Picking up exactly where the first-hours severe-TBI framework stopped, the patient has now reached the neuro-ICU with the airway secured, hemorrhage addressed or under surveillance, invasive monitoring placed when indicated, and the obvious extracranial threats corrected. The problem now becomes dynamic. Cerebral edema evolves, contusions blossom, venous outflow changes with positioning and thoracic pressure, seizures increase metabolic demand, fever raises cerebral blood flow, sedatives alter blood pressure, osmotherapy changes both brain volume and intravascular volume, and every treatment applied to lower ICP can generate a competing injury elsewhere. ICU management is therefore not the mechanical progression through an ICP ladder. It is a repeated cycle of identifying the mechanism of a physiologic deterioration, selecting the least hazardous intervention capable of reversing it, and confirming that the intervention improved the brain rather than merely changing a number.

The familiar treatment threshold remains an ICP above 22 mm Hg, but the threshold should initiate an assessment, not an automatic sequence of medications. A sustained ICP of 24 mm Hg with worsening waveform compliance, falling CPP, declining PbtO₂, or progressive tissue shift is more consequential than a brief spike to 35 during suctioning. Before escalating treatment, confirm that the value is real: verify leveling and zeroing, inspect the waveform, exclude catheter obstruction, and determine whether the monitor was moved relative to the tragus or whether the arterial transducer and ICP transducer now reference different hydrostatic levels. Examine the pupils, review the preceding trend rather than the latest minute, and ask whether the rise followed coughing, agitation, ventilator dyssynchrony, fever, turning, a seizure, or an EVD clamp. An unexplained or treatment-resistant rise should prompt repeat imaging because no combination of propofol, hypertonic saline, and hyperventilation definitively treats an expanding hematoma, enlarging contusion, acute hydrocephalus, venous infarction, or surgically remediable mass effect. The ICP monitor is an alarm within a diagnostic system; it is not a substitute for source control.

The Seattle International Severe Traumatic Brain Injury Consensus Conference, or SIBICC, is best understood as a risk-stratification framework. Treatments within a tier are not necessarily biologically equivalent, every therapy in a lower tier does not have to be exhausted before moving upward, and higher tiers represent increasing treatment hazard rather than increasing certainty of benefit. Basic care continues regardless of the ICP: head elevation when hemodynamically tolerated, neutral cervical alignment, relief of restrictive collars or endotracheal-tube ties, adequate analgesia, oxygenation, normocapnia, temperature control, seizure surveillance, euvolemia, and avoidance of hypotonic fluids. When the head is elevated, remember that MAP measured at the heart overestimates arterial pressure at the brain unless the arterial transducer is referenced appropriately. A calculated CPP of 65 may therefore be several millimeters lower at the circle of Willis than the bedside monitor suggests. The Brain Trauma Foundation continues to recommend a general CPP target of 60–70 mm Hg and cautions against aggressively maintaining CPP above 70 with fluid and vasopressors because pressure is not flow and systemic complications can erase the cerebral benefit. These remain population-level boundaries rather than a claim that every injured brain has the same autoregulatory range.

Brain Trauma Foundation severe-TBI guidelinesSIBICC consensus algorithm

---

Sedation is the first major example of a therapy with competing effects. Pain, agitation, coughing, shivering, and ventilator dyssynchrony increase sympathetic activity, cerebral metabolic demand, thoracic pressure, and cerebral venous pressure. Appropriate analgesia and sedation can therefore reduce ICP by lowering CMRO₂, cerebral blood flow, and cerebral blood volume while improving ventilation and venous drainage. But a sedative that lowers ICP by 5 mm Hg while lowering MAP by 15 mm Hg may worsen CPP and cerebral oxygen delivery. Sedation depth should consequently be indication-driven. A patient with controlled ICP and a potentially informative examination should not remain deeply sedated merely because severe TBI is present. A patient with pressure waves, coughing-induced crises, or evolving swelling may appropriately require a depth of sedation that would be excessive in a general ICU patient.

Propofol is often favored early because it is rapidly titratable, suppresses cerebral metabolism and seizures, and permits a comparatively rapid examination when discontinued. It lowers ICP effectively, but it has not been shown to improve long-term neurologic outcome and its vasodilatory and myocardial effects can produce exactly the hypotension the injured brain cannot tolerate. Escalating propofol indefinitely is particularly dangerous. High-dose, prolonged exposure—especially with catecholamines, corticosteroids, critical illness, carbohydrate depletion, or mitochondrial vulnerability—raises concern for propofol infusion syndrome. New metabolic acidosis, rising lactate, rhabdomyolysis, hyperkalemia, acute kidney injury, hepatomegaly, bradyarrhythmia, or cardiovascular collapse should trigger immediate reassessment rather than another increase in the infusion.

