耐药结核病的危险因素分析及风险预测模型的建立

摘要

Objective To investigate the risk factors for drug resistance‐tuberculosis(DR‐TB) ,and to establish a clinical risk predictive model .Methods A total of 126 cases of DR‐TB patients and 126 cases of non‐DR‐TB patients treated in our hospital from January 2014 to January 2015 were included in this study .The clinical data of these patients were collected .We used univariate and multivariate logistic regression analysis to determine the independent risk factors and established a risk predictive model .The calibration and discrimination of the model were assessed by the H‐L test and the area under the ROC curve ,respectively .Results Statistical analysis showed that the risk factors included previous treatment ,a duration of first treatment of more than 8 months ,ad‐verse effects of anti‐TB medication ,more than three TB foci in the lung and diabetes mellitus .H‐L statistic(χ2 =8 .760 ,P=0 .363) . The area under the ROC was 0 .826 ,95% CI(0 .766 ,0 .886) .Conclusion Logistic regression model established in the study can pre‐dict the incidence of DR‐TB with high prediction accuracy .%目的：筛选耐药结核病（DR‐TB）的危险因素，并构建风险预测模型。方法选择该院2014年1月到2015年1月确诊的126例DR‐TB患者为病例组和126例非耐药性结核病患者为对照组。回顾性收集纳入患者的临床资料。采用Logistic回归分析筛选危险因素，建立预测模型，并用 H‐L χ2检验来检验模型的拟合优度，用ROC曲线下面积来评价模型的预测效能。结果 Logistic回归分析结果显示，复治结核、第1次治疗时间大于8个月、抗结核药物不良反应、肺结核病灶数大于3个、合并糖尿病是DR‐TB的独立危险因素。 H‐L χ2检验（χ2＝8．760，P＝0．363），ROC 曲线下面积为0．826，95％ CI（0．766，0．886）。结论研究中拟合的Logistic回归模型预测准确率较高，对DR‐TB发生风险的评估有一定的参考价值。