796 respectively, which indicated the well fit.The final model showed that both information and motivation can affect prevention behavior via behavioral skills. The information’s direct effect on prevention behavior was insignificant while the motivation got a strong influence on prevention behavior. Information’s effect on behavioral skills were almost equal to the motivation’s but the later’s effect on prevention behavior was much greater. The standard coefficient illustrated as follow: Information’s direct effects on behavioral skills and prevention behavior were 0.503 and 0.327 respectively; Motivation’s direct effects on behavioral skills and prevention behavior were 0.510 and 0.369 respectively; Motivation’s total effect on prevention behavior was 0.7 and behavioral skills’ direct effect on prevention behavior was 0.650.Conclusion:1. Low motivation and behavioral skills level are showed in FCSWs in Jinan which explains the low rate of condom use.2. The main influencing factors of prevention behaivor among FCSWs are motivation and behavioral skills, while the information’s effect is weak.3. Structural equation modeling which is considered as the organic combination of mathematical knowledge and speciality knowledge possess many advantages compared with conventional statistical methods when handling data from behavior research. But it has its own limitations and shortages. For example, the theoretical model and definition of the latent variables in this study are based on subjective assumption in terms of relevant theory. When drawing any conclusions, being prudential is preferred.4. IMB theoretical model was proved according to this study which illustrates us that only knowledge intervention can’t promote behavior change. An integrated manner that covers information, motivation and behavioral skills should be adopted when conducting behavior intervention.更多还原