Optic nerve sheath diameters in nontraumatic brain injury: A scoping review and role in the intensive care unit

Abstract

BACKGROUND

Neuromonitoring in medical intensive care units is challenging as most patients are unfit for invasive intracranial pressure (ICP) modalities or unstable to transport for imaging. Ultrasonography-based optic nerve sheath diameter (ONSD) is an attractive option as it is reliable, repeatable and easily performed at the bedside. It has been sufficiently validated in traumatic brain injury (TBI) to be incorporated into the guidelines. However, currently the data for non-TBI patients is inconsistent for a scientific recommendation to be made.

AIM

To compile the existing evidence for understanding the scope of ONSD in measuring ICP in adult non-traumatic neuro-critical patients.

METHODS

PubMed, Google Scholar and research citation analysis databases were searched for studies in adult patients with non-traumatic causes of raised ICP. Studies from 2010 to 2024 in English languages were included.

RESULTS

We found 37 articles relevant to our search. The cutoff for ONSD in predicting ICP varied from 4.1 to 6.3 mm. Most of the articles used cerebrospinal fluid opening pressure followed by raised ICP on computed tomography/magnetic resonance imaging as the comparator parameter. ONSD was also found to be a reliable outcome measure in cases of acute ischaemic stroke, intracerebral bleeding and intracranial infection. However, ONSD is of doubtful utility in septic metabolic encephalopathy, dysnatremias and aneurysmal subarachnoid haemorrhage.

CONCLUSION

ONSD is a useful tool for the diagnosis of raised ICP in non-traumatic neuro-critically ill patients and may also have a role in the prognostication of a subset of patients.

Key Words: Intracranial hypertension; Intracranial pressure monitoring; Medical intensive care unit; Neuro-monitoring; Non-traumatic brain injury; Optic nerve sheath diameter; Optic nerve ultrasound

Core Tip: Neuromonitoring in critically ill patients is challenging as many patients are unfit for invasive intracranial pressure (ICP) monitoring or unstable to transport for imaging. Bedside ultrasonography based optic nerve sheath diameter (ONSD) has been proven to be a reliable option in patients with traumatic brain injury (TBI). However, it’s efficacy has not been extensively evaluated in neuro-medical patients. In this review, we analyzed data from 37 articles which had compared ONSD with other established modalities of measuring ICP in non-TBI patients. The analyzed data suggests that ONSD may be a useful tool to detect raised ICP and predict outcome in patients with acute ischemic stroke, intracerebral bleed and intracranial infection. However, further large-scale randomized trials are required, especially in patients with septic metabolic encephalopathy, dysnatremias and aneurysmal subarachnoid hemorrhage, before it is routinely employed in managing neuro-medical patients with elevated ICP.

INTRODUCTION

A number of pathologies lead to increased intracranial pressure (ICP) in medical intensive care unit (ICUs). A new onset altered sensorium can be due to septic or metabolic encephalopathy, new intracranial vascular event, anoxic brain injury, newly diagnosed intracranial malignancy, seizures, posterior reversible leukoencephalopathy syndrome (PRES), intracranial infections, amongst others[1]. Raised ICP in most of these cases is underdiagnosed as the patient is unstable to be shifted for radiological imaging like computed tomography (CT) or magnetic resonance imaging (MRI) or has contraindications for invasive monitoring. There are added issues of accessibility and expenses related to the invasive modalities. Ultrasonography-based optic nerve sheath diameter (US-ONSD) is a quick and safe bedside technique with reasonable accuracy, reproducibility, and a smaller learning curve, even in novice operators[2,3]. Additionally, the inter-observer variation has been reported to be minimal[4]. The procedure has been standardized and can be performed using the bedside ultrasonography machines, now routine ICU equipment, making it a useful point-of-care test. The data on traumatic brain injury (TBI) patients is expansive and convincing enough for The Brain Trauma Foundations to adapt it into their guidelines[5]. However, there is a dearth of data in non-traumatic causes of increased ICP. Most data in this cohort of patients comes from prospective cohort or case-control studies, retrospective case series, and individual case reports. Also, the patients included have a heterogeneous aetiology, making it difficult to pool the data while adjusting the baseline variables. The comparator parameters in these studies also vary from invasive ICP monitoring, cerebrospinal fluid (CSF) opening pressure and CT/MRI imaging. Hence, we conducted a scoping review to analyze the available data.

MATERIALS AND METHODS

The review was conducted in accordance with the Joanna Briggs Institute methodology for scoping reviews[6].

Aims and objectives

The objective of this review was to describe the extent and type of evidence for monitoring ICP in non-TBI using US-ONSD. Our primary outcomes for the study were: To assess the strength of evidence of ONSD's accuracy in non-traumatic cases compared to other modalities of ICP monitoring, both invasive and noninvasive. Evaluate cut off for ONSD in non-traumatic increased ICP. Narrative review of the role of ONSD in medical ICUs.

Inclusion criteria: (1) Studies including patients over 18 years old with raised ICP on whom US-ONSD was performed; All the studies irrespective of the study site (emergency department, ICU, neurology or neurosurgery wards); (2) Studies which evaluated the predictive value of ONSD for determining the ICP using any of the comparator parameters to detect elevated ICP (intraventricular catheter, intraparenchymal monitor, CSF opening pressure on lumbar puncture (LP), radiological signs of cerebral oedema on CT/MRI) either as primary objective or secondary outcomes were included; (3) Studies on neurosurgical patients who did not have any history of trauma were also included; and (4) Studies in the English language.

Exclusion criteria: (1) Studies which included patients with TBI; (2) Studies with ill-defined aetiology of raised ICP; (3) Literature reviews, letters to editors, editorials and clinical opinions; and (4) Studies which compared ONSD to indirect evidence of raised ICP like papilloedema on fundoscopy or optic disc height (ODH) on ultrasonography, Glasgow Coma Score (GCS), automated pupillometry, transcranial doppler (TCD), or ophthalmic artery indices.

Concept: In patients with TBI, raised ICP is the norm. However, not all patients with altered sensorium in a medical ICU have increased ICP. Untreated intracranial hypertension (ICH) can lead to worse patient outcomes. The points of interest in our review included the primary pathology involving the central nervous system, methodological details, the diagnostic accuracy of ONSD in comparison to the different methods in detecting ICH, the sensitivity and specificity of a specific ONSD method and ONSD cut off, if determined.

Search strategy: A search in PubMed, Google Scholar and Research citation analysis aimed to find studies published between January 2010 and March 2024. This expansive period was considered with an effort to integrate maximum number of studies. Only studies after 2010 were included to remain relevant to technological advances and updated clinical practices that have occurred in the last two decades. 109 studies were screened, 37 were excluded as they did not address the research question. Two reviewers reviewed 72 articles; 35 were excluded, as shown in Figure 1. A total of 37 articles which satisfied the inclusion and exclusion criteria and passed the review were included in the study.

Figure 1 Scheme for data extraction.

RESULTS

We included 37 publications that answered our search question according to our inclusion and exclusion criteria. Table 1 outlines the characteristics of the reviewed studies. The most common method for detecting increased ICP was CSF opening pressure on LP (20 studies, 54.05%), followed by the presence of cerebral oedema on CT/MRI scans (14 studies, 37.83%) and intraventricular catheter (3 studies, 8.1%).

Table 1 Basic characteristics of the studies on optic nerve sheath diameter in patients with non-traumatic increased intracranial pressure, n (%).

Study characteristic	Number of studies, n = 37
Year of publication
2013	1 (2.7)
2014	3 (8.1)
2015	5 (13.5)
2016	3 (8.1)
2017	4 (10.8)
2018	2 (5.4)
2019	4 (10.8)
2020	1 (2.7)
2021	3 (8.1)
2022	2 (5.4)
2023	7 (18.9)
2024	2 (5.4)
Country of origin
China	8 (21.6)
India	7 (18.9)
Turkey	5 (13.5)
South Korea	3 (8.1)
Iran	2 (5.4)
Brazil	2 (5.4)
Italy	2 (5.4)
United States	2 (5.4)
Uganda	1 (2.7)
Spain	1 (2.7)
Saudi Arabia	1 (2.7)
Greece	1 (2.7)
South Africa	1 (2.7)
Germany	1 (2.7)
Comparator parameter
Intraventricular ICP monitoring	3 (8.1)
CSF opening pressure for lumbar puncture	20 (54)
MRI/CT scans	14 (37.8)
Type of study
Prospective observational study	17 (45.9)
Prospective case control study	12 (32.4)
Prospective cohort study	4 (10.8)
Retrospective case study	2 (5.4)
Case reports	2 (5.4)

ICP: Intracranial pressure; CSF: Cerebro-spinal fluid; MRI: Magnetic resonance imaging; CT: Computed tomography.

Seventeen studies were prospective observational studies, 12 were prospective case–control, 4 were prospective cohort, and 2 were retrospective case series. Two case reports were also included. The details of the studies, sample size, study population, comparator measure, cut-off for ONSD with Area under receiver operating curve (AUROC), when available, and limitations of the study have been described in Table 2[1,7-42].

Table 2 Summary of studies on optic nerve sheath diameter in patients with non-traumatic raised intracranial pressure.

