Ophthalmic POCUS to Evaluate the Pupillary Light Reflex in Critically Ill Children

Vaishnavi Divya Nagarajan, MBBS1; Kurt M. Drury, MD MCR2; Katherine Barton, MD3; Courtney L. Kraus, MD4; Misun Hwang, MD5; Becky J. Riggs, MD1,6*

(1) Division of Pediatric Critical Care Medicine; Department of Pediatrics, Oregon Health & Science University, Portland, OR, USA

(2) Division of Pediatric Critical Care Medicine; Department of Pediatrics, University of Chicago, Chicago, IL, USA

(3) Department of Diagnostic Radiology, Oregon Health & Science University, Portland, OR, USA

(4) Wilmer Eye Institute, Johns Hopkins University; Baltimore, MD, USA

(5) Department of Radiology; Children’s Hospital of Philadelphia, University of Pennsylvania, Philadelphia, PA, USA

(6) Division of Pediatric Anesthesiology and Critical Care Medicine, Johns Hopkins University, Baltimore, MD, USA

*Corresponding Author: Dr. Becky J. Riggs (email: riggsbe@ohsu.edu)


Download article PDF – POCUS Journal 2026;11(2):132-142

DOI: https://doi.org/10.24908/pocusj.v11i02.20720

Supplementary Materials: Video S1, Videos S2


Abstract

Background: This study evaluated ophthalmic point of care ultrasound (O-POCUS) in critically ill children with difficult-to-assess pupils as a method for quantifying pupillary constriction in response to light, compared with infrared pupillometry and direct visualization. Methods: This is a multi-center prospective observational study performed within the pediatric intensive care units at two pediatric academic hospitals. All critically ill children admitted to the pediatric intensive care unit with difficult-to-assess pupils who were less than 18 years old were eligible for enrollment. Children who were wards of the state, had penetrating ophthalmic trauma, or had congenital ophthalmic disorders were excluded. After enrollment, nine sequential measurements of the pupillary light reflex (PLR) were obtained from each eye, including three measurements by direct visualization, three by infrared pupillometry, and three by O-POCUS. Results: Forty-six children, aged 7 days to 17 years old, were enrolled, with a total of 92 individual eyes evaluated. Determining PLR was successful in 40 patients (87%) by direct visualization, 32 patients (70%) by infrared pupillometry, and all 46 patients (100%) by O-POCUS. A mixed-effects model demonstrated that ultrasound and pupillometry produced higher mean estimates of pupillary constriction by about 0.18 mm each but did not differ from each other on average. Bland-Altman analysis showed small biases between methods, and modality-specific residual standard deviations revealed that ultrasound had the lowest residual variability at 0.32 mm, compared with direct visualization at 0.46 mm and pupillometry at 0.44 mm. Conclusion: O-POCUS was the only modality able to determine the PLR in all patients; therefore, O-POCUS should be considered in children where direct visualization or pupillometry is not possible.

Introduction

Assessing the pupillary light reflex (PLR) in critically ill children is an essential part of evaluating and monitoring their neurological status [1]. Changes in a child’s PLR may be one of the earliest signs of a new or evolving neurologic catastrophe [2]. Altered PLR is a strong predictor of adverse outcomes after traumatic and ischemic brain injuries [1,2]. Direct visualization is the most clinically utilized method for evaluating the PLR in most pediatric patients [1,3]. Along with direct visualization, many adult neurocritical care units utilize infrared pupillometry as an objective tool to measure, record, and trend the PLR [4,5]. Both direct visualization and infrared pupillometry measurements of the PLR require the patient’s eyelid to be opened, and the pupil to be visualized by the examiner. In patients with severe periorbital edema, facial trauma, corneal opacity, corneal burns, eyelid sutures, or anterior segment pathology, immediate direct visualization of the pupil is challenging, if not impossible. Ophthalmic point of care ultrasound (O-POCUS) is a promising tool that can measure the PLR in critically ill infants and children even when the eyelids cannot be retracted and the pupil cannot be visualized [6–10]. Separate adult studies have shown strong agreement (87%) between direct visualization and ophthalmic ultrasound, as well as strong agreement between ophthalmic ultrasound and pupillometry to evaluate the PLR [6,11]. The aim of this prospective observational study is to compare the PLR determined by O-POCUS and infrared pupillometry in critically ill children with difficult-to-assess pupils against the most clinically utilized method of direct visualization.

Methods

All children 0–17 years of age being cared for in the pediatric intensive care unit (PICU) at Johns Hopkins Children’s Hospital from October 2020 until January 2021 and in the PICU at Doernbecher Children’s Hospital from January 2022 to May 2024 with difficult-to-assess pupils were eligible for enrollment in this prospective observational cohort. Difficult-to-assess pupils were defined as pupils that could not easily be visualized and assessed for their PLR within 1 to 2 minutes of attempting to do so by the patient’s front-line provider or assigned PICU nurse. For example, difficult-to-assess pupils were often identified as the inability of the bedside provider to fully retract the eyelid of the eye, preventing full visualization of the pupil. Children who were wards of the state, had penetrating ophthalmic trauma, eyelid denudement, congenital ophthalmic disorders, or whose pupils could be easily assessed were excluded [12]. Basic de-identified demographic and clinical support data were obtained and recorded for all enrolled patients (Table 1).

