[MS2-2B-4] “Amplitude-integrated EEG with raw trace” compared with “conventional video EEG” for seizure detection in term neonates: A diagnostic accuracy study
[Background] Role of amplitude-integrated electroencephalography (aEEG) for detection of neonatal seizures is a topic of debate.
[Method] A prospective study comparing “aEEG with raw trace” with the “gold-standard” conventional video electroencephalogram (c-VEEG) in term infants at risk of seizures was conducted. Simultaneous recording of aEEG and c-VEEG were done for 24 hours for each infant. “aEEG with raw trace” was interpreted by two neonatal readers; c-VEEG was interpreted by a neurologist independently.
[Results] Thirty-five infants were enrolled in the study. All seven infants who had seizures on c-VEEG were also diagnosed to have seizures on aEEG, resulting in a sensitivity of 100%. However, seven other infants were incorrectly diagnosed to have seizures on aEEG and hence specificity was 75% (95% CI 59, 91).
Among the seven infants with confirmed seizures, there were 169 “individual seizure” episodes, of which only 57 were picked up on aEEG. For the detection of these “individual seizures,” sensitivity and specificity were 33% (95% CI 26, 41) and 70% (95% CI 64, 77) respectively. There was a fair inter-observer agreement between the aEEG readers for detection of individual seizures (Kappa=0.38).
[Conclusions] “aEEG with raw trace” has excellent sensitivity to detect “patients with seizures” but comes with a cost of over-diagnosis. It also failed to identify majority of the “individual seizures”. Therefore, aEEG is a good tool for identifying infants who would benefit from continuous c-VEEG monitoring. aEEG with raw trace cannot be recommended as a mainstay diagnostic tool for continuous monitoring and treatment of neonatal seizures.
[Method] A prospective study comparing “aEEG with raw trace” with the “gold-standard” conventional video electroencephalogram (c-VEEG) in term infants at risk of seizures was conducted. Simultaneous recording of aEEG and c-VEEG were done for 24 hours for each infant. “aEEG with raw trace” was interpreted by two neonatal readers; c-VEEG was interpreted by a neurologist independently.
[Results] Thirty-five infants were enrolled in the study. All seven infants who had seizures on c-VEEG were also diagnosed to have seizures on aEEG, resulting in a sensitivity of 100%. However, seven other infants were incorrectly diagnosed to have seizures on aEEG and hence specificity was 75% (95% CI 59, 91).
Among the seven infants with confirmed seizures, there were 169 “individual seizure” episodes, of which only 57 were picked up on aEEG. For the detection of these “individual seizures,” sensitivity and specificity were 33% (95% CI 26, 41) and 70% (95% CI 64, 77) respectively. There was a fair inter-observer agreement between the aEEG readers for detection of individual seizures (Kappa=0.38).
[Conclusions] “aEEG with raw trace” has excellent sensitivity to detect “patients with seizures” but comes with a cost of over-diagnosis. It also failed to identify majority of the “individual seizures”. Therefore, aEEG is a good tool for identifying infants who would benefit from continuous c-VEEG monitoring. aEEG with raw trace cannot be recommended as a mainstay diagnostic tool for continuous monitoring and treatment of neonatal seizures.