Abstracts

Rationale: Sudden unexpected death in epilepsy patients (SUDEP) is an increasingly concerning cause of death among patients suffering from epilepsy. Though great inroads have been made in searching for mechanisms and risk factors associated with SUDEP, many of these studies have been done on adult patients, with considerably less research on pediatric SUDEP patients.  Many risk factors have been implicated in contributing to SUDEP, though only a few of these have been consistently shown in the pediatric population. One such risk factor, decreased heart rate variability, has been correlated with a higher SUDEP-7 risk score but has not yet been studied in actual SUDEP cases. The purpose of this study is to develop a predictive model for at-risk pediatric patients through a matched case-control study of the pediatric SUDEP cases at our tertiary epilepsy center. Methods: Using the SUDEP criterion of Nashef et al. (1, 2), we identified 11 SUDEP cases occurring between 2007 to 2017 from the Medical College of Georgia and 53 matched controls, each case matched for age, epilepsy duration, and gender.  Using a conditional logistic regression model, we evaluated 9 predictor variables: mental retardation, seizure frequency, seizure type, prior status epilepticus, number of antiepileptic drugs (AEDs), prior epilepsy surgery, vagus nerve stimulator (VNS) therapy, seizure progression and awake interictal heart rate variability.  For each predictor variable, we determined the odds ratio (OR) with 95% confidence intervals (95% CI), and log likelihood ratio test p-value. We measured heart rate variability using root-mean square differences of successive R-R intervals (RMSSD).  We identified the optimum predictor models using Akaike's information criterion (AIC) and evaluated model performance using receiver operating characteristic (ROC) area under the curve (AUC). Results: Prior status epilepticus (OR 7.83, 95% CI 1.91-32.16, p = 0.0043), prior epilepsy surgery (OR 4.23, 95% CI 1.23-14.54, p = 0.022), and higher number of AEDs (OR 4.7, 95% CI 1.61-13.71, p = 0.0046) were significant predictors for SUDEP.  Heart rate variability (RMSSD) was not a significant predictor (OR 0.984, 95% CI 0.96- 1.01, p = 0.17).   Seizure frequency, seizure type, VNS therapy, and mental retardation were also not found to be significant.  Using AIC criteria to select the best 3 models, the best model used number of AEDs and prior epilepsy surgery (AUC 0.855).  The second-best model used number of AEDs and prior status epilepticus (AUC 0.839).  The third-best model used a single predictor variable, number of AEDs alone (AUC 0.807). Conclusions: These findings suggest that higher number of AEDs used and prior epilepsy surgery may have a higher risk of SUDEP in the pediatric epilepsy population.  1. Epilepsia. 1997 Nov;38(11 Suppl):S6-82. Epilepsia. 2012 Feb;53(2):227-33 Funding: None