Unilateral ultra long‐term subcutaneous EEG monitoring in drug‐refractory idiopathic generalized epilepsy

Abstract

Objective

Seizure self‐reporting is known to be unreliable, particularly with non‐convulsive seizures. There is increasing interest in long‐term monitoring systems to detect and count seizures, including the use of the minimally‐invasive CE‐marked subcutaneous unilateral electroencephalography (EEG) recording system, which has been shown to accurately detect ipsilateral focal seizures.

Methods

We implanted the subcutaneous EEG system in nine patients with refractory idiopathic generalized epilepsy (IGE), following initial characterization of their inter‐ictal and ictal EEG abnormalities, with in‐patient video‐EEG monitoring as gold standard comparison. We compared the in‐patient video‐EEG findings to the abnormalities detected on the long‐term subcutaneous system in seven of the nine patients. We compared EEG signal characteristics for convulsive seizures, and for spike and wave discharges of duration of <3 s, between 3 to 10 s, and greater than 10 s.

Results

Three generalized convulsive seizures were recorded in total during video‐EEG monitoring, and all were easily detected by the long‐term subcutaneous EEG system. Ninety percent of significant spike and wave discharges lasting >3 s were identified using the subcutaneous EEG device. There were minimal adverse effects from use of the device; the patients were satisfied that the device was easy to use and were likely to recommend the use of the device to other patients with epilepsy, as measured by Likert scales. The device implantation was judged to be easy by our epilepsy neurosurgeons.

Significance

This study demonstrates that ultra long‐term monitoring with unilateral subcutaneous EEG can reliably detect all convulsive and most non‐convulsive generalized seizures in patients with refractory IGE. The device is well tolerated and easy to insert, and it can be used as a useful adjunctive seizure monitoring system in select patients. Furthermore, it may guide therapy to optimize seizure control, and it may confer cost savings by avoiding the need for in‐patient video monitoring in appropriate patients.

Key points.

• Detection and quantification of seizures is known to be unreliable in many patients with epilepsy, particularly in those with difficult drug‐resistant epilepsy.

• Routine ambulatory electroencephalography (EEG) is difficult to perform beyond a period of a few days, and, worldwide, there are limited resources for in‐patient video‐EEG monitoring.

• It can be difficult to monitor and detect absence seizures in patients with refractory idiopathic epilepsy, with important implications for decisions such as granting of driving privileges.

• Here we show that a minimally invasive unilateral subcutaneous 24/7 EEG device can reliably detect convulsive and most (90%) absence seizures in patients with difficult idiopathic generalized epilepsy.

• The 24/7 device is easy to insert, and it is well tolerated and acceptable to selected patients.

1. INTRODUCTION

The epilepsies are a group of neurological conditions, with an estimated point prevalence of 1% in the general population ¹ .Epilepsies are classified into electroclinical syndromes, based on seizure type and electroencephalography (EEG) characteristics. Idiopathic generalized epilepsies (IGEs) have recently been classified as a subgroup of genetic generalized epilepsies (GGEs), which include childhood absence epilepsy (CAE), juvenile absence epilepsy (JAE), juvenile myoclonic epilepsy (JME), and epilepsy with generalized tonic clonic seizures alone (GTCA). ² Although the seizure generation mechanisms in IGE remain unclear, they are thought to involve the fronto‐thalamic networks. ³ Some of the seizure types encountered with IGE, such as generalized tonic–clonic seizures (GTCS), are easy to detect and record. Other seizure types, such as absence seizures, are subtle, due to the lack of positive ictal semiology and brief duration, and may remain unreported by patients and their families, or may be overreported, leading to potential overtreatment. ⁴ Thus, exact seizure counting and determination of seizure control in IGE remains challenging in a substantial proportion of patients. Approximately 30% of patients with IGE are medically refractory, ⁵ as defined by the International League Against Epilepsy (ILAE). ⁶ The National Institute for Health and Care Excellence (NICE) epilepsy treatment guidelines emphasize the importance of regularly assessing whether anti‐seizure medication (ASM) is effective in reducing the occurrence of seizures (NICE guideline [NG217]).

The EEG hallmark of IGE is bisynchronous, symmetric, and generalized spike–wave complexes occurring at a frequency between 2.5 and 6 Hz. Additional EEG features include polyspikes, polyspike–wave discharges, occipital intermittent rhythmic delta activity, and a photoparoxysmal response ⁷ . These abnormalities usually occur on a normal EEG background and are more frequently detected during non–rapid eye movement (REM) sleep and after awakening.

