Abstract
Theodore WH, Porter RJ, Penry JK. Neurology. 1983;33(9):1115-1121. doi: 10.1212/wnl.33.9.11515. PMID: 6684245 Videotape analysis of 163 complex partial seizures in 40 patients showed that the mean duration of the attack was 128 seconds. Automatisms occurred in 159 seizures (97%) and involved more than the face and arms in 132 (80%). Most automatisms were simple, stereotypic, or aimless movements. Postural tone increased in 24 seizures and decreased in 62. Clonic movements of the eyelids occurred in 19 attacks, and clonic movements of the extremities in 4. Only nine patients reported auras. Distinct ictal and postictal phases could be distinguished in 132 seizures (80%); in these, the mean ictal duration was 54 seconds and the mean postictal duration 89 seconds. Videotape analysis provides objective criteria by which complex partial seizures may be differentiated from other seizure types.
Devinsky O, Sato S, Kelly K, Porter RJ, Theodore WH. Neurology. 1988;38:1347-1352 The clinical and electroencephalographic features of 87 simple partial seizures in 14 patients were studied with video-EEG telemetry. The patients were able to respond to verbal stimuli during all seizures and, later, could clearly recall ictal events. To determine whether the EEG changes in simple partial seizures could be reliably observed, a reader blindly reviewed four EEGs of equal duration for each seizure. These EEGs consisted of one ictal and three nonictal recordings obtained at predetermined times before the seizure. There were 27 motor seizures (mean duration, 86 seconds; range, 2 to 250 seconds), all involving clonic movements of the head and/or upper extremities; 8 (30%) of these had a sensory component (pain in 6, paresthesia in 2). An EEG change, usually localized spikes or sharp waves over the contralateral or both rolandic regions, was identifiable in nine (33%) of the motor seizures. The 60 nonmotor seizures (mean duration, 63 seconds; range, 8 to 375 seconds) involved a variety of symptoms, including somatosensory/special sensory (3 seizures), autonomic (26 seizures), cognitive (1 seizure), affective (14 seizures), and mixed, or more than one category of nonmotor symptoms (16 seizures). In only nine (15%) of the nonmotor seizures was there an ictal EEG change, usually localized spikes or paroxysmal theta activity over the temporal region. Overall, among the 87 simple partial seizures, only 18 (21%) revealed ictal EEG changes. Thus, a normal EEG is common during simple partial seizures and does not exclude the diagnosis.
Theodore WH, Porter RJ, Albert P, Kelley K, Bromfield E, Devinsky O, Sato S. Neurology. 1994;44:1403–1407. We studied 120 generalized tonic–clonic seizures (GTCSs) in 47 patients with video-EEG telemetry. GTCSs were preceded by antecedent seizures, including 13 simple partial, 70 complex partial, 17 simple partial leading to complex partial, seven tonic, seven clonic, and one typical absence. We divided GTCSs into the following phases: onset of generalization, pretonic clonic, tonic, tremulousness, and clonic. The mean GTCS duration was 62 seconds. There was a non-significant trend toward longer duration on reduced antiepileptic drug doses. Marked heterogeneity in GTCS phenomenology was present; only 27% of seizures included all five phases. Individual phase duration and clinical expression, including tonic and clonic phases, was highly variable. The clinical phenomena suggest that multiple cortical and subcortical routes of spread may exist. When GTCSs last longer than 2 minutes, intravenous antiepileptic drug treatment should be initiated.
Commentary
In 1983, Theodore, Rogers, and Penry published a seminal paper on complex partial seizures using a combination of clinical observation of video recordings and simultaneous EEG “video with EEG telemetry” and defined differentiating characteristics from absence and pseudo-seizures. 1
The NINDS epilepsy group later published similar video EEG telemetry studies of simple partial and secondarily generalized seizures.2,3
These papers advanced understanding of the three primary seizure types in focal epilepsy—simple partial, complex partial, and secondarily generalized seizures, now termed focal preserved consciousness, focal impaired consciousness, and bilateral tonic–clonic seizures, and fueled the development of video EEG telemetry technology, which remains central to our evaluation and surgical assessment for patients with epilepsy today.
Despite leaps in diagnostic technology and now artificial intelligence (AI), we still place principal value in our descriptive powers to localize, diagnose, and teach, through the history, neurological examination, and observation of clinical seizure symptoms and signs; “semiology.” The advent of video-EEG telemetry in the 1970s, initially in select units such as NINDS, and others, opened a gateway into a new observational science of seizure semiology analysis. The NINDS investigational EEG lab started in the 1950s and had already demonstrated the usefulness of combined video and EEG telemetry and ambulatory EEG in obtaining ictal recordings beneficial for diagnosis.4,5 Through the ensuing decades, a multitude of semiology studies from video-EEG telemetry have contributed to our current understanding of epilepsy network localization.
The authors describe the clinical features of 163 complex partial seizures (for this commentary, we will use the historical terms) in 40 adults who had been referred to the Clinical Epilepsy Section at the NINDS with medication-resistant seizures and who underwent video-EEG telemetry. The authors emphasize the critical role of videotaping seizures rather than the historical use at the time of descriptive commentary or chart notes. With the ubiquity of handheld video recording devices, we are now asking our patients to do this themselves.
The study population was heterogeneous with a broad range of onset (12–58 years), long duration of epilepsy (mean 13 years), and one-quarter with intellectual impairment.