Ref.	Country	Type of study	Number of participants	Patient characteristics	Comparator parameter and P value, r value	ONSD cut off, AUROC	Results	Limitations
Amini et al[7], 2013	Iran	Descriptive prospective	50	Non-traumatic patients requiring lumbar puncture	CSF pressure on LP, (P = 0.05; r = 0.88)	5.5, NA	The ONSD of greater than 5.5 mm predicted an ICP of ≥ 20 cmH2O with sensitivity and specificity of 100% (95%CI: 100-100) (P = 0.001)	Small sample size
Caffery et al[8], 2014	United States	Prospective observational trial	51	Non-traumatic causes of raised ICP	Opening pressure on LP, r = 0.53	> 5.0 mm, 0.69	Sensitivity 0.75, specificity of 0.44	Use of a convenience sample could introduce bias. Sample size was small with large confidence intervals. One physician with specialized training, patients were not matched for demographic variables such as age or sex
Nabeta et al[9], 2014	Uganda	Prospective descriptive study	57	HIV positive, ART naïve adults suspected with meningitis	CSF opening pressure, P < 0.001	5	Sensitivity 86% and specificity 63% for predicting a CSF ICP > 200 mm PPV was 77% and NPV was 75%. Also, in ONSD > 5mm had a RR of 2.39 for IICP > 200 cmH2O	Inter-operator variability, with training being essential
Shirodkar et al[10], 2014	India	Prospective, observational case control study	101, 60 study, 41 control	Non traumatic cause of increased ICP	Increased ICP on CT/MRI, P < 0.001	4.71, 0.986	Sensitivity of 77.8% and specificity of 100%	Small size
Wang l et al[11], 2015	China	Prospective observational cohort study	279	Non-traumatic cause of increased ICP	CSF opening pressure, P < 0.001	4.1, 0.965	Sensitivity of 95% and a specificity of 92%	Average of 8 measurements of ONSD to decrease variability. May not be feasible practically
Du Toit et al[12], 2015	South Africa	Prospective observational study	76	Meningitis	CSF opening pressure, Cohen’s kappa was 0.41	4.8, 0.73	Sensitivity of 50% and specificity of 89.8% PPV of 54.8% and NPV of 88.3% PLR of 4.92 and NLR of 0.56	The study was unable to establish inter-observer variability owing to the large number of operators and the small number of patients with increased ICP
Sangani et al[13], 2015	India	Prospective observational study	25	Tubercular meningitis	CSF opening pressure, P < 0.001	NA	Those patients with TBME had a mean ONSD of 5.81 mm	Small sample size
Singleton et al[14], 2015	United States	Case report	1	Idiopathic ICH	CSF opening and closing pressure which was 36 cm H2O and 19.5 cm H2O after removal of 19 cc of CSF	NA	Pre-LP ONSD of left and right eye were 72 and 6.8 mm, respectively. Second study after 30 minutes left and right ONSD were 58 and 6.2 mm, respectively
Karzamni et al[15], 2015	Iran	Prospective case control study	60, 30 cases and 30 controls	Intracranial SOL and ICH	Increased ICP on CT, P < 0.001	4.53	Sensitivity and specificity of 100%. ONSD was the most sensitive and specific parameter, followed by RI, PI and EDV. ONSD correlated significantly with GCS (r = −0.40, P = 0.003) and ventricular shift on CT images (r = 0.37, P = 0.02)	Small size, lack of direct ICP measurement
Komut et al[16], 2016	Turkey	Prospective case control study	100	Nontraumatic intracranial event in ED	Increased ICP on CT, P < 0.05	5.3, 0.728	Sensitivity 70%, specificity 74%	Lack of direct ICP measurement
del Saz-Saucedo et al[17], 2016	Spain	Prospective case control study	30	IIH	CSF opening pressure, P = 0.005	6.3 to predict CSF pressure of 25, 0.93	Sensitivity 94.7%, specificity 90.9% and PLR of 10.4. After a therapeutic lumbar puncture an 87% of cases had a partial reduction of ONSD values	Small size
Salahudd et al[1], 2016	Saudi Arabia	Prospective cohort study	102	Non traumatic raised ICP	Increased ICP on CT, P < 0.001	5.7, 0.785	Sensitivity 84 % and specificity 71%. PLR = 2.89, NLR = 0.22	Study did not include a detailed neurological exam or record any specific localizing neurologic signs, individual GCS
Jeon et al[18], 2017	Korea	Prospective case control study	62	Nontraumatic cases requiring EVD placement	Opening pressure on EVD insertion, P < 0.01	> 5.6, 0.936	Sensitivity of 93.75% and a specificity of 86.67% for identifying increased ICP	To reduce selection bias, patients with severely increased ICP which required emergency surgical decompression before ICP monitor insertion. Study reflects increased ICP due to moderate hematoma in a Korean population
Gökcen et al[19], 2017	Turkey	Retrospective comparative study	191	Acute ischemic stroke	Raised ICP on CT, P < 0.001	Right ONSD 5.4, 0.941, Left ONSD 5.3, 0.922	CVD subgroups were compared with the control group the highest ONSD was in TACI group and the lowest was in LACI group (P < 0.001)	Unequal number of cases in the subgroups and adjustment of baseline charecteristics not mentioned
Wang et al[20], 2017	China	Prospective case control study	316	Nontraumatic increased ICP requiring LP	CSF opening pressure, r = 0.758, P < 0.001	NA	Xing and Wang mathematical equation for predicted ICP = −111.92 + 77.36 × ONSD (Durbin-Watson value = 1.94)	Equation may underestimate the true ICP value in patients with extremely high ICP. The Bland-Altman analysis in this study suggested that any estimate might be deviate by as much as ± 80mmH2O
Liu et al[21], 2017	China	Prospective observational study	110	Non-traumatic increased ICP requiring LP	CSF opening pressure, P < 0.001	5.6, 0.861	Sensitivity of 86.2% and specificity of 73.1%	5%–15% of the cases were classified
Wang et al[22], 2018	China	Prospective case and control study	60	Nontraumatic causes of IICP requiring LP	CSF opening pressure, P < 0.001. The ultrasonographic ONSD and ICP were measured on admission and follow-up	NA	ONSD was strongly correlated with ICP (r = 0.702, P < 0.001)	Small size, ONSD cut off not obtained
Canakci et al[24], 2018	Turkey	Prospective case control study	100	Non-traumatic headache presenting to ER	Raised ICP on CT, P < 0.001	5.5, NA	ONSD value in the ipsilateral side with the lesion was significantly higher than the contralateral side (P < 0.001). Discharge, clinical hospitalization, referral, ICU stay, emergency surgery	ER based study including patients with nontraumatic headache not exclusively patients with clinical features of raised ICP. No AUROC calculated
Naldi et al[24], 2019	Italy	Prospective case control study	46 cases, 40 controls	Primary ICH	Increased ICP on CT, P < 0.01	5.6, 1.0	Sensitivity 100%, Specificity 100%	ICP was presumed to be normal in control, limited predictive value of abnormal CT findings. Second CT scan was performed not on a given day, but depending on clinical conditions
Gupta et al[25], 2019	India	Prospective observational study	100	Raised ICP requiring LP	CSF opening pressure, P < 0.001	6.3 to predict CSF pressure of > 20 cm of water	Sensitivity 77.3%, specificity 92.3%, PLR = 10.05, NLR = 0.25	Did not include any condition causing a mass effect, malignant infarcts, ICH or obstructive hydrocephalus
Gupta et al[26], 2019	India	Retrospective case series study	100	Raised ICP requiring LP	CSF opening pressure, P < 0.001	4.8	Sensitivity of 85% and specificity of 88%	Single center, retrospective, small size
