Intracranial pressure monitoring

1. Bedside vignettes: why this matters in the PICU#

Vignette A. Severe pediatric TBI day 1, the waveform changes shape before the number does#

A 4-year-old falls from a third-floor balcony. GCS 5 on arrival; CT shows bifrontal contusions, a small subdural, and effacement of the basal cisterns. A right frontal Camino is placed within the hour; baseline ICP 14 mmHg, clean three-peak P1 > P2 > P3 waveform. By hour 18 the mean has crept to 28 mmHg sustained, and the waveform morphology has flipped: P2 > P1, rounded contour, RAP 0.8. You start the tiered escalation: head-up to 30 degrees, propofol bolus and increase infusion, hypertonic saline 3% at 3 mL/kg, normothermia, normocapnia. ICP falls to 22 over 30 minutes. The team begins to plan early decompressive craniectomy because compliance is gone, not because the absolute number is alarming. 1 2 3

Vignette B. Neonate with hydrocephalus, no invasive monitor available#

A 3-week-old with congenital aqueductal stenosis presents with a tense, bulging anterior fontanelle, sun-set sign, and head circumference crossing two centiles in 10 days. The neurosurgeon is two hours away. There is no ICP monitor on the unit. You scan ONSD bedside: 5.8 mm bilaterally (well above the 4.0 mm pediatric cutoff for under-1s), TCD-PI 1.6 with low diastolic flow, NIRS rSO2 falling from 70 to 58%. The clinical and non-invasive picture converges on raised ICP without ever placing an invasive monitor. The infant is intubated, head elevated, and transferred for emergent third ventriculostomy. 4 5

Vignette C. Adolescent SAH day 6, EVD draining 5 mL/h, but TCD MFV is rising#

A 12-year-old with a ruptured posterior-communicating aneurysm post-clipping, day 6. EVD set to drain at 10 cm H2O, mean ICP 8 mmHg, output 5 mL/h. The bedside nurse notes the right MCA TCD MFV has risen from 95 to 160 cm/s over 12 hours, with a Lindegaard ratio of 3.8. The team asks the question that ICP alone cannot answer: is this vasospasm or rising ICP? The ICP trace is reassuring; the EVD is patent (good pulsatility, respiratory swing of 2 mmHg); the TCD signature with LR > 3 is classic vasospasm. The team escalates haemodynamics (lift MAP, induce mild hypertension), schedules angiography, and the ICP trace remains stable while the spasm is treated. The ICP measured what ICP measures, and the TCD answered what the ICP could not. 6 7

2. What ICP is, and what it is not#

ICP is the pressure of the supratentorial CSF + parenchymal compartment transmitted to a probe inside the skull, in mmHg. The reading reflects three superimposed signals:

1. The mean, set by total intracranial volume against cranial compliance (the Monro-Kellie doctrine, Marmarou exponential P-V curve). The mean is the number nurses chart and physicians act on.
2. The pulse amplitude (AMP), transmitted from the arterial pulse and modulated by compliance. AMP rises as compliance falls, often before the mean ICP does.
3. The respiratory swing, transmitted from intrathoracic pressure through the venous compartment. A 1–4 mmHg swing confirms catheter patency.

The Monro-Kellie equation is the ground truth:

Once compensation is exhausted, the Marmarou exponential becomes the dominant truth:

where PVI (pressure-volume index) is the volume change that would multiply ICP by ten. PVI in healthy adults ~25 mL; in severe TBI it falls to ~10–15 mL, and the next +1 mL of oedema can double the ICP. 8 9

ICP equals brain-tissue pressure only when the waveform shape is healthy and the probe is communicating with the whole intracranial volume. In compartmentalised oedema (e.g., temporal lobe contusion behind a closed tentorium), the regional pressure can be much higher than the supratentorial probe reads. This is the canonical "ICP is normal but the pupil blew" presentation.

What ICP cannot do. It cannot localise the lesion, distinguish vasogenic from cytotoxic oedema, confirm whether autoregulation is intact, detect non-convulsive seizures driving the rise, or tell you the perfusion delivered to the tissue. It is the bedrock, not the building.

