基于碳青霉烯耐药革兰阴性菌检出率与抗菌药物使用强度相关性的线性回归模型构建

目的考察碳青霉烯耐药革兰阴性菌(CRGN)检出率与抗菌药物使用强度(AUD)的相关性,建立预测CRGN检出率的回归模型。方法回顾性调查2014年第1季度至2020年第1季度我院碳青霉烯耐药大肠埃希菌(CREC)、碳青霉烯耐药肺炎克雷伯菌(CRKP)、碳青霉烯耐药铜绿假单胞菌(CRPA)和碳青霉烯耐药鲍曼不动杆菌(CRAB)检出率,PASS临床药学管理系统提取我院抗菌药物使用强度数据,应用SPSS 19.0软件分析耐碳青霉烯菌株检出率和AUD随时间变化趋势,并分析耐药菌株检出率与AUD的相关性。以检出率为因变量,AUD为自变量,对相关性显著的变量建立线性回归模型,应用模型预测2020年第2季度和第3季度CRGN检出率。结果CRKP检出率随时间呈上升趋势(P<0.05)。CRKP检出率与喹诺酮类AUD(r=0.447,P<0.05)、头孢他啶AUD(r=0.504,P<0.05)呈正相关。CRPA检出率与喹诺酮类AUD(r=0.548,P<0.01)、碳青霉烯类AUD(r=0.435,P<0.05)呈正相关。CREC检出率和CRAB检出率与抗菌药物使用强度无显著相关性(P>0.05)。基于以上相关性共建立了4个回归模型,结果显示,模型均通过显著性检验(F_(1)=5.759,P_(1)=0.025;F_(2)=5.749,P_(2)=0.025;F_(3)=5.364,P_(3)=0.030;F_(4)=9.893,P_(4)=0.004)。将2020年第2季度和第3季度抗菌药物AUD分别带入回归模型,并对GRGN检出率预测值和实际值作进行比较,结果显示,CRGN实际检出率在95%预测区间内。结论我院CRGN检出率与某些抗菌药物使用强度存在相关性,通过纳入抗菌药物AUD作为自变量建立回归模型可用于预测CRGN检出率。

Establishment of linear regression model based on correlation between detection rate of carbapenem-resistant gram-negative bacteria and antibacterial use density

Objective To investigate the correlation between detection rate of carbapenem-resistant gram-negative bacteria(CRGN)and antibacterial use density(AUD)and establish the regression models for predicting the detection rate of CRGN.Methods The detection rate of carbapenem-resistant Escherichia coli(CREC),carbapenem-resistant Klebsiella pneumoniae(CRKP),carbapenem-resistant Pseudomonas aeruginosa(CRPA)and carbapenem-resistant Acinetobacter baumannii(CRAB)from the first quarter of 2014 to the first quarter of 2020 were retrospectively investigated,and AUD were collected by PASS clinical pharmacy management system at the same time.SPSS 19.0 software was used to analyze the trend of detection rate of CRGN and AUD with time going,and the correlation between detection rate of CRGN and AUD was analyzed.Linear regression models were established for variables with significant correlation,with detection rate of CRGN as dependent variable and AUD as independent variable.The detection rate of CRGN of the second and third quarter in 2020 were forecasted by the linear regression models.Results The detection rate of CRKP had an upward tendency over time(P<0.05).The detection rate of CRKP was positively correlated with the AUD of quinolones and ceftazidime(r=0.447,P<0.05;r=0.504,P<0.05).The detection rate of CRPA was positively correlated with the AUD of quinolones and carbapenem(r=0.548,P<0.01;r=0.435,P<0.05).No significant correlation was found between AUD and the detection rate of CREC or CRAB(P>0.05).Four regression models were established based on the above correlations,all of which passed the significance test(F₁=5.759,P₁=0.025;F₂=5.749,P₂=0.025;F₃=5.364,P₃=0.030;F₄=9.893,P₄=0.004).The AUD of the second and third quarter in 2020 were entered into the regression models,and predictive values and actual values of detection rate of CRGN were compared.The results showed that actual value of detection rate of CRGN was within 95%of the prediction range.Conclusion The AUD of some antimicrobials has correlation with the detection rate of CRGN in our hospital,the detection rate of CRGN can be predicted by regression models with AUD as the independent variable.