A home-grown 4 or 8 channel telemetered EEG with split screen video was obtained on each patient, combined with hyperventilation, photic stimulation, and nasopharyngeal electrodes. 5 Video recordings were analog reel-to-reel tape with 2 TV cameras for the body and face (W.H. Theodore, personal communication). Spatial coverage was limited to only 2 or 4 channels on each side of the head in that era. Video-EEG technology rapidly advanced, with up to 128 channels achievable by the early 1990s. 6 A minimum of 25 scalp EEG electrodes is now recommended by the International Federation of Clinical Neurophysiology, and up to 256 electrodes can be recorded in high-density or intracranial EEG. However, 40 years later, we have circled back to the analysis of low-density EEG recorded from implanted long-term subcutaneous or intracranial EEG recording devices.
Three neurologists and one technician reviewed all seizures using a standardized scoring form and the 1981 ILAE classification of seizures. 7 Video was initially reviewed separately from the EEG. Clinical seizures by video analysis were defined from the onset of activity arrest or automatisms. The post-ictal phase was defined by the first response to command and ended when there was an accurate and complete response to questions.
Of the 163 clinical seizures recorded, the most common observed behavior was automatisms. These were seen most frequently involving the trunk and legs in 132 of 163 seizures. The next most common automatism was of the face and arms in a further 19 seizures. Notably, all 40 patients with recorded complex partial seizures demonstrated some form of automatism.
Automatisms were stereotypical an followed the same order from seizure to seizure; this observation of the stereotypy of seizures is a pillar of video-EEG telemetry practice. Twelve patients had less typical automatisms with complex behaviors, and two had violent, probably hypermotor behavior. Automatisms could extend into the postictal phase. Seizure clusters had a similar seizure semiology to single seizures.
An aura was the second most frequent behavioral manifestation of seizures observed in this population, with an aura documented in 14 of 163 seizures and in 9 patients.
In 132 of 163 complex seizure seizures, language was disturbed, without consistent lateralization identified in either the ictal or post-ictal state, potentially due to variable language lateralization, limited EEG coverage, and bilaterally active patients. The authors note the challenges of defining ictal from postictal phases clinically, and postictal aphasia from other postictal phenomena.
The mean total duration of the complex partial seizure was short at 128 s, with only 14% under 30 s—duration was measured by clinical analysis and not EEG. Distinct ictal and postictal phases were identifiable in 37 out of 40 patients. In those cases, mean ictal duration was 54 s and postictal phase 89 s, with a direct correlation between the length of both.
The EEG showed bilateral or multifocal interictal epileptiform discharges in 20 of the 40 patients. Another 16 had focal temporal discharges. At ictal onset, only 17 patients had focal temporal ictal discharges, 12 had bilateral sharp and slow waves, 5 had diffuse slowing only, and 6 with no change in interictal spiking. The limited spatial coverage and patients with bilaterally active epilepsy or intellectual disability likely contributed to the relative lack of focality in the ictal EEG.
The authors include useful tables to differentiate between partial onset and absence seizures or pseudoseizures. They highlight that intensive video-EEG telemetry is important to differentiate between complex partial seizures and pseudoseizures, a core principle in epileptology practice. However, they note that the diagnosis of complex partial seizures can still be made clinically by the presence of a consistent aura, automatisms, and postictal state.
The study highlights many astute clinical observations classically taught in epileptology today—stereotypy of seizures, prevalence of automatisms, seizure phases, and distinguishing the ictal from postictal states. Analyses of the sequence of semiology or “phases” have led to a greater understanding of the epileptogenic zone with incorporation into subsequent seizure classification schemes.8,9 Accurate definitions of complex partial seizures and differentiation from generalized or functional seizure types had long-lasting implications, including access for patients to targeted therapeutics, clinical trials, and defining candidates for epilepsy surgery.
In a similar fashion, Devinsky and Theodore analyzed 87 simple partial seizures in 14 patients correlated with 16-channel EEG. 2 Twenty-seven were motor and 60 were non-motor simple partial seizures. An ictal EEG change was seen in 33% of motor seizures near Rolandic regions and only 15% of nonmotor seizures near temporal regions. Although numbers are small, the analysis was blinded and highlighted that not all seizures cause an ictal EEG change, another core principle of video-EEG analysis. More recent series with advanced EEG technology increased the yield up to 33% for nonmotor auras, 10 with potential to increase further with the aid of AI-assisted scalp EEG analysis. 11
In the third paper, Theodore analyzed 140 generalized seizures (of which 100 were focal onset) from 47 patients and phenomenologically described them in 7 phases. 3 They reported a mean duration of 62 s with none more than 2 min, leading to a recommendation that rescue anti-seizure medication be given after 2 minutes, a standard practice in use in epilepsy monitoring units and in the community today.
Prior to the onset of video EEG telemetry, seizure classification was obtained by history and observing clinical behavior, leaving conclusions open to subjectivity and potential bias. The authors of these papers utilized the latest technology of the day to generate electro-clinical understanding of seizures, which enhanced our understanding of focal seizure semiology. Even in 2025, there are still many patients with epilepsy without access to video-EEG telemetry, a pressing need that demands attention.
Contemporary analyses of video-recorded seizure semiology and EEG using statistical methods or AI can augment conclusions. 12 However, the powers of visual observation of the recorded seizure video, as demonstrated in these papers, mean AI will assist but not replace the human expert observer in analyzing seizure semiology.
