UT Neuroanesthesia / Neuro 1 · Day 3
Unit Intracranial pressure Read 14 min Reviewed Sep 1, 2026 · JP Next review Aug 2027

The Number and the Waveform

Everyone knows the threshold. Almost nobody knows that the only randomized trial of monitoring against it was neutral, that the number is a dose rather than a line, and that the waveform which would actually tell you about compliance has never been shown to change anything.

Physiology · established Clinical claim · uncertain 1 RCT · 1 meta-analysis (n = 24,792) · 1 cohort (n = 2,395)
01

The pearl

Mean ICP is the number on the screen and the least informative thing the monitor knows. The same mean can come from a brain with room left in it and from one with none, and the difference shows in the waveform: P1 taller than P2 in a compliant brain, P2 climbing above P1 as compliance runs out. The honest position is that this morphology is well described and that acting on it has never been shown to change an outcome — which is roughly where the threshold itself stands.

02

Why it is true

mechanism

Normal mean intracranial pressure in a healthy, resting, supine adult is 5 to 10 mm Hg. Pressure above 20 to 25 is widely considered to require treatment, and the current Brain Trauma Foundation guideline recommends treating above 22. Those numbers are worth knowing because you will be held to them. It is worth knowing what is behind them, which is less than the confidence with which they are quoted.

What the monitor is, and what it is not

An intraparenchymal microtransducer sits 1.5 to 2 cm into the brain, placed through a cranial access device or at craniotomy. The intraventricular catheter has historically been called the gold standard, but intraparenchymal devices give equivalent pressure measurement with lower risk of haematoma and infection. Their weakness is a different one: zero drift over several days, with no way to recalibrate in vivo. The ventricular catheter’s real advantage is not accuracy but that it can drain.

Then there is the arithmetic that gets done with the number. Cerebral perfusion pressure is mean arterial pressure minus intracranial pressure, and it is the main driver of oxygen and substrate delivery. For that subtraction to mean anything the arterial transducer has to be referenced at the level of the foramen of Monro, taken at the tragus. Zero it at the phlebostatic axis, sit the patient up thirty degrees, and the displayed CPP overestimates the real one by the height difference. This is the most ordinary way a patient ends up being run at a lower perfusion pressure than the chart records, and nobody notices, because the chart looks fine.

The waveform

Each cardiac cycle produces three peaks. P1, the percussion wave, is arterial pulsation transmitted from the choroid plexus. P2, the tidal wave, reflects the compliance of the brain itself. P3, the dicrotic wave, follows aortic valve closure. In a brain with reserve they descend in a staircase: P1 > P2 > P3.

As compliance falls, the brain stops damping the transmitted pulse, P2 rises, and eventually climbs above P1 while the whole trace becomes rounded. The important part is that this can happen while the mean pressure barely moves. Two patients with an ICP of 15 are not necessarily in the same situation, and the waveform is where the difference is visible.

That is a genuinely useful thing to be able to see, and here is where honesty is required about how far it goes. Waveform analysis has been studied specifically to predict intracranial hypertension before compensation is exhausted — in theory, warning early enough to intervene in time. Clinical translation of the technique remains unproven. Use it the way you would use any unvalidated signal: to look harder, not to act.

Pressure is a dose

The most useful reframing on this page is that intracranial pressure is not a line to stay below but an exposure with a magnitude and a duration. The CENTER-TBI high-resolution data show that episodes of higher pressure are tolerated for shorter periods than more modest elevations, and that impaired cerebrovascular autoregulation or a reduced CPP shortens what the brain will tolerate further.

An ICP of 30 for two minutes and an ICP of 23 for six hours are different exposures, and no fixed threshold distinguishes them. That is the argument Chesnut and Videtta make for situational rather than fixed thresholds, and it is the position the evidence supports.

Where the noninvasive measures fit

Optic nerve sheath diameter can be measured at the bedside with a linear probe across a closed eyelid, and a diameter above about 0.5 cm has been reported as a cut-off for detecting intracranial hypertension in a binary, present-or-absent way. Automated pupillometry gives a Neurological Pupil Index from 0 to 5, where 3 or above is normal and a value below 3 has been shown to indicate an ICP above 20 mm Hg. Transcranial Doppler pulsatility index is an imprecise estimate of ICP with large intra- and inter-operator variability.

All three are screening tools. They are worth having when invasive monitoring is unavailable or contraindicated — a severe coagulopathy, for instance — and worth using to decide who needs a monitor. None of them gives you a pressure.