Table 1. Study sample baseline characteristics. Values are presented as numbers and percentages.
Patient Characteristics<1 year old 1-4 years old5-9 years old10-14 years old15-17 years old All patients
Demographics
Age group18 (39%)5 (11%)6 (13%)9 (20%)8 (17%)46 (100%)
Female5 (11%)1 (2.2%)2 (4.4%)6 (11%)2 (4.4%)16 (35%)
Male13 (28%)4 (9%)4 (9%)3 (6.7%)6 (13%)30 (65%)
Admitting Diagnosis
ARDS1 (2%)03 (6.5%)1 (2%)3 (6.5%)8 (17%)
Cardiac arrest/HIE3 (6.5%)2 (4%)002 (4%)7 (15%)
Uncompensated Shock3 (6.5%)001 (2%)2 (4%)6 (13%)
Congenital heart disease4 (9%)2 (4%)0006 (13%)
Trauma or burn1 (2%)1 (2%)1 (2%)3 (6.5%)06 (13%)
Status epilepticus1 (2%)01 (2%)2 (4%)04 (9%)
Surgical non-cardiac1 (2%)02 (4%)2 (4%)05 (11%)
Bronchiolitis3 (6.5%)00003 (6.5%)
Diabetic ketoacidosis0001 (2%)01 (2%)
Clinical Support
Awake not sedated3 (6.5%)2 (4%)3 (6.5%)3 (6.5%)5 (11%)16 (35%)
Intubated and sedated11 (24%)3 (6.5%)3 (6.5%)5 (11%)3 (6.5%)25 (54%)
ECMO and sedated4 (9%)001 (2%)1 (2%)6 (13%)
Pupillary Light Reflex Evaluations
Cannot open eyelids*3 (6.7%)01 (2%)1 (%)05 (11%)
All modalities successful8 (17%)5 (11%)4 (9%)8 (17%)7 (16%)32 (70%)
Successful direct visualization15 (33%)5 (11%)5 (11%)8 (17%)7 (16%)40 (87%)
Successful pupillometry8 (17%)4 (9%)4 (9%)9 (20%)7 (16%)32 (70%)
Successful Ophthalmic POCUS18 (39%)6 (13%)6 (13%)9 (20%)7 (16%)46 (100%)
Cannot perform direct visualization*3 (6.7%)1 (2%)1 (2%)1 (%)06 (13%)
Cannot obtain pupillometry reading*10 (22%)1 (2%)2 (4.4%)1 (%)014 (30%)
Cannot obtain pupillometry despite full pupil exposure*7 (15%)01 (2%)008 (17%)
Mortality5 (11%)01 (2%)01 (2%)7 (15%)
ARDS, acute respiratory distress syndrome; ECMO, extracorporeal membrane oxygenation; HIE, hypoxic ischemic encephalopathy; POCUS, point of care ultrasound.
*Unsuccessful after attempting for a minimum of 5 minutes for each eye.

After enrollment, nine sequential measurements of the PLR of each eye were obtained by three distinct methods: three measurements each by direct visualization, infrared pupillometry, and bedside O-POCUS. Each of these nine measurements was taken uninterrupted in this same order to ensure a similar amount of ambient light exposure for each modality. Each method of evaluation was performed and recorded by different providers in hopes of minimizing carryover bias. Direct visualization measures of the PLR were obtained from each eye by the patient’s assigned PICU nurse. The nurse shone a flashlight into the open eye being examined for 1 to 2 seconds before removing the light source. The same standard Eveready D-powered PICU-stocked flashlights were used by the bedside nurse as the light source to obtain the direct PLR. The nurse recorded the pupillary reactivity, followed by the size of the pupil before shining the light (maximal dilation) and the size of the pupil when light was shone (maximal constriction); it was also noted if no change in the pupil occurred or if direct visualization was not successful. Direct visualization was attempted for at least 5 minutes per eye by the PICU nurse before determining if it was unsuccessful and moving on to the next PLR evaluation modality. If the PICU nurse could not visualize the pupil, additional attempts at direct visualization were made by a pediatric neurointensivist (BR) to visualize the pupil for an additional 5 minutes after pupillometer and O-POCUS evaluations were completed. Three additional measurements of the PLR were then obtained for each eye by a PICU fellow (VDN) or PICU nurse using an NPi-300 pupillometer (Neuroptics, Irvine, CA). Maximal pupil dilation, maximal pupil constriction, pupil reactivity, and whether the pupillometer was unable to obtain a reading were recorded. Pupillometer readings were obtained per the manufacturer’s instructions [4]. Pupillometer readings were attempted for at least 5 minutes per eye before determining if no measurements could be obtained and moving on to O-POCUS evaluation. If the PICU fellow (VDN) or nurse could not obtain pupillometry readings, additional attempts were made by a pediatric neurointensivist (BR) for an additional 5 minutes after all three modalities of PLR had been attempted. Finally, three O-POCUS measurements of the PLR of each eye were obtained by the same pediatric neurointensivist (BR), with over 12 years of O-POCUS experience. Again, Maximal pupil dilation, maximal pupil constriction, pupil reactivity, and whether the pupillometer was unable to obtain reactivity were recorded. O-POCUS was attempted for at least 5 minutes per eye before determining that no measurements could be obtained by ultrasound.

