RESUMO
BACKGROUND AND OBJECTIVE: For neonates and infants receiving intermittent vancomycin infusions, the area under the concentration-time curve during 24 h (AUC24) is often estimated with Bayesian forecasting using one or more measured vancomycin concentrations. When practical peak and trough concentrations are measured at steady state, AUC24 can also be calculated with first-order steady-state equations for a one-compartment model (Sawchuk-Zaske method), but previously this method has been applied only for adults. The objective of this study was to compare AUC24 values obtained with the Sawchuk-Zaske method and two Bayesian models. METHODS: AUC24 values were estimated retrospectively for 18 neonates and infants with steady-state peak and trough concentrations using traditional compartmental analysis with a one-compartment model (reference method), the Sawchuk-Zaske method, and Bayesian forecasting with two previously published models. In Bayesian forecasting, both original and modified residual error models were used. In the modified models, the residual error was reduced by setting the additive residual error to zero and the proportional error to 15%. RESULTS: AUC24 estimates obtained with the Sawchuk-Zaske method differed - 2.7 to 0.9% from the reference method. When both peak and trough concentrations were used in Bayesian forecasting, 61% and 33% of AUC24 estimates obtained with two original models differed less than 15% from the reference method, and these fractions increased to 83% and 72% with the modified models, respectively. CONCLUSION: When practical peak and trough concentrations are measured at steady state, the simple Sawchuk-Zaske method is very useful for AUC24 estimation in neonates and infants. In Bayesian forecasting, the reduced residual error model can be used to improve the model fit.