Start with Monro-Kellie. Within an intact adult skull, brain tissue, blood, and cerebrospinal fluid occupy a nearly fixed total volume. Brain tissue contributes roughly 1.4 liters, while blood and CSF contribute on the order of 150 mL each. An increase in one compartment initially does not have to produce much increase in ICP because compensatory mechanisms move CSF into the spinal compartment and displace predominantly venous blood out of the cranium. The essential concept is therefore not merely intracranial volume, but intracranial compliance: delta volume divided by delta pressure. Early on the pressure-volume curve is relatively flat. Once reserve is exhausted, the curve becomes steep, and a few additional milliliters of hematoma, edema, venous blood, or CSF can produce a disproportionately large rise in ICP. This is why a patient can remain stable for hours and then appear to deteriorate catastrophically after what seems radiographically like only modest additional swelling. Deranged Physiology’s treatment of ICP emphasizes exactly this pressure-volume relationship and the compensatory roles of CSF and venous displacement.

That compliance curve also explains why the same ICP of 19 means different things in different patients. A patient sitting on the flat portion of the curve may tolerate an additional small change in volume without consequence. A patient already at the elbow of the curve may have an ICP of 19 with progressively abnormal waveform morphology and then move rapidly to 30 or 40 with stimulation. So when you look at an ICP monitor, you should mentally ask three separate questions: what is the absolute ICP, what is the trend and pressure-time burden, and what is the underlying compliance?

The ICP pulse waveform gives some information about the last question. The classic waveform contains P1, the percussion wave related predominantly to arterial pulsation; P2, often called the tidal wave and more reflective of intracranial compliance; and P3, the dicrotic wave. In a reasonably compliant cranial compartment P1 generally exceeds P2. As compensatory reserve deteriorates, P2 may rise relative to P1. This is not a perfectly specific diagnostic test, but a progressively dominant P2 in the appropriate clinical setting is a physiologic warning that the patient may be moving onto the steep portion of the pressure-volume curve. Plateau waves take that concept further. A sudden sustained rise in ICP can result from changes in cerebrovascular volume as vascular tone changes, demonstrating that ICP is not merely a measure of how swollen the brain is; it is also a dynamic vascular signal.

Now add cerebral perfusion pressure. At the bedside we use CPP equals MAP minus ICP. More precisely, the effective downstream pressure is whichever is greater, intracranial pressure or cerebral venous pressure, so a more physiologic formulation is MAP minus the greater of ICP or CVP. This becomes clinically relevant when venous pressure is high. A patient with elevated PEEP, severe right-sided heart failure, tension physiology, abdominal hypertension, a tight cervical collar, jugular obstruction, extreme Trendelenburg positioning, or even vigorous coughing may worsen intracranial venous drainage and increase ICP without any meaningful increase in brain water. Deranged Physiology summarizes cerebral blood flow as the pressure gradient divided by cerebrovascular resistance and specifically emphasizes that CPP depends on MAP relative to ICP or venous pressure.

But even CPP is only a pressure gradient. It is not cerebral blood flow. Cerebral blood flow is approximately CPP divided by cerebrovascular resistance, and the rapidly adjustable part of that equation is vascular radius. This is where autoregulation enters the story. In an intact brain, arterioles constrict as perfusion pressure rises and dilate as it falls, attempting to maintain relatively stable cerebral blood flow. The traditional teaching diagram shows a broad plateau somewhere around a MAP of 50 to 150 mm Hg, but that diagram should never be interpreted as a universal human operating range. The limits vary among individuals, shift with chronic hypertension, and can change over hours in an injured brain. Traumatic brain injury, ischemia, global hypoxic injury, intracranial mass lesions, meningitis, hepatic encephalopathy, and other insults can all impair autoregulatory behavior.

This distinction determines how blood pressure affects ICP. Imagine that autoregulation is intact and MAP falls. Cerebral arterioles dilate to preserve CBF. Cerebral blood volume increases. Because cerebral blood is an intracranial compartment, ICP can increase. Now CPP falls further because CPP equals MAP minus ICP. That can provoke still more vasodilation and further increase blood volume. Conversely, raising MAP within an intact autoregulatory range may produce vasoconstriction, reducing cerebral blood volume and sometimes lowering ICP. This is one reason a hypotensive patient with elevated ICP may improve when you give norepinephrine rather than worsen.

Now imagine autoregulation is lost. Cerebral vessels no longer compensate normally. Blood flow becomes more pressure-passive. Increasing MAP may directly increase cerebral blood flow and blood volume, potentially increasing ICP; decreasing MAP may decrease ICP but at the cost of cerebral ischemia. That is why indiscriminately treating systemic hypertension simply because the number looks uncomfortable can be dangerous in acute brain injury. A blood pressure of 190 may be pathological, compensatory, or both. You need to know the disease, ICP, CPP, autoregulatory state, imaging, and clinical trajectory before deciding what the systemic pressure means.

