Optic nerve sheath diameter trajectories and mortality in children with clinically relevant elevated intracranial pressure

Abstract

BACKGROUND

The optic nerve sheath diameter (ONSD) measured by ultrasound has emerged as a significant noninvasive method for detecting elevated intracranial pressure (ICP), guiding timely interventions, and monitoring treatment response. Previous studies have shown that the baseline ONSD at admission is a prognostic indicator of mortality in adult patients with cerebrovascular events, traumatic brain injury, hepatic encephalopathy, and acute stroke. However, pediatric data on the dynamic changes in ONSD remain limited.

AIM

To study the association between within-48 hours admission dynamic ONSD changes and mortality in children with clinically relevant elevated ICP.

METHODS

This single-institution prospective study was performed at a tertiary Children’s Hospital in Vietnam, between November 2023 and August 2024. The primary outcome was in-hospital mortality rate. ONSD data were measured at admission, 24 hours, and 48 hours post-admission to pediatric intensive care unit (PICU). Linear mixed-effects models accounting for repeated measures within individuals were used to analyze the association between ONSD changes and in-hospital mortality.

RESULTS

A total of 69 PICU-admitted children with clinically relevant raised ICP were enrolled and included in the analysis. The median patient age was 6 years (interquartile range: 1-12), and males accounted for 54% of all patients. The in-hospital mortality rate in children with clinically relevant raised ICP was 23.2%. Traumatic brain injury, sepsis-associated encephalopathy, and septic shock were the main causes of death in this cohort. Linear mixed-effects analysis showed that dynamic variability in ONSD values upon PICU admission and during the first 48 hours later correlated significantly with increased mortality. Nonsurvivors had a 5.3% increase in the mean ONSD at 48 hours compared to baseline levels, while the survivors showed a 5.6% reduction in ONSD.

CONCLUSION

Serial ultrasound-based ONSD measurements within 48 hours of admission better predicted mortality than baseline data in critically ill children, offering a practical, noninvasive tool for early prognosis in elevated ICP.

Key Words: Glasgow coma scale; Intracranial pressure; Mortality; Optic nerve sheath diameter; Pediatric intensive care unit

Core Tip: The in-hospital mortality rate in children hospitalized with clinically relevant raised intracranial pressure was 23.2%. Nonsurvivors had a median 5.3% increase in ultrasound-measured optic nerve sheath diameter (ONSD), whereas survivors showed median 5.6% reduction during the first 48 hours of admission. Early changes in ONSD, as measured by ultrasound during the first 48 hours of admission, were strongly correlated with neurological deterioration and mortality risk.

INTRODUCTION

Elevated intracranial pressure (ICP) is a critical factor contributing to secondary brain injury, often leading to poor neurological outcomes, and is life-threatening in severe cases[1,2]. Raised ICP can arise from various conditions, such as severe traumatic brain injury (TBI), intracranial infections, hydrocephalus, brain tumors, cerebrovascular disorders, and hepatic encephalopathy[3]. Secondary brain damage typically develops within minutes to hours following the initial insult and is driven by a series of pathological processes that impair cerebral perfusion, oxygenation, and chemical metabolism. Early recognition and intervention for elevated ICP are essential for managing neurologically compromised patients in critical care settings[4]. Conventional ICP measurement methods are regarded as the gold standard for diagnosing and continuously monitoring ICP. However, these techniques are invasive procedures, carry hazardous risks such as bleeding and infection, and have limited availability in healthcare facilities. These limitations highlight the pressing need for noninvasive and cost-effective alternatives to evaluate elevated ICP. Emerging evidence in adult populations suggests that optic nerve sheath diameter (ONSD) measurement using noninvasive imaging modalities such as computed tomography (CT), magnetic resonance imaging (MRI), and bedside ultrasound may offer a viable substitute for conventional invasive ICP assessment. However, CT and MRI scans require patient preparation and transport to radiology centers, which may be laborious and impractical for critically ill individuals[5]. Nevertheless, the ultrasound-based ONSD method offers various advantages, considering that it can be performed repeatedly at the bedside and yields rapid results. This is particularly beneficial for unstable patients in the pediatric intensive care unit (PICU) who require ongoing ICP assessment[6]. Furthermore, several studies have indicated that ONSD measurements upon hospital admission may serve as a prognostic marker for predicting the risk of mortality in adult patients presenting with such neurological conditions as TBI, cerebrovascular diseases, and hepatic encephalopathy[7-10]. However, to date, there are limited data on the association between dynamic ultrasound-measured ONSD and survival outcomes in PICU-admitted pediatric patients. Hence, we hypothesized that dynamic changes in ONSD levels in children with clinical manifestations consistent with increased ICP would be associated with poor survival outcomes. Therefore, this study aimed to investigate the relationship between early dynamic ONSD changes during the first 48 h of PICU admission and mortality in PICU-admitted children with clinically relevant elevated ICP.

