1.Construction of a risk prediction model for high plasma concentra-tion of voriconazole
Juxiang ZHOU ; Yanfei LI ; Fangjun LV ; Daitian LI ; Jihong ZHANG ; Jichu WU
Chinese Journal of Clinical Pharmacology and Therapeutics 2024;29(6):653-660
AIM:To develop and validate a predic-tive model for the risk of high plasma concentra-tion of voriconazole,and to guide clinical individual-ized medication of voriconazole.METHODS:Based on the real-world data from the hospital Informa-tion system(HIS),the clinical data of hospitalized patients who received voriconazole treatment and underwent voriconazole plasma concentration monitoring in our hospital from August 2017 to Au-gust 2021 were collected.Univariate and multivari-ate logistic regression analysis were performed on the included influencing factors.At the same time,in order to minimize the potential collinearity and overfitting between variables,the least absolute shrinkage and selection operator regression were used to screen the potential predictors.Logistic re-gression analysis was used to construct a predic-tion model for the risk of high plasma concentra-tion of voriconazole.C-index,calibration chart and clinical decision curve analysis were used to evalu-ate the discrimination,consistency and clinical ap-plicability of the model,and a nomogram was drawn.RESULTS:A total of 147 patients were en-rolled in this study.Plasma albumin and procalcito-nin were selected as predictive variables for Logis-tic regression analysis,and the prediction model was established.Draw predict voriconazole nomo-gram risk blood drug concentration on the high side.The receiver operating characteristic curve showed that the AUC of the prediction model for predicting the risk of high plasma concentration of voriconazole was 0.787(95%CI 0.663-0.911).Vori-conazole blood drug concentration was high inci-dence of cut-off value was 33.06%,sensitivity was 63.64%,87.65%and 58.33%positive predictive val-ue,negative predictive value of 89.87%.The cali-bration curve showed good consistency,and the clinical decision curve showed that the model had a positive net benefit when the threshold probabili-ty was between 6.67%and 99.99%.CONCLUSION:The predictive model for the risk of high plasma concentration of voriconazole has good predictive efficacy,which can provide guidance for clinical in-dividualized medication of voriconazole.