3. EVD versus intraparenchymal: choosing the probe#

Fig. 1

The decision tree. Choose EVD when you want to drain CSF (communicating hydrocephalus, IVH, SAH with hydrocephalus, neuro-deterioration with raised ICP and patent ventricles). Choose IPM when ventricles are slit or unavailable, when coagulopathy raises bleed concern with deeper catheter passes, or when fast bedside placement is the priority. Many units default to EVD for SAH and to IPM for diffuse TBI. 12 1 13

4. The ICP waveform: anatomy of a single cardiac cycle#

A clean ICP trace has three readable peaks per cardiac cycle:

1. P1, the percussion wave, transmitted directly from the arterial systole via the choroid plexus and large basal arteries. Sharp leading edge. In a healthy compliant brain, P1 is the tallest of the three.
2. P2, the tidal wave, a reverberation through the brain parenchyma. Its height tracks compliance: low-compliance brain dampens the parenchymal recoil and P2 grows. P2 > P1 is the hallmark of exhausted compliance.
3. P3, the dicrotic wave, following the aortic valve closure transmission. P3 is the smallest in normal compliance; it can merge with P2 as morphology deteriorates.

The amplitude (AMP) is the peak-to-trough of the cardiac pulse, usually 1–4 mmHg in healthy compliance, rising to 6–10 mmHg as compliance falls.

RAP, the compensatory reserve index, is the moving correlation between AMP and mean ICP over a 4-minute window:

• RAP ≈ 0: good compensatory reserve (changes in volume produce little change in pressure, so AMP and mean ICP are uncorrelated).
• RAP → +1: compensatory reserve exhausted (every volume change moves both AMP and mean ICP together).
• RAP falling toward 0 or negative at high ICP: the Marmarou curve has crested; further volume increase brings pressure-volume decompensation. This is a sinister change. 14 15 3

5. The numbers to record (the six-pack)#

For every patient, on every nursing shift, record this six-pack and trend over time:

The dose is the single most important addition to the chart over the past decade. ICP minutes spent above threshold correlate with neurological outcome at 6 months in adult and pediatric severe TBI more tightly than peak ICP or mean ICP. The KidsBrainIT pediatric dataset showed a clear dose-response curve: cumulative ICP > 20 mmHg burden over the first 72 hours predicted GOS-E 6 months later, with a knee at ~8 mmHg·hours below which outcome was unchanged. 16 17

This is why bedside teams should think in area under the curve, not point measurements. A patient who spent 6 hours at 25 mmHg has a worse forecast than one who spiked to 35 mmHg for 10 minutes and resolved.

6. What is normal? Age-banded reference values#

Sources: 1 11 18. The guideline target is ICP below 20 mmHg across all ages (treat above 20; Kochanek 2019, PBTF, Level III); there are no separate age-stratified ICP treatment thresholds. Emerging paediatric evidence points lower: sustained ICP above approximately 14 to 15 mmHg is associated with worse outcome, and the field is moving toward ICP dose (intensity x duration) rather than a single cutoff. 19 20 Pair the number with clinical exam, fontanelle palpation, ONSD, and trend. CPP, by contrast, IS guideline-age-banded (Level III): an individualised CPP target that rises with age, infants at the lower end and adolescents at or above the upper end (see the CPP column above). The neonate column has the weakest evidence base; treat the bedside picture and the fontanelle as much as the number.

7. What is abnormal? A pattern library#

Lundberg taxonomy comes from the original 1960 continuous-monitoring monograph that established the modern bedside language. 22 Plateau waves are pathognomonic of low compliance; their amplitude can briefly exceed the operating MAP, producing a brief CPP crash and risk of herniation.

8. Try it: interactive widgets#

9. ICP-guided CPP management and tiered escalation#

This is where the ICP number enters the therapeutic cascade. The pediatric severe-TBI playbook (PBTF 4th edition) organises interventions into three tiers, with thresholds for escalation that depend on time spent above ICP threshold, not on a single peak.

