摘要
目的建立万古霉素血药浓度预测的定量评价模型,探讨预测结果与疗效和不良反应之间的相关性,指导临床万古霉素个体化用药。方法基于医院信息系统(HIS)的真实世界数据,收集某三甲医院2017年1月1日至2018年12月31日期间开展万古霉素血药浓度监测的住院患者病例信息。对纳入的影响因素进行单因素和多因素Logistic回归分析,筛选影响万古霉素血药浓度的预测因素,构建列线图并进行内部验证。最后利用收集的数据根据建立的列线图模型进行评分,将得分从低到高排序,分为5组,每组35例患者,探讨不同组别的风险分布情况及与疗效和不良反应的相关性。结果共136例患者(175个血药浓度监测值)纳入本次研究。单因素分析和多因素Logistic回归分析后,患者的年龄、感染类型、体温、谷氨酰转肽酶、血清肌酐、美罗培南、溶媒和治疗方案合理性等8个因素差异有统计学意义(P<0.05),纳入模型构建。模型验证中校准曲线显示拟合度较好,敏感度为92.1%,特异度为78.1%,ROC曲线下面积为0.9195(95%CI 0.8741-0.9649)。不同评分组别之间的风险概率值、疗效和不良反应之间差异有统计学意义(P<0.05)。结论构建的列线图可直观、高效地预测患者使用万古霉素后血药浓度异常风险程度,同时根据列线图评分可大致推断患者的疗效和不良反应发生情况,可为临床万古霉素的个体化用药提供指导。
Objective To establish a quantitative evaluation model for the prediction of vancomycin blood concentration,explore the correlation between the prediction results and the efficacy and the adverse reactions,and guide the clinical personalized vancomycin medication.Methods Based on the real-world data of the Hospital Information System(HIS),the case information of inpatients with vancomycin blood concentration monitoring were collected in a tertiary hospital from January 1,2017 to December 31,2018.Single-factor analysis and multi-factor Logistic regression analysis were performed on the included influencing factors to screen the predictive factors affecting the blood concentration of vancomycin and construct a nomogram for the internal verification.Finally,the collected data were scored according to the established nomogram model,and the scores were sorted from low to high and divided into 5 groups with 35 patients in each group.The risk distribution of different groups and its correlation with efficacy and adverse reactions were discussed.Results A total of 136 patients(175 blood drug concentration monitoring values)were included in this study.After univariate analysis and multivariate Logistic regression analysis,8 significant factors(P<0.05)were included in model construction,including the patient’s age,the infection type,the body temperature,the levels of glutamyl transpeptidase,serum creatinine and merope-nem,the vehicle,and the rationality of the treatment plan.In model verification,the calibration curve showed a good fit,with a sensitivity of 92.1%,a specificity of 78.1%,and an area under the ROC curve of 0.9195(95%CI 0.8741-0.9649).There was significant difference in the risk probability value,the curative effect and the incidence of adverse reactions between different scoring groups(P<0.05).Conclusion The constructed nomogram can intuitively and efficiently predict the risk degree of abnormal blood concentration of vancomycin,and roughly infer the curative effect and adverse reactions of patients according to the score of nomogram,which provides a guidance for the clinical individualized use of vancomycin.
作者
陈海琴
石庆平
孔令提
桑冉
朱裕林
朱锦秀
刘哲
CHEN Haiqin;SHI Qingping;KONG Lingti;SANG Ran;ZHU Yulin;ZHU Jinxiu;LIU Zhe(Department of Pharmacy,First Affiliated Hospital of Bengbu Medical College,Bengbu 233000,China;Faculty of Pharmacy,Bengbu Medical College)
出处
《山西医科大学学报》
CAS
2020年第12期1397-1406,共10页
Journal of Shanxi Medical University
基金
安徽省教育厅高等教育振兴计划人才项目[皖教秘人[2014]181号]
北京医卫健康基金资助项目(YWJKJJHKYJJ-B183073)
安徽省重点研究与开发计划项目(202004j07020008)
蚌埠医学院研究生科研创新计划项目(Byycx1960)。
关键词
万古霉素
血药浓度
列线图
疗效
不良反应
预测模型
vancomycin
blood concentration
nomogram
efficacy
adverse reactions
predictive model
