The management of patients with seizures relies on self-reported seizure frequency. The goal of therapy for patients with epilepsy is seizure freedom, but this requires patient awareness for accurate reporting. Amnesia for epileptic seizures can occur, resulting in impaired recognition (Schultz et al., 1995) and therefore can delay appropriate treatment (Tatum et al., 1998). In-patient video-EEG monitoring has revealed that unrecognized seizures are frequent, and that patient recognition is poor (Blum et al., 1996). A routine 30-minute EEG recording has negligible ability to capture seizures (Ajmone–Marsan and Zivin, 1970); however, computer-assisted ambulatory EEG (CAA-EEG) has demonstrated good success rates for clinically useful information in people with epilepsy (Morris et al., 1994). In one study, 46 patients with a clinical diagnosis of epilepsy had outpatient CAA-EEG that was superior to routine EEG, with 7 of 46 records (15.2%) containing seizures, including 3 patients with multiple unidentified seizures (Liporace et al., 1998). The goal of this study was to quantify and detail patients with unreported seizures in an outpatient population using CAA-EEG.
All patients were evaluated with CAA-EEG at a tertiary-care, university-affiliated hospital for diagnostic purposes. Patient age ranged from 1 month to 93 years (mean age, 35.3 years). Adults 18 years and older accounted for 320 patients (70.1%) studied, and 81 of 552 children (14.7%) younger than 12 years were studied. A total of 305 female and 238 male patients were evaluated with CAA-EEG. The most common reason for referral listed was for evaluation of “seizures.” CAA-EEG was the chosen procedure based on the clinical choice of the physician. Demographic information available included name, age, medical record number, date performed, reason for referral, and a brief description of events. Additional information regarding pertinent past medical history, results of neuroimaging, and prior EEG information were included individually.
An EEG technologist instructed patients and/or caregivers on the recording system and the goal of the monitoring session. Patients were also given detailed instruction regarding push-button activation (PBA), and were asked to press the push-button for all typical events and for any events in question. Review of the monitoring session for “events” was performed by the technologist at termination of the recording. Because many patients were not driving, additional individuals were likely to be present on application or removal. Furthermore, patients were evaluated in person or by telephone no more than every 24 hours after the onset of monitoring. Telephone contact was always available for any unexpected situations. A portable 16-channel computer-assisted EEG (DigiTrace Care Services, Boston, MA) with spike and seizure automated detection algorithms (Gotman, 1982;Gotman and Gloor, 1976) was used. One of two montages was used, either a bitemporal, biparasagittal longitudinal bipolar montage or a bitemporal longitudinal bipolar montage with an anterior temporal–coronal chain that included T1 and T2 electrodes. Electrocardiographic readings were monitored routinely. Descriptions of recording specifications (Gilliam et al., 1999) and technical application (Morris et al., 1994) are provided elsewhere. Electroencephalography was stored for 120 seconds before and after each PBA. Computer-captured seizures were stored either as a “weak” (45 seconds) or “strong” (120 seconds) detection based on the duration of activity. Interictal samples were obtained for 20 seconds every hour during the day, and every 10 minutes during the night.
All records analyzed were interpreted by board-certified (American Board of Clinical Neurosphysiology) neurophysiologists at a single dedicated epilepsy center. A second review of all CAA-EEGs representing electrographic seizures without PBA was performed. A diary provided by the patient and/or caregiver contained hand-written descriptions of events and the time they occurred. At the end of the CAA-EEG recording, an EEG technologist verified events as typical of the habitual events when they occurred. Seizures were categorized as either generalized, lateralized left, lateralized right, or electrographically nonlateralized. Generalized seizures were defined as ictal EEG changes with clinical symptoms. Prolonged interictal discharges were considered separately when EEGs with 3-Hz spike–wave discharges of more than 3 seconds occurred without clinical symptoms, suggesting unidentified generalized seizures. Generalized spike–wave discharges of less than 3 Hz while awake were evaluated as seizures only when clinical symptoms were identified, given the inherent difficulty in differentiating ictal from interictal states.
