摘要
目的：探索结构方程模型（SEM）方法与中医证候研究的理论联系；探索SEM框架下进行中医证候研究的可行性路径；应用SEM实务操作平台IBM SPSS AMOS20.0，验证、揭示冠心病的证候特征、证候要素、病机演变规律；在理论、实践上论证结构方程模型是合理、量化、有效中医证候研究方法。
材料与方法：依托冠心病证候要素、证候特征、证候病机演变规律临床专家调查问卷三维结构化关联数据库，按照早期、发作期、缓解期、恢复期的病程进行分析，通过perl语言区分题项，通过IF语句区分选项，对数据库进行结构化处理，使数据转换达到AMOS要求。
通过结构方程模型，对冠心病临床专家调查问卷三维结构化关联数据库进行数据挖掘。在前期探索性因子分析（EFA）的基础上，应用测量模型（measured model）的检验与修正，建立冠心病证候特征模型；应用二阶验证性因素分析（second-order CFAmodel），通过分析冠心病证候特征与证候要素的关系，提取有效证候要素；应用潜在变量路径分析（PL-AV），探索冠心病各分期间病机发展的因果联系，进而明确冠心病病机演变规律。
结果：
1．测量模型检验结果显示，冠心病四期十二证的证候特征模型均具有良好的模型适配度。早期气滞心胸证模型由胸闷、胸痛、心烦、手足不温、背痛、脘腹痞闷、胁痛、叹息要素构成，叹息与胁痛、胸闷与叹息之间关联有意义；早期气阴两虚证模型由气短、心悸、胸闷隐痛、背痛、头晕、乏力、心烦要素构成，其中气短与心悸、心悸与乏力、胸闷隐痛与乏力之间具有关联；发作期气滞痰阻证模型由胸闷如窒、头晕、心烦、手足不温、背痛、脘腹痞闷、食欲不振、恶心呕吐、叹息、胁痛、喘促要素构成，其中手足不温与食欲不振、食欲不振与脘腹痞闷、叹息与胁痛之间具有关联；心阳不振证模型由胸闷、心烦、胸痛、背痛、喘促、手足不温、食欲不振、脘腹痞闷、恶心呕吐要素构成，其中胸闷与胸痛、手足不温与食欲不振、食欲不振与脘腹痞闷之间具有关联；寒凝心脉证模型由胸闷如窒、喘促、痛引背肩、背痛、手足不温、心烦、腰膝浮肿、食欲不振、恶心呕吐要素构成，其中手足不温与食欲不振之间具有关联；瘀血阳微证模型由心悸、胸闷如窒、痛引背肩、背部刺痛、喘促不平卧、冷汗淋漓、手足不温、腰膝酸软、下肢浮肿、头晕、乏力、口干咽干要素构成，其中痛引背肩与背部刺痛、痛引背肩与喘促不平卧、喘促不平卧与下肢浮肿、腰膝酸软与口干咽干、心悸与乏力、痛引背肩与乏力之间具有关联；缓解期肝脾不调证模型由胁痛、脘腹痞闷、食欲不振、叹息、呃逆、胸闷、胸痛、背痛、心烦、头晕要素构成，其中胁痛与叹息、食欲不振与心烦、胁痛与胸痛、胸闷与胸痛、心烦与头晕之间具有关联；心肾阳虚证模型由胸闷、胸痛、背痛、气短、乏力、腰膝酸软、手足不温、感冒、汗出、下肢浮肿、头晕、喘促要素构成，其中胸闷与胸痛与气短、气短与乏力与汗出、感冒与手足不温、感冒与喘促、头晕与气短之间具有关联；心肺气虚证模型由心悸、气短、乏力、胸闷、喘促、背痛要素构成，其中喘促与胸闷与背痛、乏力与喘促、乏力与背痛、心悸与背痛之间具有关联；恢复期心气虚证模型由心悸、气短、乏力、胸闷、胸隐痛、失眠、头晕、心烦、汗出、浮肿要素构成，其中头晕与心烦、浮肿与胸闷、气短与乏力、心悸与气短之间具有关联；阳虚气滞证模型由胸闷、胸隐痛、背痛、手足不温、感冒、心烦、胁痛、叹息、腰膝酸软、喘促、浮肿、脘腹痞闷要素构成，其中感冒与叹息、胁痛与叹息、感冒与喘促、感冒与浮肿、脘腹痞闷与腰膝酸软、胁痛与喘促、叹息与喘促、喘促与浮肿之间具有关联；气阴两虚证模型由胸闷、胸隐痛、背隐痛、心悸、乏力、自汗、心烦、干咳、干呕、口干不欲饮、腰酸痛、下肢浮肿要素构成，其中胸隐痛与背隐痛、胸闷与心悸与乏力、胸闷与心烦、心悸与心烦、乏力与心烦、心烦与下肢浮肿之间具有关联。
2.证候要素的二阶CFA显示，发作期气滞痰阻证可取证候要素气滞{手足不温，叹息，胁痛，心烦}、痰饮{头晕，喘促，胸闷如窒，背痛}、脾气虚{脘腹痞闷，食欲不振，恶心呕吐}，因素载荷量分别为1.027、0.958、1.033；发作期寒凝心脉证可提取证候要素心阳虚{胸闷如窒，喘促，手足不温，心烦}、脾阳虚{食欲不振，恶心呕吐}、寒邪{背痛，痛引肩背，腰膝浮肿}，因素载荷量分别为1.023、0.982、0.889；发作期阳虚血瘀证可提取证候要素血瘀{痛引肩背，背部刺痛，紫绀}、心阳虚{喘促不能平卧，冷汗淋漓，手足不温，胸闷如窒}、心气虚{头晕，心悸，乏力}、肾气虚{腰膝酸软，下肢浮肿，口干咽干}，因素载荷量分别为0.608、1.033、0.888、0.828；缓解期心肺气虚证可提取证候要素心气虚{心悸，乏力，胸闷，背痛}、肺气虚{气短，喘促}，因素载荷量分别为0.985和0.897；恢复期气阴两虚证二阶可提取证候要素心气虚{心悸，乏力，自汗，胸闷，胸隐痛，背隐痛，心烦}、心阴虚{干呕，干咳}、肾阴虚{口干不欲饮，腰酸痛，下肢浮肿}，因素载荷量分别为0.194、0.499和1.869。
3．PA-LV模型路径分析得出证候演变规律是：发病早期→发作期：气滞心胸转化为瘀血阳微、气滞痰阻、寒凝心脉、心阳不振证路径系数分别为0.67、0.69、0.00和0.45，气阴两虚转化为瘀血阳微、气滞痰阻、寒凝心脉、心阳不振证路径系数分别为0.25、0.46、0.17和0.40；发作期→缓解期：瘀血阳微转化为心肾阳虚、心肺气虚、肝脾不调证路径系数分别为1.00、0.80和0.69，气滞痰阻转化为心肾阳虚、心肺气虚、肝脾不调证路径系数分别为0.80、0.38和0.87，寒凝心脉转化为心肾阳虚、心肺气虚、肝脾不调证路径系数分别为0.92、0.43和0.92，心阳不振转化为心肾阳虚、心肺气虚、肝脾不调证路径系数分别为0.89、0.38和0.90；缓解期→恢复期：心肾阳虚转化为气阴两虚、心气虚、阳虚气滞路径系数分别为0.23、0.78和0.81，心肺气虚转化为气阴两虚、心气虚、阳虚气滞路径系数分别为0.16、0.88和0.72，肝脾不调转化为气阴两虚、心气虚、阳虚气滞路径系数分别为0.27、0.67和0.81。病机演变规律显示：冠心病病机演变规律显示，因虚致实、因实致虚是冠心病病机转化的基本规律。心阳不振可导致寒邪、痰饮、气滞、瘀血发作，标实去后本虚又现，阳虚诸证又显，故心阳不振为冠心病的基本病理基础，寒邪、痰饮、气滞、瘀血为主要的病理要素。
结论：
1．结构方程模型是中医证候研究有效方法。证候“内实外虚，动态时空，多维界面”的特点与结构方程模型体现了系统性原理，将SEM方法引入中医证候研究，提供了合理、有效、量化的证候研究工具。