MATERIALS AND METHODS

Study design, setting and participants

This single-center prospective observational study was conducted on all children admitted to the PICU of Children’s Hospital 2, Vietnam between November 2023 and August 2024. The eligibility criteria were PICU-admitted children aged 1 month to 16 years who clinically presented with elevated ICP. Patients were excluded from the study if they had orbital trauma, periorbital tumor, abnormal intraocular pressure, known optic nerve atrophy, intraocular space-occupying lesions, respiratory failure with severe hypercapnia, concurrent comorbidities including critical pulmonary hypertension, pulmonary embolism, and mechanical ventilation requiring a positive end-expiratory pressure of more than 10 cm H₂O.

Study definitions

Patients with clinically relevant elevated ICP in the presence of at least three of the following clinical signs and symptoms and abnormal radiological findings on CT and/or MRI scans[3,11-13]: Presence of Cushing’s triad (hypertension, bradycardia, irregular respiration), Glasgow Coma Scale (GCS) score ≤ 8, Seizures attributable to cerebral pathology, abnormal motor postures (decorticate, decerebrate rigidity), unequal pupil sizes, papilledema. Abnormal neuroimaging findings included intracranial mass lesions, cerebral edema, midline shift, or compression of the basal cisterns.

Primary outcome and covariates

The primary study outcome was in-hospital mortality rate. The pre-specified covariates included baseline ONSD levels upon admission and ONSD measurements (at 24 hours and 48 hours post- admission), GCS scores, and vasoactive inotropic score (VIS).

Measurement of ONSD

Ocular ultrasound was used to measure ONSD and all scans were solely performed by a highly skilled certified radiologist. A Vivid 7 Dimension ultrasound system, equipped with a high-frequency linear M12 L small part probe (optimized 10–12 MHz), was used for imaging. The transducer, using the anterior axis approach, was placed transversely on the closed upper eyelid (axial plane) and oriented toward the temple to visualize the optic nerve in the sagittal plane. The optic nerve appeared as a hypoechoic tubular structure extending posteriorly from the globe and was encased in a hyperechoic sheath. ONSD was measured 3 mm posterior to the retina, from the inner edge of the optic nerve sheath, using the optic disc as a landmark[14]. Two perpendicular diameters of the optic nerve sheath were measured, 3 mm posterior to the retina-vertical diameter (D1) and horizontal diameter (D2). The mean values of D2 measurements were calculated for further analysis (Figure 1). Measurements were taken from both eyes, with at least three readings per eye, and the average of these values was used to determine the final ONSD for each time point. Ultrasound assessments were conducted at three specific intervals during the PICU stay: Upon admission (baseline), 24 hours and 48 hours post-admission. In addition, CT and/or MRI were performed during the first 24 hours of PICU admission to confirm ONSD measurement values.