O-POCUS measurements were obtained using either a Butterfly iQ+ (Butterfly Network, Burlington, MA) ultrasound probe (1–10 MHz) attached to a HIPAA-compliant iPhone or an L10-22-RS ultra-high frequency (10–22 MHz) linear array probe attached to a Venue Go™ (GE HealthCare, Chicago, IL) ultrasound machine. The Venue Go TM ultrasound machine was preferentially used for all scans. The Butterfly iQ+ probe was used when the Venue GoTM ultrasound machine was not available due to updates or repairs by the vendor. Each B-mode scanning ultrasound probe was set in ophthalmic safety mode with a depth of 1.5–3 cm depending on the age of the patient. In addition to utilizing ophthalmic safety mode, the as low as reasonably achievable (ALARA) principle was followed by minimizing total ophthalmic ultrasound scan time. The ultrasound probe was coated with GenTeal eye lubricant (Novartis Pharmaceuticals Corporation, East Hanover, NJ) and gently placed in the longitudinal/sagittal plane lateral to the closed eyelid (Figure 1B). The probe beam obtained a cross-sectional image of the pupil (Figure 2). If a cross-sectional image of the pupil (Figure 2) could not be obtained, the probe was then gently placed in a transverse plane directly over the closed eyelid (Figure 1A). The probe beam would then obtain a direct image of the pupil (Figure 3). After either a direct (Figure 3) or cross-sectional image (Figure 2) of the pupil was obtained by ophthalmic POCUS, a PICU nurse shone a flashlight beam (same bedside Eveready flashlight used for direct visualization) through the closed eyelid that was not being examined by the ultrasound for 1 second. A 6-second ultrasound cine clip capturing either a direct view (Supplementary Material – Video S1) or a cross-sectional view of the pupil (Supplementary Material – Video S2) was recorded. This captured 1 to 2 seconds before the light was shone, during the 1 to 2 seconds of light shining in the contralateral eye, and about 2 to 3 seconds after the light was removed. After obtaining three cine clips for each eye, the videos were reviewed, and measurements were obtained using the ultrasound machine calipers to quantify maximal dilation (Figure 2A, Figure 3A), maximal constriction (Figure 2B, Figure 3B), and reactivity of the pupil (Supplementary Material – Video S1, Supplementary Material – Video S2).

Figure 1. A) Ultrasonographer using a high-frequency linear array probe gently placed over the closed right eyelid in the transverse plane directly over the closed eyelid, with the probe beam obtaining a direct image of the pupil. The ultrasonographer’s right hand is securely anchored on a towel next to the patient’s face to minimize pressure placed on the eye and maximize control of the probe. B) Ultrasonographer using a high-frequency linear array probe gently placed over the closed right eyelid in the longitudinal plane lateral to the closed eyelid, with the probe beam obtaining a cross-sectional image of the pupil.
Figure 2. A) Sagittal plane ultrasound image showing a cross-sectional view of the iris surrounding the central black pupil (vertical line transection) with a pupillary diameter measuring 0.37 cm (3.7 mm) before light exposure (GE HealthCare, Chicago, IL). B) Cross-sectional ultrasound image of the same eye showing the iris surrounding the central black constricted pupil (vertical line transection) after light exposure, showing a pupillary diameter of 0.15 cm (1.5 mm) (GE HealthCare, Chicago, IL).
Figure 3. A) Transverse plane ultrasound image showing a direct image of the iris surrounding the central black pupil (line transecting) with a pupillary diameter measuring 0.19 cm (1.9 mm) before light exposure (GE HealthCare, Chicago, IL). B) Ultrasound image of the same eye showing the iris surrounding the central black constricted pupil (line transection) after light exposure, showing a pupillary diameter of 0.1 cm (1 mm) (GE HealthCare, Chicago, IL).

Statistical Analysis

We used RStudio, Version 15 for all analyses and considered a p-value of <0.05 as statistically significant [13]. For each eye, we recorded three separate values of the PLR for each of the three modalities: direct visualization, pupillometry, and O-POCUS. We performed descriptive statistics and reported counts with percentages to describe the patient population and ability to detect a PLR. Changes in pupil size were summarized using descriptive statistics (mean, standard deviation (SD), median, and interquartile range (IQR)).