Video‐EEG monitoring (VEM) is the internationally accepted gold standard for detecting and recording seizures. However, this investigation is particularly time and resource intensive and difficult to access and use for seizure quantification in most jurisdictions. The cost per night of a VEM admission varies among juristictions, from $1500–$3000 in the United States ⁸ to €100–€2200 across various European health systems. ⁹ Thus costs rapidly accumulate if a prolonged admission is necessary. On the other hand, an implanted subcutaneous EEG (sqEEG) system frontloads most of the cost, allowing continuous recording for up to 36 months from the time of the initial implantation. In addition to the substantial demands and reduced availability of VEM, seizures are unpredictable, and therefore, despite medication reduction and EEG activating procedures, normal EEG is often recorded for several days prior to the first recorded seizure and between subsequent seizures. Although ambulatory VEM equipment has been demonstrated to be as effective as inpatient VEM, ¹⁰ this recording modality still requires frequent hospital visits for data download and electrode adjustments, thus limiting the overall recording duration to days or weeks.

Subcutaneous EEG (or sqEEG) is a novel method designed to record high‐quality EEG for prolonged periods, ¹¹ thus resolving the time and cost constraints encountered with typical long‐term EEG monitoring in a VEM unit. The UNEEG solution (UNEEG Medical, Denmark) is currently the only CE‐marked device for ultra long‐term sqEEG monitoring approved in Europe. This device was previously validated and used in focal epilepsies. ¹² This is the first study on its unilateral use for monitoring generalized epilepsies.

In this study, we compared the sensitivity for seizures and interictal epileptiform discharges (IEDs) between ultra long‐term sqEEG and scalp video‐EEG monitoring in patients with medically refractory IGE. In addition to this primary endpoint, we examined the automated seizure detection during the outpatient recording period and the safety, tolerability, and satisfaction of the subjects with this ultra long‐term sqEEG monitoring system.

2. MATERIALS AND METHODS

The study was approved by the Royal College of Surgeons of Ireland and Beaumont Hospital Research and Ethics Committee. All participants provided written informed consent at their recruitment. The study was registered with clinicalstrials.gov (NCT05241678) and the study followed ISO14155.

2.1. Study design and participants

This was a 12‐week prospective study that aimed to (1) compare the sensitivity for seizures and IEDs in subjects with drug‐refractory IGE between ultra long‐term sqEEG in comparison to scalp video‐EEG monitoring in our epilepsy monitoring unit (EMU), and (2) evaluate the acceptability of everyday use in this patient population (Figure 1).

FIGURE 1.

Study schedule.

The subjects were recruited from the Beaumont Hospital Epilepsy clinic. The main inclusion criteria were an established diagnosis of drug‐refractory IGE and a minimum of one typical known seizure per month. The expanded inclusion and exclusion criteria are summarized in Table S1.

Fourteen patients were initially recruited based on the eligibility criteria, of which nine decided to proceed with sqEEG implantation and completed the study. Of the nine patients, two were excluded from EEG data analysis due to the reclassification of their epilepsy syndrome after the EMU admission. Patient 003 could not tolerate the EMU admission and discharged himself against medical advice after 1 h of simultaneous sqEEG recording. The EEG record for Patient 007 was normal during monitoring, and thus active IGE was deemed unlikely. Patient 004 was reclassified to symptomatic generalized epilepsy; however, the EEG data were included in the analysis, given that the epileptiform abnormalities were generalized rather than focal. The subjects excluded from the EEG data analysis were still able to provide information on the safety and tolerability of the device, which was included in that specific sub‐analysis.

2.2. Data collection

Patients were recruited at Visit 1, underwent implantation at Visit 2, and commenced sqEEG recording at Visit 3. The 24/7 EEG SubQ device (UNEEG Medical, Denmark) was used to collect all sqEEG data as part of this study. The sqEEG implant consists of a three‐contact electrode (yielding two‐channel bipolar EEG) and a small housing, implanted unilaterally under local anesthesia. An external recorder connects to the implant housing via an inductive link, powering the implant and recording data. We opted to implant the device as depicted in Figure 2, with the distal contacts of the lead overlying the frontotemporal region. The devices were implanted by our epilepsy neurosurgeons (D.O'B. and K.S.) during day procedures. The patients underwent implantation on their dominant side, to ensure subsequent ease of use with pairing the recorder to the implant housing. An EMU admission was planned within 4 weeks of the sqEEG implantation, and sqEEG outpatient data were downloaded at the time of the EMU admission. Simultaneous sqEEG and scalp EEG were undertaken during the EMU admission. Medication reduction was not performed, and patients were recorded for between 24 and 72 h. Scalp and sqEEG data were downloaded at the end of the EMU admission period, and the patients underwent another outpatient data transfer visit prior to the sqEEG device explantation, which occurred on Visit 6. A visual representation of the data collection process is detailed in Figure 1.

