Vancomycin Area under the Concentration-Time Curve Estimation Using Bayesian Modeling versus First-Order Pharmacokinetic Equations: A Quasi-Experimental Study

Abstract

Aim: To evaluate the efficiency of Bayesian modeling software and first-order pharmacokinetic (PK) equations to calculate vancomycin area under the concentration-time curve (AUC) estimations. Methods: Unblinded, crossover, quasi-experimental study at a tertiary care hospital for patients receiving intravenous vancomycin. Vancomycin AUC monitoring was compared using Bayesian modeling software or first-order PK equations. The primary endpoint was the time taken to estimate the AUC and determine regimen adjustments. Secondary endpoints included the percentage of vancomycin concentrations usable for AUC calculations and acute kidney injury (AKI). Results: Of the 124 patients screened, 34 patients had usable vancomycin concentrations that led to 44 AUC estimations. Without electronic health record (EHR) integration, the time from assessment to intervention in the Bayesian modeling platform was a median of 9.3 min (quartiles Q1–Q3 7.8–12.4) compared to 6.8 min (Q1–Q3 4.8–8.0) in the PK equations group (p = 0.004). With simulated Bayesian software integration into the EHR, however, the median time was 3.8 min (Q1–Q3 2.3–6.9, p = 0.019). Vancomycin concentrations were usable in 88.2% in the Bayesian group compared to 48.3% in the PK equation group and there were no cases of AKI. Conclusion: Without EHR integration, Bayesian software was more time-consuming to assess vancomycin dosing than PK equations. With simulated integration, however, Bayesian software was more time efficient. In addition, vancomycin concentrations were more likely to be usable for calculations in the Bayesian group.

Keywords

• vancomycin
• AUC
• Bayesian
• therapeutic drug monitoring