Midazolam is reasonable when hemodynamic tolerance or seizure control is more important than rapid awakening, but its context-sensitive half-life lengthens substantially with prolonged infusion. Active metabolite accumulation in renal dysfunction, tissue sequestration, tachyphylaxis, and delirium can transform an intended short period of sedation into days of diagnostic uncertainty. A 2025 multicenter observational analysis found no reliable causal difference in ICP-lowering treatment requirements between midazolam-based and other sedation strategies after adjustment for center preference, although midazolam use was associated with a longer ICU stay. That study should not be read as proof that midazolam prolongs admission, because drug choice is entangled with injury severity and the perceived need for prolonged deep sedation. It instead reinforces how little comparative outcome evidence exists.

2025 midazolam comparative-effectiveness study

Ketamine should no longer be reflexively rejected because of the historical claim that it inevitably raises ICP. That concern arose largely from small studies of spontaneously breathing patients in whom ventilation and sympathetic responses were not controlled. In adequately sedated, mechanically ventilated patients, contemporary observational and physiologic data do not show a consistent ICP increase, and ketamine may be useful when analgesia and sedation are needed without further systemic hypotension. What remains unproven is whether adding ketamine reduces treatment intensity or improves outcome. The placebo-controlled BIKe trial is testing adjunctive ketamine at 1 mg/kg/h, but results are not expected until after enrollment is completed; its 2025 publication was a protocol, not an efficacy result. Ketamine is therefore a reasonable individualized agent, not an established neuroprotectant. Dexmedetomidine can facilitate lighter sedation and later ventilator liberation, but bradycardia, hypotension, and inadequate depth make it poorly suited as the sole agent during uncontrolled intracranial hypertension.

BIKe trial protocol

Sedation interruption deserves the same physiologic discipline. Daily awakening is beneficial in many general ICU populations, but an abrupt sedation holiday during active intracranial hypertension can provoke coughing, hypertension, cerebral vasodilation, seizures, and an ICP crisis. The decision should incorporate the recent ICP burden, current treatment tier, pupil stability, imaging, EVD status, seizure risk, and the likelihood that the examination will change management. When the brain is stable, sedation should be reduced incrementally while watching ICP, CPP, respiratory pattern, and, when available, PbtO₂. When the brain is unstable, the examination is obtained through pupils, brainstem reflexes not obscured by the drugs, imaging, EEG, and multimodality monitoring until the risk of awakening falls.

Neuromuscular blockade is similarly a diagnostic therapeutic trial rather than an automatic escalation. If deeply sedated paralysis abolishes coughing or ventilator dyssynchrony and produces a meaningful, reproducible ICP reduction, continued blockade may be justified temporarily. If ICP does not improve, the paralytic has removed the motor examination and increased the risks of weakness, corneal injury, venous thrombosis, pneumonia, and unrecognized seizures without providing cerebral benefit. Paralysis must never substitute for sedation, and continuous EEG becomes particularly important because clinical seizure activity is no longer visible.

---

CSF drainage is often the most anatomically direct next intervention. Unlike osmotherapy, it removes volume from the intracranial compartment without relying on renal function or creating systemic hypertonicity. When an EVD is present and the ventricles remain accessible, drainage can rapidly lower ICP and provide evidence that the CSF compartment is contributing to the crisis. The Brain Trauma Foundation notes low-quality evidence that continuous drainage may reduce ICP burden more effectively than intermittent drainage, particularly in the most severely injured patients. The monitoring configuration matters: a continuously open EVD cannot simultaneously provide an accurate pressure unless ICP is measured through a separate parenchymal probe; if the EVD is the only monitor, it must be temporarily closed and allowed to equilibrate for pressure measurement. Every position change requires releveling, and unexpectedly poor drainage should trigger assessment for catheter obstruction, ventricular collapse around the catheter, malposition, or a pressure problem driven primarily by tissue swelling rather than CSF. Drainage should be governed by a prescribed pressure level and clinical objective, not by an arbitrary daily volume target. Excessive drainage can cause ventricular collapse, subdural hygroma or hemorrhage, and dangerous pressure gradients. Lumbar drainage is a specialized option only in carefully selected patients with patent basal cisterns and no obstructive lesion or threatening compartmental shift; it is not a routine substitute for ventricular drainage.

---

Hyperosmolar therapy remains a bolus rescue treatment whose success should be measured, not presumed. Hypertonic saline and mannitol both lower ICP, but neither has demonstrated improved neurologic outcome, and the Brain Trauma Foundation finds insufficient comparative evidence to endorse one agent universally. The Neurocritical Care Society conditionally favors hypertonic sodium solutions for initial treatment in TBI, based on low-quality evidence and their relative preservation of intravascular volume and CPP. Mannitol remains entirely reasonable in a hemodynamically stable patient, especially when severe hypernatremia or volume overload limits additional sodium. Conversely, hypertonic saline is often preferable when hypotension, hypovolemia, or mannitol-associated diuresis would be hazardous. The correct bedside question is not which agent wins globally, but which agent’s extracranial effects are safer in this patient at this moment.