Compliant brain Compliance exhausted
height of a normal P1 A · COMPLIANT BRAIN P1 P2 P3 percussion tidal dicrotic P1 > P2 > P3 A staircase. Room left in the box. B · COMPLIANCE EXHAUSTED P1 P2 P3 P2 > P1 Rounded. The staircase is gone. The morphology is well described, and the mean pressure can be identical in both panels. Whether acting on it changes outcome has not been shown.
Figure 1. The same mean pressure, two different brains. P1 is the arterial percussion wave, P2 the tidal wave that reflects the compliance of the brain itself, P3 the dicrotic wave following aortic valve closure. As compliance falls, P2 rises and the descending staircase is lost. Traces are idealized and carry no calibration; they show morphology, not patient data. The caption below the figure is the part that matters: waveform analysis has been used to try to predict intracranial hypertension before compensation is exhausted, and clinical translation of the technique remains unproven.
03

The evidence

1 RCT · 1 meta-analysis (n = 24,792) · 1 cohort (n = 2,395)

BEST:TRIP

Chesnut RM, Temkin N, Carney N, et al. A trial of intracranial-pressure monitoring in traumatic brain injury. N Engl J Med. 2012;367(26):2471-2481. No benefit
Design
Multicenter RCT in Bolivia and Ecuador
Population
324 patients aged 13 or older with severe traumatic brain injury
Intervention
Guidelines-based management using a protocol for monitoring intraparenchymal intracranial pressure
Control
A protocol in which treatment was based on imaging and clinical examination, without an ICP monitor
Primary
A composite percentile across 21 measures of survival, consciousness, functional status at 3 and 6 months, and neuropsychological status at 6 months
Result
56 in the pressure-monitoring group vs 53 in the imaging–clinical examination group · P = 0.49
Mortality
39% vs 41% at six months · P = 0.60

What it does and does not settle It is the only randomized trial of ICP monitoring in traumatic brain injury, and it is routinely misread in both directions. It did not show that intracranial pressure does not matter, and it did not show that monitors are useless. What it showed is that a management protocol anchored to a monitor and a threshold produced outcomes no better than a protocol anchored to serial imaging and examination — in a setting with limited access to prehospital care and neuroimaging, which is a fair objection to generalizing it and not a reason to dismiss it.

Meta-analysis of ICP monitoring and mortality

Yuan Q, Wu X, Sun Y, et al. Impact of intracranial pressure monitoring on mortality in patients with traumatic brain injury: a systematic review and meta-analysis. J Neurosurg. 2015;122(3):574-587. No association
Design
Systematic review and meta-analysis of 14 studies
Population
24,792 patients with traumatic brain injury
Question
Does ICP monitoring reduce mortality compared with management without it?
Result
Pooled OR 0.93 (95% CI 0.77–1.11), P = 0.40 overall · in the 7 studies published after 2012 (12,944 patients), pooled OR 0.56 (95% CI 0.41–0.78), P = 0.0006 favouring monitoring

Why the split by year is the interesting part A result that reverses when you slice by publication date is telling you something about how practice changed, or about how studies were selected and reported, rather than delivering a clean answer. Both readings are live. The observational studies pooled here cannot separate the monitor from the intensity of care that accompanies it.

SYNAPSE-ICU

Robba C, Graziano F, Rebora P, et al. Intracranial pressure monitoring in patients with acute brain injury in the intensive care unit (SYNAPSE-ICU): an international, prospective observational cohort study. Lancet Neurol. 2021;20(7):548-558. Benefit
Design
International prospective observational cohort
Population
2,395 patients with acute brain injury (TBI, intracerebral haemorrhage, subarachnoid haemorrhage) in 146 ICUs across 42 countries; 55% were monitored
Result
Six-month mortality 34% (441/1,318) with ICP monitoring vs 49% (517/1,049) without · in patients with unreactive pupils, HR 0.35 (95% CI 0.26–0.47)

The reason this is not the answer This is observational, and the exposure is not randomly assigned. Patients get a monitor because a clinician decided they were worth escalating for, and the authors themselves note that monitoring was associated with a more intensive therapeutic approach. A fifteen-point mortality difference in that design is a description of who gets treated aggressively at least as much as a measurement of what the monitor does.

Synthesis. Set the three side by side and the shape is familiar. The one randomized comparison found no difference. The pooled observational data found no overall mortality effect but a strong one in the more recent studies. The largest modern cohort found a large mortality difference in a design that cannot separate the monitor from the decision to escalate care.

That pattern — null when randomized, impressive when observed — is what confounding by indication looks like from the inside, and it should be the reflex reading whenever a monitor appears to save lives.

What survives all three is narrower and more useful than a threshold. Intracranial pressure matters; the CENTER-TBI high-resolution data show that outcome depends on both the magnitude and the duration of raised pressure, so higher pressures are tolerated only for shorter periods, and impaired autoregulation or a low CPP shortens that tolerance further. Pressure is a dose. The number that has never been validated is the line drawn across it.

This is also why the Brain Trauma Foundation's 22 mm Hg and the widely used 20 to 25 mm Hg are best held as prompts to look rather than as commands to treat. Chesnut and Videtta's argument for situational rather than fixed thresholds is the position the evidence actually supports.

04

What you actually do

in the room
  1. Reference the arterial transducer at the foramen of Monro, level with the tragus, whenever you are calculating CPP. Zeroed at the phlebostatic axis with the head of the bed up, measured CPP overestimates real CPP by roughly the height difference. This is the most common way a patient is quietly run at a lower perfusion pressure than the chart says.
  2. Read the ICP as a dose, not as a line crossed. Magnitude and duration interact: brief high pressures and sustained moderate ones are different exposures. A single number with no time attached to it is half the information.
  3. Look at the waveform before you accept the mean. P2 rising above P1 with loss of the descending staircase says the pressure-volume reserve is spent, at a mean that may look acceptable. Use it to raise your suspicion; do not use it as an indication in its own right, because that step has not been validated.
  4. Treat a threshold as a prompt to look for a cause, not as an instruction to treat the number. The guideline figure of 22 mm Hg and the conventional 20 to 25 are consensus values, and the one randomized trial of managing to a threshold was neutral. The question is always what changed.
  5. Do not assume the intraventricular catheter is the better monitor because it is called the gold standard. Intraparenchymal devices give equivalent pressure measurement with lower haematoma and infection risk. Their real weakness is different: zero drift over days, with no way to recalibrate in vivo. The ventricular catheter's advantage is that it can also drain.
  6. Interpret ICP alongside the other monitors rather than on its own. The same pressure means different things depending on autoregulatory state, CPP, brain tissue oxygen and the examination. Individualized treatment is the recommendation the evidence supports; a universal threshold is not.
  7. Use the noninvasive measures as screening, not as substitutes. Optic nerve sheath diameter above about 0.5 cm and a Neurological Pupil Index below 3 both flag an ICP above 20, and transcranial Doppler pulsatility index is an imprecise estimate with wide operator variability. They tell you to look; they do not tell you the pressure.
05

The pitfall

Treating the monitor

The characteristic failure is not missing intracranial hypertension. It is the slow drift into managing the number as if the number were the patient — an ICP of 23 gets a bolus of mannitol, an ICP of 19 gets nothing, and nobody asks what changed in the last twenty minutes or what the waveform is doing.

Two facts should keep that drift in check. The first is that a monitor is not a treatment: BEST:TRIP compared a protocol built around one against a protocol built around serial imaging and examination and found no difference in outcome or in six-month mortality. The second is that the threshold itself has no trial behind it. Both 20 to 25 and the guideline's 22 are consensus figures, and the argument for situational rather than fixed thresholds comes from the people who ran the trial.

The mirror-image error, and the more dangerous one on a neuroanesthesia rotation, is to take all of that as licence to ignore the number. Pressure does injure brain, in proportion to how high and for how long. Nothing here says the pressure does not matter. It says that a line drawn across it, treated as a rule, has never been shown to be the thing that helps.

06

Sources

primary first

Full text · Preston goes through LibKey to whatever Preston holds — the PDF where there is one — and works on a phone. The doi link always reaches the publisher's record, whether or not we hold a subscription. If you are reading on a computer, LibKey Nomad adds the same one-click PDF button to PubMed and publisher pages everywhere else; it is a desktop browser extension, so it does nothing on a phone or tablet.

07

Check yourself

3 items · one attempt each
Item 1 Analysis Neuro 1 · Day 3

The only randomized trial of intracranial pressure monitoring in traumatic brain injury found no difference in outcome or six-month mortality. A large international observational cohort found six-month mortality of 34% with monitoring versus 49% without. Which reading is best supported?

Item 2 Application Neuro 1 · Day 3

A patient with an intraparenchymal ICP monitor is nursed with the head of the bed at 30 degrees. The arterial transducer is zeroed at the phlebostatic axis. The monitor displays a CPP of 65 mm Hg. What should you conclude?

Item 3 Recall Neuro 1 · Day 3

An ICP trace shows three peaks per cardiac cycle. Over several hours, the second peak rises above the first and the waveform becomes rounded, while the mean pressure changes little. What does this indicate?

↻

Returning items

from earlier this rotation

No returning items before day 4. The retrieval window opens once there is something far enough back to be worth retrieving.