Wang et al[27], 2019	China	Case reports	2	Venous sinus stenosis and venous sinus thrombosis	CSF opening pressure	NA	Case 1 A predicted ICP by ONSD was 346 mmH2O. and CSF opening pressure was 355 mmH2O. Case 2 ONSD was 5.95 mm with CSF opening pressure higher than 400 mmH2O
Zoerle et al[28], 2020	Italy	Prospective observational study	20	Aneurysmal SAH with EVD	Intraventricular ICP, P > 0.05	NA	ONSD measurements were accurate, very similar to the diameters measured by MRI (the mean difference in the Bland–Altman plot was 0.08 mm, 95% limits of agreement: −1.13; + 1.23 mm). No clear relationship was detectable between the ICP and ONSD, and a linear regression model showed an angular coefficient very close to 0 (P > 0.05). US-ONSD and ICP values were in agreement after CSF drainage and shifts in ICP in a limited number of patients	Measured ICP in the ICU after the patients were stabilized, the aneurysm repaired, and large intracerebral hematomas surgically removed, with EVD and CSF drainage. As a consequence, the ICP values in our cohort were relatively low for the majority of cases
Sahu et al[29], 2021	India	Prospective, double blinded observational study	30	Nontraumatic increased ICP	Direct intraventricular ICP, P = 0.01	5.5 to predict ICP > 20 mmHg, 0.904	Sensitivity 100% and specificity 75%. The ONSD values predicting ICP at 25-, 30-, and 35-mm Hg were was 6.3 mm, 6.5 mm, and 6.7 mm, respectively	Small number of patients having ICP > 30 mm of Hg, appropriate ONSD values could not be predicted
Yildiz et al[30], 2021	Turkey	Prospective, observational study	82	Acute ischaemic stroke	Increased ICP on CT, P < 0.05	NA	ONSD on the 3rd day and 5th day was larger (> 5 mm) than on first day (P < 0.05). In the patients who received tPA right eye ONSD on the 5th day were significantly raised P < 0.05)	ONSD only after the symptoms started, and were also not measured during the decline periods and response to treatment
Kim et al[31], 2021	South Korea	Prospective, observational study	199	Suspected raised ICP	Increased ICP on CT, P < 0.001	5.3, 0.903	Sensitivity of 75.4%, specificity of 90.8%, PPV of 76.8%, and NPV of 90.2%	Single centre, 2 observers hence there can be variability
Qamar Akhtar et al[32], 2022	India	Prospective case control study	100	Non traumatic emergencies with suspected raised ICP	Raised ICP on CT/MRI (P = 0.05; r = 0.88)	≥ 6.3, 0.956	Sensitivity of 100%, specificity of 89.2%, PPV of 83.3%, NPV of 100%, and diagnostic accuracy of 93% for detection of raised ICP by bedside USG ONSD measurement compared to CT/MRI brain	CT or MRI brain scan which is an indirect indicator of raised ICP, and use of a high ONSD mean value (mm) cut-off
Oliveira et al[33], 2022	Brazil	Prospective observational study	40	Malignant MCA infarct requiring decompressive craniotomy	Increased ICP on CT, P: NA	5.4 mm, ROC for, Right eye: 0.82, Left eye: 0.77	Post craniectomy, there was a decrease in the mean value of 1.04mm in the right eye 086 mm in left. (P = 0.003)	Small size, CT unreliable for increased ICP. DC individualized is routinely adopted at this center, the neurosurgical team was allowed to perform surgery using individual interpretations of criteria, with controversial decisions on some patients
Roemer et al[34], 2022	Germany	Prospective observational study	23	Increased ICH	CSF opening pressure, P = 0.9	NA	No correlation between CSF opening pressure and ONSD was found	Small size, results could be biased by the ongoing treatment of the patients
Bhide et al[35], 2023	India	Prospective observational study	114	Non-traumatic causes of raised ICP	Increased ICP on CT/MRI, P < 0.001	5.75, 0.844	Sensitivity and specificity of 77.55% and 89.06%. PLR and NLR of 7.09 and 0.25	Comparator used was CT or MRI brain scan which is an indirect indicator of raised ICP, and use of a high ONSD mean value (mm) cut-off
Yu et al[36], 2023	China	Prospective observational study	107	Non traumatic increased ICP requiring LP	CSF opening pressure, P < 0.001	6.3 mm	73% sensitivity and 83% specificity, ODH with ONSD showed the highest value under the receiver operating characteristic curve of 0.965 with a sensitivity of 93% and a specificity of 92%	Single lumbar puncture
Batur et al[37], 2023	Turkey	Prospective case control study	105	Acute ischemic stroke	Features of raised ICP on MRI (P < 0.001)	5.05, 0.978	Sensitivity 96.8%, specificity 95.6%. The cut-off for need for treatment 4.95 mm with AUC of 0.807 (sensitivity = 71.4%, specificity = 79.6%)	Single-centered study. Although 30-day mortality rates were recorded, a detailed information about the outcome could be given by monitoring the neurological healing rate and time of the patients
Li et al[38], 2023	China	Prospective observational study	56	Suspected encephalitis	CSF pressure, r = 0.769, P < 0.01	NA	Both ODH and ONSD had the ability to predict ICP (P < 0.05), but with time factors, ONSD displayed a stronger ability to predict ICP than ODH	Single-center design and small sample size. Cut-off value with AUROC not calculated
de Moraes et al[39], 2023	Brazil	Prospective observation study	18	Acute stroke (ischemic and hemorrhagic)	A 5-point visual scale for n raised ICP on CT and two parameters (time-to-peak and P2/P1 ratio) of a noninvasive ICP wave morphology monitor (r = 0.29)	5.2, 0.69	Sensitivity was 71.4%, the specificity was 70.4%, the PPV was 43.5%, and the NPV was 88.6%	Small size, assessment intervals varied, Non blinded, correlation modest to moderate strengths
Cheng et al[40], 2023	China	Prospective cohort study	223	Non-traumatic causes of raised ICP requiring LP	CSF opening pressure, P < 0.001	5.47, 0.933	ICP values were strongly correlated with ONSD, ONSD, and ONSD/ETD. ONSD and OND combined model predicted ICP = 139.394 × ONSD-112.428 × OND267.461 prediction accuracy was the highest. (ICC = 0.88)	Underestimated the ICP in very high cases, the maximum limit of our ICP values was 330 mmH2O, and values greater than 330 mmH2O were counted as 330 mmH2O
Bakola et al[41], 2024	Greece	Prospective center case-control study	31 case and 34 controls	Idiopathic ICH	CSF opening pressure on LP, (r = 0.716, P < 0.001)	5.15, 0.914	Sensitivity and specificity of TOS for diagnosis of IIH were 85% (95%CI: 66%-95%) and 90% (95%CI: 76%-98%), respectively. PPV 83% (95%CI: 74%-96%), NPV 94% (95%CI: 83-98%)	Subsequent measurements to estimate the potential treatment response using TOS were not part of our study protocol.
Kim et al[42], 2024	Korea	Retrospective analysis of prospectively gathered data ONSD measurements were conducted using a handheld ultrasonography device during the course of endovascular treatment	126	Aneurysmal SAH	CSF opening pressure on LP, (P < 0.001), the association between ONSD and ICP was validated through the application of a linear regression machine learning model. The correlation between ICP and various factors was explored through the modeling	5.45, 0.90, SHAP 5.58	Sensitivity 92.50, specificity 78.00, PPV 82.70 NPV 90.20	Small size, single center

