欢迎访问《中国临床药理学与治疗学》杂志官方网站,今天是 分享到:

中国临床药理学与治疗学 ›› 2017, Vol. 22 ›› Issue (8): 904-909.

• 定量药理学 • 上一篇    下一篇

基于PLSR算法的CR-KP检出率与抗菌药物使用相关性研究

蒋 程,李芳琼,叶佐武,孙云峰,张美玲,李功华   

  1. 浙江省立同德医院,杭州 310012,浙江
  • 收稿日期:2016-11-03 修回日期:2017-01-03 出版日期:2017-08-26 发布日期:2017-08-18
  • 通讯作者: 李功华,男,本科,副主任药师,研究方向:医院药学。 Tel:0571-89972239 E-mail:ligonghua88@163.com
  • 作者简介:蒋程,女,博士,主管药师,研究方向:临床药学。 Tel:0571-89972240 E-mail:jiangcheng0818@126.com
  • 基金资助:

    浙江省公益技术应用研究计划项目(2016C33127);浙江省中医药科技计划项目(2015ZQ007);浙江省药学会医院药学专项科研资助项目(2014ZYY10)

Correlation analysis between the detection rate of CR-KP and the utilization of antibacterial drugs based on PLSR algorithm

JIANG Cheng, LI Fangqiong, YE Zuowu, SUN Yunfeng, ZHANG Meiling, LI Gonghua   

  1. Tongde Hospital of Zhejiang Province, Hangzhou 310012, Zhejiang, China
  • Received:2016-11-03 Revised:2017-01-03 Online:2017-08-26 Published:2017-08-18

摘要:

目的: 采用偏最小二乘回归(PLSR)算法分析碳青霉烯耐药肺炎克雷伯菌(CR-KP)检出率与抗菌药物使用的相关性,为抗菌药物临床的合理使用提供指导。方法: 以不同季度抗菌药物的用药频度(DDDs)为自变量,以同期及滞后不同时期(1~4个季度)CR-KP的检出率为因变量,分别建立PLSR模型,分析CR-KP检出率比抗菌药物使用的滞后性。根据各变量的回归系数,筛选与CR-KP检出率相关性较显著的抗菌药物,明确不同品种对CR-KP检出率的影响。结果: CR-KP的检出率滞后于抗菌药物的使用两个季度,与阿莫西林-克拉维酸、磺苄西林的DDDs正相关性较显著,与伊曲康唑的DDDs负相关性较显著。阿莫西林-克拉维酸、伊曲康唑、磺苄西林的回归系数分别为0.085 2、-0.083 9、0.076 3。结论: PLSR算法可用于同时分析CR-KP检出率与多种抗菌药物使用的相关性,可为CR-KP检出率控制提供依据。

关键词: 偏最小二乘回归, 碳青霉烯耐药肺炎克雷伯菌, 耐药性, 抗菌药物, 用药频度

Abstract:

AIM: To provide guidance for the reasonable clinical usage of antibacterial drugs by employing the partial least squares regression (PLSR) algorithm to analyze the correlation between the detection rate of carbapenem-resistant Klebsiella pneumoniae (CR-KP) and the utilization of antibacterial drugs.  METHODS: Different PLSR models were established with the defined daily doses (DDDs) of antibacterial drugs in different quarters as independent variables and the detection rate of CR-KP with same and different periods (lagged 1-4 quarters) as dependent variables. The lag time between the detection rate of CR-KP and the utilization of antibacterial drugs was investigated. The antibacterial drugs, which were highly correlated to the detection rate of CR-KP were screened. The influences of the antibacterial drugs on the detection rate of CR-KP were then investigated according to the regression coefficients. RESULTS:The detection rate of CR-KP lagged about two quarters behind the utilization of antibacterial drugs. It was significantly positively correlated with the DDDs of amoxicillin-clavulanic acid and sulbenicillin, while negatively correlated with the DDDs of itraconazole. The regression coefficients of amoxicillin-clavulanic acid, itraconazole and sulbenicillin were 0.085 2, -0.083 9 and 0.076 3, respectively. CONCLUSION: It is effective to employ PLSR algorithm to simultaneously analyze the correlation between the detection rate of CR-KP and the utilization of multiple antibacterial drugs, which provides a new scientific tool for controlling the detection rate of CR-KP.

Key words: partial least squares regression, carbapenem-resistant Klebsiella pneumoniae, drug resistance, antibacterial drugs, defined daily doses

中图分类号: