Adherence patterns in antiseizure medications influencing risk of sudden unexplained death in epilepsy: A data linkage study using dispensed prescriptions.

Objective: Medication adherence is considered an important risk factor for sudden unexpected death in epilepsy (SUDEP), although measurement accuracy is difficult. Using prescription dispensations, this study aims to estimate antiseizure medication (ASM) adherence and identify adherence patterns that influence epilepsy mortality.

Methods: This is a retrospective cohort study of tertiary epilepsy outpatients seen at St Vincent's Hospital Melbourne, Victoria, Australia, from January 1, 2012 until December 31, 2017. Privacy-preserving data linkage with the Australian national prescription, death, and coroner's databases was performed. We fitted a four-cluster longitudinal group-based trajectory model for ASM adherence from recurring 90-day windows of prescription dispensations during a 3-year "landmark period" from January 1, 2012 to December 31, 2014, using the AdhereR package. We estimated the risk of SUDEP and all-cause death for each adherence pattern during an "observation period" from January 1, 2015 to December 31, 2017. The Cox proportional hazards and logistic regression models were adjusted for age, sex, socioeconomic status, epilepsy duration, comorbidity, drug resistance, and inadequate seizure control.

Results: One thousand one hundred eighty-seven participants were observed for a median of 3.2 years (interquartile range = 2.4-4.0 years). We observed 66 deaths with 10 SUDEP cases during the observation period. We identified four patterns of ASM adherence: good, 51%; declining, 24%; poor, 16%; and very poor, 9%. Declining adherence was associated with an increased risk for SUDEP, with hazard ratio (HR) = 8.43 (95% confidence interval [CI] = 1.10-64.45) at 1 year and HR = 9.17 (95% CI = 1.16-72.21) at 3 years. Compared to no ASM therapeutic change, the addition of a second to fourth ASM offered increased protection against SUDEP in patients with continuing drug-resistant epilepsy.

Significance: ASM nonadherence was observed in half of outpatients with epilepsy. A declining pattern of adherence, observed in a quarter of patients, was associated with more than eight times increased risk of SUDEP. Any ongoing medication interventions must include strategies to maintain and improve ASM adherence if we are to reduce the risk of SUDEP.