摘要
目的分析耐多药结核病患者治疗成功因素,并建立路线图模型预测其治疗成功概率。方法回顾性分析2019年11月至2022年5月成都市公共卫生临床医疗中心结核科接诊的耐多药结核病患者263例。患者按7∶3分为模型组184例,验证组79例。根据疗效结果将模型组分为治疗成功组与不良结局组,比较两组患者临床参数、CT检查结果及实验室指标,以最小绝对收缩与选择算子(LASSO)回归法筛选潜在因素后行多因素Logistic回归分析,根据分析结果建立路线图模型并进行验证。结果模型组184例患者中治疗成功103例(55.98%),不良结局81例(44.02%)。LASSO回归基础上行多因素Logistic回归分析结果显示:年龄、二线抗结核药物使用史、治疗6个月后痰菌转阴、病变范围、多发空洞、红细胞沉降率及尿蛋白为耐多药结核病患者治疗成功的独立影响因素。模型组路线图模型预测耐多药结核病患者治疗成功的AUC为0.894(95%CI:0.847~0.941);验证组AUC为0.865(95%CI:0.810~0.919)。模型组与验证组预测曲线与标准曲线基本拟合。当该路线图模型预测耐多药结核病患者治疗成功的概率阈值为0.10~0.95时,患者的净受益率大于0。结论耐多药结核病患者治疗成功受年龄、二线抗结核药物使用史、治疗6个月后痰菌转阴等因素的影响,路线图模型预测耐多药结核病患者治疗成功概率具有较高的准确度与区分度。
Objective To analyze the factors related to treatment success in patients with MDRTB,and establish a nomogram model to predict the probability of treatment success.Method 263 MDR-TB patients admitted to the tuberculosis Department of Chengdu Public Health Clinical Medical Center from November 2019 to May 2022 were retrospectively analyzed.The patients were divided into model group(184 cases)and verification group(79 cases)by 7:3.According to the efficacy results,the model group was divided into successful treatment group and adverse outcome group,and the clinical parameters,CT examination results and laboratory indicators of the two groups were compared.After the potential factors were screened by LASSO regression method,multivariate Logistic regression analysis was performed,and a nomogram model was established and verified according to the results of multiple factors.Result Among the 184 patients in the model group,103(55.98%)cases were successfully treated,and 81(44.02%)cases had adverse outcomes.The results of ascending Logistic regression analysis on the basis of LASSO regression showed that age,history of second-line anti-TB drug use,sputum turning negative after 6 months of treatment,lesion scope,multiple cavity,erythrocyte sedimentation rate and urinary protein were independent factors influencing the treatment success of MDR-TB patients.The AUC for successful treatment of MDR-TB patients was 0.894(95%CI:0.847-0.941).The AUC of the verification group was 0.865(95%CI:0.810-0.919).The prediction curves of the model group and the verification group were basically fitted with the standard curves.When this column chart model predicts that the probability threshold of treatment success for MDR-TB patients is 0.10-0.95,the net benefit rate for patients is greater than 0.Conclusion The success of treatment of MDRTB patients is affected by age,history of second-line anti-TB drugs,sputum turning negative after 6 months of treatment and other factors.The nomogram model has high accuracy and differentiation in predicting the success of treatment of MDR-TB patients.
作者
巫雪琴
程耀
杨铭
袁平
李娇
孙秋萍
Wu Xueqing;Cheng Yao;Yang Ming;Yuan Ping;Li Jiao;Sun Qiuping(Department of Tuberculosis,Chengdu Public Health Clinical Medical Center,Sichuan Chengdu 610061,China)
出处
《新发传染病电子杂志》
2023年第4期30-35,共6页
Electronic Journal of Emerging Infectious Diseases
基金
2022年成都市医学科研课题(2022263)。
关键词
耐多药结核病
多因素分析
列线图
预测模型
Multidrug-resistant tuberculosis
Multi factor analysis
Nomogram
Predictive models