Figure 1 Anatomical landmarks and ultrasonographic measurements of optic nerve sheath diameter. A: Anatomical landmarks used for optic nerve sheath diameter measurements; B: Enlarged optic nerve sheath on transorbital ultrasonography in a patient with traumatic brain injury. The patient was examined in the supine position with eyes gently closed, and a copious amount of acoustic gel was applied to the upper eyelid to prevent direct pressure on the globe. The probe was positioned lightly over the closed eyelid in the axial plane to visualize the optic nerve posterior to the globe. Two perpendicular diameters of the optic nerve sheath were measured 3 mm posterior to the retina: The vertical diameter (D1 = 3 mm) and the horizontal diameter (D2 = 5 mm). The mean values of D2 were used for further analysis.

Data collection

Patients who met the inclusion criteria were enrolled in this study. Clinical and laboratory data were collected at the time of PICU admission, 24 hours and 48 hours post-admission, entered into structured case report forms for further analysis. Patient outcomes were monitored continuously during their stay in the PICU and evaluated at the time of hospital discharge.

Statistical analysis

Continuous variables were reported as medians with interquartile ranges (IQR), whereas categorical variables were presented as frequencies (n) and percentages (%). Statistical comparisons of continuous variables were conducted using either two-tailed paired t-tests or the Mann–Whitney U test, depending on data normality. Categorical data were analyzed using the χ² test or Fisher’s exact test, as appropriate. Missing data of interest, including GCS and ONSD at 24 hours and 48 hours were less than 3% and handled by multiple imputation using chained equations. Linear mixed-effects modeling was used to analyze the dynamic changes in ONSD over time according to the survival status (survivors vs nonsurvivors). ONSD measurements obtained at baseline, 24 hours, and 48 hours post-admission were regressed against time (in hours) for each patient. The slope of the regression line represents the estimated change in ONSD over the 48-hour period. A two-tailed P-value of less than 0.05 was deemed indicative of statistical significance for all analyses. Data analysis was conducted using R software (version 4.4.2; Boston, MA, United States).

RESULTS

Baseline characteristics of study participants on PICU admission and clinical outcomes at discharge

A total of 69 patients who met the inclusion criteria were enrolled in the analysis. Baseline clinical and laboratory characteristics at the time of PICU admission are summarized in Table 1. The median age of the cohort was 6 years (IQR: 1–12 years), with males comprising 54% of the study population. TBI was the leading cause of elevated ICP. The median GCS score on PICU admission was 7 (IQR: 6-7). Seizures and pupillary abnormalities were most frequently observed in 33% and 28% of the patients, respectively. Patients underwent brain radiological scans, including CT (97%) and MRI (16%). In addition, 41 children (59%) received vasopressors, most of whom were administered noradrenaline. The median VIS was 15 (IQR: 0-30). The median Pediatric Logistic Organ Dysfunction-2 (PELOD-2) score of the participants upon PICU admission was 6 (IQR: 4-9). Notably, 67 (97%) patients were treated with hypertonic (3%) sodium chloride and 35 (51%) children received mannitol 20% for raised ICP interventions. Regarding neurosurgical treatment, 39% of the patients (27/69) underwent decompressive craniectomy to reduce the increased ICP. Additionally, four (5.8%) patients with intraventricular hemorrhage were treated with external ventricular drainage, and five (7.2%) patients with acute hydrocephalus were intervened with ventriculoperitoneal shunt placement by neurosurgeons. The median ONSD of the right eye measured on PICU admission, 24 hours and 48 hours post-admission were 5.3 (IQR: 5-5.6), 5.2 (IQR: 4.9-5.5) and 5.0 (4.7-5.3) mm, respectively. Meanwhile, the median ONSD of the left eye measured on PICU admission, 24 hours and 48 hours post-admission were 5.4 (IQR: 5-5.6), 5.3 (IQR: 4.9-5.5) and 5.2 (IQR: 4.8-5.5) mm, respectively. Significantly, non-survivors had a higher ONSD levels during the first 48 hours of PICU admission than survivors, as can been seen in the Figure 2. The median duration of mechanical ventilation was 4.2 days (IQR: 3.3-5.2 days). Overall, 16 of 69 (23.2%) patients with clinically relevant increased ICP died during the PICU stay (Table 1). The main causes of death were severe TBI, sepsis-associated encephalopathy, and septic shock with multi-organ damage.