2.基于EFA因子结构的测量模型可周延性地揭示证候特征。测量模型结果显示，EFA（聚类分析）探索出的四期十二证模型结构合理，可有效反应冠心病证候特征；基于理论假设和EFA因子结构的测量模型较之单纯的EFA分析具有理论和结构的周延性。
3．二阶CFA模型是提取证候要素的有效方法。以象为素，用具有高度共线性（因素负荷量大于0.75）的“象”构成初阶潜在变量“证候要素”；以素为候，将证候特征定为高阶潜在变量，构成二阶CFA模型，进行检验估计，有效提取了12个冠心病证候要素，反映了证候研究降维升阶的思维过程。
4．潜在变量的路径分析模型是探索病机演变规律的创新方法。在“动态时空”下，以候为证，以证应机，得出的冠心病病机演变规律。PA-LV模型较之其它方法，可明确冠心病各阶段病机作用力的影响强弱、作用效果是否明显、作用途径直接抑或间接、转化预后如何，是一种创新性研究方法。
5．温阳益心法是针对冠心病阳虚病理基础的有效治法。融温阳补气益心、活血化痰通脉、清心宁神为一体的温阳益心法，是结合心脏象理论，整体把握冠心病病变全程、养护俱效的中医治法。
Purpose：Explore methods of structural equation modeling (SEM) and TCM syndromeresearch linking theory; explore the feasibility of the framework of the SEMStudy in the Syndrome analysis; with the help of IBM SPSS AMOS20.0, SEM practicaloperation platform, verify and reveal coronary heart disease syndromecharacteristics, syndrome elements, and pathogenesis evolution; demonstrate thestructural equation model is reasonable, quantitative and effective TCM syndromeresearch method, both in theory and practical application.
Material and method：On the basis of three-dimensional structure relationaldatabase of coronary heart disease syndrome elements, syndrome characteristicsand syndrome pathogenesis evolution questionnaire by clinical experts,distinguish data information in four stages of early, ictal, remission andrecovery; structurally transform and process database to meet Amos requirementby perl language and IF statement.
Explore data mining in three-dimensional structure relational database ofcoronary heart disease syndrome elements, syndrome characteristics and syndromepathogenesis evolution questionnaire by clinical experts with the help ofstructural equation modeling. Based on the former research of early exploratoryfactor analysis (EFA), launch model verification and modification of measuredmodel, to found coronary heart disease syndromes characteristic model; launchthe second-order CFA model to analyze relationship of TCM syndromecharacteristics and syndrome elements, and extract the effective syndromeelements; launch latent variable path analysis (PA-LV) to explore causalconnection of pathogenesis development between different coronary heart diseasestages, and explicit the evolution principle of coronary heart diseasepathogenesis.
Results：
1. Measurement model test results show that twelve syndrome feature model in four phases of coronary heart disease syndromes has a good model fit. In earlystage, the qi stagnation in chest syndrome model is made up of chest oppression,chest pain, dysphoria, not warm hands, back pain, abdominal swelling,hypochondriac pain and sigh, with connection between sigh and hypochondriac pain,chest oppression and sigh. Deficiency