Overnight EEG to Forecast Epilepsy Development in Children with Autism Spectrum Disorders

Objective To establish the utility of long-term electroencephalogram (EEG) in forecasting epilepsy onset in children with autism spectrum disorder (ASD). Study design A single-institution, retrospective analysis of children with ASD, examining long-term overnight EEG recordings collected over a period of 15 years, was conducted. Clinical EEG findings, patient demographics, medical histories, and additional Autism Diagnostic Observation Schedule (ADOS) data were examined. Predictors for the timing of epilepsy onset were evaluated using survival analysis and Cox regression. Results Among 151 patients, 17.2% (n=26) developed unprovoked seizures (Sz group), while 82.8% (n=125) did not (non-Sz group). The Sz group displayed a higher percentage of interictal epileptiform discharges (IEDs) in their initial EEGs compared with the non-Sz group (46.2% vs. 20.0%, p=0.01). The Sz group also exhibited a greater frequency of slowing (42.3% vs. 13.6%, p < 0.01). The presence of IEDs or slowing predicted an earlier seizure onset, based on survival analysis. Multivariate Cox proportional hazards regression revealed that the presence of any IEDs (HR 3.83, 95% CI 1.38-10.65, p=0.01) or any slowing (HR 2.78, 95% CI 1.02-7.58, p=0.046 significantly increased the risk of developing unprovoked seizures. Conclusion Long-term EEGs are valuable for predicting future epilepsy in children with ASD. These findings can guide clinicians in early education and potential interventions for epilepsy prevention.