【作者】 施学忠；

【导师】 段广才；

【作者基本信息】 郑州大学 ， 流行病与卫生统计学， 2008， 博士

【摘要】 近年来,由于西医尚无治愈艾滋病的疗法等原因,使得艾滋病的中医学研究越来越受到重视。虽然国内中医界也在积极探索对艾滋病病因、病机和证候规律的认识,但目前尚未形成理论体系。艾滋病中医证候分布规律不明、证候标准尚未建立的现状,越来越成为中医药进一步深入研究艾滋病的一个瓶颈。传统辨证是医师先通过望、闻、问、切手段归纳出某一个证候,然后对患者施治。在这个过程中通常需要分析多个因素,定量地描述某些因素之间的相互依存关系。然而,由于医师所归纳出的证候是一个不可直接测量且带有综合特性的变量,不同的医师对同一个病人可能会归纳出不同的证候,因此,如何明确艾滋病的证候规律,建立一种客观而又定量的方法规范证候标准,成为艾滋病中医证候研究的关键。尽管众多学者从不同角度对证候学进行了大量的研究和探索,但近50年来,始终未取得突破性进展,严重影响了中医的发展进程。出现此种状况的原因是多方面的,由于中医证候的特点及相关数据的复杂性而导致传统生物统计学方法应用受限是重要的原因之一。J(o|¨)reskog、Kessling等人将通径分析的思想引入到潜变量研究中,并同因子分析方法结合起来,形成了结构方程模型(Structural Equation Model,SEM)。SEM是一种复杂的因果关系模型,可以处理显在变量与潜在变量之间以及潜在变量之间的关系,同时还考虑了误差变量,近二十年来在许多领域中得到了越来越广泛的应用。由于西医诊断疾病的实质是医师对疾病基本规律的揭示,因此,以西医的病与中医的传统辨证相结合,然后应用SEM的分析方法发现证候与四诊信息之间的联系,有可能客观而又定量地辨明病的证候规律,为中医证候标准的建立提供客观依据。进行中医证候研究时的主要难题在于难以直接、精确地对证候加以测量与分析,其相关数据及属性也十分复杂,所以,单一选择某个统计方法对其进行分析是不现实的,作者认为通过统计建模手段,联合多种方法对其进行研究可能是个有效的途径。因此,该研究拟对获得的中医证候相关数据在单变量分析的基础上,通过多重对应分析、多分类logistic回归分析和探索性因子分析,探索艾滋病中医证候及与四诊信息之间的联系,结合中医临床专家意见构建证候与四诊信息的初始理论模型,应用结构方程模型对建立的模型进行拟合、验证与修正。研究结果可阐明HIV/AIDS中医证候规律和主要证候类型,寻找证候相关的四诊信息,研究证候之间的结构联系,探讨适合中医证候的统计分析方法,为制订艾滋病中医证候诊断标准、实施艾滋病辨证治疗提供依据。资料来源和方法采用多阶段分层整群抽样技术,在河南省艾滋病高发地区随机抽取1277例HIV/AIDS进行调查,调查内容包括一般人口学特征、中医四诊信息、中医证候诊断和实验室检查指标。对中医证候量表进行效度和信度评价和对数据预处理后进行统计建模,主要统计学方法包括:集中趋势和离散趋势描述、单变量分析、对应分析、多分类logistic回归分析、探索性因子分析、证实性因子分析和结构方程模型。应用SAS9.1,SPSS15.0和LISREL8.54统计软件进行统计学处理、分析与建模。结果1.问卷质量评价主要/次要中医症状体征条目的Cronbach’s Alpha为0.628/0.792;调查表具有较好的内容效度和表面效度。体检及实验室指标缺失值在6.3-20.7%之间,采用期望最大化法填充后的数据和原有数据的分布非常接近。2.单因素分析HIV和AIDS基本实证和虚证构成不同(χ~2=54.374,P<0.001;χ~2=21.790,P<0.001)。湿热蕴毒、邪结皮肤和肺脾气虚是AIDS患者的主要证候,湿热内蕴、肝郁气滞和脾气虚弱是HIV感染者的主要证候。3.对应分析舌象和脉象与中医基本实证/虚证的多重对应分析效果较理想,在两个维度上有较好的区分度。HIV携带者舌象与基本实证/基本虚证在两个维度上的特征根分别为0.149,0.097/2.603,1.801,惯量为2.680,1.752/0.145,0.100;AIDS患者舌象与基本实证/基本虚证在两个维度上的特征根分别为2.610,1.948/2.484,1.942,惯量为0.145,0.108/0.138,0.108。HIV携带者脉象与基本实证/基本虚证在两个维度上的特征根分别为2.024,1.324/2.017,1.288,惯量为0.202,0.132/0.202,0.129;AIDS患者舌象与基本实证/基本虚证在两个维度上的特征根分别为2.041,1.397/2.035,1.362,惯量为0.204,0.140/0.204,0.136。4.多分类logistic回归分析HIV感染者中,腹泻(β=1.43,P=0.03)与湿热蕴毒有联系;皮肤瘙痒(β=4.13,P<0.01)、舌下脉络细短(β=2.04,P=0.02)、舌抓痕(β=3.68,P<0.01)与邪结皮肤有关;脉弦(β=1.75,P<0.01)与肝郁气滞有关;脱发(β=0.81,P=0.04)、舌齿痕(β=1.36,P=0.02)、脉沉(β=0.73,P=0.03)与脾气虚弱有关;神疲乏力(β=2.60,P<0.01)与肺脾气虚有关;胁肋疼痛(β=2.42,P<0.01)、舌瘦薄(β=2.96,P<0.01)与气阴两虚有关;爪甲淡白(β=1.67,P<0.03)与气血亏虚有关。AIDS患者中,舌苔腻(β=1.34,P<0.01)与湿热内蕴有联系;舌疱疹(β=3.05,P<0.01)、舌抓痕(β=3.01,P<0.01)与邪结皮肤有联系;消瘦(β=2.20,P=0.02)、恶寒(β=2.28,P<0.01)、脉细(β=1.17,P<0.01)与脾气虚弱有联系;发热(β=2.13,P=0.07)与气阴两虚有联系。5.因子分析HIV携带者KMO统计量是0.84,Bartlett检验:χ~2=2193.87,P<0.01,根据特征根大于0.8的原则,提取出12个公因子,累计贡献率为68.27%。AIDS者KMO统计量是0.86,Bartlett球形检验:χ~2=2592.68,P<0.01,当特征根为0.8时,可以提取出12个公因子,累计贡献率为69.99%。6.结构方程建模HIV携带者主要以肝郁气滞,AIDS主要以邪结皮肤实证为主。中医证候实证结构方程模型拟合指数:Chi-Square=985.26(P<0.01);近似误差均方根(Root Mean Square Error of Approximation,RMSEA)=0.070;非范拟合指数(Non-Normed Fit Index,NNFI)=0.84;比较拟合指数(Comparative Fit Index,CFI)=0.87;Incremental Fit Index(IFI)=0.87;拟合优度指数(Goodness of Fit Index,GFI)=0.92;调整拟合优度指数(Adiusted Goodness of Fit Index,AGFI)=0.89。