中国药物警戒 ›› 2026, Vol. 23 ›› Issue (9): 1028-1033.
DOI: 10.19803/j.1672-8629.20260451

• 安全与合理用药 • 上一篇    下一篇

丙戊酸钠血药浓度影响因素超警戒值风险预测模型的建立

冯杰1, 张惠兰1, 刘磊2, 吴宛焰1, 李燕菊1, 李红健1,*   

  1. 1新疆维吾尔自治区人民医院药学部,新疆维吾尔自治区临床药学研究所,新疆 乌鲁木齐 830000;
    2新疆医科大学第八附属医院药学部,新疆 乌鲁木齐 830000
  • 收稿日期:2026-06-04 出版日期:2026-09-15 发布日期:2026-09-15
  • 通讯作者: *李红健,男,硕士,主任药师,精准药学。E-mail: leehongjian2006@163.com
  • 作者简介:冯杰,男,本科,主管药师,精准药学。
  • 基金资助:
    新疆维吾尔自治区自然科学基金资助项目（2023D01C69）

A risk prediction model for serum valproic acid concentrations exceeding the safety threshold

Feng Jie1, Zhang Huilan1, Liu Lei2, Wu Wanyan1, Li Yanju1, Li Hongjian1,*   

  1. 1Department of Pharmacy, Institute of Clinical Pharmacy, People's Hospital of Xinjiang Uygur Autonomous Region, Urumqi Xinjiang 830000, China;
    2Department of Pharmacy, the Eighth Affiliated Hospital of Xinjiang Medical University, Urumqi Xinjiang 830000, China
  • Received:2026-06-04 Online:2026-09-15 Published:2026-09-15

PDF (PC)

9

摘要： 目的 建立丙戊酸钠（Valproic Acid,VPA）血药浓度超警戒值的列线图预测模型,筛选VPA超警戒浓度的独立影响因素,为临床安全使用VPA提供参考。方法 将患者VPA血药浓度监测值分为达标组（50~100 μg·mL-1）和超警戒组（>100 μg·mL-1）。选用R4.3.0软件构建列线图模型,采用受试者工作特征（Receiver Operating Characteristic, ROC）曲线评估列线图模型的区分度,采用Hosmer-Lemeshow拟合优度检验评价预测准确度。结果 813例患者中VPA血药浓度达标组699例,超警戒组114例。Lasso回归分析得到4个最佳预测因素分别为性别、体重、BMI和给药途径。二元Logistic回归分析结果显示,体重（OR=0.985）和给药途径（OR=5.880）分别是VPA血药浓度达标范围的独立危险和保护因素。列线图预测模型预测患者VPA血药浓度超警戒值的ROC曲线下面积（AUC）为0.705（95%CI=0.654~0.757）,灵敏度为78.95%,特异度为55.94%,提示模型预测效能较好。Hosmer-Lemeshow检验结果显示,模型拟合优度较好（χ2=14.144,P=0.078）。决策曲线分析显示,在部分高风险阈值范围内,列线图模型预测VPA血药浓度是否在达标范围内具有良好的净获益。结论 体重和给药途径是VPA血药浓度超警戒值的独立危险因素,据此构建的列线图模型能有效预测患者VPA血药浓度超警戒风险程度。

关键词: 丙戊酸钠, 血药浓度, 超警戒值, 预测, 体重, 给药途径, 治疗药物监测

Abstract: Objective To establish a nomogram model to identify independent factors associated with serum VPA concentrations that exceed the safety threshold so as to provide a reference for safe clinical use of VPA. Methods The patients were divided into two groups based on serum VPA concentration: a therapeutic range group (50-100 μg·mL-1) and a group exceeding the safety threshold (>100 μg·mL-1). R4.3.0 software was used to establish a nomogram model. The receiver operating characteristic (ROC) curve was used to evaluate the discriminative ability of the model while the Hosmer-Lemeshow goodness-of-fit test was used to assess the accuracy of prediction. Results A total of 813 patients were enrolled, including 699 in the VPA therapeutic range group and 114 in the group exceeding the safety threshold. Lasso regression identified four top predictors: gender, body weight, body mass index, and administration routes. Binary logistic regression analysis found that body weight (OR=0.985) and administration routes (OR=5.880) were independent risk and protective factors, respectively, for maintaining serum VPA concentrations within the therapeutic range. The nomogram model was moderately accurate for predicting patients exceeding the safety threshold, with an ROC AUC of 0.705 (95%CI: 0.654-0.757), a sensitivity of 78.95%, and a specificity of 55.94%. The Hosmer-Lemeshow test indicated good model fit (χ2=14.144, P=0.078). Decision curve analysis showed that the nomogram model provided a good net benefit in predicting whether serum VPA concentrations were within the therapeutic range across a range of high-risk threshold probabilities. Conclusion Body weight and administration routes are independent risk factors associated with serum VPA concentrations exceeding the safety threshold. The nomogram model can effectively estimate the risk of elevated serum VPA concentrations.

Key words: Valproic Acid (VPA), Serum Drug Concentration, Exceed the Safety Threshold, Prediction, Body Weight, Administration Route, Therapeutic Drug Monitoring (TDM)

中图分类号: