Seizure identification by clinical description in temporal lobe epilepsy - How accurate are we?

Objective: To determine the accuracy of the clinical history performed by epileptologists in the identification of seizures in patients with suspected temporal lobe epilepsy. Methods: The clinical and EEG telemetry (EEGT) monitoring data of 88 patients with suspected refractory temporal lobe seizures referred for evaluation of epilepsy surgery were prospectively evaluated. All clinical events obtained by history in these patients were adjudicated as being a seizure or not by two blinded (without access to EEG data) independent epileptologists. Each clinical event was then matched with the corresponding clinical event recorded with EEG telemetry in the epilepsy monitoring unit (gold standard). Sensitivity, specificity, overall accuracy, predictive value, and interrater agreement for the clinical assessment were obtained. Results: Of 357 clinically different events, 175 (49%) were reproduced in the epilepsy monitoring unit. Only 10 events were misidentified by history as being a seizure or not, resulting in an overall clinical accuracy of 94%. Epileptologists' sensitivity for seizure identification was 96% (95% CI 92, 98%) but specificity was only 50% (95% CI 22, 79%). Accuracy for complex partial seizures and generalized seizures was higher than for simple partial seizures (SPS). Misidentification occurred only with SPS and nonepileptic events. Agreement beyond chance among epileptologists was good. Conclusion: In this selected group of patients with temporal lobe epilepsy, seizure identification by clinical history is highly accurate. Epileptologists rarely miss seizures (high sensitivity) but more often overcall nonepileptic events as seizures (low specificity).