Pressure reactivity monitoring tries to operationalize this. PRx examines the correlation between slow fluctuations in MAP and ICP. If MAP rises and the cerebral vasculature appropriately constricts, intracranial blood volume does not rise proportionally and the correlation tends to be near zero or negative. When autoregulation is impaired and the vasculature becomes pressure-passive, MAP and ICP tend to move together, producing a positive PRx. Plotting PRx against CPP over time can sometimes produce a U-shaped curve whose nadir has been termed CPPopt, the CPP at which pressure reactivity appears best preserved. This is one of the most attractive precision-neurocritical-care concepts because it acknowledges that a CPP of 62 may be adequate for one patient and inadequate for another.

The same caution applies to ICP itself. The current Brain Trauma Foundation guideline remains the fourth edition; BTF lists development of a fifth edition as an ongoing project. The current Level II B recommendation is to treat ICP above 22 mm Hg because values above this level are associated with increased mortality. CPP remains targeted to 60 to 70 mm Hg.

But 22 should be understood as a population-derived treatment threshold, not a biological cliff. A 2024 adult TBI study specifically attempted to reproduce the statistical methodology that helped support the 22-mm-Hg threshold. In its cohort of 331 patients, the ICP value that best discriminated outcome was closer to 18 mm Hg, and subgroup thresholds varied substantially, roughly from 16 to 30 mm Hg. The investigators therefore could not reproduce 22 as a universal critical threshold. That study does not justify changing routine adult treatment thresholds to 18; it demonstrates that looking for one biologically privileged number is probably the wrong conceptual model.

The emerging model is pressure times time. Twenty-four millimeters of mercury for three minutes while coughing is fundamentally different from 24 for six hours. Conversely, an ICP of 19 or 20 maintained for many hours in a patient with impaired autoregulation and marginal CPP may carry meaningful secondary-injury burden even though the bedside monitor never technically crosses 22. A 2025 review of time-burden physiology emphasized that lower elevations become deleterious when sustained, while substantially higher pressures can sometimes be tolerated for very short periods; the interaction also changes with age, CPP, and autoregulatory function.

So 22 remains a clinically useful trigger, but the better mental model is an ICP dose-response surface containing pressure, duration, CPP, autoregulation, and patient phenotype. If the ICP is 23 for twenty minutes and rising, P2 is becoming dominant, CPP is drifting toward 55, pupils are changing, and the CT shows worsening cisternal compression, that is very different from ICP 23 during suctioning with immediate return to 14.

Carbon dioxide is one of the fastest levers in this system. CO2 diffuses readily into the CNS, alters extracellular pH, and strongly influences cerebrovascular tone. Hypercapnia produces vasodilation, increasing CBF and cerebral blood volume, which can increase ICP. Hypocapnia produces vasoconstriction, reducing cerebral blood volume and therefore rapidly lowering ICP. Around the normal physiologic range, relatively small PaCO2 changes can produce substantial changes in cerebral blood flow. Severe hypoxemia also stimulates cerebral vasodilation, particularly once PaO2 falls below roughly 50 mm Hg.

This is why hyperventilation is simultaneously one of the most effective and one of the most dangerous acute ICP therapies. You can make an ICP of 40 become 22 in minutes by driving PaCO2 down, but you have achieved that result by constricting cerebral arteries. You have not treated edema or removed the hematoma. You have reduced intracranial blood volume by sacrificing blood flow. The Brain Trauma Foundation therefore recommends against prolonged prophylactic hyperventilation to a PaCO2 of 25 mm Hg or less. Routine management generally aims for normocapnia or low-normal PaCO2, approximately 35 to 40 mm Hg. In impending herniation, transient ventilation toward approximately 25 to 30 can be appropriate as a bridge while definitive therapy—osmotherapy, CSF drainage, hematoma evacuation, decompression—is being executed.

Cerebral oxygen delivery adds another dimension. Oxygen delivery is cerebral blood flow multiplied by arterial oxygen content. Hemoglobin therefore matters even when ICP and CPP are perfect. This was illustrated by the TRAIN randomized trial published in JAMA in 2024. Among anemic patients with acute brain injury—including TBI, SAH, and ICH—a transfusion strategy triggered at hemoglobin below 9 g/dL resulted in unfavorable neurological outcome in 62.6% versus 72.6% with a threshold below 7 g/dL, adjusted relative risk 0.86.

Brain tissue oxygen monitoring extends that concept directly into the parenchyma. PbtO2 represents highly local tissue oxygen tension around the probe, not whole-brain oxygenation. Values below approximately 15 to 20 mm Hg are generally regarded as concerning. Crucially, normal ICP does not guarantee normal PbtO2. A patient can have ICP 14 and CPP 70 while experiencing tissue hypoxia because of diffusion limitation, regional ischemia, microvascular dysfunction, anemia, hypoxemia, or impaired oxygen utilization. The phase II BOOST-II study demonstrated that PbtO2-directed treatment markedly reduced brain hypoxia burden, but it was not powered to establish functional efficacy. As of the most recent ClinicalTrials.gov update in May 2026, the definitive BOOST-3 trial remains active without posted outcome results, and the international BONANZA-GT randomized trial is also evaluating PbtO2-guided management. We therefore have excellent physiologic justification and encouraging phase II data, but no completed phase III evidence yet proving that adding PbtO2 monitoring improves functional outcome.