Figure 2 Optic nerve sheath diameter (mm) variability by survival outcome in patients with clinically diagnosed elevated intracranial pressure at various time points: At the baseline, 24 hours and 48 hours post-admission. ¹P-values from Mann-Whitney tests. ONSD: Optic nerve sheath diameter.

Table 1 Baseline clinical characteristics of study participants on pediatric intensive care unit admissions and associated outcomes at discharge (n = 69), n (%)/median (25^th-75^th percentiles).

Characteristics	Statistics
Age, (years)	6 (1-12)
Male patients	37 (54)
Traumatic brain injury	27 (39)
Cerebrovascular disease	14 (20)
Cerebral infection	3 (4.4)
Septic shock	5 (7.3)
Hydrocephalus	5 (7.3)
Hepatic encephalopathy	3 (4.4)
Asphyxia due to drowning	3 (4.4)
Abnormal pupils	19 (28)
Abnormal posture	1 (1.5)
Seizure	23 (33)
Cushing triad syndrome	2 (2.9)
GCS on admission	7 (6-7)
ONSD on right eyes, (mm)
ONSD on PICU admission	5.3 (5-5.6)
ONSD at 24 hours	5.2 (4.9-5.5)
ONSD at 48 hours	5.0 (4.7-5.3)
ONSD on left eyes, (mm)
ONSD on PICU admission	5.4 (5-5.6)
ONSD at 24 hours	5.3 (4.9-5.5)
ONSD at 48 hours	5.2 (4.8-5.5)
Variability of mean ONSD at 48 hours compared to baseline levels, (%)	-4.8 (-7.4-1)
CT scan of brain performed	67 (97)
MRI of brain performed	11 (16)
Vasopressor use	41 (59)
Vasoactive inotropic score	15 (0-30)
PELOD-2 score	6 (4-9)
Treatments
Mannitol infusion	35 (51)
Hypertonic saline 3% infusion	67 (97)
Decompressive craniectomy	27 (39)
External ventricular drainage	4 (5.8)
Ventriculoperitoneal shunt	5 (7.2)
Patient outcomes at discharge
Length of mechanical ventilation, (days)	4.2 (3.3-5.2)
Fatal outcome	16 (23.2)

GCS: Glasgow Coma Scale; ONSD: Optic nerve sheath diameter; PELOD-2: Pediatric logistic organ dysfunction-2; PICU: Pediatric intensive care unit.

Associated risk factors for in-hospital mortality in children with clinical relevant elevated ICP

As shown in Table 2, there were significant differences in septic shock, abnormal pupils, GCS at three time points (on PICU admission, 24 hours, 48 hours), VIS score, VIS > 30, and PELOD-2 scores between the survivor and nonsurvivor groups. In particular, the marked differences in ONSD values between the alive and fatal patient groups were more evidently at 48 hours after admission than at the baseline and 24 h landmarks. However, no association was observed between hyperosmolar treatment, neurosurgical intervention, and patient survival outcome.

Table 2 Risk factors associated with fatal outcome in children with clinically relevant raised intracranial pressure, n (%)/median (25^th-75^th percentiles).

