Rashmi V Shingde1,2, Garry G Graham1,3, Stephanie E Reuter4, Jane E Carland1,2, Richard O Day1,2,3, Sophie L Stocker1,2,3. 1. Department of Clinical Pharmacology & Toxicology, St Vincent's Hospital, Darlinghurst. 2. St Vincent's Clinical School, University of New South Wales, Kensington. 3. School of Medical Science, University of New South Wales, Kensington, New South Wales. 4. School of Pharmacy & Medical Sciences, University of South Australia, Adelaide, SA, Australia.
Abstract
BACKGROUND: Vancomycin pharmacokinetics are best described using a 2-compartment model. However, 1-compartment population models are commonly used as the basis for dose prediction software. Therefore, the validity of using a 1-compartment model to guide vancomycin drug dosing was examined. METHODS: Published plasma concentration-time data from adult subjects (n = 30) with stable renal function administered a single intravenous infusion of vancomycin were extracted from previous studies. The vancomycin area under the curve (AUC0-∞) was calculated for each subject using noncompartmental methods (AUCNCA) and by fitting 1- (AUC1CMT), 2- (AUC2CMT), and 3- (AUC3CMT) compartment infusion models. The optimal model fit was determined using the Akaike information criterion and visual inspection of the residual plots. The individual compartmental AUC0-∞ values from the 1- and 2-compartment models were compared with AUCNCA values using one-way repeated measures analysis of variance. RESULTS: The mean (±SD) AUC estimates were similar for the different methods: AUCNCA 180 ± 86 mg·h/L, AUC1CMT 167 ± 79 mg·h/L, and AUC2CMT 183 ± 88 mg·h/L. Despite the overlapping AUC values, AUC2CMT and AUCNCA were significantly greater than AUC1CMT (P < 0.05). The 3-compartment model was excluded from the analysis because of the failure to converge in some instances. CONCLUSIONS: Dose prediction software using a 1-compartment model as the basis for Bayesian forecasting underestimates drug exposure (estimated as the AUC) by less than 10%. This is unlikely to be clinically significant with respect to dose adjustment. Therefore, a 1-compartment model may be sufficient to guide vancomycin dosing in adult patients with stable renal function.
BACKGROUND:Vancomycin pharmacokinetics are best described using a 2-compartment model. However, 1-compartment population models are commonly used as the basis for dose prediction software. Therefore, the validity of using a 1-compartment model to guide vancomycin drug dosing was examined. METHODS: Published plasma concentration-time data from adult subjects (n = 30) with stable renal function administered a single intravenous infusion of vancomycin were extracted from previous studies. The vancomycin area under the curve (AUC0-∞) was calculated for each subject using noncompartmental methods (AUCNCA) and by fitting 1- (AUC1CMT), 2- (AUC2CMT), and 3- (AUC3CMT) compartment infusion models. The optimal model fit was determined using the Akaike information criterion and visual inspection of the residual plots. The individual compartmental AUC0-∞ values from the 1- and 2-compartment models were compared with AUCNCA values using one-way repeated measures analysis of variance. RESULTS: The mean (±SD) AUC estimates were similar for the different methods: AUCNCA 180 ± 86 mg·h/L, AUC1CMT 167 ± 79 mg·h/L, and AUC2CMT 183 ± 88 mg·h/L. Despite the overlapping AUC values, AUC2CMT and AUCNCA were significantly greater than AUC1CMT (P < 0.05). The 3-compartment model was excluded from the analysis because of the failure to converge in some instances. CONCLUSIONS: Dose prediction software using a 1-compartment model as the basis for Bayesian forecasting underestimates drug exposure (estimated as the AUC) by less than 10%. This is unlikely to be clinically significant with respect to dose adjustment. Therefore, a 1-compartment model may be sufficient to guide vancomycin dosing in adult patients with stable renal function.
Authors: Rachel Constance Yager; Natalie Taylor; Sophie Lena Stocker; Richard Osborne Day; Melissa Therese Baysari; Jane Ellen Carland Journal: BMC Health Serv Res Date: 2022-04-18 Impact factor: 2.908