ONSD: Optic nerve sheath diameter; ICP: Intracranial pressure; AUROC: Area under receiver operating curve; CSF: Cerebro spinal fluid; LP: Lumber puncture; NA: Not available; HIV: Human immunodeficiency virus; ART: Anti-retroviral therapy; PPV: Positive predictive value; NPV: Negative predictive value; RR: Relative risk; CT: Computed tomography; MRI: Magnetic resonance imaging; PLR: Positive likelihood ratio; NLR: Negative likelihood ratio; TBME: Tuberculous meningo-encephalitis; IIH: Idiopathic intracranial hypertension; LP: Lumber puncture; SOL: Space occupying lesion; ICH: Intracranial hypertension; RI: Resistive index; PI: Pulsatility index; EDV: End diastolic velocity; GCS: Glasgow coma score; EVD: Extra-ventricular drain; TACI: Total anterior circulation infarct; LACI: Lacunar infarct; SAH: Subarachnoid haemorrhage; MCA: Middle cerebral artery; DC: Decompressive craniotomy; ODH: Optic disc height; TOS: Transorbital sonography.

DISCUSSION

In our scoping review of 37 studies, US-ONSD was found to be fairly predictive of ICP. Barring the 2 case reports, 30 studies calculated the p-value for the association between US-ONSD and ICP. Twenty-eight studies showed a statistical significance between US-ONSD and ICP with P < 0.05. A prospective observational study by Zoerle et al[28] included 20 patients with aneurysmal subarachnoid hemorrhage (SAH) who underwent extra-ventricular drain (EVD) insertion and intraventricular ICP monitoring. An average of 4 measurements were taken for 15-day period for each patient. This study did not find significant correlation between US-ONSD and ICP (P > 0.5). They further conducted a dynamic test in 10 patients where ONSD values were correlated with ICP before and after CSF drainage. The correlation was inconsistent with 2 patients (20%) having completely different trends for US-ONSD and ICP[28]. Another study conducted by Roemer et al[34] in 23 patients with idiopathic ICH also did not show a significant correlation between US-ONSD and ICP (P = 0.9)[34]. Both these studies had a small sample size; hence, in the face of positive data from multiple other studies, these results must be interpreted cautiously.

Even though ONSD has been extensively used in trauma patients, there are still concerns regarding the procedure of ONSD and its clinical applications in non-traumatic patients. Hence, we have expanded on the concerns and grey areas regarding ONSD in medical ICU, outlined in Table 3.

Table 3 Pertinent concerns regarding optic nerve sheath diameter measurement in medical intensive care unit.

	Related issues
How to measure ONSD?
(1)	A scan or B scan? What is blooming effect?
(2)	CLOSED protocol?
(3)	Transverse or horizontal?
(4)	How many values before obtaining a mean?
(5)	ONSD or OND or ratio?
What is the body of evidence in various subset of patients?
(1)	Acute ischemic stroke and CVST
(2)	Acute hemorrhagic stroke
(3)	Hydrocephalus
(4)	Idiopathic intracranial hemorrhage
(5)	Meningitis
(6)	Septic metabolic encephalopathy
Can ONSD be used as a management tool?
Can ONSD be a reliable outcome measure?
(1)	Post cardiac arrest
(2)	Dysnatremia

ONSD: Optic nerve sheath diameter; OND: Optic nerve diameter; CVST: Central venous sinus thrombosis

The ONS is an anatomical continuum of the meninges. Hence, the ICP variation is reflected in the ONS, resulting in dilatation and increased ONSD. The point of measurement of ONSD lies 3mm behind the optic disc as this is where the sheath is most sensitive to ICP changes[43,44]. Most of the studies have followed this landmark for obtaining the ONSD.

Cut-off for ONSD

A single cutoff for ONSD has yet to be defined to diagnose raised ICP in neuro-medical patients. 27 studies specified the cutoff for ONSD by ROC to vary from 4.1 to 6.3 mm. A systematic review of idiopathic ICH patients reported similar concerns. Still, they noted that a cutoff of 5.15 mm had the best accuracy with the area under the curve (AUC) of 0.914 to predict idiopathic ICH[41].

A study by Wang et al[20] included 316 neurology patients who underwent LP. The group was divided using a 7: 3 ratio into modelling and test groups. The modelling group was studied for the patient's baseline characteristics, age, sex, body mass index (BMI), mean and diastolic blood pressure, and they were found not significantly associated with raised ICP. The authors then derived Xing and Wang's mathematical equation to predict ICP using US-ONSD values. Predicted ICP was calculated as = −111.92 + 77.36 × ONSD. This was further validated in the test group (n = 94), where a significant correlation was found between the observed and predicted ICP (r = 0.76, P < 0.001). The mean difference between measurements was −0.07 ± 41.55 mmH2O using Bland-Altman analysis. The intra-class correlation (95%CIs) for the calculation of noninvasive ICP using the prediction model was 0.86 (0.79–0.90). The main limitation of the equation is that in patients with very high ICP values, it might be underestimated as the 1.96 standard deviations, as per Bland- Altman analysis, mount up to +/- 80 mmH2O[20]. Further refinement of the equation may be required using larger RCTs to improve its accuracy.

Various studies describe different thresholds for the cutoff for ONSD. The blooming effect partly explains this. When the diameter is measured by decreasing the gain, it will appear larger, and increasing the gain will appear smaller. This blooming effect is the error attributed to non-standardized gain and sensitivity settings[45]. The differences between the A and B scans are summarized in Table 4. The role of A scan is limited by its availability in the ICU. If A scan is proven superior, future research implications would be developing hand-held A scan devices to accurately measure ONSD.