Characteristics	Survivors (n = 53)	Non-survivors (n = 16)	P value
Age, (years)	6 (1.4-13)	4.5 (1-9.4)	0.50
Male patients	27 (51)	10 (62)	0.42
Traumatic brain injury	24 (45)	3 (19)	0.11
Cerebral infection	1 (2)	2 (13)	0.13
Septic shock	0 (0)	5 (31)	< 0.001
Hydrocephalus	5 (9)	0 (0)	0.58
Hepatic encephalopathy	0 (0)	3 (19)	0.01
Asphyxia due to drowning	2 (4)	1 (6)	0.55
Abnormal pupils	10 (19)	9 (56)	< 0.01
Abnormal posture	1 (2)	0 (0)	0.77
Seizure	18 (34)	5 (31)	0.99
Cushing triad syndrome	1 (2)	1 (6)	0.41
Glasgow Coma Scale
GCS on PICU admission	7 (7-7)	5 (3-7)	< 0.001
GCS at 24 hours	7 (7-7)	3 (3-4)	< 0.001
GCS at 48 hours	7 (7-8)	3 (3-3)	< 0.001
ONSD on right eyes, (mm)
ONSD on PICU admission	5.2 (4.9-5.5)	5.5 (5.1-5.6)	0.35
ONSD at 24 hours	5.1 (4.8-5.3)	5.6 (5.27-5.73)	0.03
ONSD at 48 hours	5.0 (4.63-5.2)	5.74 (5.24-5.8)	< 0.001
ONSD on left eyes, (mm)
ONSD on PICU admission	5.4 (5.0-5.6)	5.4 (5.18-5.6)	0.65
ONSD at 24 hours	5.3 (4.9-5.5)	5.55 (5.19-5.8)	0.09
ONSD at 48 hours	5.1 (4.7-5.3)	5.66 (5.39-5.84)	0.003
Vasoactive inotropic score (VIS)	10 (0-20)	60 (30-90)	< 0.001
VIS > 30	4 (8)	14 (88)	< 0.001
PELOD-2 score	4 (4-6)	12 (11-15)	< 0.001
Mannitol infusion	27 (51)	8 (50)	0.99
Hypertonic saline 3% infusion	52 (98)	15 (91)	0.41
Decompressive craniectomy	21 (40)	6 (38)	0.99
Ventriculoperitoneal shunt	4 (8)	1 (6)	0.57
Length of mechanical ventilation, (days)	4.2 (3.5-5.0)	3.3 (2.9-5.9)	0.77
Length of PICU stay, (days)	4.6 (3.9-5.6)	3.3 (2.9-5.9)	0.90

GCS: Glasgow Coma Scale; ONSD: Optic nerve sheath diameter; PELOD-2: Pediatric logistic organ dysfunction-2; PICU: Pediatric intensive care unit.

Dynamic changes of ONSD values within 48 hours post-admission in children with clinically relevant raised ICP

The dynamic trajectories of ONSD measured at three time points (at admission, 24 hours, 48 hours) in alive and fatal patients are presented in Figure 3. There was a significant difference in ONSD between nonsurvivors and survivors among PICU-admitted children with clinically relevant elevated ICP. In the nonsurvivor group, ONSD continued to increase from baseline until 48 hours post-admission, whereas in the survivor group, ONSD decreased gradually.

Figure 3 Dynamic changes in optic nerve sheath diameters in the two groups of patients according to survival outcomes during the first 48 hours post-admission. There was a rising trend in optic nerve sheath diameter values among nonsurvivors, in contrast to the declining trend among survivors. ONSD: Optic nerve sheath diameter.

Linear mixed effects model for dynamic variability of ONSD associated with the survival outcome at discharge

The random- and fixed-effects models used to analyze the association between ONSD changes and survival outcomes are presented in Table 3. The fixed-effects model showed a significant correlation between ONSD variability and death adjusted for at-admission GCS scores. Nonsurvivors had significantly higher baseline ONSD values than survivors (P < 0.001). During 48 hours after admission, nonsurvivors showed a marked increase in ONSD values compared to survivors (slope estimate 0.125, P = 0.03). Notably, random effects analysis, reflecting the intra-subject variability in ONSD values, accounted for 23% of the linear mixed-effects model.

Table 3 Linear mixed effects model for dynamic changes in optic nerve sheath diameter within 48 hours of pediatric intensive care unit admission in association with the survival outcome at discharge (n = 69).

Random effect analysis1	Variance	SD
Intercept	0.084	0.289
Residual	0.047	0.217
Fixed effect analysis	Estimate	SE	95%CI	P value
Baseline ONSD levels	0.837	0.032	0.774-0.899	< 0.001
GCS at admission	-0.047	0.015	-0.076 to -0.018	< 0.01
Death (yes)	0.125	0.058	0.011-0.238	0.03

¹Random effect accounted for 23% from linear mixed effects model.

