Return of spontaneous circulation restores a pulse, not normal cerebral physiology. The term “anoxic brain injury” is commonly used, but most resuscitated patients have sustained a variable mixture of complete no-flow ischemia, low-flow hypoxia–ischemia during cardiopulmonary resuscitation, heterogeneous reperfusion, and subsequent systemic insults. Post–cardiac arrest hypoxic–ischemic brain injury is therefore usually the more accurate construct. Its severity cannot be inferred from arrest duration alone, and the comatose examination immediately after ROSC is not a prognosis. It is the starting point for a second resuscitation whose purpose is to prevent potentially salvageable tissue from being lost to hypoxemia, hypocapnia, hypotension, fever, seizures, dysglycemia, and recurrent cardiac or respiratory failure.

During circulatory arrest, cerebral oxygen delivery collapses and oxidative phosphorylation stops. ATP depletion disables membrane ion pumps, producing potassium efflux, sodium and water influx, depolarization, cytotoxic edema, glutamate release, calcium entry, and activation of proteases and phospholipases. Cardiopulmonary resuscitation supplies only partial and spatially uneven flow. ROSC then repolarizes some cells but simultaneously exposes injured mitochondria and endothelium to oxygenated blood. Reactive oxygen species, calcium overload, mitochondrial permeability transition, blood–brain barrier disruption, microvascular thrombosis, leukocyte adhesion, and neuroinflammation extend injury after circulation has been restored. Reperfusion is also heterogeneous: transient hyperemia may be followed by delayed hypoperfusion and microvascular “no-reflow,” while impaired oxygen extraction can produce tissue hypoxia even when arterial oxygen content and global cerebral blood flow appear adequate. The hippocampi, cerebral cortex, cerebellar Purkinje cells, thalami, and basal ganglia are particularly vulnerable, but injury is not anatomically or temporally uniform. The contemporary International Liaison Committee on Resuscitation framework therefore describes overlapping phases of ischemic depolarization, reperfusion repolarization, dysregulation, and recovery or repair rather than a single completed anoxic event at the moment of ROSC (Perkins et al., 2024; Sandroni et al., 2021).

The first post-ROSC decision is whether the brain was the victim of the arrest, the cause of the arrest, or both. A presumed primary cardiac arrest should be revised when the prodrome included sudden headache, focal deficit, seizure, unexplained vomiting, respiratory deterioration, trauma, or toxic exposure. Subarachnoid hemorrhage, intracerebral hemorrhage, large ischemic stroke, cerebral venous thrombosis, status epilepticus, pulmonary embolism, airway obstruction, opioid toxicity, severe metabolic disturbance, and environmental hypothermia require treatments that generic post-arrest care cannot provide. Obtain the history from witnesses and emergency personnel while reviewing the initial rhythm, no-flow and low-flow intervals, defibrillations, drugs, airway events, end-tidal carbon dioxide, and hemodynamics. A 12-lead ECG is required; early echocardiography helps distinguish myocardial stunning, focal ischemic dysfunction, right-ventricular failure, tamponade, and hypovolemia. Head-to-pelvis CT may be reasonable when the cause or resuscitation complications remain uncertain, but imaging must be directed by the suspected pathology and balanced against transport instability. Persistent ST elevation, cardiogenic shock, recurrent ventricular arrhythmia, or convincing ongoing ischemia supports emergent coronary angiography. In a stable comatose patient without ST elevation, shock, electrical instability, or ongoing ischemia, routine immediate angiography has not improved outcomes over a delayed or selective strategy (Hirsch et al., 2025).

Oxygen should initially be treated as a rescue drug and subsequently as a titrated drug. Until pulse oximetry or an arterial blood gas is reliable, the 2025 American Heart Association guideline recommends 100% inspired oxygen because an early attempt to restrict oxygen can inadvertently produce hypoxemia. Once measurement is dependable, leaving the patient indefinitely at an FIO₂ of 1.0 is not neuroprotection.

Once hemoglobin is nearly saturated, raising PaO₂ dramatically adds little oxygen content through the dissolved component, while potentially increasing reactive oxygen species and cerebral vasoconstriction. The competing danger is that an injured brain may have impaired extraction and little reserve against even brief hypoxemia.

The AHA recommends titrating to an SpO₂ of 90%–98% or PaO₂ of 60–105 mm Hg. The 2025 European Resuscitation Council–European Society of Intensive Care Medicine guideline uses a narrower SpO₂ range of 94%–98% and PaO₂ of 75–100 mm Hg. These are not evidence that the lower boundaries are desirable targets. A practical interpretation is to aim for confirmed mid-normal oxygenation, avoiding both saturation hunting at 100% and aggressive oxygen restriction near the hypoxemic threshold. Pulse oximetry may be unreliable during vasoconstriction, low flow, motion, dyshemoglobinemia, and in patients with darker skin pigmentation, in whom occult hypoxemia is more frequent. Early and repeated arterial blood gases are therefore important when vasoactive support or temperature control makes the displayed saturation uncertain (Hirsch et al., 2025; Nolan et al., 2025).