A total of 502 patients had 595 CAA-EEGs recorded between October 1993 and May 1999 at a single university-based hospital neurophysiology laboratory. A total of 552 records (92.8%) were available for review. Of the records available, 78 recordings (15.5%) were performed for more than 24 hours, and 39 patients (7.8%) were monitored more than once. The results of the recording sessions are summarized in Table 1. No EEG change was recorded for 854 PBAs (86.6%). Symptoms with PBA and no EEG change included headache, dizziness, confusion, eye flutter, stare, chest pain, psychological symptoms, and heart palpitations. An electroclinical seizure accompanied 132 PBAs (13.4%). Four patients had PBAs with notable electrocardiographic change, including third-degree arteriovenous block, supraventricular tachycardia, and bradycardia.
A total of 47 of 552 CAA-EEG recordings (8.5%) contained an electrographic partial seizure. The summary of recorded partial seizures is presented in Table 2. Seizures were categorized as either lateralized left, lateralized right, or nonlateralized (Fig. 1). Generalized epileptiform discharges with a frequency of 3 Hz or more are categorized in Fig. 2 relative to PBA. Push-button activation with electrographic partial seizures accounted for 61.7% of all partial seizures recorded. Other seizures may have been “clinical” but not recognized. Partial seizures of right hemispheric origin were most common. Eleven CAA-EEGs contained partial seizures detected solely by the computer without PBA or diary entry, and accounted for 23.4% of all seizures recorded. Equal distribution of right- and left-lateralized unidentified partial seizures were seen. Four patients had multiple unrecognized seizures, two with right lateralization and two with left lateralization. Seven CAA-EEGs contained partial seizures with and without PBA and accounted for 15.2% of all recorded seizures. All seizures with and without PBA lateralized to the same hemisphere except for one patient with six recognized partial seizures of left hemispheric onset and one unreported seizure of right hemispheric onset. A detailed patient profile of unidentified seizures is presented in Table 3.
Our study shows that unreported seizures are relatively frequent in the outpatient setting. Despite thorough instruction before outpatient seizure monitoring, 38.3% of all ictal recordings contained identified seizures. Our outpatient CAA-EEG results for patient reporting of seizures are similar to the inpatient video-EEG study of awareness reported by Blum et al. (1996). Both studies emphasize the underreporting of seizure frequency by patients. The 38.3% of our outpatient CAA-EEG recordings that contained at least one unreported seizure is less than the 63% of patients with unrecognized seizures reported by Blum et al. (1996) during inpatient video-EEG monitoring. The difference is likely multifactorial, with differences in patient population evaluated, technology of devices used in recording, and a greater potential for outpatient seizure identification. Nevertheless, CAA-EEG may provide a “real-life” setting to analyze seizure reporting. Despite the lack of continuous recording capability seen with CAA-EEG technology, continuation of antiepileptic drugs in outpatients, and the imperfect nature of automated seizure detection programs for every type of seizure, the overall seizure yield for patients recorded with CAA-EEG was 9.2% (46 of 502). This result is slightly lower than the 15.2% reported by Liporace et al. (1998) in patients with epilepsy and initial, normal routine EEGs. However, using χ analysis between 46 of 502 and 7 of 46 patients, the difference was not significant (P = 0.2380). The discrepancy may stem from a more heterogenous population evaluated in the current study. Results are similar to the 11.9% seizure yield on PBA reported by Morris et al. (1994); however, electrographic seizures were not detailed. Outpatient seizures may be recognized more easily because family or friends may assist with PBA, or environmental clues may help patients recognize that they have had a seizure when amnesia for the seizure occurs. Still, 23% of all ictal recordings contained only electrographic seizures captured by the computer, including 8.5% of records (4 of 47) with multiple unidentified seizures. We found an equal predisposition of both left- and right-hemispheric seizures in patients with unidentified partial epilepsy in contrast to prior reports demonstrating a higher incidence with left-hemispheric onset (Ajmone–Marsan and Zivin, 1970).