HIV主要以脾气虚弱,AIDS主要以肺脾气虚虚证为主。中医证候虚证结构方程模型拟合指数:Chi-Square=768.90(P<0.01);近似误差均方根(Root Mean Square Error ofApproximation,RMSEA)=0.088;非范拟合指数(Non-Normed Fit Index,NNFI)=0.82;比较拟合指数(Comparative Fit Index,CFI)=0.86;Incremental Fit Index(IFI)=0.86:拟合优度指数(Goodness of Fit Index,GFI)=0.92;调整拟合优度指数(Adjusted Goodness of Fit Index,AGFI)=0.88。结论1.HIV/AIDS具有4个主要实证和4个主要虚证:湿热内蕴,湿热蕴毒,邪结皮肤,肝郁气滞,脾气虚弱,肺脾气虚,气阴两虚,气血亏虚。湿热蕴毒、邪结皮肤和肺脾气虚是AIDS患者的主要证候;湿热内蕴、肝郁气滞和脾气虚弱是HIV感染者的主要证候。证候之间及与四诊信息存在的数量化关系可以作为制订艾滋病中医证候诊断标准的依据。2.多重对应分析模型对于探索舌象、脉象与证候间联系较为合适;无序多分类logistic回归模型在探索对证候影响较大的四诊信息方面效果较好;探索性因子分析于归纳分析支配主要症状体征的潜在因子上具有优势。它们均适合于艾滋病中医证候的研究,在一定程度上解释了四诊信息与中医证候之间的关系,在艾滋病等疾病中医证候研究中有较高的价值。3.证实性因子分析和结构方程模型可以用来验证艾滋病中医证候与四诊信息的内在结构理论和因果关系,研究不同证型(即隐变量)间的联系和演变规律。4.任何一种单一的统计分析技术均不能很好地研究艾滋病的中医证候分类和证候演变规律。多种方法联合使用,在探索性分析的基础上,结合中医专家的经验,进行证实性分析可作为艾滋病等疾病中医证候的研究模式。更多还原

【Abstract】 Study of HIV/AIDS in Chinese medicine field is being concerned more and more for no effective therapies to cure HIV/AIDS in recent years.There have been many studies to explore the cause,pathogenesis and syndromes of HIV/AIDS,but so far an overall and systematic theoretical system has not been formed yet.Unclear distribution regularity of syndrome of traditional Chinese medicine and no standardization of criterion of syndromes of HIV/AIDS has become a bottleneck in the research of HIV/AIDS.In traditional syndrome differentiation,one syndrome was induced by clinical doctor from four diagnostic methods(inspection,listening and smelling,inquiry,palpation and pulse taking), then the relevant treatment was carried out on patients.Multivariate analysis is often used during this process to describe the quantificational relationship among those factors. Syndromes induced by clinical doctor are often non-measured directly and synthetic. Different clinical doctor might induce different syndromes on the same patient.How to identify distribution regularity and establish an objective and quantitative diagnostic criterion are the key points in study of syndrome of traditional Chinese medicine in HIV/AIDS.No breakthrough progress had not been achieved in many studies on syndromes of HIV/AIDS during the past 50 years,which had impacted the development of Chinese Medicine.Characteristics of syndromes of Chinese medicine and complexity of dataset which make it difficult to use traditional bio-statistical methods accounted for this situation. J(o|¨)reskog and Kessling applied the idea of path analysis into the latent variable research, combined with factor analysis and developed a new method-structural equation modeling (SEM).SEM is a complicated causal relationship model which can measure variables and possibly latent variables that have a causal structure.It has been widely used in many fields in recent 20 years.Diagnostics of disease in western medicine is to reveal basic