Abstract
OBJECTIVES: The aim of this study was to evaluate the impact of switching population pharmacokinetic models used for model-informed precision dosing (MIPD) of vancomycin on target attainment, acute kidney injury rates, dosing, and sampling burden in 90 295 patients across nineteen hospital systems. METHODS: We assessed sites that changed their default vancomycin MIPD model on primary endpoints, including vancomycin area under the concentration-time curve (AUC) target attainment, time to target attainment, and vancomycin-associated AKI rates using an interrupted time series model. Secondary endpoints included vancomycin dosing (day 1 and total course) and sampling frequency. Hospital systems that did not change models served as controls. RESULTS: In BMI < 40 kg/m² patients, the Goti-to-Thomson model transition significantly improved AUC target attainment on day 1 (+2.1%, P < 0.001), day 2 (+3.1%, P < 0.001), and across the whole treatment course (+2.7%, P < 0.001). Small increases occurred in vancomycin sampling per course (+0.06 samples/course, P < 0.05) and time to target attainment (+0.26 hours, P < 0.05). AUC improvements occurred without statistically significant increases in AKI incidence (KDIGO Stage 1: +0.314%, P > 0.05/3; Stage 2: +0.347%, P > 0.05/3, Stage 3: +0.034%, P > 0.05/3). No significant differences were observed in the BMI ≥ 40 kg/m(2) or control cohorts. CONCLUSIONS: This large quasi-experimental study shows that switching from one pair of well-validated and highly predictive models to another improves vancomycin target attainment while maintaining low adverse event rates.