Agreement between left and right eyes for each test was assessed using a repeated-measures Bland-Altman framework via linear mixed-effects models for repeated paired measurements. We report the mean inter-eye bias, the between-participant SD of this difference, and the within-participant SD. To compare modalities, we fit a heteroscedastic linear mixed-effects model with modality as a fixed effect and direct visualization as the reference. An overall modality effect was assessed using the model-based ANOVA for the fixed effect of modality. Pairwise contrasts were used to estimate mean differences between modalities with 95% confidence intervals and multiplicity-adjusted p-values (Holm). Residual measurement variability by modality was summarized using modality-specific SD from the mixed model. Finally, agreement between modalities at the individual-measurement level was assessed using pairwise repeated-measures Bland-Altman analyses for each modality pair, with estimated bias and model-based 95% limit of agreement (LoA) reported.

Results

Forty-six critically ill pediatric patients with difficult-to-assess pupils were enrolled in this multi-center IRB-approved prospective observational cohort (Table 1). Participants ranged in age from 7 days to 17 years, with a mean age of 6.2 years (SD = 6.3). This study was initially approved by the IRB at a single institution; however, the COVID-19 pandemic interfered with enrollment, allowing for five patients to be enrolled between October 2020 and January 2021. The study did not enroll patients during 2021. During the pandemic, members of the research team relocated and continued this study at a second institution where 41 additional patients were enrolled. The study concluded enrollment in May 2024 when additional study team members left the institution.

Of the 46 total patients enrolled, 92 individual eyes were evaluated for their PLR. O-POCUS was successful in all 46 patients (92 eyes; 100%); direct visualization was successful in 40 patients (80 eyes; 87%); and infrared pupillometry was successful in 32 patients (64 eyes; 70%) (Table 1). All initial unsuccessful attempts at obtaining both direct visualization (PICU nurse) and pupillometry (VDN) repeated by the attending pediatric neurointensivist (BR) remained unsuccessful after the second 5-minute attempt. Direct visualization of the pupil was not successful in five patients (10 eyes) because the eyelids could not be opened sufficiently to visualize the pupil. In one additional patient (2 eyes), corneal burns obstructed the view of the anterior chamber, making direct visualization of the PLR impossible despite eyelid retraction. Of the five patients whose eyelids could not be opened enough to visualize the pupil, two were severely fluid overloaded and three had bilateral tarsorrhaphies (for treatment of exposure keratitis), which prevented eyelid retraction. Direct visualization was not possible in these five patients, nor was infrared pupillometry successful in these five patients. Additionally, pupillometry failed to give measurements despite being able to retract the eyelids enough to fully visualize the pupil in nine patients (18 eyes), including the patient who had corneal burns. Additional clinical characteristics of these 14 total patients for whom pupillometry readings were not successful are included in Table 2.

Table 2. Clinical characteristics of the 14 children for whom pupillometry was unsuccessful and the 6 children younger than 6 months for whom pupillometry was successful.
Age (months)WGA (weeks)Weight(kg)Sedated & IntubatedPupil Visualized DV Successful Pupillometry SuccessfulDeathPupils Difficult to Assess
0.2383.4YesYesYesNoNoFluid overload
0.55372.5YesYesYesNoNoFluid overload
1405.6NoYesYesNoNoAngioedema
1405.3NoYesYesNoNoFluid overload
2365YesYesYesNoNoFluid overload
3376YesYesYesNoYesFluid overload
4376YesYesYesNoYesFluid overload
6397.3YesNoNoNoYesFluid overload
6407.8YesNoNoNoYesTarsorrhaphies
8388.2YesNoNoNoNoTarsorrhaphies
413814YesYesNoNoNoCorneal burns
743710.2NoYesYesNoNoAngioedema
634027.1YesNoNoNoNoFluid overload
1243938.5YesNoNoNoNoTarsorrhaphies
0.2393.2YesYesYesYesNoFluid overload
1393YesYesYesYesNoFluid overload
1413.6YesYesYesYesYesFluid overload
2385YesYesYesYesNoFluid overload
2395YesYesYesYesNoFluid overload
4396YesYesYesYesYesFluid overload
DV, direct visualization; kg, kilogram; WGA, weeks gestational age at birth.

Forty-five of the 46 patients enrolled in this study had bilaterally reactive pupils. One 3-month-old patient who was later declared brain dead did not have any pupillary reactivity to light confirmed by both direct visualization and O-POCUS; however, the infrared pupillometer failed to give any readings, including a reading of non-reactivity. Descriptive statistics for pupil size change by modality are shown in Table 3. The distribution of change scores differed across modalities, with direct visualization showing the highest change and pupillometry showing the greatest dispersion. Scatter plots showing comparisons of changes in pupil size for each modality pair are shown in Figure 4. Inter-eye agreement differed by modality (Table 4). Mean inter-eye bias was small for all three methods, indicating minimal systematic left-right offset on average. The mean inter-eye bias for ophthalmic ultrasound was -0.06 mm, -0.17 mm for pupillometry, and -0.026 mm for direct visualization.

