Detection of Neonatal Seizures
EEG monitoring using conventional video-EEG (cEEG) or amplitude-integrated EEG (aEEG) is essential for reliable detection of neonatal seizures, as clinical observation alone is insufficient and misses the majority of seizures. 1
Why Clinical Observation Fails
The fundamental problem with detecting neonatal seizures is that most seizures have no visible clinical manifestations (electrographic-only seizures), and when movements do occur, differentiating true seizures from other abnormal movements is extremely difficult 1. Clinical observation alone can only reliably diagnose focal clonic and focal tonic seizures—all other seizure types require EEG confirmation 1.
Gold Standard Detection Methods
Conventional Video-EEG (cEEG)
- This is the gold standard for seizure detection 1
- Provides comprehensive brain activity monitoring
- Allows correlation of electrical activity with clinical manifestations
- Recommended by ILAE, International Neonatal Consortium, European Medicines Agency, US FDA, Brighton Collaboration, and American Clinical Neurophysiology Society 1
Amplitude-Integrated EEG (aEEG)
- Can be used in addition to or when cEEG is unavailable 1
- Important caveat: Sensitivity and specificity are variable, so it cannot be the sole method for seizure detection 1
- More practical for continuous bedside monitoring in resource-limited settings
- Requires review of the 'raw' EEG waveform when seizures are suspected on aEEG traces 2
Practical Detection Algorithm
For neonates at risk of seizures (hypoxic-ischemic encephalopathy, stroke, hemorrhage, infection):
Initiate continuous EEG monitoring using cEEG if available, or aEEG as an alternative 1
Look for these specific patterns:
Do not rely on clinical signs alone - even if you see suspicious movements, confirm with EEG 1
Be aware of electroclinical uncoupling - after starting antiseizure medication, clinical manifestations may cease while electrical seizures continue 1
Emerging Technologies
Recent advances show promise for improving detection:
- Audio playback of EEG: Playing back EEG at 60× speed generates a distinct "buzz-type sound" for seizures, with 96% sensitivity for rapid review 2
- Machine learning systems: Achieving 80-99% accuracy in automated detection, though not yet ready for standalone clinical use 4, 5
- Multi-sensor wearable platforms: Combining ECG, respiration, and movement data shows promise for reducing false alarms 6
Critical Pitfalls to Avoid
- Never diagnose or rule out seizures based on clinical observation alone - you will miss most seizures and treat non-seizure events 1
- Don't assume treatment worked based on cessation of movements - check EEG to ensure electrical seizures have stopped 1
- Don't delay EEG monitoring - only 11% of seizures are treated within 1 hour of onset, partly due to delayed detection 4
- Don't use aEEG as your only monitoring tool if cEEG is available - its variable sensitivity means seizures can be missed 1
Each neonatal unit should have a standardized pathway for seizure detection and management that incorporates EEG-based monitoring 1.