结核病患者药物性肝损伤预警模型的建立及验证研究

Development and validation of prognostic models for tuberculosis patients with drug-induced liver injury

目的建立并验证预测结核病患者服用固定剂量复合制剂后药物性肝损害发生的预测模型。方法选取208例肝损伤患者,随机分为建模样本156例和验证样本52例,建模样本用于单因素及Logistic回归分析,以寻找影响肝损害发生的影响因素,根据Logistic回归分析结果建立不同的肝损伤预警模型,通过计算ROC曲线下面积检测模型性能。验证样本用于计算不同预警模型的肝损害发生的准确率、敏感度和特异度。结果 Logistic回归分析结果显示居住环境、工作环境和诊断结果是肝损伤发生的影响因素。根据分析结果和专业性,经过比较分析,通过居住环境、工作环境、诊断结果和年龄建立的肝损害预测模型性能良好(偏卡方检验,P>0.05,ROC曲线下面积为0.884)。验证样本预测肝损害发生的准确率为84.6%,敏感度为55.56%,特异度为93.02%。结论通过居住环境、工作环境、诊断结果和年龄建立的肝损害预测模型可以早期、简单且准确地预测结核病患者服用固定剂量复合制剂后药物性肝损害的发生风险,可用于辅助临床决策的制定。

Objective To develop and validate prognostic models for tuberculosis patients who had the fixed dose compound preparation with drug-induced liver injury.Methods The clinical data of 208 patients with drug-induced liver injury were collected to develop prognostic models.The patients were randomly divided into the modeling sampling group(n=156)and the validation sampling group(n=56).Modeling sampling were used for univariate and Logistic regression analysis to identify factors related with drug-induced liver injury.According to univariate and Logistic regression analysis results,a prognostic model was established.The validation sampling was used to compute the accuracy,sensitivity and specificity of the prognostic models.Result Logistic regression analysis revealed that living environment,working environment and the diagnose result were independent risk factors for drug-induced liver injury.The prognostic model based on the risk factors of admission had favorable performance(P>0.05 for Partial chi square test,0.884 for AUC).The accuracy of the prognostic model was 84.6%,sensitivity was 55.56%and specificity was 93.02%.Conclusion The established model,which is convenient to manage,can timely and accurately predict the prognosis of tuberculosis patients with drug-induced liver injury,and can help in decision-making in clinics.