Table 3. Descriptive statistics reporting the number of successful pupillary light reflex measurements, average change in pupillary size, and median change in pupillary size in response to light separated by three different measuring modalities.
ModalityParticipants with PLR measurementsTotal eyes with PLR measurementsMean change (mm)SDMedian change (mm)IQR
Direct Visualization40 (87%)80 (87%)10.4810.67
Pupillometry32 (70%)64 (70%)0.830.640.560.65
Ultrasound46 (100%)92 (100%)0.880.440.80.64
IQR, interquartile range; mm, millimeter; PLR, pupillary light reflex; SD, standard deviation.
Figure 4. Pairwise scatter plots of change in pupil size across all imaging modalities. Dashed line is identity (y = x). Points shown for all repeats; missingness handled pairwise per panel.

Table 4. Inter-eye agreement and measurement noise within each pupillary response to light separated by three different measuring modalities.
ModalityMean inter-eye bias (mm)SD between participantsSD within participantsLower LoA (mm)Upper LoA (mm)Residual SD
Direct visualization-0.0260.220.37-0.870.820.46
Pupillometry-0.170.410.42-1.320.990.44
Ultrasound-0.060.270.4-1.020.890.32
LoA, limit of agreement; mm, millimeter; SD, standard deviation.

In the heteroscedastic mixed-effects model comparing modalities with direct visualization as the clinical standard reference, there was evidence of a modality effect (F = 13.39; p < 0.0001). Relative to direct visualization, ultrasound measurements were higher by 0.176 mm (95% CI 0.140 to 0.232, Holm-adjusted p < 0.0001) and pupillometry measurements were higher by 0.185 mm (95% CI 0.140 to 0.229; Holm-adjusted p < 0.0001). Ultrasound and pupillometry did not differ meaningfully on average (ultrasound was lower than pupillometry by -0.008 mm; 95% CI -0.047 to 0.030; Holm-adjusted p = 0.8295). Residual measurement variability differed by modality (Table 4). The estimated residual SD was 0.464 mm for direct visualization, 0.317 mm for ultrasound, and 0.437 mm for pupillometry, indicating the lowest residual variability for ultrasound and the highest for direct visualization. Pairwise repeated-measures Bland-Altman analyses are summarized in Table 5. Biases were centered near zero and small relative to the corresponding LoA, indicating that individual-level differences were larger than the average offset. Direct visualization compared to ultrasound showed a mean bias of -0.182 mm with 95% LoA -1.413 to 1.105 mm; direct visualization to pupillometry showed a mean bias of -0.185 mm with 95% LoA -1.555 to 1.185 mm; and ultrasound and pupillometry showed a mean bias of 0.003 mm with 95% LoA -1.037 to 1.043 mm. Bland-Altman plots displaying these differences are shown in Figure 5.

Table 5. Pairwise repeated-measures Bland-Altman agreement between pupillary light reflex measuring modalities based on a single measurement occasion.
PLR Measuring Method pairBias (mm)Lower LoA (mm)Upper LoA (mm)LoA width (mm)
Direct visualization – Ultrasound-0.182-1.4131.1052.518
Direct visualization – Pupillometry-0.185-1.5551.1852.74
Ultrasound – Pupillometry0.003-1.0371.0432.08
Differences were centered near zero relative to the corresponding LoA. LoA, limit of agreement; mm, millimeter; PLR, pupillary light reflex.
Figure 5. Bland-Altman plots showing agreement in pupillary change across all imaging modalities. Solid line = model-based bias. Dashed lines = model-based 95% limits of agreement (single measurement occasion).

Discussion

Direct visualization of pupillary constriction in response to light is the most clinically utilized method for evaluating the PLR in critically ill children. The eyelids of most critically ill children can quickly and easily be fully retracted to expose the pupil, allowing for rapid direct assessment of the PLR. However, in critically ill children who are severely overloaded with fluid, have angioedema, suffer facial trauma, have corneal burns, or whose eyelids are sutured shut (tarsorrhaphies), direct visualization might not be possible. While direct visualization is the most clinically utilized qualitative method for evaluating the PLR, recent research suggests that when evaluating the quantitative pupillary change (maximal dilation – maximal constriction) in response to light, infrared pupillometry might provide a more accurate objective measurement of pupillary change [4,5]. However, it is uncertain if the quantitative difference in PLR measurements obtained by direct visualization versus pupillometry is more clinically relevant compared to knowing qualitative reactivity of the pupil (reactive versus non-reactive) in a manner that would impact a patient’s medical care [14]. In current pediatric intensive care practice, identifying the initial degree of pupillary reactivity to light and trending this reactivity over time appears to have more clinical relevance than knowing the exact change in pupil size at a fixed time point; however, with the increased utilization of infrared pupillometry, this might change.