regularity of disease.Combination of Traditional Chinese Medicine and Western Medicine and application of SEM to find out relationship between syndromes and four diagnostic methods can help to quantify the regularity of disease and provide objective basis for establishment of diagnostic criterion of syndromes of traditional Chinese medicine in HIV/AIDS.The most difficult problem in studying syndromes of traditional Chinese medicine are the complexity of data and hard to measure directly.A single method could not sort out such kind of problem,while combination of multiple methods by statistical modeling might sort out.This study aims to explore the relationship between regularity of syndromes of traditional Chinese medicine and four diagnostic methods(inspection,listening and smelling,inquiry,palpation and pulse taking),to investigate the main types of syndromes of traditional Chinese medicine in HIV/AIDS and construct its standard diagnostic criterion by using multiple correspondence analysis,multinomial logistic regression,exploratory factor analysis and structural equation model.Materials and methods1277 HIV/AIDS patients were interviewed with a self-made questionnaire including demographic characteristics,four diagnostic methods of Chinese medicine,syndrome of patients and index of laboratory examination using a multistage stratified random sampling technique.The reliability and validity of questionnaire was assessed.Statistical methods in this study included description of central tedndency and dispersion,univariate analysis, correspondence analysis,multinomial logistic regression,exploratory factor analysis and structural equation model.Statistical analysis of the data was performed by SPSS software (SAS 9.1),SPSS 15.0 and LISREL 8.54.Results1.Assessment of quality of questionnaire Cronbach’s alpha was 0.63 and 0.79 respectively for the main and subordination syndrome items which indicated this questionnaire had a good construct validity and content validity.Missing values in index of laboratory examination ranged from 6.3%to 20.7%,which were imputed by expectation maximization algorithm.2.Univariate analysis The proportions of excess syndrome and syndrome of deficiency in HIV/AIDS were different(χ~2=54.37,P<0.01;χ~2=21.79,P<0.01). Syndrome of dampness-heat and accumulated toxin,syndrome of pathogens accumulation involving skin,syndrome of deficiency of lung and spleen qi were main syndromes of AIDS patients,while syndrome of dampness-heat accumulated interior,syndrome of liver and qi stagnation and syndrome of deficiency of spleen qi were main syndromes of HIV carrier.3.Correspondence analysis Correspondence analysis of tongue manifestation and pulse manifestation and excess syndrome and syndrome of deficiency had a good clearly differentiation index in two dimensions.Eigenvalue in two dimensions between tongue manifestation and basic excess