Rationale:
Treatment of Epilepsy depends on patient reporting of seizure frequency and severity. Erroneous reporting diminishes the physician's ability to properly manage the patient’s care. Similarly, clinical trials aiming to approve new medications also rely on patient reporting. Unreliable reporting may result in faulty assessments of investigational medications.
Indeed, many popular drugs that are considered to be effective in clinical practice only showed marginal efficacy in trials. Furthermore, some drugs (Brivaracetam, Lyrica) performed both poorly and successfully in separate studies, raising doubt in the accuracy of patient reporting.
Previous studies have already shown patient reporting to be unreliable, but there is room for improvement and clarification. Previous studies suffer from low patient and seizure numbers. Patient environment, study protocols, the measurement indices used to evaluate patient reporting accuracy, and the choices for statistical analysis have often been flawed.
Our study compiled and utilized a large seizure database to determine the accuracy of patient seizure reporting.
Methods:
As part of their diagnosis procedure, patients at the Arkansas Epilepsy Program underwent ambulatory EEG video monitoring for a typical period of three days. The resulting EEG was analyzed by automated Spike and Seizure software, then reviewed by a board-certified epileptologist. Ambulatory monitoring also collected patient reports. During monitoring, patients and/or caregivers were instructed to “push the button” on the ambulatory device when they perceived a seizure.
Patient EEG along with “push button events” were used to calculate two measurements of reporting accuracy: “missed seizures” and “false positives.”
This data was compiled into a database and analyzed. From ~2000 EEG monitoring reports, the study included 407 Epilepsy patients with 1605 seizures and 1726 push button events.
Results: After removing patient reports with descriptions that don’t match seizures (such as headaches), as well as Simple Partial Seizures (which don’t always appear on EEG), the average patient missed 60% of their seizures and reported 59% false positives. Reports of convulsive seizures were the most accurate (Tonic Clonic: 0% false positives, 0% missed seizures; Focal with Secondary Generalizations: 27% false positives, 30% missed seizures), while reports of seizures affecting consciousness were the least accurate (Absence: 54% false positives, 79% missed seizures; Complex Focal Seizures: 61% false positives, 56% missed seizures). Accuracy measurements are displayed in Tables 1 and 2.
Conclusions: Patient seizure reporting is inaccurate and should not be used as the sole or primary basis to determine effective treatments or the outcome of clinical trials. Multi-faceted approaches to monitoring - technology detecting motion, autonomic function, and other body characteristics - along with ambulatory EEG data may prove to be the most reliable.
Funding: None