Abstracts

RATIONALE: EEG is an important tool in identification of neonatal seizures. Seizures in newborns are topographically restricted, and it is known that extremely reduced montages may fail to identify such seizures. Current guidelines recommend either a full 10/20 montage (16 electrodes) or a reduced montage (9 electrodes) for recording EEG activity, with additional channels devoted to polygraphic variables. A comparison of these two montages for detection of neonatal seizures has not been reported. With digital technology, it is now possible to re-montage previously recorded EEGs for such a comparison.
METHODS: One hundred and fifty-one neonatal EEG records were obtained from the master database of the Clinical Neurophysiology Laboratory at Children[ssquote]s Hospital, Boston. EEGs were given a study number and identifying markings removed. All EEGs were first analyzed by 2 independent, blinded readers using a reduced montage (RM). The number, location and duration of seizures was recorded. All EEGs were then re-analyzed, by the same blinded readers, using a full montage (FM), and seizure data again recorded. Results of these analyses were compared to evaluate the sensitivity and specificity of the RM compared with the FM for the detection of electrographic seizures.
RESULTS: Electrographic seizures were identified in 31 of the 151 records analyzed. A total of 187 electrographic seizures were identified in these 31 records when the FM was used for display. Using the RM, 166 electrographic seizures were identified in 30 records. RM therefore failed to identify the single seizure occurring in 1 record. The RM was 97% sensitive and 100% specific for identification of electrographic seizures when compared to the FM. In 2 records with multiple seizures, [gt]50% were missed when reading with RM. In 1 record, a single prolonged seizure was misread by both readers as representing 4 separate short seizures when using the RM display, thus overestimating seizure number in this instance. Excluding this case, the rate of underestimation of seizure number was less than 14% for the RM in this study, and seizure activity was missed altogether in only 1 record.
Inter-reader agreement was 100% for identification of ictal records when using either RM or FM. With respect to the number of seizures identified in the ictal records, inter-reader agreement was 74%.
CONCLUSIONS: The potential advantages of using a RM for detection of neonatal seizures are: shorter electrode application time, decreased handling of vulnerable infants, and increased scalp space for performance of cranial ultrasound and IV access. This study illustrates the modest potential for both underestimation and overestimation of neonatal seizure number when using a RM compared to a FM. Despite this, the RM has a high sensitivity and specificity for seizure detection when compared to the FM, and may be adequate for diagnosis and managment of neonatal seizures in these ill infants.
At the end of this activity the participants should be able to discuss the advantages and disadvantages of using a reduced electrode montage rather than a full montage in detection of electrographic seizures in neonates.
[Supported by: Dr Tekgul was supported by a grant from the NATO Science Fellowship Program, under the auspices of The Scientific and Technical Research Council of Turkey (TUBITAK).]