Optic nerve sheath diameter (ONSD)

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

Vignette A. Suspected raised ICP at a peripheral hospital#

A 4-year-old falls from a tree at a peripheral hospital with no neurosurgical capability. GCS 7 on arrival, no obvious skull fracture, CT shows diffuse cerebral oedema with effacement of basal cisterns. Transfer is 2 hours by helicopter. The treating team has access to bedside ultrasound but no ICP monitor. They measure ONSD bilaterally: right 6.2 mm, left 6.1 mm (well above the 4.5 mm cutoff for age 1–15). The combination of clinical picture, CT, and elevated ONSD prompts intubation, head elevation, hypertonic saline 3% bolus, and helicopter transfer. ICP measured invasively on arrival is 32 mmHg. ONSD did not measure ICP, but it changed the decision to escalate before transfer. 1 2 3 4

Vignette B. Bacterial meningitis, ONSD trend over 24 hours#

A 7-year-old with pneumococcal meningitis day 1 presents with GCS 11, severe headache, photophobia. ONSD on admission 5.8 mm bilaterally. The team initiates antibiotics, supportive care, and selective ICP-monitoring decision: invasive placement deferred because GCS is improving with treatment. ONSD is repeated every 6 hours:

• 0 h: 5.8 mm (raised).
• 6 h: 5.6 mm.
• 12 h: 5.4 mm.
• 24 h: 5.1 mm.

The bedside team interprets the trend as resolving cerebral oedema with antibiotic and supportive care. GCS rises to 14 by 24 hours. No invasive monitor was placed. The ONSD trend gave the team temporal data the static measurement could not. 5 6 3

Vignette C. DKA cerebral oedema, ONSD rising 4 hours into rehydration#

A 9-year-old new-onset DKA, pH 7.05, bicarbonate 6, glucose 35. Rehydration started per PECARN-style protocol. At 4 hours, the patient reports a new severe headache and is mildly less responsive. ONSD measured: 5.2 mm right, 5.0 mm left (both above the 4.5 mm pediatric cutoff). The team treats empirically for evolving cerebral oedema with hypertonic saline 3% 5 mL/kg bolus, head elevation, and senior consult. CT is arranged. ONSD repeated at 30 minutes post-bolus is 4.8 mm bilaterally. The patient stabilises; CT confirms mild diffuse oedema without herniation. ONSD detected the change early, before pupillary signs, and the response to treatment was visible at the bedside in 30 minutes. 7 8 9 10 11

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

ONSD is the transcutaneous ultrasound measurement of the optic nerve sheath, taken 3 mm posterior to the globe along the optic nerve axis, in millimetres.

Why this works. The optic nerve is a CNS structure; its sheath is contiguous with the intracranial subarachnoid space and contains CSF. Raised ICP transmits through this CSF column and distends the sheath within minutes. The distension is most apparent in the retrobulbar segment, where the sheath is most compliant.

Three things follow.

ONSD is a surrogate, not a measurement. It does not give ICP in mmHg; it gives a probability of raised ICP. Various regression equations relate ONSD to ICP (e.g., Geeraerts 2007: ICP ≈ 7.0 × ONSD − 25), but bedside use is threshold-based, not equation-based.

ONSD has lag and inertia. A sheath that has been distended for days does not collapse immediately when ICP normalises; a sheath that has never been distended takes minutes to distend with new ICP rise. For acute ICP detection, ONSD has good sensitivity; for chronic raised ICP follow-up, ONSD lags ICP trends.

ONSD is bilateral and repeatable. A single measurement is suggestive; bilateral measurements + serial trend over hours is the bedside-useful pattern.

3. Technique#

3.1 Equipment#

• Linear high-frequency probe (7.5–15 MHz). The standard small-parts/vascular probe.
• Mechanical index < 0.23 to protect the lens; ocular preset on the ultrasound machine.
• Coupling gel in generous quantity to avoid pressure on the globe; sterile gel for closed-eye application.

3.2 Patient and probe positioning#

• Patient supine, head neutral, closed eyelids.
• Probe placed gently over the closed eyelid; transverse (axial) plane standard; sagittal plane as confirmation.
• No pressure on the globe; use coupling gel to keep the probe slightly off the skin.

3.3 Measurement#

• Identify the globe: anechoic vitreous; bright retina-choroid-sclera complex posterior.
• Identify the optic nerve: hypoechoic structure leaving the posterior globe.
• Mark 3 mm posterior to the retina-globe junction along the optic nerve axis.
• Measure the sheath diameter perpendicular to the nerve axis at this point.
• Repeat in the contralateral eye.
• Repeat in the sagittal plane if axial is ambiguous; values should agree within ~0.2 mm.