syndrome or syndrome of deficiency was 0.149,0.097 or 2.603,1.801 respectively,and inertia in two dimensions was 2.680,1.752 or 0.145,0.100 respectively in HIV carriers.Eigenvalue in two dimensions between tongue manifestation and basic excess syndrome or syndrome of deficiency was 2.610,1.948 or 2.484,1.942 respectively,and inertia in two dimensions was 0.145,0.108 or 0.13 8,0.108 respectively in AIDS patients.Eigenvalue in two dimensions between pulse manifestation and basic excess syndrome or syndrome of deficiency was 2.024,1.327 or 2.017,1.288 respectively,and inertia in two dimensions was 0.202,0.132 or 0.202,0.129 respectively in HIV carriers.Eigenvalue in two dimensions between pulse manifestation and basic excess syndrome or syndrome of deficiency was 2.041,1.397 or 2.035,1.362 respectively,and inertia in two dimensions was 0.204,0.140 or 0.204,0.136 respectively in AIDS patients.4.Multinomial logistic regression Results of multinomial logistic regression showed that in HIV carriers,diarrhea(β=1.43,P=0.03)was related with syndrome of dampness-heat and accumulated toxin.Pruritus of skin(β=4.13,P<0.01),thin and short of sublingual vessel(β=2.04,P=0.02) and scratch-printed tongue(β=3.68,P<0.01) was related with syndrome of pathogens accumulation involving skin.Stringy pulse(β=1.75, P<0.01) was related with syndrome of liver and qi stagnation.Alopecia (β=0.81,P=0.04),teeth-printed tongue(β=1.36,P=0.02) and deep pulse (β=0.73,P=0.03) were related with syndrome of deficiency of spleen qi.Spiritlessness and weakness(β=2.60,P<0.01) were related with syndrome of deficiency of lung and spleen qi.Hypochondriac pain(β=2.42,P<0.01) and thin tongue(β=2.96,P<0.01) were related with deficiency of both qi and yin.Pale nail(β=1.67,P<0.03) was related with syndrome of deficiency of both qi and blood.In AIDS patients,greasy fur(β=1.34, P<0.01) was related with syndrome of dampness-heat accumulated interior.Tongue bleb (β=3.05,P<0.0l) and scratch-printed tongue(β=3.01,P<0.0l) were related with syndrome of pathogens accumulation involving skin.Emaciation(β=2.20,P=0.02), aversion to cold(β=2.28,P<0.01) and thready pulse(β=1.17,P<0.01) were related with syndrome of deficiency of spleen qi.Fever(β=2.13,P=0.07) was related with deficiency of both qi and yin.5.Factor analysis KMO was 0.84,which indicated a good a measure of sampling adequacy in HIV carriers.Bartlett’s test of sphericity showed that correlation was appropriate for factor analysis(χ~2=2193.87,P<0.01).Based on extraction retains factors with eigenvalues >0.8,12 factors were extracted with cumulative loading of 68.27%。