GEL Live! Optic Nerve Sheath Diameter & How to G.E.L. Part 1

By Michael Prats

ONSD Denver Part 1

This Special Edition podcast is the first "live" (at the time) Ultrasound GEL podcast, recorded with our friends from Denver Health Emergency Medicine and Denver Health Emergency Ultrasound. This is Part 1 of 2. The blog post is the same for both parts.

The podcast was certainly a collaborative effort and special thanks to Matthew Riscinti for helping develop the idea and set up it up.

The entire conversation was quite long so some of the content has to be edited. Additional thanks to all of the wonderful Denver residents, fellows, and faculty who participated in the discussion, including:

Ross Orpet (from EMS Cast)

Mike Heffler

Nick Kelly

Claudia Quenelle

Cody McIlvian

Austin Meggitt

Unlike most of our typical article summary podcasts, this was a more in-depth journal-club type discussion. It is broken into 2 parts. We discuss a recent article (see below), but also develop general principles about Gathering Evidence from the POCUS Literature. Below are just some general notes regarding the article. Make sure you check out the additional thoughts from our guest Expert Reviewer Tessa Damm at the end of this post.

A Prospective Study of Optic Nerve Ultrasound for the Detection of Elevated Intracranial Pressure in Severe Traumatic Brain Injury

Critical Care Med December 2020 - Pubmed Link


NIH funded - that is interesting because that is rare in POCUS studies

This study is important because it is the latest data in this area and compares to actual invasive standard



  • Age > 18 years

  • Severe TBI

  • ONSD could be done within 48 hrs of injury

  • IMPORTANT: Needed to have invasive ICP monitor (this is a huge bias in the population, some body already decided that they need an ICP monitor. This is going to skew the population towards the sick and high ICP)


  • Unlikely to survive 48 hrs

  • Globe injury

  • Ocular diseases??

  • TIL >10 (therapy intensity level - a scale measuring the degree of interventions taken for high ICP)

Reference standard was intraparenchymal monitor. Usually removed when ICP <20 (so ensuring that all of them would have elevated ICP).

Got scanned at enrollment and then at least daily until ICP removed (would have been better if kept measuring after removal). Max 7 days.

None had prior experience with Ocular ultrasound??

  • Training in supplemental data - 1 hour powerpoint, bedside demo, 20 scans at least. Then calculated the intra-class coefficient with an expert after 50 patients - if <0.8 → remediated training with everyone.

They were testing the accuracy of the HIGHEST ONSD during admission - that is less relevant to us. Especially for something that is a dynamic process.

Compared highest ONSD for diagnosis of ICP >25 mm Hg.

  • IMPORTANT: Most other studies used ICP >20 as abnormal. They used >25 mm Hg (primary) and also 22 (secondary - because BTF guidelines revised this as the treatment threshold).

They tried to figure out what is best cut off focusing on sensitivity (shot for 98% with 80% spec)

Figure 2 - flowchart. 723 excluded from severe TBI 723/858 - that means that this is represents only about 16% of people with severe TBI.

Those with ICP elevation were younger (25 vs 34)

Minor difference in the first recorded ONSD (0.69 vs 0.64) BUT there was a different in first recorded optic disc elevation 5 vs 2 mm


N = 120

5% of videos were rejected due to poor quality - good indicator of skill of providers and challenge of the scan (they couldn’t even find a good still from their clips)

Only 18% of the population had ICP >25 at any time. Interestingly only a small percentage of the patients (5% and 9% in high and not-high groups) had elevated ICP at the time of a measured ONSD. Big problem. Buried in supplemental table 2.

AUC of ONSD for ICP >25 mm was 0.76 (0.67-0.83).

Optimal threshold was >0.72 cm → Sensitivity 83% and specificity 76%.

This means that they were unable to find a sensitivity of 98% with at least 80% spec so their backup plan was to get at least 90% sensitivity and minimum spec of 60%. Sooo….

Had to remediate - hoping for 0.8 → got 0.21 ICC. Improved on the back end.

Interestingly, optic disc elevation had AUC 0.84, >0.04 cm best cutoff, 90% sensitive, 71% specific

ONSD of 0.5 was 100% sensitive, <15% specific → this is similar to my experience

They likely had very few false negatives - because we know this is sensitive and almost everybody was sick. Few false negatives means that the sensitivity would be high.

OR 3.72 (CI 1.68-8.24) - so people with a dilated ONSD were ~4x more likely to have an ICP over 22 (remember that 22 was a secondary analysis)



  • (Largest?) Prospective, compares to actual intracranial pressures

  • Good methodology

  • Optic nerve elevation


  • Poor agreement between observers

  • Small amount of high ICP

  • Different definition of high ICP

  • Compared US to the greatest ICP elevation which was commonly not at time of ONSD

  • Most patients were only monitored for 2 days

Not spelled out but it seems that the patients were from India. See the Clinical Trials page


Prospective study of 120 patients with severe TBI and invasive ICP monitoring. AUC for ONSD to find ICP >25 was 0.76. Best they could get was 83% sens, 76% spec. Interestingly optic disc elevation has AUC 0.84

Take Home Points from the Article

1. ONSD had poor interobserver reliability in this study of novice users.

2. The test characteristics are not good enough to use for identifying ICP >25 in a population of sick TBI patients

3. Optic disc elevation was actually more accurate than ONSD and deserves further study for its use in detecting elevated TBI.

