RESUMO
OBJECTIVE: To compare all published sodium valproate population pharmacokinetic models of epileptic children in China and assess them by external validation to determine their predictive performance. METHODS: The published population pharmacokinetic model of sodium valproate in children with epilepsy in China was collected by database retrieval. By retrospectively collecting patients' information, we performed an external validation to evaluate the power of prediction. RESULTS: Four sodium valproate models were published before. The external validation of 101 samples showed that the MPE, MAE, RMSE of the four model were similar. All of them showed good adequacy between predicted concentrations and observed concentrations. Comparing with other models, the model established by Ding has better prediction error coincidence rate in most interval, which is more targeted for the prediction of individualized dosage regimen for epileptic children in our hospital. However, the overall prediction accuracy of the four models is not significantly different. CONCLUSION: By the evaluation of four published population pharmacokinetic models of sodium valproate model B performs better than others, while the overall accuracy of the four models did not differ much. Racial difference may be an important factor affecting the accuracy of the model, which needs to be further explored in subsequent studies.