Similar to direct visualization, infrared pupillometry requires that the eyelids be opened and the infrared beam aligned with the pupillary space. The pupillometer cannot successfully measure the PLR in patients whose eyelids cannot be retracted to fully expose the pupil. Additionally, infrared pupillometry measurements can be difficult to obtain if the patient is frequently blinking, moving, or looking away from the pupillometer, rendering it dependent on patient cooperation in conscious patients. It has also been suggested that pupillometry readings might be difficult to obtain in premature infants due to their underdeveloped autonomic nervous system, altered myelination, and thicker central corneas with smaller corneal diameters compared to full-term infants [15–17]. It is also possible that the poorly formed pupillary border in infants less than 6 months of age might impact pupillometry readings as well. The majority of infrared pupillometry data have been reported on adult patients [3–5]. There is little pediatric data reporting on the utility and feasibility of infrared pupillometry measuring the PLR in neonates, infants, and children. Winston et al. reported no difficulties in obtaining infrared pupillometry readings in children from 2 to 21 years old [18]. However, when de Vries et al. utilized eye tracking stimulation screens with infrared pupillometry technology to capture the PLR in infants 5 to 24 months old, they were unsuccessful with several infants that were ultimately excluded from the study [19].

In our study, infrared pupillometry failed to measure the PLR in 14 children (30%), of which five patients’ eyelids could not be retracted to expose the pupil, and one where the cornea was clouded due to burns (Table 2). Seven of the remaining eight patients (15%) were term infants (≥37 weeks gestational age) with an average age of 7.8 weeks (SD = 5.6) and an average weight of 4.5 kg (SD = 1.4). It was not reported that any of these patients fought against the pupillometer scan (no head shaking or excessive blinking). Six of these eight patients were intubated and heavily sedated. All were term babies ≤ 5 months of age, suggesting that lack of cooperation, corneal thickness, or prematurity should not have contributed to these failed readings; however, it does suggest that young age might have been a contributing factor to the failure of these readings. Small pupillary size was unlikely to be the only contributing factor to these failed readings, given that pupillometry readings were successfully obtained on six other term babies ≤ 5 months of age within our study population (Table 2). Pupillometry failed in the patient who had corneal burns, which most likely blocked the infrared pupillometry beam, causing unsuccessful pupillometry readings. The remaining patient in whom pupillometry failed was a 6-year-old with a POLR2A genetic mutation causing refractory epilepsy, significant neurological delays, severe microcephaly (44 cm) and growth retardation (weight 10.2 kg). Additional future pupillometry research can perhaps more accurately determine what contributing factors lead to failed pupillometry readings in babies < 5 months of age.

Unlike both direct visualization and infrared pupillometry, it has been reported that O-POCUS can detect the PLR in adult patients through a closed eyelid, which can be performed despite eyelids being difficult or impossible to retract [20,21]. Studies have shown both feasibility and accuracy of O-POCUS to determine the PLR compared to infrared pupillometry (R 0.83, p < 0.01) in critically ill adults [22]. Additionally, Chen et al. reported the use of an automated O-POCUS system produced by Edan Instruments (EDAM AX8 ultrasound system), where the ultrasound probe has a built-in light source that turns on and off while a B-mode cine loop is recorded, with software producing a continuous diameter-time curve in addition to key PLR metrics [11]. This automated O-POCUS system was compared to infrared pupillometry in critically ill adults, showing strong agreement with an inter-observer intraclass correlation coefficient ranging from 0.94–0.999 and an intra-observer intraclass correlation coefficient of 0.89–0.985 [11]. POCUS machines are widely available, safe, and effective at evaluating a wide variety of ophthalmic pathology [23–25]. Additionally, O-POCUS is being increasingly utilized within Emergency Departments and Critical Care Units [26–30].

Both direct and consensual PLR can be obtained through O-POCUS by shining a light directly over the closed eyelid of the eye being scanned (direct) or shining the light over the closed eyelid of the eye not being scanned (consensual) [25,26]. To obtain the direct PLR, a cross-sectional ultrasound view of the pupil must be obtained so as not to block the pupil from the light source with the ultrasound probe. To ensure more consistent shining of light through the closed eyelid and ease of scanning, a consensual PLR was obtained on all patients in this study. The light reaching the pupil is more intense when directed into an open eye than when transmitted through a closed eyelid before reaching the pupil. Therefore, it is possible that the globally decreased change in pupillary size in response to light measured by O-POCUS compared to direct visualization and pupillometry could be due to less light reaching the pupil through a closed eyelid as opposed to more light reaching the pupil through an open eye. Additionally, measuring contralateral PLR instead of direct PLR in this study assumes there is no optic nerve damage. If the optic nerve is injured, the pupil might not constrict in response to light shining in the eye with the damaged nerve. However, if light is shone directly into the contralateral eye, the pupil with nerve injury will still constrict. No nerve injury was identified in this study; however, this should be considered in future uses of this technology. Despite O-POCUS showing significantly less quantitative change in pupillary size in response to light, it was able to accurately determine qualitatively that the pupils were responsive to light bilaterally in 45 patients. Additionally, O-POCUS accurately identified bilateral non-reactive pupils in one patient.