FIGURE 2.

Positioning possibilities of the implant. The red circle indicates the implant position for patients in this study.

Two EEG quality tests were performed at Visit 3 and at Visit 6 prior to explantation. The patients were asked to perform a standardized set of tasks, such as resting with eyes closed and eyes opened, chewing gum, blinking or performing eye movements (left, right, up, and down), and turning their head with eyes closed to generate known pattern EEG artifacts, which were visualized on both scalp and sqEEG. A neurophysiology scientist (B.D. or C.O'D.) subjectively rated the generated EEG signals (scalp EEG and sqEEG) on a scale of 1–5 based on their visual assessment of the EEG signal quality.

2.3. Data analysis

Due to the wide array of data collected, multiple analyses were performed, including sqEEG sensitivity for IEDs and seizure detection, subject satisfaction questionnaire analysis, and the analysis of the quality of the EEG data recorded. Figure 3 depicts a visual workflow for analyzing the EMU and outpatient EEG records. EEG data were reviewed by a clinical neurophysiologist, and abnormalities were identified and labeled according to the most recent clinical neurophysiology nomenclature. ¹³ The scalp EEG data recorded during the VEM admission were compared to the data that were simultaneously recorded. A seizure was defined as a burst of spike and wave complexes that lasted longer than 10 s.

FIGURE 3.

EMU and outpatient EEG data analysis process.

2.4. sqEEG monitoring: IED and seizure detection rates

2.4.1. Inpatient data (EMU)

Scalp and sqEEG were recorded simultaneously during the EMU admission. Scalp EEG was recorded via the 10–20 international electrode placement system. Data collected in the EMU were analyzed in two ways:

1. Data collected through the two recording modalities were reviewed manually by a neurophysiology scientist, where IEDs and seizures were identified and classified by (1) morphology, which was categorized into spike and wave (SW), polyspike and wave (PSW), and focal sharp and wave (FSW) discharges, and (2) duration, which was categorized into <3 s, between 3 and 10 s, and longer than 10 s. To investigate the relationship between IED duration and frequency, we divided the duration into 0–1 s, 1–2 s, 2–3 s, and >3 s based on the dominant frequency of SW or PSW, as assessed by the EEG reviewer. In addition to these categories, fragmented generalized SW and PSW discharges and focal discharges were classified into ipsilateral and contralateral relative to the side of the sqEEG recorder.

The sensitivity was defined as:

$$\text{Sensitivity}=\frac{\text{Count\hspace{0.17em}sqEEG\hspace{0.17em}events}}{\text{Count\hspace{0.17em}ipsilateral\hspace{0.17em}scalp}\mathrm{EEG}\text{events}}\times 100$$

• BThe sqEEG data were analyzed with a semi‐automatic seizure and rhythmic patterns detection algorithm (UNEEG EpiSightAnalyzer v2.1), which highlights areas in the sqEEG recordings that may correspond to ictal events and displays these markings for review. The algorithm detections were compared to the manually reviewed scalp EEG, and the sensitivity was defined as above, using the scalp EEG as gold standard.

Linear mixed modeling was employed to investigate potential differences between the scalp EEG and sqEEG recording modalities for frequency activity and duration of interictal events. The analyses were performed unstratified and stratified, with the morphologies (SW and PSW) defining the strata. The unstratified models were adjusted for morphology and either duration or frequency, depending on whether frequency or duration was the dependent variable. In all cases, frequency and duration were included in the models as continuous values. All models included a random effect for subjects due to repeated measures for the subjects. P‐values for the recording modalities and duration were computed using likelihood ratio tests. Bonferroni correction was used to control for multiple testing. p < .05 was considered statistically significant. The statistical analyses were performed using R 4.4.2. An expanded description of the analysis and the analysis of variance (ANOVA) tables were included in the Supplementary Material S1.