9.1 The CPP target by age (operational)#

CPPopt by PRx is preferred where available (see PRx page and CPPopt page). The age-banded floor is a default; if a PRx-derived U-curve gives a CPPopt of 52 in a 9-year-old, target ±5 mmHg of that, not the table value. 23 24 25 26

9.2 The tiered escalation pathway#

flowchart TD
  Read[ICP reading] --> Trend{Mean &gt; threshold<br/>for &gt; 5 min<br/>or rising AMP?}
  Trend -->|No| Continue[Continue monitoring]
  Trend -->|Yes| Tier1[Tier 1: head up 30 deg,<br/>sedation analgesia,<br/>CSF drainage if EVD,<br/>normocapnia 35-40,<br/>normothermia]
  Tier1 --> Persist1{Persists &gt; 20 min?}
  Persist1 -->|No| Wean[Continue and reassess]
  Persist1 -->|Yes| Tier2[Tier 2: hyperosmolar<br/>3% NaCl 3 mL/kg<br/>or mannitol 0.5 g/kg,<br/>neuromuscular blockade,<br/>repeat CSF drainage]
  Tier2 --> Persist2{Persists?}
  Persist2 -->|No| Stabilize[Stabilise + reassess]
  Persist2 -->|Yes| Tier3[Tier 3: hyperventilation<br/>PaCO2 30-35,<br/>barbiturate coma,<br/>decompressive craniectomy,<br/>mild therapeutic hypothermia]

Tier 1 is the bedside hygiene every PICU should reflexively apply when ICP creeps. Head-of-bed at 30 degrees, neutral midline neck position, optimised sedation and analgesia, normocapnia at PaCO2 35–40, normothermia, sodium 145–155, glucose normal, seizure prophylaxis if EEG suggests need.

Tier 2 introduces hyperosmolar therapy and neuromuscular blockade. Hypertonic saline (3% NaCl bolus 3–5 mL/kg, then infusion to keep sodium 150–155) is generally preferred over mannitol in pediatrics because of stable haemodynamics and the ability to continue infusion. Mannitol 0.25–1 g/kg works as a bolus but carries diuresis and rebound risks. 1 27

Tier 3 is the rescue tier. Mild hyperventilation (PaCO2 30–35) reduces CBF acutely and lowers ICP transiently; below 30 it risks ischaemia. Barbiturate coma (pentobarbital infusion to burst suppression on EEG) reduces CMRO2 and ICP but carries refractory hypotension and immune suppression. Decompressive craniectomy is the most definitive intervention; RESCUEicp (adults) and emerging pediatric data show ICP reduction at the cost of unfavourable functional outcomes in survivors. 2 28

10. Clinical contexts: ICP across acute brain injuries#

10.1 Severe pediatric TBI#

The canonical indication. PBTF 4th-edition guidelines recommend invasive ICP monitoring for children with GCS ≤ 8 and an abnormal CT, individualised CPP target by age, and tiered therapy keyed to ICP > 20 mmHg (all ages). The KidsBrainIT cohort demonstrated the cumulative ICP-dose to outcome relationship that underpins modern "minutes above threshold" thinking. 1 16 17 11

The BEST-TRIP trial (adults, Bolivia/Ecuador) showed that protocolised ICP-guided care was not superior to imaging/exam-guided care for 6-month outcome. Critics argue the trial recruited a population where ICP monitoring was already not routine and that the result reflects context, not modality. SYNAPSE-ICU observational data (37,000 patients) showed ICP monitoring associated with lower 6-month mortality. The bedside consensus remains: monitor when you can; act on dose, not peak. 12 29

10.2 Aneurysmal SAH and IVH#

ICP via EVD is the standard placement in aneurysmal SAH with IVH or hydrocephalus, because CSF drainage is itself therapeutic. Adult AHA/ASA 2023 SAH guidelines recommend EVD for symptomatic hydrocephalus and continuous ICP monitoring in poor-grade SAH. Pediatric SAH (often AVM- or trauma-related) follows similar principles but is anatomically different (younger vessels, different aneurysm patterns). 6 7

In SAH, rising ICP must be distinguished from vasospasm-driven flow changes: ICP can be stable while TCD MFV rises (vasospasm without ICP rise) or rising independently (rebleed, acute hydrocephalus, oedema). This is the canonical "pair ICP with TCD" scenario.

