Abstracts

Rationale: The electroencephalogram (EEG) is an essential investigation after the first seizure. Epileptiform discharges provide prognostic value for seizure recurrence, as well as help classify epilepsy syndromes. Previous studies have proposed that emergent EEGs after the first seizure increase the diagnostic yield. We sought to test this hypothesis using multivariable statistical analysis.  Methods: We retrospectively reviewed patients who presented with a first-ever seizure over 6 years to a teaching hospital. We collected data on patient demographics, seizure semiology, EEG outcomes and neuroimaging results from medical records. Time from the first seizure to the EEG was calculated from the seizure onset (based on ambulance notes) to the EEG recording time. For descriptive statistics, we reported frequencies and percentages for categorical variables and mean, median, and standard deviation for continuous variables. We defined the diagnostic yield as a percentage: the number of positive EEGs divided by the total number of EEGs with 95% confidence intervals (CI). We used logistic regression to investigate epileptiform EEG outcome for these variables as potential predictors: age, the time gap between last seizure and EEG, unprovoked seizure vs provoked seizure, single first seizure vs multiple seizures within 24-hours, seizure in wakefulness vs sleep, seizure type (GTCS vs other), and epileptogenic structural abnormalities on neuroimaging. Data analyses were performed with IBM SPSS statistical software package (IBM Corporation, New York, USA).  Results: We analyzed 386 patients (147, 38%, females; median age 39.5 ± 21.3) presenting with first seizures. All patients had EEGs recorded at varying times after the first seizure. Forty patients (10%) had multiple seizures within 24 hours. 42 (11%) had status epilepticus whereas 79 (20%) and 33 (9%) had nocturnal and provoked seizures respectively. Of 355 witnessed first seizures, 305 (86%) were generalized tonic-clonic seizures. Five patients had sleep-deprived EEGs. The median time from first seizure onset to EEG was 234 hours (interquartile range 123 to 426 hours). Neuroimaging was performed in 361 patients (324 CTs, 272 MRIs); 71 (20%) had a potentially epileptogenic structural abnormality. Seventy-nine EEGs reported epileptiform activity (diagnostic yield 20%; 95% CI 17%-25%) and 37/71 (52%) had focal discharges. Figure 1 shows the distribution of positive EEGs over time. Only provoked (vs unprovoked) seizures significantly impacted the diagnostic yield for epileptiform discharges (OR=4.36, p=0.04). Table 1 highlights results of the statistical analysis.  Conclusions: The time lag from the first seizure to the EEG does not impact the diagnostic yield of EEGs, refuting a common belief that a delayed EEG is less useful. This is important for the prioritization of testing as urgent EEGs are costly.  Funding: Nil