Amnesia for the ictus is a feature of complex partial seizures with variable periods of amnesia peri-ictally (Bergin et al., 1995). In our study, partial seizures were the most common electrographic seizure type recorded with temporal onset in the majority of records. Because most of our patients were older than 18 years of age, it is not surprising that 61.7% of seizures captured were partial seizures. Our analysis may have underestimated true partial seizure incidence by the exclusion of simple partial seizures, which frequently do not have a surface EEG electrographic correlate (Devinsky et al., 1988). Similarly, extratemporal seizures with a subtle or absent ictal scalp EEG correlate (Williamson et al., 1985) may not have been detected by the computer. Still, 23% of patients had outpatient CAA-EEG recording sessions without a report of any seizures on equipment return.
Palmini et al. (1992) postulated that amnestic seizures resulted from selective ictal inactivation of the mesial temporal structures without cortical involvement. Furthermore, working memory may be impaired in patients with complex partial seizures (Hermann et al., 1997;Krauss et al., 1997). It is possible that memory impairment or failure to understand EEG recording methods may have limited a response to seizures by PBA. However, the high number of false-positive PBAs seen, as well as the lack of clear correlation between seizure awareness and memory testing (Ajmone–Marsan and Zivin, 1970) make this possibility less likely. Additionally, the recording device itself may serve as a constant signal to others to pursue PBA in the case of a behavioral event. Because of the inherent difficulty in defining generalized electrographic seizures without awareness, we did not consider records with generalized spike–wave discharges to be seizures without patient identification. Additionally, we did not analyze spike–wave discharges less than 3 Hz in duration given the difficulty of separating ictal from interictal states unless clinical symptoms were present. Fig. 2 shows a breakdown of the records with generalized spike–wave discharges that were 3 Hz or more during the burst. Approximately 30% of our patients were less than 18 years of age and contained the greatest percentage of generalized interictal epileptiform discharges. Holmes et al. (1987) considered spike–wave or multispike–wave discharges of more than 3 seconds as seizures in absence epilepsy; however, response times may be compromised during brief generalized discharges with detailed testing (Browne et al., 1974;de Foe et al., 1991), making accurate identification of “seizures” imprecise based purely on duration. Therefore, we separately designated generalized spike–wave discharges lasting at least 3 seconds as prolonged interictal discharges. When bilateral anterior-predominant spike–wave discharges were at least 3 Hz at the onset of the burst, we found that 39% of EEG recordings demonstrated prolonged interictal epileptiform discharges longer than 3 seconds without PBA. The clinical importance of generalized, prolonged interictal epileptiform discharges appears most relevant for patients with impaired cognition.
Our results suggest that patients undergoing outpatient 24-hour CAA-EEG underreport seizures identified by PBA on CAA-EEG. The limitations of our study include the retrospective design, reliance on patients to comply with a PBA system, intermittent EEG sampling using CAA-EEG, and the heterogenous population studied. Additionally, the lack of patient testing hampers definitive statements about outpatient seizures and awareness. Similarly, patient understanding of ambulatory EEG monitoring with a PBA system could raise concern about true patient recognition. Furthermore, definitive statements with respect to generalized seizures were unable to be made, given the difficulty in differentiating interictal from ictal electrographic discharges in the absence of associated clinical symptoms. Nevertheless, our study results suggest that seizures in an outpatient setting are frequently unreported or underreported even within a 24-hour period.
Although our study does not allow more definitive statements to be made regarding seizure type and the relationship to unidentified seizures, practitioners caring for patients with epilepsy must use outpatient reporting of seizure frequency for treatment. If this method of data collection underestimates seizure frequency, patients may be undertreated. Patients who live alone, have unexplained injuries, report only seizures with awareness, or patients with a history that suggests unrecognized seizures may benefit from prolonged recording using CAA-EEG. Patients with or suspected to have complex partial seizures reflect the most favorable group for CAA-EEG. Clinically subtle seizures, nocturnal seizures, or seizures in patients with compromised recognition or awareness capabilities may impair accurate seizure counts. Such clinical scenarios highlight the potential benefit of CAA-EEG. Outpatient CAA-EEG has the advantage of portability, longer recording time, and seizure and spike detection possible for a 24-hour cycle in a routine environment. The disadvantage is the absence of direct visualization, testing, or support, as with inpatient video-EEG. No immediate data are available as seen with routine EEG. Still, CAA-EEG occupies a niche for patients when a routine EEG is too limited by time and inpatient video-EEG is neither practical nor feasible.