Carbon dioxide is often the faster cerebral perfusion intervention. Hypocapnia constricts cerebral resistance vessels and can reduce blood flow to tissue already affected by no-reflow or impaired autoregulation. Hypercapnia increases cerebral blood volume and may improve perfusion, but it can also worsen intracranial pressure, acidemia, pulmonary vascular resistance, and right-ventricular function. This made mild hypercapnia physiologically attractive. The 1,700-patient TAME trial compared PaCO₂ targets of 50–55 mm Hg and 35–45 mm Hg for 24 hours. Favorable six-month neurological outcome occurred in 43.5% with mild hypercapnia and 44.6% with normocapnia, providing no evidence that routine CO₂-mediated cerebral vasodilation improves recovery. Current guidelines consequently recommend PaCO₂ of approximately 35–45 mm Hg rather than routine permissive hypercapnia (Eastwood et al., 2023).

That target requires arterial measurement. After arrest, low cardiac output, pulmonary embolism, aspiration, atelectasis, pulmonary edema, and increased dead space may produce a large or changing PaCO₂–end-tidal CO₂ gradient. An end-tidal value of 35 mm Hg can coexist with clinically important arterial hypercapnia, while rapid ventilation changes can cause cerebral vasoconstriction before the next blood gas is obtained. Use waveform capnography to follow direction and stability, but calibrate it against serial arterial gases. If hypothermia is used, know whether the laboratory reports temperature-corrected or 37°C values and use a consistent convention. Employ lung-protective ventilation with approximately 6–8 mL/kg predicted body weight, appropriate plateau pressure and PEEP, and treatment of aspiration or ARDS. PEEP should not reflexively be withheld because of theoretical ICP effects; its net cerebral effect depends on recruitment, oxygenation, PaCO₂, venous pressure, cardiac output, and intracranial compliance. Conversely, hyperventilation should not be used prophylactically. Temporary hypocapnia is a rescue maneuver for impending herniation, not routine post-arrest care.

Blood pressure is similarly a surrogate rather than the endpoint. Cerebral perfusion pressure approximates MAP minus the greater downstream pressure imposed by ICP or central venous pressure. After cardiac arrest, autoregulation may be impaired, narrowed, or shifted toward higher pressures, making cerebral blood flow more pressure-passive. Raising MAP could therefore improve perfusion, but it may also increase myocardial afterload, vasopressor exposure, arrhythmias, pulmonary edema, and oxygen consumption without correcting microvascular no-reflow or impaired oxygen diffusion.

The 2025 AHA guideline recommends a minimum MAP of 65 mm Hg, while the European guideline advises avoiding hypotension and targeting MAP above approximately 60–65 mm Hg. BOX directly compared MAP targets of 63 and 77 mm Hg in 789 comatose survivors of presumed cardiac OHCA. The higher target did not reduce death, severe disability, or coma, and did not improve neuron-specific enolase, modified Rankin Scale, Cerebral Performance Category, or cognitive results. The trial argues against a universal MAP target of 77–80 mm Hg for every post-arrest patient; it does not prove that 63 mm Hg is appropriate for a patient with chronic hypertension, persistent oliguria, rising lactate, low cerebral oxygenation, significant cerebral edema, or pressure-dependent neurological findings (Kjaergaard et al., 2022).

Hemodynamic treatment should therefore begin with phenotype. Early shock may be dominated by reversible myocardial stunning and low cardiac output; later shock may become vasoplegic as systemic ischemia–reperfusion produces endothelial dysfunction and inflammatory vasodilation. Assess ventricular function, filling, valve pathology, right-sided pressures, lactate trajectory, capillary refill, urine output, temperature gradient, and response to a small, reassessed fluid challenge. Avoid indiscriminate crystalloid loading into a stunned ventricle. Norepinephrine is commonly favored when vasoplegia predominates, while an inotrope may be required for demonstrable low-output physiology, but the AHA concludes that evidence is insufficient to recommend one specific vasopressor after cardiac arrest. MAP should be at least 65 mm Hg and then individualized upward only when the physiology suggests that additional pressure improves organ or cerebral perfusion.

Temperature management remains mandatory, but routine prolonged hypothermia does not. The 2025 AHA guideline recommends a deliberate protocol for every adult who remains unresponsive to verbal commands, permits a target between 32°C and 37.5°C, and considers at least 36 hours of total temperature control reasonable. The 2025 European guideline more specifically recommends active fever prevention at or below 37.5°C for 36–72 hours. Both reject passive indifference to temperature. Fever increases cerebral metabolic demand, excitotoxicity, inflammation, and the mismatch between energy supply and demand, although observational associations between fever and poor outcome do not by themselves prove that every lower temperature is beneficial.