of both qi and yin syndrome model is madeup of breath shortness, palpitation, chest dull pain, back pain, dizziness,fatigue and dysphoria, with connection between breath shortness and palpitation,palpitation and fatigue, chest dull pain and fatigue. In ictal stage, Qistagnation and phlegm blockage syndrome model is made up of chest oppression,dizziness, dysphoria, not warm hands, back pain, abdominal swelling, inappetence, nausea, sigh, hypochondriac pain and dyspnea, with connection betweennot warm hands and in appetence, in appetence and abdominal swelling, sigh andhypochondriac pain. Deficiency in heart yang syndrome model is made up of chestoppression, dysphoria, chest pain, back pain, dyspnea, not warm hands, inappetence, abdominal swelling and nausea, with connection between chestoppression and chest pain, not warm hands and in appetence, in appetence andabdominal swelling. Coagulated cold in heart syndrome model is made up of chessoppression, dyspnea, back and shoulder pain, back pain, not warm hands, dysphoria,waist knee edema, in appetence and nausea, with connection between not warm handsand in appetence. Blood stasis and yang lacking syndrome is made up of palpitation,chess oppression, back and shoulder pain, back pain, dyspnea, cold sweat, notwarm hands, waist and knee soreness, lower limb edema, dizziness, fatigue andthirsty, with connection between back-shoulder pain and back pain, back-shoulderpain and dyspnea, dyspnea and lower limb edema, waist and knee soreness andthirsty, palpitation and fatigue, back-shoulder pain and fatigue.In remission stage, disharmony of liver and spleen syndrome model is made upof hypochondriac pain, abdominal swellings, in appetence, sigh, hiccough, chestoppression, chest pain, back pain, dysphoria, and dizziness, with connectionbetween hypochondriac pain and sigh, in appetence and dysphoria, hypochondriacpain and chest pain, chest oppression and chest pain, dysphoria and dizziness. Yang deficiency of heart and kidney syndrome model is made up of chest oppression,chest pain, back pain, hiccough, fatigue, waist and knee soreness, not warm hands,cold, sweat, lower limb edema, dizziness and dyspnea, with connection betweenchest oppression and chest pain and hiccough, hiccough and fatigue and sweat,cold and not warm hands, cold and dyspnea, dizziness and hiccough. Qi deficiencyof heart and lung syndrome model is made up of palpitation, hiccough, fatigue,chest oppression, dyspnea, and back pain, with connection between dyspnea andchest oppression and back pain, fatigue and dyspnea, fatigue and back pain,palpitation and back pain. In recovery stage, deficiency of heart qi syndromemodel is made up of palpitation, hiccough, fatigue, chest