Prospective Study to Optimize Vancomycin Dosing in Children and Adults Using Multiple-Model Bayesian Adaptive Control

Recent guidelines to use the antibiotic vancomycin for serious, resistant gram-positive bacterial infections advocate higher plasma concentrations than are routinely achieved with conventional dosing. Moreover, there is wide interpatient variability in vancomycin plasma concentrations, even with standardized dosing. The hypothesis for this study is that dosing vancomycin assisted by computer software and Bayesian algorithms will lead to more rapid and accurate attainment of therapeutic blood vancomycin concentrations in children and adults. This study will enroll 90 patients per year for three years, totaling 270 patients. Eligible patients will be of any age and who are to be prescribed vancomycin by their clinicians for medical indications. Patients with vancomycin-resistant organisms, severe vancomycin allergies or who need dialysis will not be eligible. Participants in the first group of 90 will be treated according to standard care. The second and third groups of patients will be dosed with vancomycin according to the recommendations made by the study team using the BestDose software developed by the USC Laboratory of Applied Pharmacokinetics. The second group will be dosed with the software in its current form, and the third group with funded updates. For all groups, no additional blood samples will be drawn for research purposes; only routinely obtained clinical data will be used. The primary outcome in all groups will be the percentage of participants with appropriate vancomycin concentrations. Secondary outcomes in those who receive vancomycin for at least 72 hours will include effectiveness, toxicity rates, and costs of therapy. Participation in the study will cease at the time of hospital discharge or 72 hours after termination of vancomycin therapy.