TTM2 randomized 1,900 comatose OHCA survivors to 33°C or controlled normothermia with early treatment of fever. Six-month mortality was approximately 50% and 48%, with no neurological advantage from hypothermia; arrhythmias causing hemodynamic compromise were more common at 33°C, 24% versus 17%. A subsequent individual-patient meta-analysis combining the nonshockable-rhythm participants from TTM2 and HYPERION found no survival or functional advantage from 33°C, weakening the argument that the HYPERION signal identified a reliably responsive nonshockable subgroup (Dankiewicz et al., 2021; Taccone et al., 2024).

A defensible contemporary default for most comatose post-arrest adults is consequently active normothermia and reliable fever prevention rather than routine prolonged cooling to 33°C. The AHA still permits lower targets, and uncertainty remains in populations underrepresented in the major trials, but no arrest rhythm, downtime, or bedside severity score currently identifies a subgroup with proven benefit from 33°C. If lower-temperature control is selected, use continuous core measurement and feedback-controlled surface or endovascular therapy. Treat shivering because it raises oxygen consumption, CO₂ production, catecholamines, and metabolic demand while defeating temperature control. Analgesia, sedation, counterwarming, and selective neuromuscular blockade may be required, but paralysis must never be mistaken for seizure control or neurological improvement. Patients arriving spontaneously at 32–36°C should not be rapidly warmed to normothermia. During rewarming, avoiding rates faster than approximately 0.5°C per hour is reasonable, while monitoring potassium, glucose, hemodynamics, drug accumulation, and recurrent fever.

EEG is part of treatment surveillance, not merely later prognostication. In any patient who does not follow commands after ROSC, obtain and interpret EEG promptly and use repeated or continuous monitoring when risk remains high. Treat clinical and electrographic seizures. Do not administer routine antiseizure prophylaxis in the absence of seizures. Rhythmic or periodic patterns on the ictal–interictal continuum require integration of frequency, evolution, modifiers, background continuity and reactivity, systemic physiology, and imaging; a trial of a nonsedating antiseizure medication may be reasonable, but indiscriminate anesthetic suppression of every periodic discharge is not supported. Myoclonus without an EEG correlate is neither automatically status epilepticus nor an indication for treatment solely to improve neurological outcome. Suppressing visible jerks may occasionally facilitate ventilation or nursing care, but it can delay awakening and obscure examination without treating cortical seizure activity.

Sedation should be sufficient for ventilation, temperature control, shivering, and safety but no deeper than the physiology requires. Prefer short-acting agents, document every sedative and paralytic exposure, and reassess the need for each drug as temperature and organ function change. Hypothermia, shock, hepatic or renal failure, obesity, and prolonged infusions can make nominally short-acting drugs persist. An early sedation interruption may be unsafe during active shivering, severe hypoxemia, unstable shock, or seizures; when conditions permit, however, reducing sedation allows detection of purposeful movement and prevents drug-induced coma from being misclassified as brain failure.

Avoid glucose below 70 mg/dL and above 180 mg/dL rather than pursuing tight normoglycemia. Correct clinically important potassium, magnesium, calcium, acid–base, and volume abnormalities, recognizing that intracellular–extracellular shifts during cooling and rewarming can reverse rapidly. Treat aspiration pneumonia when clinically suspected, but routine prophylactic antibiotics have not improved patient-centered outcomes. Routine steroids and proposed pharmacologic neuroprotectants likewise lack established benefit. Head elevation, neutral neck position, venous drainage, normoxemia, normocapnia, adequate MAP, and fever control are reasonable cerebral-protection measures. Hyperosmolar therapy should be reserved for suspected or demonstrated cerebral edema with intracranial hypertension rather than administered prophylactically. Invasive ICP, brain-tissue oxygen, cerebral blood-flow, or jugular venous monitoring may reveal important physiology in selected patients, but no randomized evidence establishes routine use after cardiac arrest.

The early head CT answers questions about cause, hemorrhage, mass effect, and gross edema; it should not be used to declare a normal brain or a hopeless brain shortly after ROSC. Diffuse loss of gray–white differentiation is concerning, but a normal early CT does not exclude evolving hypoxic–ischemic injury. MRI diffusion abnormalities, EEG evolution, examination, somatosensory evoked potentials, and biomarkers mature on different timelines. Their prognostic interpretation belongs after adequate time, normothermia, and removal of confounders. During the initial resuscitative phase, they should be used to detect treatable complications and characterize injury, not to justify premature withdrawal of life-sustaining therapy.

References