JMIR Publications

ABSTRACT

Background:

Intravenous (IV) vancomycin is used in the treatment of severe infection in neonates. It is efficacious but also associated with elevated risks of developing acute kidney injury. The risk is even higher in neonates admitted to the neonatal intensive care unit (NICU) because the pharmacokinetics of vancomycin in neonates vary widely. Therapeutic drug monitoring has been an integral part of the management to guide individual dose adjustments based on observed serum vancomycin concentrations (Cs) to balance efficacy against toxicity. However, the existing trough-based approach shows poor evidence for improved clinical outcomes. The updated clinical practice guideline recommends population pharmacokinetic (popPK) model-based approaches, targeting area under curve preferably through the Bayesian approach. Since Bayesian methods cannot be performed manually and require specialized computer programs, there is an urgent need to provide the clinicians with a user-friendly interface to facilitate accurate, personalized dosing recommendations for vancomycin in critically ill neonates.

Objective:

To utilize medical data from electronic health records (EHRs) to develop a popPK model and subsequently a web-based interface to perform model-based approaches to individual dose optimization of IV vancomycin for NICU patients in local medical institutions.

Methods:

Data were collected from EHR sources, namely Clinical Information System, In-Patient Medication Order Entry, and electronic Patient Record for subjects prescribed IV vancomycin in the NICU of Prince of Wales Hospital and Queen Elizabeth Hospital in Hong Kong. Patient demographics, serum creatinine (SCr), vancomycin administration records and Cs were collected. The popPK model used comprises a two-compartment infusion model, and various covariate models were tested against body weight, postmenstrual age (PMA), and SCr for the best goodness-of-fit. A previously published web-based dosing interface was replicated and adapted to the needs in this study.

Results:

The final dataset consisted of EHR data extracted from 207 subjects, obtaining a total of 689 Cs measurements. The final model chosen explains 82% of the variability in vancomycin clearance. All parameter estimates are within the bootstrapping confidence intervals. Predictive plots, residual plots, and visual predictive checks demonstrate good model predictability. Model approximations show that the model-based Bayesian approach consistently promotes the probability of target attainment (PTA) above 75%, while only half of the subjects can achieve PTA over 50% with the trough-based approach. The dosing interface was developed with the capability to optimize individual doses with the model-based empirical or Bayesian approach.

Conclusions:

Utilizing EHRs, a satisfactory popPK model has been verified and used to develop the web-based individual dose optimization interface. The interface is expected to improve treatment outcomes of IV vancomycin in the treatment of severe infections among neonates in local NICUs. This study provides the foundation upon which to conduct a cohort study to demonstrate the utility of the new approach compared with previous dosing methods.