oppression, chest dullpain, insomnia, dizziness, dysphoria, sweat, and edema, with connection betweendizziness and dysphoria, edema and chest oppression, hiccough and fatigue,palpitation and hiccough. Yang deficiency and qi stagnation syndrome model ismade up of chest oppression, chest dull pain, back pain, not warm hands, cold,dysphoria, hypochondriac pain, sigh, waist and knee soreness, dyspnea, edema,and abdominal swellings, with connection between cold and sigh, hypochondriacpain and sigh, cold and dyspnea, cold and edema, abdominal swellings and waistand knee soreness, hypochondriac pain and dyspnea, sigh and dyspnea, dyspneaand edema. Deficiency of both qi and yin syndrome model is made up of chestoppression, chest dull pain, back dull pain, palpitation, fatigue, sweating,dysphoria, dry cough, retching, thirst, waist soreness, and lower limb edema,with connection between chest dull pain and back dull pain, chest oppressionand palpitation and fatigue, chest oppression and dysphoria, palpitation anddysphoria, fatigue and dysphoria, dysphoria and lower limb edema.
2. The results of syndrome elements second-order CFA model reveal: In ictal stage,Qi stagnation and phlegm blockage syndrome contains syndrome elements of Qistagnation {not warm hands, sigh, hypochondriac pain, dysphoria}, phlegm{dizziness, dyspnea, chest oppression, back pain}, spleen qi deficiency{abdominal swellings, in appetence, nausea}, with loading1.027,0.958,1.033.Coagulated cold in heart syndrome contains syndrome elements of heart yang deficiency {chest oppression, dyspnea, not warm hands, dysphoria}, spleen yangdeficiency {in appetence, nausea}, cold pathogen {back pain, back-shoulder pain,waist and knee edema}, with loading1.023,0.982,0.889. Blood stasis and yanglacking syndrome contains syndrome elements of blood stasis {back-shoulder pain,back pain, cyanosis}, heart yang deficiency {dyspnea, cold sweating, not warmhands, chest oppression}, heart qi deficiency {dizziness, palpitation, fatigue},kidney qi deficiency {waist and knee soreness, lower limb edema, thirst}, withloading0.608,1.033,0.888,0.828. In remission stage, qi deficiency of heartand lung syndrome contains syndrome elements of heart qi deficiency {palpitation,fatigue, chest oppression, back pain}, lung qi deficiency {hiccough, dyspnea},with loading0.985and0.897. In recovery stage, deficiency of both qi and yinsyndrome contains syndrome elements of deficiency of heart qisyndrome{palpitation, fatigue, sweat, chest oppression, chest dull pain, backdull pain, dysphoria}, heart yin deficiency{retching, dry cough}, kidney yindeficiency{thirst, waist soreness, lower limb edema}, with loading0.194,0.499and1.869.