Table 4 Differences between A scan and B scan on ultrasonography.

A scan	B scan
Amplitude modulation scan	Brightness modulation scan
8 mHertz frequency with small non focused probe	10 mHertz with a larger focused probe
One-dimensional image of spikes of varying amplitudes along a baseline	Two-dimensional image
Provides quantitative information: Ex length of eyeball before surgery.	Provides topographical information
Basis of ocular biometry	Evaluation of ocular pathology
No blooming effect	Blooming effect while measuring ONSD
Not available as bedside equipment	Part of point-of-care ultrasonography, easily available

ONSD: Optic nerve sheath diameter.

Another way of overcoming the blooming effect would be using a Color Doppler to delineate the optic disc. CLOSED protocol expands to Color Doppler–Low power examination–Optic disk clarity–Safety (short examination duration)–Elevate frequency–Dual measurements[45]. Following the ALARA principle, a 10 MHz frequency probe is used. For ophthalmic use, the Food and Drug Administration recommends a mechanical index (MI) and thermal index (TI) cutoff should be MI ≤ 0.23 and TI max ≤ 1 °C with spatial-peak temporal-average intensity of (ISPTA. 3) ≤ 50 mW/cm². The examination is done supine with head elevation up to 30°. Patients who can follow commands should be asked to look straight ahead to decrease the tortuosity of the optic sheath. For the horizontal measurement, the probe should be placed at an approximate angle of 15°–20° on the patient's closed upper eyelid and for the vertical measurement, the probe should be placed on the center of the patient's closed eye, slightly toward the corner of the nose (later-to-medial direction). Two landmarks delineate the course of the ONS on Color Doppler.

The central retinal artery and vein runs through the center of the optic nerve.

The ophthalmic artery runs along the medial border of the optic nerve and limits the optic sheath. The dicrotic notch on Doppler waves can further confirm the ophthalmic artery.

Once the course of the optic nerve is defined, the sheath diameter must be measured 3 mm behind the optic disc. The same authors used the CLOSED protocol to be validated in patients with idiopathic normal pressure hydrocephalus. ONSD calculated by CLOSED protocol had a lower scatter, and the values were lower than those calculated by US-ONSD or MRI scan[46]. ONSD calculated by CLOSED protocol needs to be validated against ICP measurement by standard techniques, using tests of diagnostic accuracy.

The optic nerve is thickest, about 3 mm behind the retina, followed by a curved path posteriorly and medially, as shown by MRI and cadaveric studies. A transverse measurement shows the optic nerve as a pyramidal structure that widens posteriorly. This widening could be due to an artefact from the lamina cribrosa or an ONS. This artefact may look significantly wider in patients with elevated ICP. A coronal view is achieved by placing the probe on the temporal aspect of the eyeball, directing it nasally and posteriorly. This produces a circular cross-section of the ONS, which is more accurate than transverse or axial orientation in healthy volunteers[47].

A blinded observational study in two tertiary teaching hospitals by Agrawal et al[48] included 20 adults with increased ICP expected to receive invasive intracranial monitoring. Axial and coronal measurements were taken and the highest and average values were recorded. The coronal values showed less variability between each eye as compared to axial measurements (0.5 mm vs 1mm; P = 0.03). For predicting ICP, axial ONSD was shown to have higher accuracy. Also, a cutoff of highest axial measurement greater than 6.2 mm in either eye or mean axial measurement of 5.6 mm had a sensitivity of 100% in predicting high ICP over the following 24 hours[48].

Horizontal diameter is more frequently used to measure ONSD due to the ease of use. Studies that used both transverse and sagittal diameters calculated the mean. There was no significant difference between the readings taken transversely or sagittaly[48]. In our review, most authors took 2–4 readings before taking the mean. Few studies went up to 8 readings in each section. Multiple readings, however, are concerned with taking more time, especially when the patient is critically ill and needs immediate intervention. However, most studies documented the time taken for the study to under a minute.

Both ONSD and OND are measured 3 mm behind the globe. Eyeball transverse diameter (ETD) is measured horizontally till maximum diameter is obtained. Ratios that can be used are ONSD/OND and ONSD/ETD, which have been found to correlate with poorer outcomes in comatose patients[49]. However, the utility of these newer indices is yet to be researched in predicting increased ICP.

Acute ischemic stroke

Acute ischemic stroke (AIS) can lead to raised ICP in two ways. Firstly, when large hemispheric infarctions have surrounding cerebral oedema significant enough to cause elevated ICP. These malignant ischemic strokes are usually secondary to occlusion of the internal carotid artery or M1 segment of the middle cerebral artery (MCA). These patients usually have progressive worsening over 36 to 48 hours, requiring decompressive craniotomy (DC). The second cause of raised ICP in ischemic strokes could be hemorrhagic transformation of the stroke. The deterioration is often sudden and unpredictable in these patients. A study of 40 patients with ischemic stroke showed an increase in ONSD on day 5 as compared to day 1 of stroke. A cutoff of 5.4 mm was found to be predictable of raised ICP. This study also showed a significant decrease in the ONSD post-DC (P = 0.003). The ONSD trend was also higher in patients who died compared to the survivors in both surgical and non-surgical groups[33]. Thus, ONSD could be helpful in prognosticating patients in malignant MCA infarcts.

In a study of 82 patients with AIS, 22 patients were administered recombinant tissue plasminogen activator (rtPA) treatment. The values of ONSD in the right eye on the 5^th day were significantly higher in thrombolysis patients than those of patients who did not receive rtPA (P < 0.05). However, no statistical difference occurred between 24, 36, and 48 hours of ONSD. Additionally, in patients who underwent DC, there was no correlation between ONSD and decompression surgery (P > 0.05). Other severity scores, such as GCS, National Institutes of Health Stroke Scale, Modified Rankin Scale, and ASPECTS scores, did not correlate to an increase in ONSD (P > 0.05)[30].