2.4.2. Outpatient data

The outpatient data were visualized via the UNEEG EpiSight Analyzer, and automatic seizure annotations were reviewed by a neurophysiologist scientist and the lead and senior authors (T.M. and N.D.). For each patient, false‐positive detection rate was calculated based on accepted and rejected EpiSight seizure markers. For the adherence calculation, the median of daily usage of the device per 24 h is reported as the adherence rate.

2.5. Subject satisfaction questionnaires

Subject satisfaction was measured at the end of the study by using a 5‐item Likert scale for the Ease‐of‐Use questionnaire, which is included in Supplementary Material S1. One question measured the overall satisfaction and how likely the patient is to recommend the same system to a person with the same chronic illness, and one question evaluated if the patient experienced discomfort related to the UNEEG SubQ implant. Perceived usability was measured with a 10‐item System Usability Scale (SUS). The SUS questionnaire is a widely accepted method for subjective assessment of perceived device usability and was used in this study to evaluate the 24/7 EEG SubQ system . ¹⁴ , ¹⁵ , ¹⁶ Adverse events were reported throughout the study.

3. RESULTS

3.1. Demographics

Nine patients were implanted with the sqEEG recording device and underwent ultra long‐term EEG monitoring. The mean age was 27 years, and the mean epilepsy duration was 13.7 years. Two thirds of patients (n = 6) were female. A detailed summary of the group demographics can be viewed in Table 1.

TABLE 1.

Baseline characteristics of the participants implanted with the subcutaneous device.

Patient	Age	Age at epilepsy onset	Sex	Duration of epilepsy(years)	Implant site (handedness)	Epilepsy syndrome	Adherence	Duration of simultaneous recording (hours)	Total sqEEG recorded (hours)
002	23	2	F	21	Left	CAE	95.8%	56.4	345
003	29	19	M	10	Right	Focal epilepsy	91.8%	NA a	1601
004	18	15	M	3	Right	SGE	95.9%	101.47	951
007	37	16	M	21	Right	IGE in remission and NES	81.8%	NA a	1423
008	37	24	F	13	Right	JME	46.9%	40.75	294
010	33	17	F	16	Right	JME	71.5%	60.93	919
012	23	11	F	12	Right	JME	98.2%	44.48	1288
013	23	15	F	8	Right	JME	6.5%	50.63	241
014	19	10	F	9	Right	JAE	75.8%	61.26	1096
Means	27	15.6		13.7			82.2%	Total: 415.92	Total: 8158

Abbreviations: CAE, childhood absence epilepsy; IGE, idiopathic generalized epilepsy; JAE, juvenile absence epilepsy; JME, juvenile jyoclonic Epilepsy; NES, non‐epileptic seizures, SGE, symptomatic generalized epilepsy.

^a Patient 003 could not tolerate the EMU admission and was discharged after 1 h of simultaneous sqEEG recording; Patient 007 had normal EEG recorded during EMU monitoring. Patients 008 and 013 had poor compliance with the sqEEG recording device during the outpatient setting, thus the low adherence rates.

Three patients classified as IGE based on clinical history were reclassified after the EMU admission: Patient 003 was reclassified as focal epilepsy, Patient 007 had normal EEG during the EMU admission with a non‐epileptic seizure recorded, and Patient 004 was reclassified as SGE. In total, seven patients were analyzed for the purposes of the comparative EEG study.

3.2. EEG results

3.2.1. EEG sensitivity (EMU admission)

Scalp EEG vs sqEEG (manual review)

Two of the nine patients had convulsive seizures recorded in the EMU (a total of three seizures). All three seizures were identified by the reviewers in both modalities.

IED sensitivity assessed by manual review

The ipsilateral sensitivity of interictal epileptiform discharges using sqEEG compared to scalp EEG was 90%. This indicates that 90% of interictal events were detected ipsilateral to the EEG‐recording subcutaneous implant. In addition, the overall detection rate never fell below 75% across different event durations or types. Figure 4 summarizes the sensitivity for all the event durations and morphologies studied for each patient, and overall.

FIGURE 4.

SqEEG performance on monitoring interictal findings across seven patients, three duration categories as <3 s, between 3 to 10 s, and more than 10 s. Results are also classified based on the interictal activity marker types as SW (spike and wave), PSW (polyspike and wave), and FSW (focal sharp and wave). *One event is missed by sqEEG compared to full‐scalp EEG.