10.3 HIE and post-cardiac arrest#

Routine ICP monitoring in HIE / post-arrest is not standard. The Eurotherm-3235 and THAPCA cohorts did not use invasive ICP. Selective use in patients with malignant cerebral oedema documented on imaging is centre-dependent. Non-invasive surrogates (ONSD, NIRS, fontanelle US in infants) are the more common bedside tools. 30 31 32

10.4 Bacterial meningitis with raised ICP#

Pediatric bacterial meningitis can produce both communicating hydrocephalus (from inflammatory blockade of CSF absorption) and cerebral oedema. EVD placement is the first-line invasive monitor when raised ICP is documented or strongly suspected. The European meningitis guidelines and IDSA encephalitis guidelines acknowledge ICP monitoring as a tier-2 intervention in severe cases. 33 34 35

10.5 Hydrocephalus and shunt malfunction#

ICP monitoring is used both diagnostically (overnight ICP recordings in suspected shunt failure, looking for B waves) and therapeutically (EVD as bridging in acute shunt failure pre-revision). The ESPVS style of overnight ICP-trend recording remains a useful diagnostic for shunt under-drainage. 10 36

10.6 DKA cerebral oedema#

A pediatric-specific case where ICP monitoring is rarely invasive in time. Cerebral oedema in DKA classically presents 4–12 hours into rehydration, with rapid clinical deterioration (headache, altered consciousness, then herniation). The PECARN FLUID trial showed that fluid-rate choice did not change cerebral-oedema rates; the bedside priority is early recognition with neurological scoring, hypertonic saline 3% bolus, and CT, rather than waiting for an invasive monitor. ONSD and clinical scoring are the practical bedside tools. 37 38 39 40

10.7 Hepatic encephalopathy with cerebral oedema#

Acute liver failure can produce ammonia-driven cerebral oedema that is the most common cause of death in fulminant hepatic failure. Invasive ICP monitoring is contentious because of the bleed risk from coagulopathy; transfused factor support before placement reduces but does not eliminate this. Many centres now favour non-invasive monitoring (ONSD, jugular bulb saturation, EEG) until coagulation supports a safer invasive placement. 41

10.8 Post-stroke malignant MCA and decompressive craniectomy#

Large-vessel ischaemic stroke with malignant oedema is a pediatric emergency in older children. Decompressive hemicraniectomy within 48 hours of clinical decline improves survival but at the cost of disability; ICP monitoring post-decompression is sometimes used to titrate medical therapy through the swelling phase. 28 2 42 43

11. Multimodal integration: ICP in the MMM/MNM stack#

ICP is the bedrock invasive number, but alone it answers only one question. Pair with the modalities below to answer the others.

12. Setup and technique: a step-by-step#

12.1 Site selection#

Kocher's point is the canonical entry: ~2.5 cm lateral to the sagittal midline at the coronal suture (or ~10 cm posterior to the nasion, ~2.5 cm lateral), on the non-dominant hemisphere (right-sided in most). This avoids the motor strip and the sagittal sinus, and the trajectory through frontal white matter has the least eloquent cortex.

Frazier's point (~6 cm above and 3 cm lateral to the inion) targets the occipital horn for posterior approaches.

12.2 Intraparenchymal probe insertion#

1. Position head neutral, skin shaved over the entry site, sterile prep and drape.
2. Local anaesthetic (lidocaine + adrenaline subcutaneously).
3. Skin incision 5–10 mm vertical stab.
4. Burr hole with twist drill or hand-cranked perforator.
5. Dural puncture through the bolt with a small spinal needle.
6. Probe self-zeroing in air (intraparenchymal probes calibrate once, pre-insertion, then never again).
7. Insertion to fixed depth (typically 1–2 cm into white matter).
8. Bolt locked to skull.
9. Display: connect to monitor, confirm three-peak waveform with respiratory swing and cardiac pulse, document baseline ICP and morphology.

12.3 EVD insertion#

Same approach through burr hole, then a silicone catheter is passed ~5–7 cm to enter the frontal horn of the lateral ventricle. CSF egress confirms placement. The catheter is fluid-coupled through saline-filled tubing to an external pressure transducer.

12.4 Zeroing, the most-missed source of error#

• Fluid-coupled (EVD): zero the transducer to atmosphere at tragus level (corresponds to foramen of Monro at supine). If the transducer drifts up or down with the patient's head position, ICP changes by ~7 mmHg per 10 cm. Re-zero every shift and after every position change.
• Intraparenchymal: self-zero before insertion, never again. Drift over days (typically 1–3 mmHg by day 5) is the trade-off.

12.5 Signal quality cues#

• Clean three-peak waveform with P1 > P2 > P3 in normal compliance: trust.
• Respiratory swing of 1–4 mmHg: confirms patency.
• Cardiac pulse present: confirms vascular coupling.
• No swing, no pulse, dead-line trace: catheter is blocked, kinked, or the transducer is disconnected. Flush per local protocol; do not act on a flat-line ICP.

