[Nomogram model for predicting risk of anti-tuberculosis drug-induced liver injury among inpatients with tuberculosis].

Objective: To explore the influencing factors of anti-tuberculosis drug-induced liver injury (ATB-DILI) in hospitalized tuberculosis patients, and to establish a risk prediction model of Nomogram. Methods: A retrospective study was conducted on 5 681 tuberculosis patients admitted to Guiyang public health treatment center from January 2017 to June 2021, including 3 342 males and 2 339 females. The inpatients with ATB-DILI were selected as the case group (214 cases) and the non-ATB-DILI patients as the control group (5 427 cases). The baseline characteristics, tuberculosis condition, behavior and disease-related data of the patients were retrospectively analyzed, and the influencing factors were screened by chi-square test and multivariate logistic regression, based on which the Nomogram model was constructed and verified. The decision curve was used to evaluate the clinical application value of the model. Results: In this study, 3.8%(214/5 681) patients developed ATB-DILI. Multivariate logistic regression analysis showed that extrapulmonary tuberculosis (OR=1.876, P<0.001), malnutrition (OR=4.411, P<0.001), complicated with underlying liver disease (OR=4.961, P<0.001) and intermittent use of hepatoprotective drugs (OR=2.137, P=0.007) were independent risk factors for ATB-DILI, while whole-course use of hepatoprotective drugs (OR=0.292, P<0.001) was protective factor. The Nomogram model was constructed based on the above five related factors. The area under the receiver operating characteristic (ROC) curve was 0.749 (95%CI:0.713-0.786), the sensitivity was 0.640, and the specificity was 0.752, respectively. The Bootstrap method was used for internal repeated sampling for 1 000 times, the average absolute error was 0.003, the correction curve and the ideal curve were basically fitted, and the predicted values were in good agreement with the actual values. Hosmer-lemeshow test showed that the model had a good degree of fit (χ2=3.068, P=0.381). The decision curve showed that the Nomogram model had certain clinical practicability in the high risk threshold range (0.10-0.68). Conclusions: The Nomogram model for risk predicting ATB-DILI among inpatients with tuberculosis in this study has good predictability, consistency and clinical practicability, and can provide a basis for clinical prevention and control of ATB-DILI and individualized treatment in the process of anti-tuberculosis treatment.目的： 探讨住院结核患者发生抗结核药物性肝损伤（ATB-DILI）的影响因素，并建立Nomogram风险预测模型。 方法： 回顾性收集2017年1月至2021年6月贵阳市公共卫生救治中心收治的5 681例住院结核病患者作为研究对象，其中男3 342例，女2 339例，将住院治疗过程中发生ATB-DILI的患者作为病例组（214例），非ATB-DILI患者作为对照组（5 427例）。回顾性分析患者的基线特征、结核病情、行为及疾病相关资料等，经卡方检验和多因素logistic 回归筛选分析影响因素，据此构建Nomogram模型并进行验证。使用决策曲线评估模型的临床实际应用价值。 结果： 本次研究中共有3.8%（214/5 681）例患者发生ATB-DILI。经多因素logistic 回归分析得出肺外结核（OR=1.876，P<0.001）、营养不良（OR=4.411，P<0.001）、合并基础肝病（OR=4.961，P<0.001）以及间断使用护肝药（OR=2.137，P=0.007）是其发生ATB-DILI的独立危险因素；全程使用护肝药（OR=0.292，P<0.001）是其保护因素。将以上5个相关影响因素构建Nomogram模型，ROC曲线下面积为0.749（95%CI：0.713~0.786），敏感度0.640，特异度0.752。使用Bootstrap法内部重复抽样1 000次进行验证，平均绝对误差0.003，校正曲线和理想曲线基本拟合，预测值和实际值一致性较好。Hosmer-lemeshow检验显示，模型具有较好拟合度（χ2=3.068，P=0.381）。决策曲线显示Nomogram模型在高风险阈值范围（0.10~0.68）时，有着一定的临床实用性。 结论： 本次研究所构建的住院结核患者抗结核药物性肝损伤Nomogram风险预测模型具有较好的预测性、一致性和临床实用性，能为临床在抗结核治疗过程中防控ATB-DILI并制定个体化治疗方案提供依据。.