Figure 5 provides a comparative analysis of the distribution of interictal epileptiform findings across frequency activity and duration for all seven patients included in the analysis. Each violin plot illustrates the similarities in frequency and duration between scalp EEG and sqEEG recording modalities in monitoring the events. The similarity further reinforces that both recording methods captured the same population of IEDs and seizures.

FIGURE 5.

The distribution of interictal events for all patients based on their frequency and duration across both full‐scalp EEG and sqEEG for SW (spike and wave) and PSW (polyspike and wave). “N” represents the number of events in each category.

Seizure detection rate assessed by seizure detection algorithm

Findings from automated seizure detection during the patient's EMU stay showed a median pre‐screened false detection rate of 3.4 detections per 24 h (range: 1.62 to 22.7). When excluding Patient 004 with symptomatic generalized epilepsy (SGE), the average false detection rate reached 2.9 per 24 h. Detailed results for each subject are included in Supplementary Material S1.

3.2.2. Outpatient sqEEG monitoring

A total of 2941 IED events were recorded across all the participants. Findings from automated seizure detection using UNEEG EpiSight Analyzer in the outpatient setting revealed a median pre‐screened false detection rate of 3.1 detections per 24 h (range: 1.4 to 16.3). When excluding Patient 004, the mean false detection rate was 2.4 per 24 h. Detailed results for each patient, including the electrographic seizure counts, are included in Supplementary Material S1.

3.2.3. Impedance stability

The electrode impedance of the implant (UNEEG SubQ) was measured continuously throughout the study for each participant. All measurements consistently remained low (below 5 kOhm) and showed high stability within individual subjects and can be visualized in Figure S2.

3.2.4. Quality ratings analysis

The average quality ratings for both scalp and sqEEG, as rated by a neurophysiology scientist, are represented in Figure S3. On average, EEG signal was clear and EEG artifact evaluation was possible (i.e., quality rating ranging between 3 and 5) in 86% of tasks for scalp EEG and 85% of tasks for sqEEG. Visually, the EEG morphology of epileptiform discharges was comparable between the two recording modalities as shown in Figure S4.

3.3. Questionnaire data analysis

3.3.1. Subject satisfaction and tolerability questionnaire data

Seven patients completed the Ease of Use, Overall Satisfaction, and SUS questionnaires at the end of study. Participants never or rarely found that the EEG recorder restricted their mobility, work life, spare time (1.7 ± .8), or sleep (2.1 ± 0.7), or made the chronic illness more visible/exposed (2.4 ± 1.5) (Questions 1, 3, and 7). A low score on these questions corresponds to a positive result (Table S2). Participants used the device in public (Question 5) (4.7 ± 0.59) and reported the EEG recorder to be highly acceptable to wear during daytime (mean score on Question 2: 3.9 ± 0.7) and comfortable to wear (Question 4) (4.1 ± 0.7). In addition to this, they reported that people reacted positively toward them while wearing the 24/7 EEG SubQ recorder (Question 6) (3.7 ± 0.8). The patients did not experience unacceptable implant discomfort (2.1 ± 1.5). Mean overall satisfaction scores were 7.4 ± 2.1 on a scale from 1 (extremely unsatisfied) to 10 (extremely satisfied). Furthermore, patients were likely to recommend the 24/7 EEG SubQ to a person with the same condition as themselves (8 ± 2.5) on a scale ranging from 1 (extremely unlikely) to 10 (extremely likely). The mean SUS score at end of the study was 80.4 ± 9.29 (N = 7). This corresponds to a rating of perceived usability between “good” and “excellent.” ¹⁵ Details are included in the Tables S2 and S3.

Three participants reported adverse events; one of those events was related to the implantation procedure and two events were related to coronavirus disease 2019 (COVID‐19). All were categorized as mild and resolved without sequelae.

3.3.2. Degree of difficulty of electrodes implantation

The neurosurgeons completed a questionnaire regarding the implantation and explanation process for all nine patients. Responses suggested that the explanation procedure was easy. The surgeons “strongly agreed” or “agreed” that implantation was easy to perform for eight of the nine patients. In one patient, the surgeon thought that the implantation was not easy to perform and commented that the implant did not slide into the introducer needle. No harm to the participant arose as a result of the more difficult implantation procedure.