3.4 Repeatability#

ONSD has inter-rater variability of ~0.3 mm in trained hands; intra-rater repeatability is better (~0.15 mm). Round measurements to the nearest 0.1 mm; trends of ≥ 0.3 mm are clinically significant, single measurements within 0.3 mm of cutoff are equivocal.

4. Age-banded cutoffs#

Sources: 1 12 2 3 13 14 15 16.

5. Pattern library: normal, acute raised, chronic raised#

6. Try it: interactive widgets#

7. The ONSD-driven clinical decision triage#

ONSD informs four bedside decisions. Each has a different threshold for action:

7.1 Intubate?#

• ONSD raised + GCS ≤ 8 = strong support for intubation.
• ONSD raised + GCS 9–12 with clinical concern = case-by-case; pair with TCD-PI and trend.
• ONSD normal + GCS ≤ 8 = intubate for airway protection regardless; ONSD informs subsequent escalation.

7.2 CT scan?#

• ONSD raised + new neurology = expedite CT (or MR for stroke or posterior fossa concern).
• ONSD raised + trauma + GCS deterioration = CT regardless.
• ONSD borderline + stable = clinical judgement.

7.3 Transfer?#

• ONSD raised + no neurosurgical capability on-site = escalate transfer (intubate before transfer; pre-treat with hypertonic saline / mannitol if available).
• ONSD normal + clinical concern = transfer based on clinical picture; ONSD has informed but not blocked the decision.

7.4 Invasive ICP monitor?#

• ONSD raised + clinical picture supporting raised ICP + need for ongoing precise titration = invasive placement.
• ONSD raised + improving clinically = monitor non-invasively (serial ONSD + TCD-PI + clinical exam).
• ONSD normal + stable = no invasive monitor needed acutely.

flowchart TD
  Read[ONSD suggestive of raised ICP] --> Setting{Setting?}
  Setting -->|Peripheral hospital| Stabilize[Intubate, hypertonic saline, head up]
  Stabilize --> Transfer[Arrange urgent transfer]
  Setting -->|Tertiary unit| Clinical{Clinical state?}
  Clinical -->|GCS ≤ 8 + deteriorating| InvasiveICP[Invasive ICP placement]
  Clinical -->|GCS 9-12 + stable| Serial[Serial ONSD + TCD-PI + clinical exam every 4 h]
  Clinical -->|GCS ≥ 13 + stable| Image[CT or MR; treat cause]
  Serial --> Reassess{Worsening?}
  Reassess -->|Yes| InvasiveICP
  Reassess -->|No| Continue[Continue non-invasive]

8. Clinical contexts: ONSD across acute brain injuries#

8.1 Suspected raised ICP without invasive monitor#

The canonical indication. Use cases: post-trauma at peripheral hospital, post-trauma awaiting theatre, post-trauma where coagulopathy contraindicates invasive placement. Negative predictive value > 90% in trained hands. 1 3 4

8.2 Bacterial meningitis with raised ICP#

ONSD trend over 24 hours guides escalation in pediatric bacterial meningitis. Rising or persistently high ONSD prompts CT, hyperosmolar therapy, or EVD placement. The European meningitis guidelines and IDSA encephalitis guidelines acknowledge ONSD as a useful adjunct. 5 6 17

8.3 DKA cerebral oedema#

Pediatric DKA cerebral oedema classically presents 4–12 hours into rehydration. ONSD can detect early sheath distension before pupillary or motor signs appear. Response to hypertonic saline can be observed at the bedside within 30 minutes. 8 7 9 10

8.4 Post-traumatic in low-resource settings#

Where CT and invasive ICP are unavailable or delayed, ONSD provides immediate bedside data. Multiple resource-limited cohort studies (Padayachy 2012, 2016) established the pediatric thresholds in exactly these settings. 12 1

8.5 Hydrocephalus and shunt failure#

ONSD raised in a child with a shunt is a useful adjunct to clinical assessment of shunt failure. A normal ONSD does not exclude shunt failure (slit ventricles can produce raised ICP without sheath distension); a raised ONSD adds support for shunt revision. 1 3

8.6 Idiopathic intracranial hypertension (IIH)#

In older children and adolescents with IIH, ONSD is raised but the sheath often shows chronic thickening. Follow-up over weeks to months shows partial reversibility with treatment (acetazolamide, weight loss, lumboperitoneal shunt). 3

8.7 SAH#

Less established. The principle (sheath distension reflects raised ICP from hydrocephalus or oedema in SAH) is sound. Most SAH ICP monitoring is via EVD, making ONSD a redundant tier. Useful in pre-procedural assessment. 18 Sparse

8.8 Stroke (malignant MCA, post-thrombectomy)#

ONSD can detect evolving oedema in malignant MCA syndrome before clinical herniation. Adjunct to clinical and imaging surveillance. 19 20 Sparse

9. Multimodal integration: ONSD in the non-invasive ICP stack#

The non-invasive ICP triple (ONSD + TCD-PI + clinical) is the practical bedside bundle when invasive monitoring is unavailable or contraindicated. Each modality is imperfect alone; concordance across the three substantially improves diagnostic confidence. 4 3 21 22

10. Setup and technique: bench training#

10.1 Probe choice#

A high-frequency linear probe (7.5–15 MHz, vascular or small-parts setting) is standard. Curved probes have insufficient near-field resolution for the 3 mm posterior measurement.

