To characterize neuroprognostication practices and provider documentation surrounding end-of-life decision-making in comatose post-cardiac arrest patients.

Withdrawal of life-sustaining therapy (WLST), often due to perceived poor neurologic prognosis (WLST-N), is the most common cause of death following cardiac arrest, regardless of arrest etiology. Current international guidelines for neuroprognostication after cardiac arrest promote a multi-modal approach, including pupillary and corneal reflexes, serum neuron-specific enolase (NSE), electroencephalography (EEG), somatosensory evoked potential (SSEP) and neuroimaging. We hypothesized that the rationale documented for WLST-N would not be supported by objective poor prognostic findings in a proportion of patients.

We performed a retrospective chart review of post-cardiac arrest patients from 2011 to 2018 at two tertiary academic medical centers, abstracting data on WLST, clinical examination findings, and electrophysiologic and neuroimaging data. Tests and practices underlying neuroprognostication were summarized by descriptive statistics.

Of 112 WLST-N patients, the median day of WLST-N was post-arrest day 6 (IQR 4-11 days). Pupillary and corneal reflexes were present on day 3 post-arrest (or post-rewarming if TTM-treated) in 49.1% and 34.8% of patients, respectively. SSEP results were cited as a rationale for WLST-N in 8.0% of patients and NSE results in 0.9%. EEG results were documented as a rationale for WLST-N in 25.0% of patients. MRI and CT results were cited as a rationale for WLST-N in 22.3% and 21.4% of patients, respectively. Variations of “unlikely to have meaningful improvement” were stated for 55.4% of WLST-N patients.

Pessimistic, often vague neuroprognostic impressions were common despite limited reference to results of neuroprognostic tools, including NSE and SSEPs, and frequent reliance on neuroimaging results, which have conflicting levels of evidence. Consequently, current practices challenge granular characterization of WLST and may perpetuate confirmation bias of poor outcomes.