3.4. Clinical relevance questionnaires

Clinical relevance questionnaires were completed by the treating neurologist, as part of the study. Overall, meaningful information regarding underlying seizures and epileptiform discharges was gained in all nine patients. Of particular importance, three patients were diagnosed with non‐epileptic seizures during the EMU admission. Additional information was gained on epilepsy self‐management (i.e., medication non‐compliance) in two patients (22%). Patient 008 reported excess seizures during the follow‐up clinic appointments. They were unable to retrieve the completed seizure diary due to extenuating personal circumstances. However, upon examining the sqEEG outpatient data we noted that only one electrographic seizure was recorded. This finding reinforced the recent dual‐diagnosis of IGE and non‐epileptic seizures, and based on this information we decided not to modify ASM. Ongoing seizures were detected during the sqEEG outpatient period in six patients (67%), prompting additional ASM changes.

4. DISCUSSION

This study represents the first investigation that uses unilateral sqEEG to characterize and quantify EEG abnormalities in patients with IGE. Our main goals were to assess the following sqEEG parameters in patients with IGE: (1) EMU and outpatient sensitivity of ictal patterns and IEDs; (2) signal quality and artifact interference; (3) impedance stability; and (4) subject satisfaction and tolerability. We provide evidence that patients with IGE can be reliably monitored for several months with minimally invasive subcutaneous EEG.

SqEEG has been used previously in the diagnosis and detection of focal seizures. ¹⁷ , ¹⁸ We extended the application of minimally invasive sqEEG to IGE. Idiopathic generalized epilepsies account for 20% of all epilepsies. ¹⁹ The ability to record ultra long‐term EEG in the outpatient setting can be a significant advance in the treatment of drug‐refractory IGE given the variability in seizure self‐reporting via seizure diaries. Over‐treatment due to over‐reporting carries risks of more adverse effects of ASMs, and under‐treatment may lead to dire consequences, such as SUDEP and driving accidents.

Clinical seizures can be difficult to define in IGE, as function and awareness can be maintained despite longer bursts of generalized spike and wave complexes. We arbitrarily divided IGEs into three groups according to their duration.

The 24/7 EEG SubQ system reliably detects abnormal brain activity. ¹² , ¹⁸ We showed that, when compared with the seizure detection gold‐standard (i.e., scalp video‐EEG monitoring) the sqEEG system was able to detect all the seizures recorded during the EMU admissions. We found that IED sensitivity and quality rating scores from manual review were excellent, with ipsilateral sensitivity rates of 90% and overall sensitivity rates of 75%. The sensitivity of IEDs remained high despite the lower spatial resolution of the three‐contact sqEEG electrode. Seizure detection rate with automated seizure detection showed false detection rate of 2.9 per 24 h in the EMU and 2.4 per 24 h during out‐patient monitoring. The EEG morphology of seizure onset and ictal rhythms was comparable between scalp EEG and sqEEG, and quality ratings indicated that 85% of sqEEG recordings showed clear signal with straightforward evaluation or evaluation close to optimal. The same quality ratings represented 86% for scalp EEG, demonstrating similar EEG signal quality for the two modalities. Our data accurately complement a recent study using 24/7 EEG SubQ, ²⁰ which showed an overall seizure detection sensitivity of 86% and false detection rate of 2.4 per 24 h for people with focal epilepsy. For comparison, scalp‐EEG‐based seizure detection algorithms provide sensitivities between 75% and 90% and false detection rates of 2.4–120 per 24 h. ²¹

Upon analyzing the subjective EEG quality ratings provided by our neurophysiology scientists, we found that the EEG signal was interpretable in 99% of the tasks across both methods. This finding suggests that sqEEG records were interpretable despite usual artifacts to an extent similar to that of scalp EEG. Our data extend the line of research that demonstrates excellent signal quality of sqEEG signal in the outpatient environment, with clear identification of ictal patterns and IEDs. Furthermore, our comparative analysis of the distribution of interictal epileptiform findings across frequency ranges showed that recordings from sqEEG do not differ from scalp EEG. When examining the frequency and duration of IEDs, the duration of PSW discharges is directly proportional to their frequency, a relationship that was described previously. ²²