KMO was 0.86 in AIDS patients,Bartlett’s test of sphericity was significant(χ~2=2592.68, P<0.001).12 factors were extracted with cumulative loading of 69.99%(eigenvalues >0.8) in AIDS patients.6.Structural equation model Among excess syndrome,syndrome of liver and qi stagnation was main syndrome of HIV carriers,and syndrome of pathogens accumulation involving skin was main syndrome of AIDS patients.The following goodness indexes of fit Statistics of SEM related with Chinese Medicine excess syndrome showed a good fit: Minimum Fit Function(χ~2=985.26,P<0.01).Root mean square error of approximation (RMSEA)=0.070,Non-Normed Fit Index(NNFI)=0.84,Comparative Fit Index(CFI)= 0.87,Incremental Fit Index(IFI)=0.87,Goodness of Fit Index(GFI)=0.92 and Adjusted Goodness of Fit Index(AGFI)=0.89.Among syndromes of deficiency,syndrome of deficiency of spleen qi was main syndrome of HIV carriers,and syndrome of deficiency of lung and spleen qi was main syndrome of AIDS patients.The following were Goodness indexes of Fit Statistics of SEM related with Chinese Medicine syndrome of deficiency, and indicating a good fit of SEM.Minimum Fit Function(χ~2=768.90,P<0.01),Root Mean Square Error of Approximation(RMSEA)=0.088,Non-Normed Fit Index(NNFI)=0.82, Comparative Fit Index(CFI)=0.86,Incremental Fit Index(IFI)=0.86,Goodness of Fit Index(GFI)=0.92 and Adjusted Goodness of Fit Index(AGFI)=0.88.Conclusion1.HIV/AIDS has four main excess syndrome and four main syndrome of deficiency: syndrome of dampness-heat accumulated interior,syndrome of dampness-heat and accumulated toxin,syndrome of pathogens accumulation involving skin,syndrome of liver and qi stagnation,syndrome of deficiency of spleen qi,syndrome of deficiency of lung and spleen qi,syndrome of deficiency of both qi and yin,syndrome of deficiency of both qi and blood.Syndrome of dampness-heat and accumulated toxin,syndrome of pathogens accumulation involving skin,syndrome of deficiency of lung and spleen qi are main syndromes of AIDS patients.Syndrome of dampness-heat accumulated interior,syndrome of liver and qi stagnation,syndrome of deficiency of spleen qi are main syndromes of HIV carriers.Quantitative relationship between syndromes and four diagnostic information can provide basis for establishing diagnostic criterion of syndromes of traditional Chinese medicine in HIV/AIDS.2.Corrrespondence analysis is appropriate for exploring relationship between tongue manifestation,pulse manifestation and syndromes.Multinomial logistic regression is more powerful in exploring the influence of four diagnostic information on syndromes. Exploratory factor analysis is better in analyzing latent variables which influence the main syndromes of AIDS.All those methods are appropriate and valuable.3.Confirmatory factor analysis and SEM can be used to validate the structure and causal relationship between syndromes of AIDS and four diagnostic information,to study relationships between different type of syndrome(latent variables).4.Any single statistical method can not study the syndromes of AIDS.Combination of different kinds of statistical method and suggestion of clinical doctors can be used as a powerful tool.更多还原