Escitalopram is widely used to treat major depressive disorder, yet drug exposure varies substantially across patients, causing the inadequate response or toxicity probability. Individualized dosing supported by therapeutic drug monitoring and published population pharmacokinetic models is promising, but independent evaluation of existing models and clinically usable decision-support tools remains limited. We systematically reviewed published population pharmacokinetic models for escitalopram and extracted key patient characteristics and model parameters. The models were evaluated in an independent real-world dataset from Chinese psychiatric patients using simulation-based or prediction-based metrics. The best-performing model was implemented in a web-based clinical decision-support tool. Ten published models were identified and evaluated using data from 309 Chinese patients, contributing 421 plasma concentrations. A priori predictions consistently underestimated observed concentrations, with median absolute prediction errors ranging from 37. 97-64. 62%. In contrast, Bayesian updating using TDM data markedly improved both accuracy and precision, reducing most of median absolute prediction errors to <30%. The best-performing model was implemented in an openly accessible Shiny application to support initial dose selection, TDM-guided dose individualization, and management of missed or delayed doses with remedial dosing recommendations (https: //escitalopram-liux-v1. shinyapps. io/EscitalopramMIPDTool/). This study provides a comprehensive external evaluation of escitalopram population pharmacokinetic models in a Chinese psychiatric cohort and presents a freely accessible, clinically oriented precision dosing tool to support individualized escitalopram therapy for Chinese patients.
Liu et al. (Wed,) studied this question.