A clinically normal pupillary response to light is typically reported as “Pupils Equal, Round, and Reactive to Light and Accommodation” (PERRLA). A clinically normal PLR of the right and left eye should be equal unless there is a unilateral defect or injury to the optic nerve, unilateral neural pathway, or afferent pupillary defect. None of these injuries were identified in our cohort; therefore, the research team assumed that the PLR of each patient’s right and left eye was the same. Inter-eye analyses showed small mean left-right bias within each modality (Table 4). Based on these results, direct visualization (performed by several different PICU nurses) most consistently and reliably reported equal quantitative measurements of the PLR between the right and left eye. However, observation bias might have influenced this result, with the bedside nurses knowing that the PLR should be equal in both eyes. Each modality had similar within-participant SD, reflecting minimal differences in measurement repeatability.

In the mixed-effects model comparing PLR measuring modalities, there was a significant overall method effect (F = 13.39 with p < 0.0001), indicating systematic differences in mean estimated constriction across all modalities. Both pupillometry and ultrasound differed from direct visualization by 0.18 mm on average (Table 5), whereas ultrasound and pupillometry did not differ meaningfully from one another. However, wide limits of agreement in the Bland-Altman analyses suggest that further data are needed to determine whether these are interchangeable for individual-level decision-making that would be driven by magnitude of pupillary constriction. There were clear modality-dependent differences in measurement noise. Ultrasound demonstrated the lowest residual variability, whereas direct visualization showed the highest residual SD. While direct visualization is sensitive to subjective estimation, ultrasound benefits from standardized acquisition and clear anatomical measurements, highlighting the potential benefits of this approach. Pupillometry’s residual variability fell between the other two methods, reflecting a combination of device precision and real-world acquisition challenges such as alignment and patient movement that are not as obvious to the operator as with ultrasound. These findings suggest that O-POCUS may provide a feasible and reproducible quantitative measurement approach, but further studies using standardized acquisition protocols are needed before conclusions can be drawn regarding diagnostic accuracy or superiority over existing methods.

The largest limitation of this study is the small sample size, with only 46 total patients and 92 eyes evaluated. Three separate individuals utilized a single modality to obtain PLR measurements for each patient in hopes of minimizing carryover bias. However, for patients in whom pupillometry and/or direct visualization failed on the first attempt, the same pediatric neurointensivist performed the repeat assessments. Thus, this individual was unblinded to the results of the initial attempt. Carryover bias might have influenced that neurointensivist’s subsequent attempts to obtain the PLR through pupillometry and/or direct visualization. However, in all second attempts, the pediatric neurointensivist (BR) was unable to obtain measurements of the PLR from direct visualization or infrared pupillometry in situations where others had failed. The same critical care fellow (NDV) performed 41 infrared pupillometry readings (91%), and the same critical care attending (BR) performed all of the O-POCUS exams for every patient. However, different PICU nurses obtained direct visualization measurements for each patient and five pupillometry readings. Variation in the skill level of individual nurses obtaining direct visualization of the PLR might have influenced accuracy. Finally, this study was performed at two separate institutions where different patient demographics, environmental factors, and nursing skill levels might have influenced the outcomes.

All O-POCUS exams were performed by a pediatric neurointensivist (BR) with over 12 years of frequent ophthalmic ultrasound experience, allowing for quick and gentle acquisition of O-POCUS measurements. It is possible that a less experienced sonographer might not have successfully obtained PLR measurements in all 46 patients (92 eyes) within the same timeframe. An additional limitation of this study was that 90 of the 92 pupils evaluated were reactive to light. As a result, the ability of these three modalities to detect non-reactive pupils was not sufficiently evaluated. A final limitation was that all children were critically ill; if these modalities were attempted in more vigorous or rowdy children, the PLR detection might have been less successful for some or all modalities.

O-POCUS should not be utilized to detect the PLR in patients with pupils that are easy to assess, given that it is more time-consuming and requires greater skill compared to direct visualization or pupillometry. However, in clinical scenarios where the pupil is difficult to visualize or assess, O-POCUS should be considered to determine the presence or absence of the PLR in critically ill children.

Conclusion

In this small prospective observational study, we were able to detect pupillary reactivity to light identified by O-POCUS in more critically ill children with difficult-to-assess pupils than infrared pupillometry or direct visualization. Ophthalmic ultrasound proved the most successful tool in obtaining the PLR in our cohort. For critically ill children with difficult-to-assess pupils, O-POCUS should be considered as an additional modality to determine pupillary reactivity to light.

Ethics Statement

Ethical approval for this study was obtained from Institutional Review Boards (IRBs) at both institutions approved this study (IRB: 61747 approved March 2019; IRB:25584 approved May 2022). Parental consent was obtained prior to all PLR evaluations, and patient assent was obtained whenever possible.

Disclosure Statement

The authors do not have any conflicts of interest to disclose.

Funding

The authors did not receive any financial support for the study and have no financial relationships relevant to this article to disclose.

