中医中风病证候的多元统计分析

Multivariate Analysis of TCM Syndrome of Stroke

目的 :探讨中医中风病证候的多元统计分析方法。方法 :对大规模现场流行病学资料 (390 9例 ) ,分析了克朗巴赫系数α(Cronbachα)和分半信度 (split halfreliability)后 ,分别求出每一变量卡方值 (χ2 )、概率值 (P)、优势比 (OR)、95 %CI。在变量的相关性及多重共线性分析并结合文献的系统分析及临床经验的基础上 ,应用SAS软件对有显著性意义的危险因素进行回归分析 (logistic)、聚类分析及主成分分析。结果 :构成证候的症状 (变量 )之间的多重共线性形成了复杂的多元非线性关系 ,多元统计分析可客观地评价各症状在证候中所处的地位和作用。结论 :多元统计分析是揭示中医证候复杂关系的有效方法 ,值得深入研究。

Objective: To study the multivariate analysing technique for TCM Syndrome assay of stroke. Methods: Analysis of Cronbach α and split half reliability was conducted on the epidemiologic data of 3909 subjects and values of Chi square (χ 2), probability (P), odds rate (OR) and 95% corrected increment (95%CI) of each variance were calculated. Then based on the analysis of relationship between variances and multi collinear, combined with systematic analysis of literatures and clinical experience, the logistic regression, clustering and principal component analysis to the risk factors with evident significance were carried out using SAS software. Results: The multiple collinearity of variances (symptoms that composed in the syndrome) formed a complex multiple nonlinear relationship, and the position and action of each variance in syndrome could be evaluated objectively by multivariate analysis. Conclusion: Multivariate analysis is an effective method to reveal the complexity of symptoms and worth further study.