The utility of ONSD in AIS can be questioned in two aspects. Firstly, which eye would be more sensitive to predict the increased ICP. Oliveira et al[33] found a simultaneous increase in ONSD in both eyes in patients with severely increased ICP. Furthermore, when the injured hemisphere was left, the left ONSD had higher sensitivity; when the right hemisphere was affected, the contralateral ONSD was higher[33]. Thus, these authors found left ONSD a reliable marker for elevated ICP in most cases. On the contrary, Yildiz et al[30] did not find any statistical difference (P > 0.05) between the right and left eyes on the 1^st, 3^rd, and 5^th days after AIS[30].

Secondly, the site of ischemic stroke may be a confounding factor in the utility of ONSD. Cerebral hemispheric infarcts are more likely to influence the ONS than infratentorial lesions. Two studies expanded on this by first dividing the strokes using Oxfordshire Community Stroke Project classification on the basis of their maximum neuro deficit into total anterior circulation infarct (TACI), partial anterior circulation infarct (PACI), posterior circulation infarct (POCI), or lacunar infarct (LACI). Gökcen et al[19] found that ONSD was higher in all the subgroups of strokes than in the cohorts. Also, the TACI group had the highest ONSD values, and LACI was found to have the lowest values. Right and left ONSD had a cutoff of 5.4 and 5.3 mm, respectively, to predict the presence of cerebrovascular disease[19]. Similar findings were also seen in a study by Batur et al[37] with TACI having higher ONSDs than the others (TACI: 5.27 mm; PACI: 4.73 mm; POCI: 4.77 mm; and LACI: 4.64 mm, P < 0.001). They also found that the patients with TACI had an increased need for ICU admission. Raised ONSD also significantly correlated with mortality in this study (P < 0.001)[37].

We also found one case report of a patient with cerebral venous sinus thrombosis (CVST) who had raised ONSD of 5.95 mm on presentation with a CSF opening pressure of 400 cmH2O. Further, follow up of patient after treatment showed a concurrent decline in ONSD and CSF opening pressure[27]. Another case reported a 32-year-old female who underwent a caesarian section, following which she developed a post-dural puncture headache. The patient initially had low ICP syndrome. After a progressive increase in symptoms, a repeat radiological examination indicated CVST. ONSD showed raised values of 6.0 and 6.2 mm, leading to the suspicion of raised ICP. An ophthalmologic examination revealed papilledema, and the patient was treated for elevated ICP and CVST[50].

Acute hemorrhagic stroke

A study in ICH patients showed higher median binocular ONSD, resistive index and retinal venous pulsation values compared to the control group. Median binocular ONSD cut-off of ≥ 5.6 mm showed a higher accuracy for the detection of increased ICP with AUROC of 1.0. At the onset of ICH, ONSD also showed a good correlation with haemorrhage volume (r = 0.677, P = 0.0002). As with AIS, both right and left ONSD eye were found to have good accuracy (0.92, 95%CI: 0.87–0.96), suggesting a small side-to-side variation in a study of multimodal ICP monitoring in ICH patients. The cut-off for predicting increased ICP was 5.2 mm with an AUROC of 0.69. However, the correlation between the mean ONSD value and the mean ICP was found to be weak (r = 0.29)[24].

A large study of 529 patients with ICH showed that the ONSD in patients with poor outcomes was significantly higher than in survivors (P < 0.001). There was a significant correlation between hematoma volume and ONSD measurements (r = 0.529,P = 0.012)[51].

In another large study of 126 patients of aneurysmal SAH undergoing endovascular coiling, raised ONSD correlated significantly with an ICP > 20 cmH2O (5.9 mm vs 4.8 mm, P = 0.00). Similar findings were seen at an ICP cut-off of 25 cmH2O. A linear regression model was used to generate a line of best fit. It used a predictive equation of ICP (cmH2O) = 79.70 × ONSD (cm) − 21.60. The coefficient of determination was 0.54 as per this analysis[42].

However, other studies have found ONSD to be a weak predictor of increased ICP in aSAH. A study by Zoerle et al[28] included patients with aneurysmal SAH who had an EVD system placed. There was no clear correlation between the ICP and US-ONSD and a linear regression model also concurred with a lack of statistical significance (P > 0.05)[28].

Intracranial infections

Intracranial infections are a common cause of raised ICP in medical ICUs. Nabeta et al[9] conducted a study on 98 human immunodeficiency virus (HIV) infected patients with suspected meningitis. They found a moderate correlation between increased ICP and US-ONSD (ρ = 0.44, P < 0.001) and reported that ONSD > 5 mm was likely to have an elevated ICP (> 200 mmH2O). Eighty-one percent of these patients had cryptococcal meningitis. A restricted analysis of this subset of patients showed a significant correlation of ONSD with ICP (P < 0.001) with a cut-off of 5.4 mm. There was no statistical correlation between fungal burden by quantitative CSF culture and ONSD. Hence, evidence of increased ONSD, along with relevant clinical features can be a trigger for therapeutic CSF drainage[9].

In a prospective case study of 56 patients with encephalitis, ONSD and CSF opening pressure were measured on admission and after 2 weeks. As per the etiology of encephalitis, 19 patients had viral, 16 had tuberculous, 9 had pyogenic and 22 had autoimmune or other non-infective encephalitis. There was a moderate correlation between ONSD and ICP on admission (r = 0.769; P < 0.01) and on follow up at 2 weeks and 1 month. ONSD also showed a better predictability of ICP as compared to ODH[38].

Another prospective, observational study included 25 patients with suspected tubercular meningoencephalitis (TBME) who underwent MRI followed by LP. In this study, there was a significant difference between ONSD in TBME patients as compared to the control group (5.81 mm vs 4.37 mm, P < 0.001)[13].