GCS: Glasgow Coma Scale; ONSD: Optic nerve sheath diameter.

Bivariate logistic analysis for association between the within-48 hours post-admission ONSD variability and in-hospital mortality

Nonsurvivors exhibited a median ONSD increase of 5.3% (95%CI: 1.3%-6.9%) 48 hours after admission compared to baseline levels; in contrast, the survivors showed a reduction of -5.6% (95%CI: -8% to -3.6%) (Table 4). This indicates that the surviving patients experienced a marked improvement in ultrasound-based ONSD measurements within 48 hours post-admission, in contrast to the fatal group. Bivariate logistic analysis showed that each 1% rise in ONSD corresponded to an increase of odds ratio of 31% (P < 0.001).

Table 4 Bivariate logistic analysis for association between the within 48 hours post-admission optic nerve sheath diameter variability and in-hospital mortality.

Group	ONSD variability (%)	OR (95%CI), P value
Survivors (n = 53)	-5.6 (-8 to -3.6)	1.31 (1.14–1.5), P < 0.001
Nonsurvivors (n = 16)	5.3 (1.3-6.9)

OR: Odds ratio; ONSD: Optic nerve sheath diameter.

Associations between dynamic variability of ONSD and risk factors for mortality

The fixed-effects model showed significant associations between dynamic ONSD changes within 48h of admission, GCS scores at discharge and PELOD-2 score (Table 5). Nevertheless, vasoactive inotrope score (> 30) did not correlate with dynamic changes in ONSD (P = 0.94). The random-effects analysis of approximately 45% indicated marked intra-subject variability in this linear mixed-effects model.

Table 5 Linear mixed effects model for dynamic variability in optic nerve sheath diameter within 48 hours of pediatric intensive care unit admission in patients with clinically relevant elevated intracranial pressure (n = 69).

Random effect1	Variance	SD
Intercept	0.048	0.219
Residual	0.047	0.216
Fixed effect	Estimate	SE	95%CI	P value
Baseline ONSD levels	0.841	0.029	0.784-0.898	< 0.001
GCS at discharge	-0.015	0.007	-0.029 to -0.001	0.03
VIS > 30	-0.004	0.059	-0.119 to 0.112	0.94
PELOD-2 score	-0.018	0.006	-0.029 to -0.006	0.002

¹Random effect accounted for approximately 45% from the linear mixed effects model.

GCS: Glasgow Coma Scale; ONSD: Optic nerve sheath diameter; PELOD-2: Pediatric logistic organ dysfunction-2; PICU: Pediatric intensive care unit; VIS: Vasoactive inotropic score.

DISCUSSION

Baseline ONSD has been reported as a potential prognostic marker for mortality in patients with TBI, cerebrovascular diseases, and hepatic encephalopathy[7-11]. Despite the growing interest in this noninvasive indicator, limited data exist regarding the prognostic role of dynamic ONSD changes among pediatric patients admitted to the PICU, particularly in relation to neurological conditions and in-hospital mortality. This study sought to address this gap by investigating the clinical significance of ONSD variability within the first 48 hours of PICU admission.

