Page 77 - 《中国药房》2025年20期
P. 77
·药物与临床·
替加环素相关药物性胆汁淤积性肝病的影响因素分析及风险预测
模型构建
Δ
刘丽娜 ,王建青 ,张 伦,俞 军[安徽省公共卫生临床中心(安徽医科大学第一附属医院北区)药学部,合肥
*
#
230012]
中图分类号 R969.3 文献标志码 A 文章编号 1001-0408(2025)20-2555-06
DOI 10.6039/j.issn.1001-0408.2025.20.12
摘 要 目的 分析替加环素(TGC)相关药物性胆汁淤积性肝病(DIC)的影响因素,并建立风险预测模型。方法 收集2022年8
月-2024年8月于我院接受TGC治疗的707例住院患者的临床资料,按8∶2的比例将其随机分为训练集(n=566)和测试集(n=
141)。通过最小绝对收缩和选择算子回归分析筛选预测变量,采用多因素Logistic回归分析筛选TGC相关DIC的独立危险因素,
并基于上述因素绘制列线图预测模型;采用受试者操作特征曲线(ROC曲线)及其曲线下面积(AUC)评估模型的预测性能,采用
Hosmer-Lemeshow拟合优度检验和校准曲线评估模型的准确度,采用决策曲线分析评估模型的临床净收益。结果 707例患者中,
93例患者发生DIC,发生率为13.15%。性别、年龄、TGC大剂量给药、入住重症监护室、TGC用药时间、合并使用抗真菌药伏立康
唑是TGC相关DIC发生的独立危险因素(P<0.05)。训练集模型ROC曲线的AUC为0.745(95%置信区间为0.687~0.801),灵敏
度为 76.6%,特异度为 60.3%;测试集模型 ROC 曲线的 AUC 为 0.762(95% 置信区间为 0.650~0.900),灵敏度为 81.3%,特异度为
72.0%。训练集、测试集模型Hosmer-Lemeshow拟合优度检验的χ 分别为5.187、9.980,P分别为0.737、0.266;校准曲线的平均绝
2
对误差分别为0.012、0.038;风险阈值范围分别为4%~45%、4%~28%。结论 患者的年龄、性别、TGC大剂量给药、入住重症监护
室、TGC用药时间、合并使用抗真菌药伏立康唑是TGC相关DIC发生的独立危险因素;基于以上因素所建的TGC相关DIC预测模
型的预测性能、准确度均较好。
关键词 替加环素;药物性胆汁淤积性肝病;药物性肝损伤;影响因素;预测模型
Analysis of influential factors and the construction of a risk prediction model for tigecycline-related drug-
induced cholestatic liver disease
LIU Lina,WANG Jianqing,ZHANG Lun,YU Jun[Dept. of Pharmacy, Anhui Public Health Clinical Center (the
First Affiliated Hospital of Anhui Medical University North District), Hefei 230012, China]
ABSTRACT OBJECTIVE To analyze the influential factors of drug-induced cholestatic liver disease (DIC) related to
tigecycline (TGC), and establish a prediction model for the risk of this adverse reaction. METHODS Data of 707 hospitalized
patients who received TGC treatment in our hospital from August 2022 to August 2024 were collected and randomly divided into
training set (n=566) and test set (n=141) at a ratio of 8∶2. Prediction variables were screened using the least absolute shrinkage
and selection operator regression analysis. Multivariate Logistic regression analysis was used to screen the independent risk factors
for TGC-related DIC, and a nomogram prediction model was drawn based on the above factors. The prediction performance of the
model was evaluated by the receiver operator characteristic curve (ROC curve) and its area under the curve (AUC). The accuracy
of the model was assessed by the Hosmer-Lemeshow goodness-
Δ 基金项目 国家自然科学基金项目(No.82073566);安徽省卫生
of-fit test and calibration curves. The clinical net benefit of the
健康科研项目(No.皖卫传〔2023〕515号);安徽省教育厅高等学校省级
质量工程项目(No.皖教秘高〔2024〕55号);安徽省公共卫生临床中心 prediction model were evaluated by decision curve analysis.
(安徽医科大学第一附属医院北区)院内临床重点专科(培育专科)建 RESULTS Among the 707 patients, 93 patients developed
设项目(No. 院医字〔2023〕17 号);院科研培育基金项目(No. 院字 DIC, with an incidence rate of 13.15%. Gender, age, high-
〔2022〕4号) dose administration of TGC, intensive care unit (ICU)
*第一作者 主管药师,硕士。研究方向:临床药学。E-mail:
admission, duration of medication of TGC, and concurrent
2146158060@qq.com
# 通信作者 主任药师,教授,博士。研究方向:临床药学。E-mail: use of antifungal drug voriconazole were independent risk
jianqingwang81@126.com factors for the occurrence of TGC-related DIC (P<0.05). The
中国药房 2025年第36卷第20期 China Pharmacy 2025 Vol. 36 No. 20 · 2555 ·