Idiopathic intracranial hypertension

In a study of a patient suspected of idiopathic ICH by Bakola et al[41], ONSD significantly correlated with ICP measured using CSF opening pressure (r = 0.716, P < 0.001). For predicting idiopathic ICH, the optimal ONSD cutoff value was 5.15 mm, with an AUC of 0.914. The authors then conducted a systematic review which included 14 studies and 415 patients of idiopathic ICH. The meta-analysis of pooled patient data from 8 studies which studied relation between ONSD and CSF opening pressure found a moderate correlation between ONSD and idiopathic ICH (r = 0.44; P for Cochran Q < 0.02). The cutoff values used in each study ranged from 4.8 to 6.3 mm, while the optimal cutoff point for idiopathic ICH discrimination was detected at 5.0 mm with AUC of 0.878[41].

We found 3 studies apart from those included in the systematic review and meta-analysis. In a study of 23 patients with idiopathic ICH, no correlation was found between CSF opening pressure and ONSD[34]. Another case control study of 30 patients found a positive but moderate (Spearman's rho = 0.500) correlation between the values of ONSD and CSF opening pressure (P = 0.005). The study also found a significant decrease in ONSD post-therapeutic LP[17]. A 25-year-old patient with an idiopathic ICH case report followed real-time ONSD changes. When measured before the LP, the left and right eye ONSD were 7.2 and 6.8 mm. The opening pressure was 36 cmH₂O, and the closing pressure was 19.5 cmH₂O after removal of 19 cc of clear, colorless CSF. The post-LP ultrasound was performed after 30 minutes, showing an ONSD of 5.8 and 6.2 mm in the left and right eyes, respectively[14].

Sepsis associated encephalopathy

This is a known complication of sepsis, with the risk factors being uremia, hypoglycemia, hyperglycemia, hypercapnia, hypernatremia, elderly age and higher acute physiology and chronic health evaluation score. A prospective observational study included 123 patients with sepsis, out of which 58 developed sepsis associated encephalopathy (SAE). ONSD was measured on admission (ONSDo) and alternate days. The highest ONSD was documented during illness and was reported as ONSDmax. Both ONSDo and ONSDmax were higher in SAE group, with a cutoff of 5.4 mm and 5.8 mm, respectively, having ROC curves of 0.801 and 0.829 in the prediction of occurrence of SAE. However, the correlation with ICP was not determined[52].

A study by Yang et al[53] included 90 patients with sepsis, for whom 142 ONSD measurements were carried out. ONSD was higher in SAE group but did not correlate with patient outcomes[53]. Another prospective case series of 10 patients with septic shock requiring mechanical ventilation and sedation took multiple ONSD values during the ICU stay. A value > 5.7 mm was taken as a cut-off for raised ONSD. Forty nine out of 80 measurements (62%) were raised. There was no correlation between ONSDs and C-reactive protein, highest daily lactate or sequential organ failure assessment score during the study period[54].

ONSD as a management tool

A case report of aneurysmal SAH showed that a combination of TCD and ONSD was used as a trigger for the evaluation of vasospasm. This helped in the early ventriculoperitoneal shunt and patient management[55].

Post cardiac arrest

Neuro-prognostication of post cardiac arrest patients is imperative for forming a plan of care. However, most patients may not be stable to be shifted for CT or MRI scans. Other tests like electroencephalography, somatosensory evoked potential and pupillary reflex may be confounded by metabolic derangements and sedatives.

A retrospective study of 86 adult patients with cardiac arrest underwent radiological imaging to assess for markers of poor outcomes post cardiopulmonary resuscitation. However, no correlation was found between ONSD and poor outcomes[56]. Another study, which combined ONSD and grey-white matter differentiation on CT scan, showed better sensitivity for predicting outcomes in these patients[57].

In a prospective longitudinal cohort study, which included 100 adult patients with cardiac arrest, ONSD was measured daily for 3 consecutive days or until awakening or death, by trained personnel in the first 3 days after cardiac arrest, or until decease or awakening. Good inter observer variability was found with intraclass correlation coefficient for offline and real-time measurements (inter-observer reliability) was 0.872[58].

Dysnatremias

A prospective observational study of 54 patients with serum sodium below 135 mEq/L presenting to the emergency department was conducted to test the efficacy of ONSD as a guide to correct hyponatremia. ONSD was documented on admission and on discharge. The change in ONSD did not correlate with the variation in sodium levels[59].

On the contrary, in a prospective cohort study that included 65 patients (35 with hypernatremia and 30 with hyponatremia) and 14 healthy volunteers (control group), ONSD values were found to be higher in the hypernatremia and hyponatremia groups comparing to the control group (P < 0.001). The right and left ONSD detected hypernatremia with 91.4% and 88.6% sensitivity and 92.9% and 85.7% specificity, respectively. For hyponatremia, sensitivity was 83.3% and 93.0%, and specificity was 92.9% and 86.0% for right and left ONSD, respectively. Furthermore, in patients with hypernatremia, ONSD was found to be an independent predictor of mortality. An increase of one mm ONSD in the right eye increased the probability of mortality by 6.21, and a similar rise in ONSD in the left eye increased the likelihood of mortality by 4.21[60].

CONCLUSION

ONSD was found to have a significant association with raised ICP in most of the studies in non-traumatic, neuro-critically ill patients. In a patient with altered sensorium, it can be used as a bedside screening tool for raised ICP. This can lead to early diagnosis of a significant intracranial event. However, the cutoffs still need to be defined, and larger RCTs will be required to determine this value. Due to the heterogenicity of the patient population in a medical ICU, disease-wise studies are pertinent. The quantitative equation for predicting ICP using US-ONSD opens up promising possibilities for neuromonitoring, as ONSD could also be a handy outcome measure tool in select patient subgroups.

Firstly, as this was a scoping review, no statistical analysis was done. Secondly, we excluded all the studies in a mixed population of traumatic and non-traumatic cases to avoid trauma being the confounding factor. These consisted of 30 out of 109 studies (27.5%) reviewed, indicating a large data set left unassessed. We included all the causes of non-traumatic increased ICP to improve the generalizability of medical ICUs. The subset population with AIS, ICH, hydrocephalus, meningitis and septic encephalopathy will require relevant RCTs, followed by systematic review and meta-analysis before ONSD can be incorporated to change the clinical practices or management guidelines.