Our findings demonstrated that dynamic changes in ONSD during the initial 48 hours of admission were significantly associated with in-hospital mortality in children with clinically relevant elevated ICP. In our cohort, the overall mortality rate was 23.2%, with the leading causes of death including severe TBI, sepsis-associated encephalopathy, and septic shock. These results emphasize the importance of early and continuous monitoring of the intracranial dynamics in critically ill children. The pathophysiology of severe TBI involves two insults: Primary injury from direct trauma to the brain, and secondary injury from a cascade of biochemical, cellular, and metabolic responses to direct brain injury[15]. Intracerebral swelling may progress from 24 hours to 72 hours after brain injury. Consequently, cytotoxic and vasogenic edema cause increased ICP, which potentially compromises cerebral perfusion, aggravating ischemia, brain swelling, and herniation, eventually leading to death[15]. Similarly, sepsis-associated encephalopathy can exacerbate intracranial hypertension through blood–brain barrier dysfunction and cerebral edema[16,17]. Given these mechanisms, bedside ONSD measurement using point-of-care ultrasound (POCUS) is a valuable tool for early ICP assessment. POCUS-measured ONSD is noninvasive, rapid, and can be performed repeatedly with minimal patient disturbance, making it ideal for monitoring unstable pediatric patients[6-10]. Although conventional ICP monitoring methods are the gold standard, they are invasive, time-consuming, and carry risks of bleeding and infection, particularly in PICU-admitted children[18]. Our center has been utilizing the POCUS for assessing and monitoring fluid resuscitation and hemodynamic status, including in children with severe dengue, with positive impacts on clinical outcomes[19,20].

While prior studies in adults have shown correlations between baseline ONSD values and poor prognosis in conditions such as TBI, intracranial hemorrhage, and sepsis[6,7,10,21,22], data on children remain scarce. Sekhon et al[6] reported that each 1 mm rise in ONSD doubled mortality risk in patients with severe TBI. Additionally, Boran et al[7] showed an association between higher ONSD values and poor outcomes in cases of intracranial hemorrhage and cerebrovascular events. The ONSD was reported to have a high predictive accuracy for diagnosing sepsis-associated encephalopathy in adults (area under the curve: 0.801–0.993)[21,22]. In hepatic failure, where coagulopathy limits invasive ICP monitoring, ONSD has also proven useful[8]. Nevertheless, other studies have reported that ONSD values measured by CT and MRI are not significantly associated with neurological and survival outcomes in patients with intracerebral hemorrhage, cardiac arrest, hemicraniectomy, and sepsis-associated brain dysfunction[23-26]. Such discrepancies may be attributed to heterogeneous methodologies, diverse patient demographics, and differences in the timing of ONSD assessments[4,7,10,23-27]. This underscores the need for standardized, prospective longitudinal studies to refine the clinical utility of ONSD. Our study provides new insights by focusing on dynamic ONSD changes in the pediatric population. At the time of PICU admission, there was no significant difference in ONSD between survivors and nonsurvivors. However, significant divergence emerged at 48 hours post-admission, with nonsurvivors exhibiting a median of 5.3% increase in ONSD despite receiving osmotherapy, while survivors showed a reduction. These findings suggest that serial ONSD monitoring may be more informative than single time point assessments.

TBI was the predominant cause of death in our study. Although hyperosmolar agents such as mannitol and hypertonic saline are known to reduce ICP[28-30], their efficacy in improving neurological outcomes remains uncertain. Previously reported meta-analyses have been inconclusive, and limited by heterogeneity, poor blinding and diverse etiologies among the studies[28-30]. In our cohort, most patients were managed with hyperosmolar agents of mannitol, hypertonic saline, and/or combined therapies if patients had no indications. Nevertheless, minimal clinical improvement was observed among nonsurvivors, despite receiving hyperosmolar therapy. Possible explanations include high severity of brain injury at admission, delayed initiation of treatment, or late surgical intervention. Furthermore, the presence of recurrent profound shock in patients with septic shock may result in cerebral edema and blunted response to treatment.

