Epileptiform discharges, electrographic seizures, and electroclinical seizures during paediatric sevoflurane anaesthesia: a systematic review and proposal for standard definitions

Editor—Since the advent of sevoflurane administration to children, case reports have described unexpected, involuntary behaviours suggestive of epileptic seizures.¹ Subsequent studies obtaining EEG correlates2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14 have relied on nomenclature established in the neurology–epileptology literature,¹⁵ which is unfamiliar to most anaesthesiologists. We performed a systematic review of studies of epileptiform EEG during sevoflurane to provide clear definitions of terms used in this literature and to help contextualise the clinical significance of the findings.

Our targeted clinical research question was formulated using the patient/population, intervention, comparison, and outcomes (PICO) framework and registered a priori in the International Prospective Register of Systematic Reviews (PROSPERO) on March 19, 2021 (CRD42021237719). A Boolean search string was generated incorporating ‘pediatric’ (age 0–18 yr), ‘sevoflurane’, and ‘electroencephalogram changes’ to search PubMed, Embase, Cochrane Library, Web of Science Core Collection, ClinicalTrials.gov, and Google Scholar. Literature was managed using COVIDENCE (Cochrane Collaborative Group, London, UK) and EndNote 20.2 (Clarivate, Philadelphia, PA, USA). There were 495 references imported for screening. After title and abstract screening, 63 full-text references were selected for further assessment, of which 50 were excluded most commonly because of utilisation of (i) wrong outcome measures (58%), (ii) inappropriate study design (22%), or (iii) inappropriate patient population (10%) (Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) diagram; Supplementary Fig 1). The final systematic review consisted of 13 studies.2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14 An assessment for risk of bias was made using the Newcastle–Ottawa Quality Assessment Scale.

The majority of studies were observational^3,5, 6, 7, 8, 9, 10, 11, 12^,14 (10/13; 76.9%); three randomised controlled studies assigned children to different induction techniques,² ventilation strategies,⁴ or anaesthetic agents¹³ (Table 1). Heterogeneous EEG outcome classifications resulted in variable findings. Overall, abnormal EEG patterns were detected in 192 children out of a pooled sample size of 621 subjects (prevalence: 30.9%; 95% confidence interval: 27.3–34.5%). The precision of the prevalence estimate was affected by heterogeneity and bias that exists within and between studies, particularly with regard to midazolam premedication and the number of channels recorded. For example, amongst children who were administered midazolam, one study detected no abnormalities² whilst another study observed multifocal polyspikes in 95% of children⁶ (overall prevalence: 115/293; 39.2%).^2,5, 6, 7^,9,10,12 In contrast, the prevalence of EEG abnormalities observed amongst children who did not receive midazolam was 77/328 (23.5%).^{3,4,8,11,13,14}

Table 1.

Studies selected for inclusion in the systematic review. ∗Information about sevoflurane concentrations provided if reported in individual studies. ^†Major epileptoid signs: rhythmic polyspikes, rhythmic polyspikes and burst suppression, periodic epileptiform discharges, and electrographic seizures. ^‡Age range not reported. ^¶Seven subjects with EEG abnormalities after anaesthesia and surgery not counted in prevalence estimate. ^§Minor epileptoid signs: spikes, spike and wave, or polyspikes.

Study (year of publication)	n	Age	Sevoflurane∗	Premedication	EEG recordings	EEG outcomes	Seizures (electrographic or electroclinical)
Constant and colleagues2 (1999)	32	2–12 yr	Induction (various dosing)	Midazolam (0.5 mg kg−1 per rectum)	16 channels	No epileptiform discharges detected	None detected
Conreux and colleagues3 (2001)	20	1–8 yr	Induction (8% in O2 100%)	None	Bipolar eight channels	Two (10%) abnormal myoclonic movements with paroxysmal spike waves or polyspikes	None detected
Vakkuri and colleagues4 (2001)	31	2–12 yr	Induction (8% in N2O 66.6%)	None	Bipolar four channels	17 (54.8%) interictal epileptiform discharges	None detected
Nieminen and colleagues5 (2002)	30	3–8 yr	Induction and maintenance	Midazolam (0.1 mg kg−1 i.v.)	Five channels	No epileptiform discharges detected	None detected
Sonkajärvi and colleagues6 (2009)	20	4–10 yr	Induction (8% in N2O 50%)	Midazolam (0.4 mg kg−1 p.o.)	21-channel international 10–20	19 (95%) multifocal polyspikes	None detected
Moran and colleagues7 (2011)	23	3–4.5 months	Maintenance	Midazolam (unknown dose)	Eight-channel referential	Six (26.1%) epileptiform activities	None detected
Gibert and colleagues8 (2012)	76	3–11 yr	Maintenance (N2O 50%)	Hydroxyzine (1 mg kg−1 p.o.)	Two-channel bispectral index	33 (43.4%) major epileptoid signs†; one (1.3%) tonic–clonic seizure	One subject
Schultz and colleagues9 (2012)	70	0.6–8 yr	Induction (8% in O2 100%)	Midazolam (0.5 mg kg−1 p.o.)	Two-channel Narcotrend	14 (20%) with abnormalities; 10 delta with spikes; 10 rhythmic polyspikes; four periodic epileptiform discharges	None detected
Kreuzer and colleagues10 (2014)	100	4.6 [3] yr‡	Induction (6% or 8% in O2 100%)	Midazolam (0.5 mg kg−1 p.o.)	Two-channel Narcotrend	64 (64%) epileptiform activities; 38 in sevoflurane 8%; 26 in sevoflurane 6%	None detected
Stolwijk and colleagues11 (2017)	111	0–32 days	Induction (0.04–2.5%)	None	Two-channel BrainZ	11 (9.9%) epileptic activities; four during anaesthesia/surgery; seven after surgery¶	None detected
Koch and colleagues12 (2018)	18	0.5–8 yr	Induction (6%)	Midazolam (0.7 mg kg−1 p.o.)	Two-channel Narcotrend	12 (66.7%) epileptiform discharges	None detected
Rigouzzo and colleagues13 (2018)	36	5–18 yr	Induction and maintenance	Hydroxyzine (1 mg kg−1 p.o.)	Two-channel bispectral index	17 minor epileptoid signs§; 12 major epileptoid signs†	None detected
Chao and colleagues14 (2020)	54	0.1–3 yr	Induction (8% in O2 100%)	None	21-channel international 10–20	Four (7.5%) with abnormalities; three interictal epileptiform discharges; one electrographic seizure	One subject