3．The results of PA-LV model indicates pathogenesis evolution rules as follows:Early→Ictal: qi stagnation in chest syndrome will turn into blood stasis andyang lacking syndrome, Qi stagnation and phlegm blockage syndrome, coagulatedcold in heart syndrome, deficiency in heart yang syndrome with loading0.67,0.69,0.00and0.45. Deficiency of both qi and yin syndrome will turn into bloodstasis and yang lacking syndrome, Qi stagnation and phlegm blockage syndrome,Coagulated cold in heart syndrome, Deficiency in heart yang syndrome with loading0.25,0.46,0.17and0.40. Ictal→Remission: Blood stasis and yang lackingsyndrome will turn into yang deficiency of heart and kidney syndrome, qideficiency of heart and lung syndrome, disharmony of liver and spleen syndromewith loading1.00,0.80and0.69. Qi stagnation and phlegm blockage syndromewill turn into yang deficiency of heart and kidney syndrome, qi deficiency ofheart and lung syndrome, disharmony of liver and spleen syndrome with loading0.80,0.38and0.87. Coagulated cold in heart syndrome will turn into yang deficiency of heart and kidney syndrome, qi deficiency of heart and lung syndrome,disharmony of liver and spleen syndrome with loading0.92,0.43and0.92.Deficiency in heart yang syndrome will turn into yang deficiency of heart andkidney syndrome, qi deficiency of heart and lung syndrome, disharmony of liverand spleen syndrome with loading0.89,0.38and0.90. Remission→Recovery: yangdeficiency of heart and kidney syndrome will turn into deficiency of both qiand yin syndrome, deficiency of heart qi syndrome, yang deficiency and qistagnation syndrome with loading0.23,0.78and0.81, qi deficiency of heartand lung syndrome will turn into deficiency of both qi and yin syndrome,deficiency of heart qi syndrome, yang deficiency and qi stagnation syndrome withloading0.16,0.88and0.72, disharmony of liver and spleen syndrome will turninto deficiency of both qi and yin syndrome, deficiency of heart qi syndrome,yang deficiency and qi stagnation syndrome with loading0.27,0.67and0.81.As a result, heart yang deficiency will cause cold pathogen, phlegm, qi staginess,blood staginess. Therefore, heart yang deficiency is the base of coronary heartdisease, and cold pathogen, phlegm, qi staginess, blood staginess are the mainsyndrome elements.
Conclusion：
1. Structural equation model is the effective method for syndrome analysis.Syndromes within the real outside virtual, dynamic space-time,multi-dimensional interface characteristics and structural equation model bothshow the influence of systematic principle. The introduction of the SEM methodmade it a reasonable, effective and quantifiable syndrome research tool.
2. Measurement model based on the EFA factor structure can reveal syndromecharacteristics comprehensiveness. The measurement model results show that thetwelve syndrome models from four stages that the EFA (cluster analysis) exploredare reasonable, and can effectively verify coronary heart syndromescharacteristics. Measurement model based on theoretical assumption and EFAfactor analysis is more effective and comprehensive than pure EFA analysis.
3. The second-order CFA model is an effective method to extract the syndromeelements. Factor loadings greater than0.75with a high degree of co linearityis recognized as syndrome elementary, with syndrome as latent variables. TheCFA models effectively extract the12syndrome elements and reflect the thinkingprocess of the syndromes of dimensionality reduction degree elevation.
4. Latent variable path analysis model is to explore innovative approaches tothe evolution of the pathogenesis. Compared to other methods, PA-LV model canclearly identify affect the strength force of the various stages of coronaryheart disease, whether the effect is obvious, the direct or the indirect pathway,and how will the prognosis transformed, which is an innovative research method.
5. The Wenyang beneficial heart treatment is the effective treatment method forcoronary heart yang deficiency. Combined the Wenyang qi and heart nutrition,blood circulation and vessel unblock and spirit releases together, on the basisof cardiac phenomenological theory, the treatment will overall grasp coronaryheart lesions and effective for both treating and healing.
引文
[1]饶克勤.卫生统计方法与应用进展.北京:人民卫生出版社，2008:118-127。
[2]张连文,袁世宏.隐结构模型与中医辨证研究(Ⅰ)——隐结构法的基本思想及隐结构分析工具.北京中医药大学学报,2006,29(06):365-369.
[3]张连文,袁世宏,陈弢等.隐结构模型与中医辨证研究(Ⅱ)——肾虚数据分析.北京中医药大学学报,2008,31(09):584-587.
[4]王阶,李海霞,王燚等.基于隐结构模型的血瘀证症状体征研究.世界科学技术,2006,8(06):23-26+62.
[5]袁世宏,张连文,陈弢等.隐结构模型与中医辨证研究(Ⅲ)——模型辨证与专家辨证.北京中医药大学学报,2008,31(10):659-663.
[6]龚燕冰,张连文,高思华等.隐结构模型对2型糖尿病病机规律的探讨.世界科学技术（中医药现代化）,2009,11(04):516-521.
[7]王天芳,张连文,赵燕等.隐结构模型及其在中医证候研究中的应用.北京中医药大学学报,2009,32(08):519-526.
[8]张连文,袁世宏,王天芳等.隐结构分析与西医疾病辨证分型(Ⅰ):基本原理.世界科学技术(中医药现代化),2011,13(03):498-502.
[9]张连文,许朝霞,王忆勤等.隐结构分析与西医疾病的辨证分型(Ⅱ):综合聚类.世界科学技术(中医药现化),2012,14(02):1422-1427.
[10]许碧云.结构方程模型在医学研究中的应用：(博士学位论文).成都：四川大学，2004
[11]徐秀娟.结构方程模型在医学研究中的应用：(硕士学位论文).太原：山西医科大学，2004
[12]李建明.结构方程模型及其隐变量交互作用分析在社会心理流行病学中的应用：(博士学位论文).太原：山西医科大学，2006
[14]李国春,李春婷,黄蓝平等.结构方程模型探讨慢性萎缩性胃炎证候分型规律.南京中医药大学学报,2006,22(04):217-220.
[13]王媛.医学生心理健康影响因素的结构方程模型分析：(硕士学位论文).天津：天津医科大学，2006
[15]韩潞潞.健康人生物学年龄积分及生物学衰老结构方程模型的统计建模研究：(博士学位论文).沈阳：中国医科大学，2010
[16]华丽.2型糖尿病患者自我管理行为的社会生态模型：结构方程模型的验证：(硕士学位论文).杭州：浙江大学，2010
[17]李建明,曲成毅.结构方程模型中隐变量非线性效应的分析建模.数理医药学杂志,2005,18(03):193-195.
[18]史雅翼.用线性结构方程模型对居民两周患病与就诊影响因素的分析：(硕士学位论文).杭州：浙江大学，2002
[19]杨书,张强,叶韵.结构方程模型在中医临床疗效评价中的应用.中国卫生统计,2009,26(06):565-568.
[20]韩煊.流动儿童健康状况及其影响因素的结构方程模型研究：(博士学位论文).武汉：华中科技大学，2009
[21]申春悌,陈启光,张华强等.结构方程模型在异病同证研究中的运用.中国中医基础医学杂志,2008,14(12):953-955+958.
[22]张芳.艾滋病中医虚证的结构方程建模：(硕士学位论文).郑州：郑州大学，2008
[23]吕德可.二阶证实性因子分析在高血压病中医证候要素研究中的运用：(硕士学位论文).南京：南京中医药大学，2010
[24]王静.基于结构方程模型分析的中医五脏相关理论研究：(硕士学位论文).广州：广州中医药大学，2009
[25]常艳鹏.冠心病证候要素、证候特征及证候演变规律研究：(博士学位论文).沈阳：辽宁中医药大学，2010