JMIR Publications

ABSTRACT

Background:

Adherence to antiseizure medications (ASMs) constitutes a cornerstone of effective epilepsy management. However, current consensus guidelines for assessing medication adherence via therapeutic drug monitoring (TDM) may neglect individual patient characteristics, thereby compromising the accuracy of adherence assessment.

Objective:

Therefore, this study proposes an innovative Bayesian-based pharmacokinetic framework integrated with TDM to address these limitations, with a focus on 14 widely prescribed ASMs, including brivaracetam, carbamazepine, clobazam, eslicarbazepine acetate, lacosamide, lamotrigine, levetiracetam, oxcarbazepine, perapamel, phenobarbital, topiramate, valproic acid, vigabatrin and zonisamide.

Methods:

A comprehensive clinical trial simulation was conducted to investigate the pharmacokinetics of ASMs in patients with ASMs monotherapy under conditions of full adherence versus various non-adherent dosing behaviors, including omission of the last dose and consecutive missed doses before sampling. Bayesian posterior probabilities of these dosing behaviors were derived by integrating validated population pharmacokinetic (PPK) models, individual patient demographics (e.g., age, weight, creatinine clearance), dosing history, prior probability and TDM measurements. Additionally, the influence of patient characteristics, concomitant medications, sampling times, and prior adherence probability on model outcomes was systematically evaluated.

Results:

The Bayesian-based pharmacokinetic approach demonstrated robust discriminative capacity for adherence assessment, achieving exact retrodiction of recent dosing events (1-2 preceding doses) across all 14 ASMs and partial retrodiction of extended non-adherence trajectories for six ASMs. Concentration thresholds for adherence classification were found to vary significantly between drugs and were influenced by patient-specific factors, concurrent medications, sampling time, dosing intervals, and prior full adherence probability. To effectively generate posterior probability outputs based on above information, an adaptable dashboard (https://mipd.shinyapps.io/adherence_ASM/) was engineered in Rshiny flatform to realize these insights, which could enable real-time precise assessments of medication adherence through user-driven adjustments.

Conclusions:

This study introduces an innovative clinical framework to bridge the gap between standardized TDM and medication adherence assessment by integrating Bayesian theory and population pharmacokinetic modeling. The accompanying open-access dashboard could translate complex pharmacokinetic principles into practical solutions, enabling quantitatively individualized evaluation of medication-taking behaviors. These findings demonstrate the feasibility of shifting ASMs management paradigms from population-based to patient-specific adherence profiling.