Sepsis-associated encephalopathy and septic shock were the second most prominent contributors of mortality in our cohort. Sepsis-associated encephalopathy frequently results from blood-brain barrier (BBB) disruption, allowing inflammatory mediators and neurotoxins to enter the brain, leading to cerebral edema and elevated ICP[17]. Studies have reported increased ONSD in septic shock patients, including those who did not survive[31]. Although direct evidence linking ONSD to BBB integrity and mortality is limited, elevated ONSD levels may reflect severe neuroinflammation and intracranial hypertension. Persistent ONSD elevation in nonsurvivors likely signifies extensive neurological injury secondary to systemic sepsis. In particular, the sensitivity of ONSD in predicting poor clinical outcomes in children arises from its direct association with ICP, anatomical and physiological considerations unique to the pediatric population[32]. The optic nerve sheath, a continuation of the meninges, expands in response to elevated ICP, making the ONSD a valuable surrogate marker. In children, the presence of open cranial sutures and greater intracranial compliance enhances the distensibility of the optic nerve sheath, resulting in more pronounced diameter changes in the setting of raised ICP. These developmental characteristics increase the sensitivity of ONSD measurement in pediatric patients with severe neurological conditions. Hence, ONSD has been increasingly recognized as a reliable, noninvasive surrogate marker for elevated ICP in patients with sepsis and cerebral infections[33]. The prognostic value of ONSD in this context is underscored by a meta-analysis conducted by Lee et al[34], which demonstrated its utility in predicting elevated ICP in adult patients. Our findings extend this evidence to the pediatric population, supporting the wider applicability of the ONSD as a longitudinal monitoring tool for ICP elevation across various critical illness contexts.

The clinical implications of ONSD are significant, as timely recognition of increased ICP in septic patients is crucial for preventing neurological deterioration and optimizing outcomes[33]. Elevated ONSD values have been shown to correlate with the severity of cerebral infections and are associated with poor prognostic indicators[14]. Therefore, the integration of serial ONSD measurements with clinical neurological assessments, such as the GCS score, may offer a valuable, noninvasive strategy to support early recognition and intervention in cases of evolving intracranial hypertension. This therapeutic approach may enhance the early detection of intracranial hypertension, ultimately supporting prompt and targeted therapeutic intervention. In our cohort, we observed a significant difference in GCS scores between survivors and nonsurvivors during their PICU stay, along with a strong association between dynamic changes in the ONSD and GCS at hospital discharge. These findings are in line with prior research demonstrating an inverse relationship between ONSD and GCS scores, suggesting that increasing ICP is associated with worsening neurological function[35]. Notably, Badri et al[36] reported that a higher average ICP within the first 48 hours of monitoring was associated with reduced functional and neuropsychological outcomes, as well as increased mortality, with each 10 mmHg rise in ICP corresponding to a 3.12-fold increase in the risk of death. While these findings have primarily been established in adult populations, our study extends this relationship to critically ill children, highlighting the relevance of the ONSD as a dynamic neuromonitoring tool in pediatric practice. By addressing this evidence gap, our results underscore the potential utility of combining bedside ONSD monitoring with clinical scoring systems to inform timely therapeutic decisions and improve neurological outcomes in the PICU setting.

Collectively, our results suggest that ONSD variability over the first 48 hours is a more meaningful prognostic marker than the baseline measurement alone. However, our study had several limitations. As this was a single-center observational study with a relatively small sample size, the generalizability and statistical power were constrained. A high level of random effects also raises concerns regarding intersubject consistency in ultrasound measurements. Additionally, center-specific clinical protocols may influence outcomes, further limiting external validity. Despite these limitations, our study has addressed the critical gap in pediatric critical care research. Its longitudinal design offers valuable insights into temporal ICP dynamics and their clinical implications. Future multicenter trials with larger cohorts and standardized imaging protocols are warranted to validate these findings and to establish dynamic ONSD monitoring as a routine part of pediatric neurocritical care.

CONCLUSION

Dynamic changes in the ONSD within 48 hours of PICU admission are significantly associated with in-hospital mortality in children with clinically relevant elevated ICP. Unlike baseline measurements, serial ONSD assessments provide more accurate prognostic information, particularly in cases of TBI and sepsis-associated encephalopathy. This study highlights the potential of ONSD trajectory within 48 h of admission as a noninvasive, repeatable, and practical marker for early risk stratification and outcome prediction in critically ill pediatric patients. Further multicenter studies are needed to validate these results and establish standardized monitoring protocols.

ACKNOWLEDGEMENTS

We are grateful to the patients, research staffs, particularly Drs Dat HVT, Khanh ND, Tuong TTH for their support in this study.