Abstract
OBJECTIVE: This study aims to evaluate the predictive performance of the published population pharmacokinetic (PopPK) model of escitalopram in Chinese patients using an external validation method. METHODS: PubMed, Embase, and Web of Science databases were searched to identify PopPK models. Clinical data collected from Chinese patients treated with escitalopram were used to evaluate these models. The predictive performance of the models was evaluated using both prediction-based diagnostic methods and simulation-based diagnostic methods. RESULTS: Ten published PopPK models were included in the external validation. A total of 241 serum concentration samples were collected from 193 Chinese patient. Among all evaluated models, the Poweleit 2023 model exhibited the optimal predictive performance, with the PE of -2.14% and the RMSE of 22.27% at the individual level, and corresponding values of 14.13% and 104.19% at the population level, followed the model by Liu 2022. While the predictive performance of the other models was unsatisfactory. CONCLUSION: Published PopPK models of escitalopram showed wide variations in predictive performance among our patient cohort. External models should be accurately evaluated before their application in clinical practice.