BACKGROUND:Vagus nerve stimulation (VNS) is a neuromodulatory procedure most extensively studied as an adjunct to medically refractory epilepsy. Despite widespread adoption and decades of clinical experience, clinical predictors of response to VNS remain unclear.OBJECTIVE:To evaluate a retrospective cohort of pediatric patients undergoing VNS at our institution to better understand who may benefit from VNS and identify factors which may predict response to VNS.METHODS:We conducted a retrospective cohort study examining pediatric patients undergoing VNS over nearly a 20-year span at a single institution. Presurgical evaluation, including demographics, clinical history, and diagnostic electroencephalogram, and imaging findings were examined. Primary outcomes included VNS response.RESULTS:Two hundred ninety-seven subjects were studied. The mean age at surgery was 10.1 (SD = 4.9, range = 0.8-25.3) years; length of follow-up was a mean of 4.6 years (SD = 3.5, median = 3.9 years, range 1 day-16.1 years). There was no association between demographic factors, epilepsy etiology, or genetic basis and VNS outcomes. There was an association between reduction in main seizure type with positive MRI finding. Of all MRI findings analyzed, brain atrophy was significantly associated with worse VNS outcomes, whereas dysplastic hippocampus and chronic periventricular leukomalacia findings were found to be associated with improved outcomes. Increased seizure semiology variability and seizure type were also associated with improved seizure outcomes.CONCLUSION:Predicting response to VNS remains difficult, leading to incompletely realized benefits and suboptimal resource utilization. Specific MRI findings and increased seizure semiology variability and type can help guide clinical decision making and patient counseling.