Post-Cardiac Arrest Neuroprognostication
Key Takeaways
Defer formal prognostic assessment until at least 72 hours after ROSC, or 72 hours after rewarming when hypothermia was used, and longer when confounders persist.
Bilaterally absent pupillary responses at the appropriate time and technically valid absent N20 responses at least 48 hours after ROSC are among the more reliable findings in the NCS guideline.
Absent corneal reflexes, poor motor response, early myoclonus without EEG, and an isolated NSE value are not reliable standalone predictors.
Use delayed EEG and imaging within supported time windows and interpret findings together.
An indeterminate prognosis warrants continued observation and communication consistent with patient goals.
Prognosis Is Different From Diagnosis of Death
Neuroprognostication estimates the likelihood of later functional outcomes after hypoxic-ischemic injury. It is not a determination of death by neurologic criteria and does not provide an exact personal forecast. Define the outcome and time horizon being discussed: survival, awareness, communication, independence, and need for assistance are different outcomes. A score of mRS 3 includes need for some help and should not be called complete functional independence.
The major methodological hazard is a self-fulfilling prophecy: a test seems to predict death when clinicians withdraw treatment because of that test. Studies with early treatment withdrawal, incomplete blinding, small samples, or poorly controlled confounders may overestimate certainty. Prefer validated findings and communicate both their evidential strength and their practical limits.
Timing and Confounders
The NCS 2023 guideline recommends deferring formal prognostic assessment until at least 72 hours after ROSC if hypothermia was not used, or at least 72 hours after rewarming if it was. This is a minimum framework, not a deadline to declare the prognosis hopeless. Sedative accumulation, kidney or liver dysfunction, seizures, metabolic disease, and systemic instability can require a longer period. Persistent coma at that point does not itself establish poor outcome.
Review drug dose, duration, clearance, active metabolites, temperature exposure, and residual neuromuscular blockade. A nominal half-life in healthy volunteers may underestimate the effect of a prolonged infusion in critical illness. Correct severe glucose, electrolyte, gas-exchange, endocrine, and acid–base disturbances. Examine repeatedly under appropriate conditions rather than interpreting one early assessment as final.
Communication can begin immediately. Explain uncertainty, the evaluation plan, known patient preferences, and what additional observation may resolve. A minimum prognostic waiting period is not a ban on family meetings or on considering a patient's established treatment limits. Decisions based on premorbid wishes or nonneurologic illness require their own explicit reasoning.
Clinical Examination and Evoked Potentials
| Finding | Interpretation in the appropriate setting |
|---|---|
| Bilaterally absent pupillary light response at least 72 hours after ROSC | A reliable poor-outcome predictor under the guideline's assumptions; exclude ocular and technical confounders and confirm carefully |
| Absent corneal reflexes alone | Not sufficiently reliable as a standalone predictor |
| Absent or extensor motor response alone | Not sufficiently reliable as a standalone predictor |
| Bilaterally absent cortical N20 at least 48 hours after ROSC | Reliable when peripheral and cervical responses are preserved and the study is technically adequate |
Automated pupillometry can improve objectivity, particularly with small pupils, but no universal NPi cutoff should be advertised as producing zero false positives in every patient. Drugs and ocular pathology still require consideration. If manual testing suggests absent responses, confirm the observation using appropriate expertise and equipment where available.
For somatosensory evoked potentials, verify that stimulation reaches peripheral and cervical pathways and that artifact does not obscure the cortical response. A technically uninterpretable study is not an absent response. Severe hypothermia and other limitations matter. An absent N20 finding is a prognostic result, not proof of death or a substitute for the rest of the assessment.
EEG, CT, and MRI
Suppressed or burst-suppression EEG background, with or without periodic discharges, at least 72 hours after ROSC or rewarming and without sedation or other confounders is a moderately reliable poor-outcome finding in the NCS framework. Earlier patterns are useful for seizure assessment and trends but should not be assigned the same delayed prognostic meaning. Lack of reactivity, periodic discharges, or electrographic seizures alone do not justify a guaranteed outcome statement.
On CT obtained at least 48 hours after ROSC, diffuse loss of gray-white differentiation with sulcal effacement across vascular distributions can be moderately reliable. A normal early CT does not guarantee recovery. A gray-white ratio depends on acquisition and measurement methods; do not treat one universal decimal cutoff as an invariant biological boundary.
MRI showing diffuse restricted diffusion involving cortex and deep gray structures across vascular territories, assessed approximately 2–7 days after ROSC, can also be moderately reliable. Seizures and other causes of restriction must be considered. Quantitative ADC methods are not automatically interchangeable among scanners and studies. Transport and device safety determine whether the diagnostic benefit justifies obtaining MRI.
Biomarkers and Myoclonus
Neuron-specific enolase (NSE) can add information in selected multimodal approaches, but hemolysis, assay differences, and extracerebral sources complicate interpretation. NCS does not endorse a universal NSE threshold measured within 72 hours as a reliable standalone predictor. A value above 60 or 80 micrograms/L must not be represented as proving irreversible cortical necrosis with essentially no error.
Early myoclonus without EEG is not a reliable standalone marker of poor outcome. Assess whether movements have an electrographic correlate, the EEG background, sedation, and the evolving clinical state. Action myoclonus in an awake or recovering patient may represent Lance–Adams syndrome and can coexist with meaningful recovery. Timing alone does not perfectly separate these syndromes.
Integrate and Communicate
Seek concordant information from independent modalities, investigate disagreements, and avoid counting several manifestations of the same confounder as separate proof. If prognosis remains indeterminate, explain that uncertainty and consider extended observation when consistent with the patient's goals. Describe possible functional states concretely and avoid equating disability with death or assuming the family shares the clinician's valuation of dependency.
Source checked October 17, 2026: NCS 2023 neuroprognostication guideline and AHA 2025 post-arrest algorithm.
A hypothermia-treated patient remains comatose 24 hours after rewarming while sedative clearance is uncertain. What is appropriate?
Defer formal prognostic assessment, address confounders, and continue appropriate observation and communication
Declare a certain poor neurologic outcome
Use one NSE value as definitive
Diagnose BD/DNC from coma alone
Which finding is more reliable in the NCS framework when technically valid and appropriately timed?
Absent corneal reflex alone
Bilaterally absent N20 at least 48 hours after ROSC with preserved peripheral and cervical responses
Early myoclonus without EEG
One high NSE concentration regardless of hemolysis
What should be done when prognostic modalities disagree?
Count every abnormality as independent proof
Choose the most pessimistic early result automatically
Reassess confounders, technical validity, timing, and the need for further observation
Declare exact individual certainty
Sections you finish are checked off in the contents.