10.2 Patient and probe positioning#

• Patient supine, head neutral, eyes closed.
• Generous coupling gel to avoid pressure on the globe; sterile gel for protection of the eye.
• Probe orientation: transverse (axial) primary; sagittal as confirmation.
• Probe angle: perpendicular to the optic nerve axis; tilt slightly until the longest visible nerve length appears.

10.3 Measurement protocol#

1. Position the gain so the vitreous is anechoic and the retinal complex is bright.
2. Identify the retina-globe junction at the posterior pole.
3. Measure 3 mm posterior along the optic nerve.
4. Measure the sheath diameter perpendicular to the nerve axis at this point.
5. Record to 0.1 mm.
6. Repeat in the contralateral eye.
7. Repeat in the sagittal plane if axial is ambiguous.
8. Average if multiple measurements per side; report the higher of the two sides if they differ by > 0.3 mm.

10.4 Training and competency#

• Inter-rater agreement: ~0.3 mm with formal training; ~0.5 mm without.
• Repeatability bench: 10 supervised measurements typically required for competency.
• Validation against simultaneous invasive ICP (when available) is the gold-standard training metric.

10.5 Safety#

• Mechanical index < 0.23 to protect the lens (standard ocular preset).
• No pressure on the globe; coupling gel slack.
• Avoid in patients with suspected open globe injury or recent eye surgery.

11. Pitfalls and artefacts#

• Thickened chronic sheath: chronic ICP elevation produces sheath fibrosis; the sheath does not collapse immediately when ICP normalises. ONSD lags in chronic states.
• Optic neuritis: inflammation thickens the sheath without raised ICP. False positive.
• Papilloedema visible on fundoscopy: confirms raised ICP and supports a raised ONSD; absence of papilloedema does not exclude raised ICP (papilloedema also lags).
• Asymmetric ONSD with unilateral disease: orbital tumour, optic nerve sheath meningioma, unilateral compartment syndrome. Investigate further.
• Operator dependence: inter-rater variability ~0.3 mm; trained operators only.
• Open fontanelle in infants: reduces sensitivity; pair with fontanelle US.
• Eye movement artefact: optic nerve appears off-axis when eye is rolled; have patient look straight ahead (or eyes closed, gaze straight).
• Excessive probe pressure: compresses the sheath; measure with light contact.
• Different reference values across cohorts: 4.5 mm in Padayachy pediatric cohort vs 5.0 mm in some adult cohorts; know the local standard.
• Sheath asymmetry > 0.5 mm: requires investigation; not always pathological but warrants attention.
• Ultrasound machine settings: ocular preset, mechanical index < 0.23, gain optimised for vitreous.

12. Combine with…#

• Non-invasive ICP: the wider nICP family (B4C waveform, TCD-PI, ONSD together).
• ICP: for confirmatory invasive measurement.
• TCD: for the PI non-invasive ICP comparator.
• Fontanelle ultrasound: for the infant pair.
• Pupillometry: for the brainstem function pair.
• NIRS: for the rSO2 perfusion pair.

13. Evidence summary and recent literature#

13.1 Evidence summary#

13.2 Recent literature (2022–2025)#

• Robba 2018 ONSD systematic review (cited heavily through 2025): the modern reference for thresholds and use cases; pediatric subsection establishes the 4.5 mm cutoff. 3
• Cardim 2023 multicentre nICP: cross-validation of ONSD with other non-invasive ICP estimators against invasive measurement. 28
• Rasulo 2024 B4C nICP: waveform-based non-invasive ICP estimator (Brain4Care) validated against invasive measurement; ONSD complementary. 29
• Padayachy 2016 pediatric reference: establishes age-banded cutoffs for pediatric ONSD; cited as gold-standard pediatric reference. 1
• Lovett 2022 pediatric ONSD review: contemporary synthesis of pediatric ONSD use cases and limitations. 32
• Figaji 2025 Pediatric MNM consensus: endorses ONSD as tier-1 bedside non-invasive tool, especially in resource-stratified pediatric centres. 33

14. Self-check#