Neurocritical Care Society cerebral-edema guideline

Each dose should be followed by the magnitude and duration of ICP response, CPP, urine output, sodium, chloride, measured osmolality when relevant, acid–base status, and renal function. A diminishing response should not automatically trigger shorter dosing intervals; it may mean the blood–brain barrier is becoming permeable, the expanding compartment is not osmotically responsive, or definitive source control has been delayed. Scheduled mannitol and routine continuous hypertonic infusion convert a rescue maneuver into cumulative exposure. The COBI randomized trial showed that prophylactic continuous 20% hypertonic saline did not improve six-month neurologic outcome. Hypernatremia is therefore a tolerated consequence or deliberately created gradient in selected patients, not an independent outcome target. The goal is ICP control with the lowest osmotic burden that accomplishes it.

---

Ventilation manipulates ICP rapidly because carbon dioxide changes arteriolar tone. In most patients, PaCO₂ should initially remain near the low-normal range, commonly about 35–38 mm Hg. When ICP remains elevated despite initial measures, carefully controlled mild hypocapnia around 32–35 mm Hg may reduce cerebral blood volume. More aggressive hyperventilation is a short bridge for impending herniation or otherwise uncontrollable ICP while definitive treatment is being mobilized. It should not become background ventilation. The Brain Trauma Foundation specifically recommends against prolonged prophylactic PaCO₂ values of 25 mm Hg or below and urges particular caution during the first 24 hours, when cerebral blood flow may already be reduced. If substantial hyperventilation is unavoidable, PbtO₂ or jugular venous oxygen monitoring can identify the patient in whom an attractive ICP reduction is being purchased with cerebral hypoxia.

---

CPP management becomes more sophisticated when autoregulation is considered. The initial 60–70 mm Hg target is a safe population scaffold. Beyond that, a monitored MAP challenge can test whether additional arterial pressure helps. MAP is increased modestly—often by approximately 10 mm Hg for no more than about 20 minutes—while ICP, CPP, PbtO₂, and systemic responses are observed. If arterioles constrict appropriately, cerebral blood volume and ICP may remain stable or fall while CPP and tissue oxygenation improve. If autoregulation is impaired, the pressure increase may be transmitted passively to the cerebral circulation, raising cerebral blood volume and ICP. The latter patient should not be subjected to ever-higher vasopressor targets simply because the arithmetic CPP looks better.

A 2025 study of 809 invasively monitored patients found that deviations below individualized CPPopt and the estimated lower autoregulatory limit were more strongly associated with poor outcome than comparable deviations above CPPopt. The authors proposed viewing CPPopt more as a lower boundary than as a single ideal number. That is physiologically persuasive, but it remains observational. It does not prove that prospectively raising every patient’s CPP to CPPopt improves outcome, and it does not invalidate the pulmonary, cardiac, or hemorrhagic risks of high MAP. COGiTATE established feasibility of autoregulation-guided CPP targeting, not outcome efficacy. PRx, CPPopt, the MAP challenge, waveform compliance, PbtO₂, and the CT must therefore be treated as convergent evidence rather than allowing one derived index to dictate vasopressor therapy.

2025 CPP-target reappraisal

---

When first- and second-tier measures fail, the decision is no longer simply how to lower ICP. It is which high-risk intervention offers the most acceptable trade between survival, disability, and treatment toxicity. Before declaring failure, repeat the CT, reconsider an evacuable lesion, confirm EVD function, exclude seizures and fever, reassess venous drainage and abdominal or thoracic pressure, and ask whether the monitor reflects the threatened tissue. High-dose barbiturates, decompressive craniectomy, and carefully selected temperature reduction are rescue therapies with fundamentally different mechanisms and consequences.

Barbiturates lower CMRO₂ and couple metabolic suppression to reduced cerebral blood flow and cerebral blood volume. They are not recommended prophylactically, but the Brain Trauma Foundation supports high-dose therapy for ICP refractory to maximal standard medical and surgical treatment, provided hemodynamic stability can be maintained. Pentobarbital is loaded and then titrated with continuous EEG toward adequate burst suppression and ICP control. The objective is the lowest dose that controls the physiologic crisis; deeper and deeper electrical silence is not itself therapeutic. Hypotension is the dominant immediate danger, often requiring norepinephrine and invasive hemodynamic attention. Ileus, immunosuppression, pneumonia, hepatic accumulation, prolonged coma, hypothermia, and loss of the neurologic examination follow. Drug clearance can take days, so absent motor responses during and after therapy are not interpretable without pharmacokinetic context.

