- After a one-hour training session, neuro-ICU nurses could detect seizures in adults with moderate sensitivity using compressed spectral arrays
- The false-positive rate for nurses was about twice as high as that for neurophysiologists
- Nurse review of spectrograms may offer earlier seizure detection by virtue of more frequent evaluation
Subscribe to the latest updates from Neuroscience Advances in Motion
For detecting seizures in adults, the gold standard is a full-montage review of continuous electroencephalogram (EEG) data by an experienced electroencephalographer. That work is quite burdensome, though, in terms of mental strain and the time required.
Hospitals have been working for years to find ways to extract the most important data from continuous EEG and compress it for a more rapid review. One result is compressed spectral arrays (CSAs), which put hours of EEG activity into three-dimensional images that demonstrate EEG trends in real time.
Edilberto Amorim, MD, Craig Williamson, MD, and colleagues, mentored by Neurologist M. Brandon Westover, MD, PhD, investigated continuous EEG records from 30 adults cared for at Massachusetts General Hospital between September 2011 and February 2012. Two electroencephalographers reviewed the raw EEG data and scored each two-hour segment for the presence or absence of seizures. These segments were converted to 40 CSA displays.
Nurses have the most frequent interactions with patients, and those using CSA for bedside EEG review might bring seizures to physician attention earlier. However, for both nurses and neurophysiologists who perform CSA screening, there is a risk of an increase in false alarms and resultant "alarm fatigue."
A few small studies have explored whether ICU nurses can detect seizures reliably using CSAs. In the largest study to date, nurses did well on a test of CSA screening after a brief training session. The results of the study were reported in the Journal of Clinical Neurophysiology.
Thirty-three neuro-ICU nurses at the University of Michigan and Mass General volunteered for the study. Their training, conducted by one of the investigators, consisted of listening to a 15-minute presentation, reading 10 CSA displays as a pre-test and then discussing those CSAs with the trainer.
Along with four neurophysiologists, the nurses took a web-based test that involved viewing the 40 CSA displays and deciding whether a seizure had occurred. Raw EEG data was not made available. To mimic actual ICU practice, the set of images included many cases with a significant artifact and periodic patterns on the ictal–interictal continuum.
The researchers compared the test-takers' decisions with the decisions made by the experts (the two encephalographers who read the raw EEG data). The overall accuracy for the presence or absence of seizures was 55.7% for nurses and 67.5% for neurophysiologists.
Notably, nurses and neurophysiologists had comparable sensitivities, but the false-positive rate for nurses was about twice as high as for neurophysiologists.
The use of CSA displays without conventional visual analysis of raw EEG might be inadequate for detecting seizures. On the other hand, more comprehensive training of nonexperts, including real-time practice sessions, might improve accuracy.
Visit the Neurology Department
Refer a patient to the Neurology Department