摘要
目的:探讨耐多药和利福平耐药肺结核(MDR/RR-TB)住院患者的治疗结局及影响因素,构建并评价Nomogram风险预测模型。方法:收集2019年1月1日至2019年12月31日昆明市第三人民医院MDR/RR-TB住院患者临床资料并电话随访。采用χ^(2)检验和多因素Logistic回归分析筛选治疗结局的影响因素,构建Nomogram模型并验证。结果:141例患者中,治疗成功79例(占56.03%),不良转归62例(占43.97%)。多因素Logistic回归分析结果显示,糖尿病、血液系统疾病和农村居住是MDR/RR-TB患者治疗结局的危险因素,初中及以上文化程度是MDR/RR-TB患者治疗结局的保护因素。基于上述4个相关影响因素构建Nomogram模型,受试者操作特征曲线下面积为0.752(95%CI:0.673~0.832),敏感度为0.667,特异度为0.831。采用Bootstrap法重抽样1000次行内部验证,平均绝对误差为0.041,预测值和实际值一致性较好。Hosmer-Lemeshow检验结果表明,模型拟合度较好(χ^(2)=8.119,P=0.322)。决策曲线显示Nomogram模型在高风险阈值范围(0.14~0.81)时,临床实用价值较好。结论:构建的Nomogram模型具有较好的预测性、一致性及临床实用性,可为临床治疗MDR/RR-TB提供一定的科学参考。
Objective:To investigate the treatment outcomes and influencing factors of multidrug-resistant/rifampicin-resistant tuberculosis(MDR/RR-TB),and construct a Nomogram risk prediction model and evaluate it.Methods:Clinical data of MDR/RR-TB inpatients in Kunming Third People's Hospital from January 1,2019 to December 31,2019 were collected and followed up by telephone.χ^(2) test and multiple Logistic regression analysis were used to screen the influencing factors of treatment outcome,and a Nomogram model was established for verification.Results:Among 141 MDR/RR-TB patients,79(56.03%)were successfully treated,while 62(43.97%)had adverse treatment outcomes.Multivariate Logistic regression analysis showed that diabetes mellitus,hematological diseases and rural residence were risk factors for treatment outcomes of MDR/RR-TB patients,while junior high school education and above was protective factor for treatment outcomes of MDR/RR-TB patients.A Nomogram model was constructed based on the above four factors.The area under the receiver operator characteristic curve was 0.752(95%CI:0.673-0.832),with the sensitivity of 0.667 and specificity of 0.831.After 1000 Bootstrap resamples for internal verification,the average absolute error was 0.041,and the predicted value was consistent with the actual value.The Hosmer-Lemeshow test showed that the model had a good fit(χ^(2)=8.119,P=0.322).The decision curve showed that the Nomogram model had certain clinical practical value in the high risk threshold range(0.14-0.81).Conclusion:The established Nomogram model has good predictability,consistency and clinical practicability,and can provide a scientific reference for the clinical treatment of MDR/RR-TB.
作者
李硕兰
王淑娴
徐源
刘才
李明武
Li Shuoan;Wang Shuxian;Xu Yuan;Liu Cai;Li Mingwu(College of Public Health,Dali University,Dali,Yunnan 671000,China;No.2 Department of Tuberculosis,Kunming Third People's Hospital,Kunming 650041,China)
出处
《大理大学学报》
2024年第4期1-7,共7页
Journal of Dali University
作者简介
第一作者:李硕兰,硕士研究生,主要从事疾病预防与控制研究;通信作者:李明武,主任医师,E-mail:ynkmlmw@sina.com。