The 2025 Triple-T TBI study examined 408 patients who received a tier-three therapy. Secondary decompression was associated with lower mortality, whereas barbiturate treatment was associated with higher mortality and worse outcome. Those findings should not be interpreted as proof that barbiturates cause poor outcome or that surgery is universally superior. This was a retrospective cohort selected only after rescue therapy was required; barbiturates are often chosen for patients with diffuse, surgically inaccessible swelling or the worst refractory physiology, while craniectomy requires a patient judged operable. Residual confounding by indication is unavoidable. The study’s more defensible message is that only 55% received a guideline-adherent staircase before tier-three treatment and that rescue-therapy selection remains highly variable.

Triple-T TBI study

Temperature management has been refined rather than settled. Fever increases cerebral metabolism, cerebral blood flow, and frequently ICP. A 2024 ESICM/NACCS consensus strongly supported continuous temperature monitoring and controlled normothermia between 36.0 and 37.5°C during first- and second-tier management. That is different from routine prophylactic hypothermia. POLAR found no outcome benefit from early sustained prophylactic cooling, and Eurotherm3235 showed that adding hypothermia to treat intracranial hypertension lowered ICP but worsened functional outcome. If temperature reduction is used as rescue therapy after other measures fail, the 2024 consensus recommends staying as close to normothermia as possible rather than reflexively cooling to 32–33°C. Shivering, sedation and paralysis requirements, hypotension, arrhythmia, electrolyte shifts, insulin resistance, coagulopathy, infection, and rebound ICP during rewarming must all be included in the treatment cost.

2024 targeted-temperature consensus

---

Decompressive craniectomy is the most consequential decision because it rapidly and durably changes cranial mechanics but cannot be withdrawn. It lowers ICP and prevents pressure-mediated death; it does not reverse diffuse axonal injury, brainstem injury, ischemia already sustained, or the primary traumatic lesion. The apparent conflict between DECRA and RESCUEicp becomes clearer when the enrolled physiologies are separated. DECRA used relatively early decompression for diffuse injury after ICP exceeded 20 mm Hg for more than 15 minutes within an hour despite initial treatment. Surgery lowered ICP and shortened ICU treatment but did not improve functional outcome, and more survivors remained severely disabled or vegetative. RESCUEicp studied a later rescue decision after ICP exceeded 25 mm Hg for one to twelve hours despite tier-one and tier-two measures. Decompression markedly reduced mortality, but survival was redistributed across vegetative state, severe disability, and moderate disability rather than producing more good recoveries.

At 24 months in RESCUEicp, mortality was 33.5% with surgery and 54.0% with medical treatment. For every 100 patients assigned to surgery, approximately 21 additional patients survived: four in a vegetative state, nine with severe disability, and eight with moderate disability. Rates of good recovery were almost identical, although patients in the surgical arm were more likely to improve by at least one GOS-E category between six and 24 months. This is why decompression cannot be described simply as beneficial or harmful. It is highly effective at preventing death from pressure, but the neurologic state in which death is prevented varies. The 2020 Brain Trauma Foundation update recommends secondary decompression for late refractory ICP, advises against early secondary decompression for improving outcome in the DECRA-type population, and favors a large frontotemporoparietal opening of at least 12 by 15 cm over a small opening. “Early” and “late” here describe the trial populations and treatment intensity, not a simple clock cutoff.

Brain Trauma Foundation decompression updateRESCUEicp 24-month outcomes

The attending-level decompression discussion therefore includes the injury pattern, age, preinjury function, brainstem examination, imaging evidence of irreversible injury, trajectory, duration of refractory physiology, and the patient’s values regarding survival with dependence. It should occur before the crisis whenever possible. After surgery, ICP control does not end the problem: hemorrhagic contusions can expand when tamponade is removed; external cerebral herniation can occur through an inadequate opening; venous compression can develop at the bone edge; hygromas, hydrocephalus, seizures, infection, sinking-flap physiology, and later cranioplasty complications become part of the disease course.

---

The final discipline is daily de-escalation. A patient can survive the ICP crisis and still be injured by cumulative chloride, hypernatremia, renal failure, vasopressor ischemia, propofol toxicity, barbiturate accumulation, infection, ventilator-associated lung injury, ileus, malnutrition, venous thrombosis, pressure wounds, and ICU-acquired weakness. Each morning the team should ask which treatments remain necessary, which can be reduced, and what physiologic evidence supports continuing them. Sedation should not remain deep because it was required yesterday. Hyperosmolar therapy should not remain scheduled because the sodium has not yet reached a customary target. A vasopressor should not maintain yesterday’s CPP goal after autoregulation, ICP, or systemic tolerance has changed. The EVD should not remain open or closed by habit. Treatment intensity is itself an exposure.