At the same time, remove reversible impediments to venous drainage. Keep the neck neutral. Ensure the cervical collar is not strangulating jugular outflow. Treat coughing, agitation, pain, and ventilator dyssynchrony. Make sure PEEP and intrathoracic pressures are not unnecessarily excessive. Treat fever and seizures because both increase metabolic demand and cerebral blood flow. Head elevation is usually reasonable, but it should not become ritualistic. Raising the head frequently lowers ICP by improving venous drainage, yet it may simultaneously lower arterial pressure at brain level and reduce CPP. The correct endpoint is therefore the integrated ICP-CPP response, not adherence to 30 degrees as if it were pharmacology.

Then protect systemic perfusion. In severe TBI the current BTF target remains CPP 60 to 70 mm Hg, with age-dependent systolic pressure minima also intended to avoid hypotension. If CPP is 48 because MAP is 65 and ICP is 17, giving osmotherapy solely to make ICP 10 is conceptually inferior to correcting the systemic perfusion problem. If the patient is hypovolemic, restore volume. Once euvolemic, vasopressors—often norepinephrine—are usually the more controllable method of raising MAP. Conversely, pushing liters of crystalloid merely to manufacture a CPP above 70 may create pulmonary edema, impair oxygenation, raise venous pressures, and eventually worsen the cerebral situation you were trying to improve.

If ICP remains pathologically elevated, osmotherapy shifts water from brain toward the vascular compartment by increasing plasma tonicity, provided the blood-brain barrier retains enough effective osmotic reflection. The longstanding Neurocritical Care Society cerebral-edema guideline conditionally favors hypertonic sodium solutions over mannitol for initial treatment of TBI-related ICP or cerebral edema, but the recommendation is explicitly based on low-quality evidence, and neither agent has demonstrated that lowering ICP translates into improved neurological outcome.

For an acute crisis, practical bolus regimens might include 3% saline in a several-mL/kg bolus, concentrated 23.4% saline—commonly 30 mL in an adult—or mannitol roughly 0.25 to 1 g/kg, depending on urgency, hemodynamics, renal function, prior therapy, and institutional protocol. The physiologic advantage of hypertonic saline in a hypotension-prone patient is obvious: unlike mannitol, it does not induce the same osmotic diuresis and may expand intravascular volume. Mannitol remains entirely legitimate when appropriately selected, particularly when sodium or chloride burden makes hypertonic saline less attractive.

If an EVD is present and hydrocephalus or elevated CSF volume contributes meaningfully to ICP, CSF drainage can be both diagnostic and therapeutic. BTF notes that continuous CSF drainage through an EVD may reduce ICP burden more effectively than intermittent drainage in severe TBI, although the recommendation is Level III. Remember, however, that ICP may not be spatially homogeneous. A patient with a focal posterior fossa lesion or large unilateral mass can develop dangerous pressure gradients and herniation even when a remotely positioned ventricular or parenchymal monitor gives a deceptively reassuring number. Treat the patient and anatomy, not the transducer.

The 2024 ACS revision of its TBI best-practice guidance reinforces a tiered approach to intracranial hypertension and greater use of advanced neuromonitoring rather than treating every elevation with maximal therapy from the outset. That is the correct physiologic architecture. Low-risk reversible measures come first. More aggressive interventions should require progressively stronger evidence that secondary injury is ongoing.

When ICP becomes refractory, the therapeutic logic changes. Moderate additional hypocapnia can temporarily reduce cerebral blood volume. Neuromuscular blockade may eliminate coughing or dyssynchrony but removes your motor examination and requires adequate sedation and EEG awareness where relevant. High-dose barbiturates can suppress cerebral metabolic rate and cerebral blood flow, reducing ICP, but at the cost of myocardial depression, hypotension, immunologic effects, prolonged awakening, and an almost mandatory loss of useful examination. Decompressive craniectomy directly increases intracranial compliance and can be lifesaving, but survival and functional outcome are not synonymous; trial data in TBI have repeatedly demonstrated that reducing ICP is easier to prove than improving the quality of neurological survival. Hypothermia is perhaps the cleanest example of the difference between physiology and patient-centered evidence: cooling reduces cerebral metabolic demand and frequently lowers ICP, yet prophylactic or ICP-directed therapeutic hypothermia in severe TBI has not produced the expected neurological benefit and in some settings has caused harm. Current practice therefore should not equate a therapy’s ability to lower ICP with proof that it helps the patient.

So when you hear, “The ICP is 24,” the attending-level response is not automatically, “Give 23.4%.” Your internal response should be:

• Twenty-four for how long?
• What was it an hour ago?
• What does the waveform look like?
• What is the CPP?
• Where is the arterial line referenced?
• What are the pupils?
• What does the imaging show?
• Is this focal or global?
• What is the PaCO2?
• Is the patient febrile, seizing, coughing, or obstructing venous drainage?
• Is autoregulation likely preserved?
• Is the EVD functioning?
• Is PbtO2 normal?
• What therapeutic burden have we already imposed, and is the next therapy more dangerous than the ICP we are treating?