We conclude that unidentified seizures are seen commonly during routine outpatient CAA-EEG. Ictal recordings may have a profound clinical impact on accurate diagnosis and treatment. Optimized therapy maybe limited for patients with epilepsy when treatment relies on outpatient self-reporting of seizure frequency alone.
1. Ajmone–Marsan C, Zivin LS. Factors related to the occurrence of typical paroxysmal abnormalities in the EEG records of epileptic patients. Epilepsia 1970; 11: 361–81.
2. Bergin PS, Thompson PJ, Fish DR, Shorvon SD. The effect of seizures
on memory for recently learned material. Neurology 1995; 45: 236–40.
3. Blum DE, Eskola J, Bortz JJ, Fisher RS. Patient awareness
. Neurology 1996; 47: 260–4.
4. Browne R, Penry J, Porter R, Dreifuss F. Responsiveness before, during, and after spike–wave paroxysms. Neurology 1974; 24: 659–65.
5. de Feo MR, Mecarelli O, Ricci GF, Rina MF. The utility of ambulatory EEG
monitoring in typical absence seizures
. Brain Dev 1991; 13: 223–7.
6. Devinsky O, Kelley K, Porter R, Theodore WH. Clinical and electroencephalographic features of simple partial seizures
. Neurology 1988; 38: 1347–52.
7. Gallassi R, Morrcale A, Di Sarro R, Lugaresi E. Epileptic amnesic syndrome. Epilepsia 1992; 33 (suppl 6): S21–5.
8. Gilliam F, Kuzniecky R, Faught E. Ambulatory EEG
monitoring. J Clin Neurophysiol 1999; 16: 111–5.
9. Gotman J. Automatic recognition of epileptic seizures
in the EEG. Electroencephalogr Clin Neurophysiol 1982; 54: 530–40.
10. Gotman J, Gloor P. Automatic recognition and quantification of interictal epileptic activity in the human scalp EEG. Electroencephalogr Clin Neurophysiol 1976; 41: 513–29.
11. Hermann BP, Seidenberg M, Schoenfeld J, Davies K. Neuropsychological characteristics of the syndrome of mesial temporal lobe epilepsy
. Arch Neurol 1997; 54: 369–76.
12. Holmes G, McKeever M, Adamsom M. Absence seizures
in children: clinical and electroencephalographic features. Ann Neurol 1987; 21: 268–71.
13. Krauss GL, Summerfield M, Brandt J, Breiter S, Ruchkin D. Mesial temporal spikes interfere with working memory. Neurology 1997; 49: 975–80.
14. Liporace J, Tatum WO, Morris GL, French JA. Clinical utility of sleep-deprived versus computer-assisted ambulatory 16-channel EEG in epilepsy
patients: a multi-centered study. Epilepsy
Res 1998; 32: 357–62.
15. Morris GL, Galezowska J, Leroy R, North R. The results of computer-assisted ambulatory 16-channel EEG. Electroencephalogr Clin Neurophysiol 1994; 91: 229–31.
16. Palmini AL, Gloor P, Jones–Gotman M. Pure amnestic seizures
of temporal lobe epilepsy
: definition, clinical symptomatology, and functional anatomic considerations. Brain 1992; 115: 749–69.
17. Schulz R, Luders HO, Noachtar S, et al. Amnesia
of the epileptic aura. Neurology 1995; 45: 231–5.
18. Tatum WO, Ross Cole AJ. Epileptic pseudodementia. Neurology 1998; 50: 1472–5.
19. Williamson PD, Spencer DD, Spencer SS, et al. Complex partial seizures
of frontal lobe origin. Ann Neurol 1985; 18: 497–504.