Assessing neurological prognosis in post-cardiac arrest patients from short vs plain text EEG reports: A survey among intensive care clinicians

Abstract

Background: Electroencephalography (EEG) patterns are predictive of neurological prognosis in comatose survivors from cardiac arrest but intensive care clinicians are dependent of neurophysiologist reports to identify specific patterns. We hypothesized that the proportion of correct assessment of neurological prognosis would be higher from short statements confirming specific EEG patterns compared with descriptive plain text reports.

Methods: Volunteering intensive care clinicians at two university hospitals were asked to assess the neurological prognosis of a fictional patient with high neuron specific enolase. They were presented with 17 authentic plain text reports and three short statements, confirming whether a "highly malignant", "malignant" or "benign" EEG pattern was present. Primary outcome was the proportion of clinicians who correctly identified poor neurological prognosis from reports consistent with highly malignant EEG patterns. Secondary outcomes were how the prognosis was assessed from reports consistent with malignant and benign patterns.

Results: Out of 57 participants, poor prognosis was correctly identified by 61% from plain text reports and by 93% from the short statement "highly malignant" EEG patterns. Unaffected prognosis was correctly identified by 28% from plain text reports and by 40% from the short statement "malignant" patterns. Good prognosis was correctly identified by 64% from plain text reports and by 93% from the short statement "benign" pattern.

Conclusion: Standardized short statement, "highly malignant EEG pattern present", as compared to plain text EEG descriptions in neurophysiologist reports, is associated with more accurate identification of poor neurological prognosis in comatose survivors of cardiac arrest.

Keywords: ACNS nomenclature; Cardiac arrest; EEG; Neurological prognosis; Prognostication.