# Development and validation of a population pharmacokinetic model of vancomycin for patients of advanced age

## TL;DR

This study develops a more accurate model for predicting vancomycin clearance in elderly patients by incorporating serum albumin levels and creatinine clearance.

## Contribution

A new population pharmacokinetic model for vancomycin in elderly patients that includes serum albumin as a covariate.

## Key findings

- The new model showed improved predictive accuracy with lower mean absolute and squared errors compared to previous models.
- Including serum albumin levels along with creatinine clearance enhanced the model's performance for elderly patients with low serum creatinine.

## Abstract

Population pharmacokinetic (PPK) models of vancomycin (VCM) commonly use creatinine clearance (CLcr) as a covariate for clearance (CL). However, relying on CLcr in patients of advanced age may lead to inaccuracies in estimating VCM clearance. Therefore, this study aimed to develop and validate a new PPK model specifically for patients aged 75 years and older.

PPK analysis was performed based on the blood concentrations of VCM (n = 159 patients). The predictive performance of the developed model was compared with that of previous models using mean absolute error (MAE) and mean squared error (MSE) for another dataset.

The PPK analysis optimized a two-compartment model using CLcr and the Alb levels as covariates at the central compartment of VCM clearance. The final model was as follows: CL (L/h) = 1.96 × (CLcr/3.09) 0.63 × (Serum albumin (Alb) /2.3) 0.22 × exponential (0.11). Clearance between the central and peripheral compartments (L/h) = 4.86. Central compartment volume of distribution (L) = 31.78. Peripheral compartment volume of distribution (L) = 53.64. The validation study revealed that compared with those of previous models (ranging from 0.67 to 0.79 L/h and from 0.81 to 1.11 (L/h)2, respectively), the final model demonstrated the smallest MAE of 0.60 L/h and MSE of 0.65 (L/h)2 for patients of advanced age with serum creatinine levels of < 0.6 mg/dL.

The PPK model of VCM for patients of advanced age was optimized by adding the Alb levels and CLcr as covariates for CL. The predictive accuracy of the PPK model for patients with an SCr of < 0.6 mg/dL tended to be higher than those of previous models based just on CLcr. Thus, dosage is suggested to be adjusted based on CLcr and Alb levels for patients with an SCr of < 0.6 mg/dL.

The online version contains supplementary material available at 10.1186/s40780-025-00423-8.