By Peter B. Forgacs, MD
Assistant Professor of Neuroscience and Neurology, Feil Family Brain and Mind Research Institute and Department of Neurology, Weill Cornell Medical College; Visiting Assistant Professor of Clinical Investigation, The Rockefeller University, New York
Dr. Forgacs reports no financial relationships relevant to this field of study.
SYNOPSIS: Researchers found that EEG-R testing, by itself, is not sufficiently reliable to predict neurological outcomes after cardiac arrest.
SOURCE: Admiraal MM, van Rootselaar AF, Homeijer J, et al. EEG reactivity as predictor of neurological outcome in postanoxic coma: A multicenter prospective cohort study. Ann Neurol 2019; May 24. doi: 10.1002/ana.25507. [Epub ahead of print].
Neurological prognostication in patients who regain consciousness immediately after cardiac arrest remains challenging. Current standard clinical practice guidelines recommend a multimodal approach in assessment of neurological prognosis after cardiac arrest, including bedside examination (i.e., presence of brainstem reflexes), evidence of cortical (N20) response on somatosensory-evoked potential (SSEP) examination, laboratory markers of neuronal injury (i.e., levels of neuron specific enolase [NSE]), and imaging evidence (CT and/or MRI) of overwhelming neuronal injury. In addition, the value of many electroencephalographic (EEG) features increasingly is explored in assessment of comatose postcardiac arrest patients, particularly since continuous EEG monitoring became standard of care as part of various targeted temperature management (TTM) protocols. Among these features, lack of EEG-R is considered one such important indicator for poor outcome. In fact, all major U.S. and European guidelines include EEG-R as a prognostic marker after cardiac arrest.
However, none of these guidelines, including the American Clinical Neurophysiology Society (ACNS) Standardized Critical Care EEG Terminology, provide specific descriptions of stimulus administration during testing or precise definitions for determining presence or absence of EEG-R. Furthermore, most studies assessing the relationship of EEG-R and clinical outcomes either have been relatively small or designed retrospectively with variable results. Consequently, the value of EEG-R in neurological prognostication after cardiac arrest remains unclear.
In this large, multicenter, prospective cohort study, Admiraal et al used a rigorous standardized protocol for testing of EEG-R. A total of 160 patients were enrolled in three Dutch hospitals, and EEG-R was assessed twice daily while patients underwent continuous EEG monitoring. The protocol for EEG-R testing included a fixed set of auditory, visual, tactile, and noxious stimuli employed three times in a row at each evaluation. Three experienced EEG readers blinded to all clinical variables and patient outcomes independently assessed EEG-R, defined as a change in EEG amplitude or frequency at least twice in response to any of the stimuli. Increased muscle activity or stimulus-induced rhythmic or periodic discharges (SIRPIDS) were not considered as EEG-R. If the raters disagreed, a majority vote was used to decide the presence of EEG-R.
As a secondary analysis, EEG-R also was re-evaluated in a consensus meeting in cases without unanimous decision. Thresholds for accurate prediction of good or poor outcomes were predefined based on the presence or absence of EEG-R, respectively, both using EEG-R alone or added to a multimodal prediction algorithm. Multimodal assessments included brainstem reflexes, N20 response of SSEP at 72 hours, and graded EEG categories based on background abnormalities in addition to EEG-R.
The main findings of the study showed that the absence of EEG-R predicted poor outcome with a specificity of 82% (below the predefined > 95%) and a sensitivity of 73%, while the presence of EEG-R predicted good outcome with a specificity of 73% (below the predefined > 80%) and a sensitivity of 82%. When EEG-R was added to a multimodal model, specificity of poor outcome prediction increased only marginally (from 98% to 99%), and specificity of good outcome prediction increased moderately (from 70% to 89%).
Notably, while inter-rater reliability was relatively good, there was poor agreement between the majority vote vs. the consensus meeting (ICC of 0.40). Thus, the authors concluded that EEG-R testing alone is not sufficiently reliable for neurological outcome prediction after cardiac arrest. In addition, EEG-R adds no substantial value to multimodal assessments for poor outcome prediction, but it may add value to the prediction of good outcomes.
This was the first, prospectively designed, large, multicenter study assessing the value of EEG-R in neurological prognostication after cardiac arrest. Even though EEG-R is recommended by practice guidelines as an appropriate indicator for outcomes in patients who remain comatose after severe anoxic brain injury, there have been no previous studies of this scale assessing its prognostic value using a standardized, prospectively designed protocol. The results of this study suggest that even using a carefully executed protocol with a systematic approach, EEG-R is not sufficiently reliable to predict neurological outcomes in post-cardiac arrest patients.
Major efforts are devoted to find early but accurate tools for assessing neurological recovery after cardiac arrest. Recent advancements in acute medical care and novel therapeutic interventions, such as various targeted temperature protocols (including therapeutic hypothermia), have led to improved survival and better neurological outcomes after severe anoxic brain injuries. Nevertheless, in current clinical practice, withdrawal of life-sustaining therapy (WLST) decisions continue to drive mortality in patients who do not regain consciousness readily after cardiac arrest.
Therefore, the results of most studies assessing prognosis carry the risk that self-fulfilling prophecies may affect the outcomes and limit the interpretation of results. While the ratio of WLST was relatively low and EEG-R findings were not used in clinical decision-making, the results should be interpreted with caution.
This study underscores the immense continued need for additional studies to develop highly precise and reproducible clinical or diagnostic assessments for accurate early neurological prognostication of comatose post-cardiac arrest patients.