Background:Vancomycin is commonly used for nosocomial bacterial pathogens causing late-onset septicemia in preterm infants. We prospectively collected pharmacokinetic data aiming to describe PK and determine covariates contributing to the variability in neonatal vancomycin pharmacokinetics. Further, we aimed to use the model to compare AUC24h,SS/MIC of several intermittent and continuous dosing regimens. Methods: Newborns receiving vancomycin for suspected or confirmed late-onset sepsis were included. Peak and trough concentrations for intermittent vancomycin dosing and steady-state concentrations for continuous vancomycin dosing were measured. NONMEM 7.3 was used for population pharmacokinetic analysis. Monte Carlo simulations were performed to compare dosing schemes. Results: Data from 54 infants were used for model development and from 34 infants for the model evaluation of a median (range) 29 (23.4-41.9) weeks and 28 (23.4-41.7) weeks corrected gestational age (GA), respectively. The final model was a 1-compartment model. Weight and postmenstrual age were included a priori; and after that no additional covariate significantly improved the model fit. Final model parameter estimates (mean (standard error)): CL 5.7 (0.3) L/h/70kg, V 39.3 (3.7) L/70kg. Visual predictive check of the evaluation dataset confirmed the model can predict external data. Simulations using MIC of 1 mg/Lshowed that for neonates with GA ≤25 weeks and postnatal age ≤2 weeks AUC24h,SS/MIC was lower with the intermittent regimen (median 482 versus 663). Conclusions:A population PK model for continuous and intermittent vancomycin administration in infants was developed. Continuous administration might be favourable for treating infections caused by resistant microorganisms in very young and immature infants.

BackgroundVancomycin is a commonly used antimicrobial to treat hospital acquired infections in neonates. Although both intermittent and continuous dosing regimens are used, there is limited data on population pharmacokinetics (PK) for continuous vancomycin usage to inform correct dosing.AimTo determine the population PK parameters of vancomycin in neonatal patients with a wide range of gestational ages and birth weights receiving different dosing regimens.MethodsData were collected prospectively from 56 newborns who were receiving vancomycin for late onset sepsis (32 on continuous infusion and 24 on intermittent dosage) following approval from Barts Health Clinical Effectiveness Unit. Peak and trough vancomycin concentrations were collected from infants on intermittent dosage, and random levels for continuous infusion. An enzymatic assay on the COBAS 702 platform was used to measure vancomycin (linear range 1.7–80 μg/ml). Population PK analysis was performed by simultaneously modelling both intermittent and continuous infusion data using nonlinear mixed-effects modelling (NONMEM 7.3).ResultsThere were 183 vancomycin samples available for analysis (n = 81 from the intermittent group, and n = 102 from infants on continuous infusion). The median (range) postnatal age at baseline was 26 (1–156) days; and gestational age 29 (23.7–41.9) weeks. The final model that provided the best fit to the data was a 1-compartment model. Allometric weight scaling and postmenstrual age (PMA) driven sigmoidal maturation function were included a priori and no further covariate provided a significant improvement in the model fit. The model was internally evaluated using basic diagnostic plots and a visual predictive check, which indicated that the model is able to describe the data and had good predictive power. The final parameter estimates (mean (relative standard error)) of clearance (CL) and volume of distribution (V) were 5.6 L/h/70 kg (5.3%) and 40.0 L/70 kg (9.3%), respectively. For a typical infant from the studied population (weight=1.7 kg, PMA=35.7 weeks), CL was 0.10 L/h, and V was 0.97 L.ConclusionsA population PK model was developed for both intermittent and continuous vancomycin dosage newborns and was shown to have good descriptive and predictive properties. This model will be used to develop a new dosing scheme which will then be prospectively evaluated.