Take Home Points from GEL-ing

1. Know the background literature for context/comparison

2. For methodology consider - is this true POCUS or just parts of the I-AIM being studied?

3. Indication problems → lead to problems with external validity (I can’t apply this to who I want to) Acquisition and interpretation problems → also external validity but in a different way (I am not sure if I do this, I will get same results). Medical decision making errors → affects the relevance and practical application (will this change my practice?).

Guest Expert Reviewer for this Post

Tessa W. Damm, DO @DrDamm

Intensivist, Medical Critical Care & Neurocritical Care, SSM Health Dean Medical Group. Madison, WI

Reviewer's Comments

How exciting to have a large, NIH-funded, prospective, blinded study evaluating the accuracy of point of care optic nerve sheath diameter (ONSD) measurement compared to invasive intracranial pressure (ICP) monitoring! But are the findings practice changing? Here are my initial reactions and considerations:

Patient Population: This study used a homogeneous patient population of adults admitted with severe TBI (defined as GCS 8). Significantly this implies that all included patients uniformly had high pretest probability of elevated intracranial pressure (ICP) warranting placement of an invasive intracranial monitor. This differs from how I use ONSD measurement in clinical practice, where I select patients who do not otherwise meet criteria for invasive ICP monitoring (including a GCS 8) to serially assess their neurologic exam paired with ONSD measurement to determine when invasive monitoring and/or initiation of treatment for elevated ICP may be indicated. In the study analysis, investigators notably found ONSD predictive of ICP dichotomized at less than or equal to 22 mm Hg versus greater than 22 mm Hg, supporting the aforementioned clinical practice.

Timing: This study compared the accuracy of the single highest ONSD measurement during admission, rather than dynamically correlating ONSD measurement with real-time acute elevations of invasively measured ICP; the latter clearly being of more clinical relevance. One of the many benefits of POCUS in all domains is the ability of imaging to be repeated over time in attempt to identify and discriminate dynamic findings. In this study, investigators performed ONSD imaging on patients once daily.

Type of Invasive ICP Monitor: Interesting choice by investigators to use intraparenchymal probes rather than ventriculostomy (commonly referred to as external ventricular drain or EVD) as the standard invasive measurement tool. EVD remains the gold standard in many centers as it allows for concurrent treatment of ICP elevation by way of CSF diversion. In practice, intraparenchymal probes are typically a resort used if ventricular effacement in a TBI patient makes ventriculostomy challenging or higher risk. Intraparenchymal probes also do not allow for recalibration in situ which has historically limited their use (1). In a prior study evaluating the specific intraparenchymal device used in this study, the Codman Microsensor (CMS), with ICP measurements obtained from a ventriculostomy the average ICP measured with ventriculostomy was 18.3 0.3 mm Hg and with the CMS was 19.0 0.2 mm Hg (2).

Experience of Investigators: This study had four investigators performing bedside ultrasound, none of whom had prior experience with ONSD image acquisition or interpretation. There was initial poor inter-rater agreement among the bedside clinician investigators compared with the experts; an interclass correlation coefficient that improved with remedial training. Optic disc elevation of greater than 0.04 cm was found to have acceptable sensitivity and sensitivity (90% and 71%, respectively). Of note, this latter measurement was only obtained by the expert investigators. In addition to serving as indication for further study of the utility of optic disc measurement, these findings suggest that technique and experience rather than the modality of ONSD measurement itself may be to blame for the findings of less-than-optimal sensitivity and specificity. The study did, after all, reveal expert measured ONSD achieved area under the curve >0.8 for detection of concurrent ICP > 22 mm Hg; a finding germane to clinical practice as ICP > 22 mm Hg is widely regarded as a standard treatment threshold.

Practical Point: In general, the superficial location and fluid-filled nature of the eye render it an ideal target for POCUS. Acquiring on-axes images, however, often proves challenging. Traditionally an image qualifies as being on-axis when the lens can be viewed in the same frame as the optic nerve. Much like distortion of cardiac chambers size with off-axis imaging, accurate optic nerve sheath diameter measurement is dependent upon viewing the structure on-axis (3). Proper image acquisition and interpretation takes practice.

Summary: While this publication adds to the body of literature pertaining to routine use of ONSD as a non-invasive means of identifying patients with elevated ICP, it does not change my clinical practice. ONSD measurement remains a useful adjunct to serial clinical neurologic examinations in patients who otherwise do not meet criteria for invasive intracranial monitoring. Performing this exam regularly to maintain expert status also appears to be of value.


  1. Akbik OS, Carlson AP, Yonas H. The Roles of Ventricular and Parenchymal Intracranial Pressure Monitoring. Curr Neurobio. 2016; 7 (1): 1-6.

  2. Koskinen LO & Olivecrona M. Clinical Experience with the Intraparenchymal Intracranial Pressure Monitoring Codman MicroSensor System. Neurosurgery. 2005; 56: 693.

  3. Damm T. (2020, July 13). A Neurointensivist’s Approach to POCUS for Increased ICP. Retrieved from

Cite this post as

Michael Prats. GEL Live! Optic Nerve Sheath Diameter & How to G.E.L. Part 1. Ultrasound G.E.L. Podcast Blog. Published on December 14, 2020. Accessed on May 06, 2021. Available at
Published on 12/14/20 04:00 AM
comments (0)