Objective To investigate the relationship between multidrug-resistant TB ( MDR-TB) and its' influencing factors. Methods Using the global MDR-TB surveillance datas, based on latent variable analysis theory and the spatial statistics method,we first extracted the integrated potential impact factors from the influencing factors of MDR-TB by confirmatory factor analysis; secondly, used kriging method to estimate the spatial distribution of the integrated potential impact factors and MDR-TB rate; then, constructed multiple linear regression model (global model) and geographical weighted regression model(GWR) (local model) about MDR-TB rate and the integrated potential impact factors to explore the relationship between MDR-TB rate and the integrated potential impact factors. Results Six potential impact factors were extracted by confirmatory factor analysis. Global model showed that six potential factors were all related with MDR-TB rate( P <0. 01) ;GWR model displayed R2 and parameter estimation of each local model appeared significant spatial variability, and its' fitting effect was better than the global model. Conclusion The effect of the integrated potential impact factors on MDR-TB rate present spatial variability; so we should formulate local MDR-TB monitoring planning and prevention and control strategies based on the spatial characteristic of the integrated potential impact factors and the local relationship with MDR-TB.%目的 探讨耐多药结核病及其影响因素间的关系.方法 利用全球耐多药结核病监测资料,基于潜变量分析理论和空间统计学方法,首先通过证实性因子分析从耐多药结核病影响因素中提取综合潜在影响因子,然后采用Kriging法估计综合潜在影响因子及耐多药率的空间分布,在此基础上分别构建耐多药率与其综合潜在影响因子之间的多元线性回归模型(全局模型)和GWR模型(局部模型),探讨耐多药率与综合潜在影响因子间的关系.结果 经证实性因子分析共提取了6个综合潜在影响因子.全局模型显示,6个潜在影响因子均与耐多药率有关(P值均小于0.01)；GWR 局部模型显示,各模型的R2及参数估计值均呈显著的空间变异,其拟合效果优于全局模型.结论 GWR模型揭示综合潜在影响因子对结核病耐多药率的影响存在空间变异性；应根据各综合潜在影响因子的空间特征及其与耐多药率间的局域关系制定区域化的耐多药结核病监测规划和防控策略.
展开▼