Additional sources of heterogeneity included subject age and phase of anaesthesia. Most studies (10/13; 76.9%) included older children/teenagers,2, 3, 4, 5, 6^,8, 9, 10^,12,13 but some included infants and neonates.^7,9,11,12,14 Changes in excitation/inhibition balance during brain development¹⁵ can markedly affect the incidence and types of epileptiform discharges during sevoflurane anaesthesia and the anti-epileptic effect of midazolam premedication. Whilst most studies assessed sevoflurane induction,2, 3, 4, 5, 6^,9, 10, 11, 12, 13, 14 others assessed the period of anaesthetic maintenance.^5,7,8,13 Differences in sevoflurane dosing, rapidity of administration, and recording duration also contributed to heterogeneity of EEG changes.

Abnormal EEG patterns have been well studied in paediatric neurology–epileptology,¹⁶ and some are associated with clinical epilepsy syndromes; their clinical significance and prognostic implications have been elucidated over time. Any extrapolation to the EEG during anaesthesia should be made with caution. Even in EEGs in the absence of anaesthesia, a finding of epileptiform discharges in isolation cannot diagnose epilepsy and is not synonymous with future epilepsy. The clinical significance of EEG abnormalities during sevoflurane anaesthesia in otherwise normal children with no known history of neurological disorder is unclear and warrants further research.

Recording devices with a low number of channels (e.g. bispectral index or Narcotrend) may be limited in their ability to detect epileptiform changes because they collect data from a smaller region of the head. Fewer channels also limit the ability to distinguish electrode artifacts from true EEG findings and whether findings are focal or generalised. Most studies used recordings from less than eight EEG channels (8/13; 61.5%),^4,5,8, 9, 10, 11, 12, 13 whereas fewer studies recorded data from at least eight channels (5/13; 38.5%).^2,3,6,7,14 Studies monitoring the entire head suggest the prevalence of abnormal EEG patterns may be relatively low (7.5%) in low-risk populations.¹⁴

We present the following EEG definitions and propose a simplified version of The International League Against Epilepsy (ILAE)¹⁷ and the American Clinical Neurophysiology Society (ACNS)¹⁸ classifications to enhance the rigour and reproducibility of paediatric anaesthesia EEG research. Future studies could utilise these standard definitions of EEG changes during anaesthesia and classify them accordingly:

• (1)
  No interictal epileptiform discharges or electrographic seizures

  • (1a)
    Normal EEG
  • (1b)
    Focal or generalised slowing, including rhythmic delta activity (see below)
• (2)
  Interictal epileptiform discharges, defined as spikes, polyspikes, sharp waves, or spike and wave complexes, reflecting possible areas of cortical irritability and potential epileptogenicity that are

  • (2a)
    Isolated/sporadic epileptiform discharges, meaning they occur singly without repetition or periodic recurrence
  • (2b)
    Repetitive/periodic epileptiform discharges, meaning they occur repetitively, sometimes regularly at a specific frequency
• (3)Electrographic or electroclinical seizures, with electrographic seizure referring to observed salient EEG findings alone without clinical manifestations (e.g. no convulsions or other motor behaviours), in contrast to electroclinical seizures, which are characterised by both salient EEG features and clinical manifestations, such as tonic or clonic movements.¹⁸ Despite this distinction, electrographic and electroclinical findings are included in the same category because both entities ultimately describe seizure activity, regardless of whether obvious clinical manifestations are present.

Focal slowing in the EEG, defined as abnormal slower frequencies occurring over part of the head and abnormal generalised slowing, including non-evolving rhythmic delta (0–3 Hz) activity (Category 1b), have been classified as epileptiform findings in some studies. However, this is not the case, and these patterns should not be classified as epileptiform discharges.^17,18

This simplified classification system provides a clear framework to define epileptiform EEG changes. This standardisation reduces heterogeneity and enhances the ability to combine or pool findings that might be important in future systematic reviews and meta-analyses. It would also help ensure that salient neurophysiological features during anaesthesia can be rigorously and reproducibly identified to assess associations with short-term or longer-term clinical outcomes of interest and importance to the paediatric anaesthesia community. In this way, multidisciplinary paediatric anaesthesia research can continue to bridge the gaps in knowledge of the effects of anaesthesia in the developing brain and their clinical significance.

Funding

Supported by a Foundation for Anesthesia Education and Research (FAER) Medical Student Anesthesia Research Fellowship (MSARF, to DI) and US National Institutes of Health (NIH) National Center for Advancing Translational Science (NCATS) Einstein-Montefiore Clinical and Translational Science Awards (CTSA) Program Grant Numbers UL1 TR002556 and KL2 TR002558 (to JC).

Acknowledgements

The authors thank the authors of the studies selected for inclusion in this systematic review for responding to questions clarifying their methodology, data, and findings.

Declaration of interest

The authors declare no conflicts of interest.

Appendix A. Supplementary data

The following is the Supplementary data to this article: