Abstract: AIM: The population pharmacokinetic model (PKK) of linezolid was constructed with the retrospective data of linezolid therapeutic drug monitoring from the 3 hospitals of ENZE medical center, which could be used to predict individualized pharmacokinetic parameters with Bayesian feedback method based on single point trough concentration and support scientific experimental method for rational drug use of linezolid in the future. METHODS: A total of 115 monitoring serum concentration data of 72 patients from Mar. 2016. to Dec. 2018 were included in this study. Stepwise regression method was used to screen the concomitant variable (age, weight, blood routine examination, biochemical index and drug combination etc.) for Vd and K by kinetica software. The internal and external validation were analyzed with maximum likihood method and Bayesian feedback. RESULTS: As the final model Vd=25.864－0.034×Fur(mg) shown, combination use of furosemide has a significant effect on Vd of linezolid. The level of age, Scr and the burn status of the patients have a significant effect on K of linezolid and the final model was K=0.324－0.0003×Scr－0.003×age +0.04×burn. Finally, the population mean value of Vd and K were 29.719 L(5.32, 52.36), 0.160 h-1 (0.05, 0.23) and 25.322 L (2.50, 52.51), 0.193 h-1 (0.06, 0.32) in basic model and final model. The mean absolute prediction error rate of external validation was 0.620 (0.001, 4.153) in basic model and 0.588 (0.014, 3.942) in final model. CONCLUSION: The final PPK model from the present study could well response the heterogeneous PPK characteristic of the patients from ENZE medical center, which could support scientific experiment method for improving the linezolid therapeutic effect and reducing adverse reaction rate.

Key words: Kinetica, linezolid, population pharmacokinetic