Hesdorffer, Dale C.

Summary Objective Limited data are available regarding the evolution over time of the rate of sudden unexpected death in epilepsy patients (SUDEP) in drug‐resistant epilepsy. The objective is to analyze a database of 40 443 patients with epilepsy implanted with vagus nerve stimulation (VNS) therapy in the United States (from 1988 to 2012) and assess whether SUDEP rates decrease during the postimplantation follow‐up period. Methods Patient vital status was ascertained using the Centers for Disease Control and Prevention's National Death Index (NDI). An expert panel adjudicated classification of cause of deaths as SUDEP based on NDI data and available narrative descriptions of deaths. We tested the hypothesis that SUDEP rates decrease with time using the Mann‐Kendall nonparametric trend test and by comparing SUDEP rates of the first 2 years of follow‐up (years 1‐2) to longer follow‐up (years 3‐10). Results Our cohort included 277 661 person‐years of follow‐up and 3689 deaths, including 632 SUDEP. Primary analysis demonstrated a significant decrease in age‐adjusted SUDEP rate during follow‐up ( S = −27 P = .008), with rates of 2.47/1000 for years 1‐2 and 1.68/1000 for years 3‐10 (rate ratio 0.68; 95% confidence interval [CI] 0.53‐0.87; P = .002). Sensitivity analyses confirm these findings. Significance Our data suggest that SUDEP risk significantly decreases during long‐term follow‐up of patients with refractory epilepsy receiving VNS Therapy. This finding might reflect several factors, including the natural long‐term dynamic of SUDEP rate, attrition, and the impact of VNS Therapy. The role of each of these factors cannot be confirmed due to the limitations of the study.

Purpose: No previous population-based study has addressed the contribution of activation procedures to the yield of epileptiform abnormalities on serial EEGs. We assessed yield of activation-related epileptiform abnormalities and predictors of finding an activation-related abnormality with multiple EEGs in a population-based study of newly diagnosed epilepsy. Methods: We used the resources of the Rochester Epidemiology Project to identify 449 residents of Rochester, Minnesota with a diagnosis of newly diagnosed epilepsy at age 1 year or older, between 1960 and 1994, who had at least one EEG. Information on all activation procedures (i.e., sleep, hyperventilation, and photic activation) and seizure/epilepsy characteristics was obtained by comprehensive review of medical records. Results: At the first EEG, the yield of epileptiform abnormalities was greatest for individuals 1 to 19 years of age at diagnosis, for each activation procedure. The yield in patients aged 1 to 19 versus ≥20 years was 21.6% versus 10.3% for sleep, 6.5% versus 3.3% for photic stimulation, and 10.3% versus 5% for hyperventilation. Among young people (aged 1–19 years), sleep was associated with an increased likelihood of finding an activation-related abnormality on any EEG. The likelihood of finding an activation-related abnormality on any EEG was decreased for postnatal symptomatic and for unknown etiology. Conclusions: Among activation procedures, sleep showed the highest yield of epileptiform abnormalities. There was a low yield for photic stimulation and hyperventilation. Within each activation procedure, younger age at diagnosis had the greatest yield. Sleep is the most effective activation procedure, especially in younger patients, and should be performed when possible.