One of the key advantages of sqEEG is the ultra‐long duration of the recordings, which is impossible to obtain with either in‐patient or outpatient scalp EEG. Although continuous long‐term recording of cardiac electrophysiological activity has been firmly established in clinical practice for detection of arrhythmias, continuous EEG recording with conventional scalp EEG is limited to several days. Here, we show that high‐quality EEG signal can be safely recorded over many months. The impedance of sqEEG remained below 5 kOhm throughout the recorded duration of 110 days, thus demonstrating high‐quality EEG signal, in agreement with previous findings, ²³ and also in agreement with the ILAE recommendation for routine and sleep EEG recording standards. The main advantage of ultra long‐term EEG recording constitutes accurate objective seizure quantification, which is paramount to ensure adequate seizure control, by bringing potential unreported absence seizures to the fore and allowing additional discussion regarding antiseizure treatment with the patient. The reverse also applies, and this device can detect seizure over‐reporting or a concurrent diagnosis of non‐epileptic seizures, thus allowing ASM rationalization and improving adverse effect profiles. Finally, ultra long‐term EEG recording may have a role in quantifying the risk of driving in people with IGE, although this will require additional studies focused on this matter.

This study demonstrates overall positive participant satisfaction and perceived ease of use of the recording device. The patients had a positive experience and felt confident using the device. The reported device discomfort was rated very low, and the participants were likely to recommend the 24/7 EEG SubQ system to other patients with epilepsy. Questionnaire analysis showed the patients were satisfied with the device, with a mean overall satisfaction score of 7.4, and the system was easy to integrate into their daily routine. Finally, our epilepsy neurosurgeons who performed the implantations found the UNEEG SubQ to be easy to implant, in line with previous reports that the implantation and explantation of sqEEG electrodes are easy to perform. ²⁴ Japaridze et al. used a wearable EEG recording device to develop an automatic algorithm for automatic seizure detection. ²⁵ Although the algorithm had high sensitivity for absence seizures—overall sensitivity for seizures of 79.12%, the device used had limited continuous recording capacity and was less suitable for use in day‐to‐day life due to the device size and its appearance. Compared to the continuous EEG devices with intracranial leads, the subcutaneous EEG is less invasive and has less post‐surgical complications. Although intracranial EEG provides high signal quality, this technique is quite invasive, with higher costs and higher risks of complications. The most common serious adverse event with intracranial EEG recording with the RNS system (NeuroPace, CA, USA) was infection of the implantation site according to a comprehensive safety assessment study. ²⁶ This has not occurred in our study and it was only rarely reported in other subscalp devices. ²⁷ Furthermore, given that the skull is not breached, the risk of disseminating infection to the intracranial structures is intuitively lower. In summary, patients and clinicians found the sqEEG to be safe, well tolerated, and easy to implant.

5. CONCLUSIONS

We complement previous findings supporting the tolerability and signal quality of the ultra long‐term sqEEG device. We also provide the first evidence that sqEEG can be used in people with IGE in an outpatient settings with reliable detection of abnormal brain signals over several months. We believe that sqEEG is a minimal risk and low‐cost alternative to long‐term monitoring compared to in‐patient video‐EEG recordings in select patients. Furthermore, this study shows both the utility and feasibility of using unilateral sqEEG in select patients with IGE to guide seizure detection and treatment adjustment. Further studies and experience in this cohort of patients are warranted.

AUTHOR CONTRIBUTIONS

Tudor Munteanu – First draft preparation, subject recruitment, data collection and processing, data analysis, manuscript revisions. Bryoni Daly and Cara O'Donnell: EEG review and manuscript review. Annette Breen, Tenzin Choekyi, and Breege Staunton‐Grufferty: Subject recruitment, pre‐ and post‐implantation care, data collection, and manuscript review. Donncha O'Brien and Kieron Sweeney: Device implantation and manuscript review. Ronan Kilbride, Peter Widdess‐Walsh, and Hany El‐Naggar: Patient recruitment, EEG review, and manuscript review. Norman Delanty: Study concept, design, supervision, and manuscript review.

CONFLICT OF INTEREST STATEMENT

None of the authors has any conflict of interest to disclose.

ETHICS STATEMENT

We confirm that we have read the Journal's position on issues involved in ethical publication and affirm that this report is consistent with those guidelines.

Supporting information

Data S1:

Munteanu T, Daly B, O’Donnell C, Breen A, Choekyi T, Staunton‐Grufferty B, et al. Unilateral ultra long‐term subcutaneous EEG monitoring in drug‐refractory idiopathic generalized epilepsy. Epilepsia. 2026;67:396–407. 10.1111/epi.18644

DATA AVAILABILITY STATEMENT

The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.