摘要
目的:验证基于万古霉素群体药动学模型的个体化给药决策支持系统SmartDose对万古霉素血药稳态谷浓度(C ss,trough)的预测能力,评估该工具用于辅助临床治疗的可行性。方法:收集在2018年1月—2019年10月就诊于广州市某三甲医院且接受过万古霉素血药浓度监测患者的年龄、性别、体重、万古霉素给药方案、肌酐清除率及万古霉素稳态谷浓度实测值,应用SmartDose系统预测万古霉素稳态谷浓度,应用Spearman相关分析统计稳态谷浓度预测值和实测值的相关性,采用平均预测误差(MPE)和平均绝对误差(MAE)评估预测效能的准确度和精密度。结果:纳入分析的患者共有172例、万古霉素谷浓度实测值242个,实测C ss,trough为(18.84±12.28)μg/ml;应用SmartDose计算得240个谷浓度预测值,预测C ss,trough为(14.68±9.25)μg/ml,实测值与预测值呈显著正相关(ρ=0.632,P<0.01)。在预测准确度和精密度方面,基于已有C ss,trough监测结果的调整预测值大幅优于初始预测值(MPE分别为-1.29和-5.25,MAE分别为2.38和7.54),65岁以上患者优于18~65岁患者,男性患者优于女性患者。结论:在已获得万古霉素稳态谷浓度监测结果的基础上,SmartDose决策支持系统对下一次C ss,trough的预测能力较好,可用于辅助临床决策,不建议直接应用该系统进行初始稳态谷浓度的预测。
Objective:This study aim to verify the predictive capability of SmartDose,an individualized drug delivery decision support system based on the population pharmacokinetic model of vancomycin.Methods:172 patients from a tertiary hospital in Guangzhou from January 2018 to October 2019 were included.Age,gender,body weight,vancomycin administration regimen,creatinine clearance rate and measured steady-state trough concentration(C ss,trough)of vancomycin were collected.Predicted Css,trough of vancomycin was calculated by SmartDose system.Spearman correlation analysis was used to calculate the correlation between the predicted value and the measured value.Mean prediction error(MPE)and mean absolute error(MAE)were used to evaluate the accuracy and the precision of prediction efficiency.Results:172 patients with 242 vancomycin trough concentration were included in the analysis,and the measured C ss,trough was(18.84±12.28)μg/ml;the predicted C ss,trough was(14.68±9.25)μg/ml.There was a significant positive correlation between the measured and the predicted C ss,trough(ρ=0.632,P<0.01).In terms of prediction accuracy and precision,the adjusted predictive value was better than the initial predictive value(the MPE were-1.29 and-5.25,respectively,and the MAE were 2.38 and 7.54,respectively),the prediction in patients over 65 years old was better than 18~65 years old patients,and the prediction in male patients were better than female patients.Conclusion:SmartDose decision support system had good prediction capability for the next C ss,trough,based on the measured C ss,trough.It is not recommended to directly use the system to predict the initial C ss,trough.
作者
赵丽岩
吴海燕
陈仰青
ZHAO Liyan;WU Haiyan;CHEN Yangqing(Department of Pharmacy,the First Affiliated Hospital,Sun Yat-Sen University,Guangzhou City,Guangdong Province 510080)
出处
《医学理论与实践》
2021年第22期3867-3869,共3页
The Journal of Medical Theory and Practice