12.6 Infection prevention#

Bundle elements with the strongest evidence: chlorhexidine prep, full-barrier draping, antibiotic-impregnated catheters (silver-coated or rifampicin/minocycline-impregnated), tunnelled subcutaneous track ≥ 5 cm, closed drainage system with no breaks, no routine prophylactic systemic antibiotics. Infection rate climbs from ~2% in week 1 to ~10% by week 2. Plan exchange or removal at day 5–7 if still needed and clinical state allows. 13 46

12.7 EVD drain management#

The two competing modes:

• Continuous drainage with set level: drain set at 10–20 cm H2O above tragus; CSF drains continuously when ICP exceeds that. ICP is not readable while draining.
• Intermittent drainage with monitoring: drain closed for monitoring, opened for 5-minute intervals when ICP exceeds threshold or per shift schedule. ICP is readable during closed periods.

Most centres now favour intermittent monitoring with set-level drainage to allow ICP measurement and CPP calculation; trial data is mixed and unit protocols vary widely. 13

13. Pitfalls and artefacts#

• Drift in fibre-optic probes (typically < 3 mmHg/day, but cumulative over a week).
• Damping of fluid-coupled EVDs from bubbles, kinks, or clots; characteristic loss of waveform pulsatility.
• Open EVD continuously zeroes ICP to the drainage chamber height; cannot read true ICP through an open drain.
• Transducer height changes: every 10 cm above tragus reads ICP ~7 mmHg too low; every 10 cm below reads it ~7 mmHg too high.
• Coughing, suctioning, bucking: transient ICP spikes via thoracic pressure transmission. Note in trace; do not act unless sustained or recurrent.
• Sedation effects on baseline: deep propofol reduces CMRO2 and lowers baseline ICP by ~3–5 mmHg; lightening sedation should prompt a re-baseline.
• Compartmentalised oedema (e.g., temporal lobe contusion behind a closed tentorium): regional pressure can be much higher than the supratentorial probe reads. ICP can be "normal" while the pupil blows.
• Slit ventricles: EVD placement fails ~5% from inability to cannulate; use intraparenchymal instead.
• Sodium driving ICP: hyponatremia (< 130) raises ICP via brain oedema; correct cautiously to avoid osmotic demyelination.
• PEEP transmission: PEEP > 10–12 transmits to ICP via venous outflow resistance, especially with reduced lung compliance. See PEEP-ICP-MAP demo.

14. Combine with…#

• CPP: driving pressure = MAP − ICP, the partner to every ICP measurement.
• PRx: autoregulatory state through this very ICP signal.
• CPPopt: individualised target derived from the PRx-CPP relationship.
• RAP: what the ICP waveform tells you about compliance.
• Non-invasive ICP: when invasive placement is contraindicated or unavailable.
• ONSD: bedside ultrasound surrogate.
• TCD: flow consequence of the same physiology and a spasm-vs-ICP discriminator.
• PbtO2: tissue oxygenation pair (the BOOST triplet).
• Pupillometry: brainstem early warning.

15. Evidence summary and recent literature#

15.1 Evidence summary#

15.2 Recent literature (2022–2025)#

• COGiTATE phase II (Beqiri 2024): adult severe TBI trial demonstrating CPPopt-guided care is feasible and safe, with secondary endpoints suggesting outcome benefit. The pediatric extension is awaited but Tas 2024 single-centre data already supports the concept in children. 24 26 51
• Pediatric MNM consensus 2025 (Figaji): formalises ICP as the anchor invasive modality and stratifies tier-2 modalities (PbtO2, microdialysis, PRx-derived CPPopt) by resource level. 44
• Non-invasive ICP estimators: Brain4Care-style waveform-derived nICP (Brasil 2021, 2022; Cardim 2023; Rasulo 2024). Useful for triage and trend-following but not a replacement for invasive measurement in patients with established intracranial hypertension. 52 53 54 55
• ICP dose vs outcome (Guiza 2015 + follow-on): the cumulative-dose framework is now the standard outcome-mapped metric and is increasingly built into ICU-grade ICP monitoring systems.
• SYNAPSE-ICU follow-on: ICU-level variability in ICP monitoring practice remains substantial; calls for international standardisation of indications and thresholds. 29

16. Self-check#