Author Contributions

VN: conceptualization, methodology, formal analysis, investigation, writing-original draft, writing-review & editing. KD: conceptualization, methodology, formal analysis, investigation, writing-original draft, writing-review & editing. KB: data curation, writing-review & editing. CK: data curation, writing-review & editing. MH: data curation, writing-review & editing. BR: conceptualization, methodology, formal analysis, investigation, writing-original draft, writing-review & editing. All authors approved the final manuscript.

References

1. Kim, T.J. Quantitative assessments of pupillary light reflexes in neurocritically ill patients, Journal of Neurocritical Care, 2022;15(2):79-87. doi: https://doi.org/10.18700/jnc.220085

2. Rajagopalan S, Sarwal A. Neuromonitoring in Critically Ill Patients. Crit Care Med. 2023;51(4):525-542. doi: 10.1097/CCM.0000000000005809neuromonitoring_in_critically_ill_patients.9.aspx

3. Lee H, Choi SH, Park B, Hong YH, Lee HB, Jeon SB. Quantitative assessments of pupillary light reflexes in hospital-onset unresponsiveness. BMC Neurol. 2021;21(1):234. doi: 10.1186/s12883-021-02275-9

4. Hsu CH, Kuo LT. Application of Pupillometry in Neurocritical Patients. J Pers Med. 2023;13(7):1100. doi: 10.3390/jpm13071100

5. Zafar SF, Suarez JI. Automated pupillometer for monitoring the critically ill patient: a critical appraisal. J Crit Care. 2014;29(4):599-603. doi: 10.1016/j.jcrc.2014.01.012

6. Modi P, Bhoi S, Aggarwal P, Nayer J, Sinha TP, Ekka M, Mishra PR, Kumar A, Pandey S. Feasibility of USG-Guided Objective Initial Assessment of Pupillary Size and Pupillary Reflexes Versus Clinical Examination in Patients With Altered Mental Status: A Cross-Sectional Study. J Ultrasound Med. 2024;43(2):335-346. doi: 10.1002/jum.16366

7. Fisher YL. The current status of ophthalmic B-scan ultrasonography. J Clin Ultrasound. 1975;3:219-223.

8. Németh J, Végh M, Horóczi Z, Süveges I. Examination of the orbit using a non-ophthalmologic ultrasound equipment. Orv Hetil. 1992;133:2563-2565.

9. Schmidt FA, Ruprecht K, Connolly F, Maas MB, Paul F, Hoffmann J, Harms L, Schreiber SJ. B-mode ultrasound assessment of pupillary function: feasibility, reliability and normal values. PLoS One. 2017;12(12):e0189016. doi: 10.1371/journal.pone.0189016

10. Nagarajan VD, Hunt MF, Hwang M, Kraus CL, Riggs BJ. Pupillary Light Reflex Detected by Ophthalmic Ultrasonography in Critically Ill Children With Difficult to Assess Pupils. J Ultrasound Med. 2024;43(11):2213-2216. doi: 10.1002/jum.16534

11. Chen W, Jiang X, Guo X, Lin J, Dou N, Tang M. Comparison of automated ultrasound pupillometry assessment vs. infrared pupillary in critically ill patients. Front Med (Lausanne). 2026;12:1722645. doi: 10.3389/fmed.2025.1722645

12. Roth FS, Koshy JC, Goldberg JS, Soparkar CN. Pearls of orbital trauma management. Semin Plast Surg. 2010;24(4):398-410. doi: 10.1055/s-0030-1269769

13. Team RC. R: a language and environment for statistics. R Foundation for Statistical Computing; 2017.

14. Couret D, Boumaza D, Grisotto C, Triglia T, Pellegrini L, Ocquidant P, Bruder NJ, Velly LJ. Reliability of standard pupillometry practice in neurocritical care: an observational, double-blinded study. Critical Care. 2016;20(1):99. doi: 10.1186/s13054-016-1239-z

15. Cerritelli F, Frasch MG, Antonelli MC, Viglione C, Vecchi S, Chiera M, Manzotti A. A Review on the Vagus Nerve and Autonomic Nervous System During Fetal Development: Searching for Critical Windows. Front Neurosci. 2021;15:721605. doi: 10.3389/fnins.2021.721605

16. Olivieri B, Rampakakis E, Gilbert G, Fezoua A, Wintermark P. Myelination may be impaired in neonates following birth asphyxia. Neuroimage Clin. 2021;31:102678. doi: 10.1016/j.nicl.2021.102678

17. Sehrawat P, Beri S, Garg R, Datta V, Shandil A. Central corneal thickness and corneal diameter in preterm and term newborns and preterm neonates at term. Indian J Ophthalmol. 2019;67(10):1575-1578. doi: 10.4103/ijo.IJO_1988_18

18. Winston M, Zhou A, Rand CM, Dunne EC, Warner JJ, Volpe LJ, Pigneri BA, Simon D, Bielawiec T, Gordon SC, Vitez SF, Charnay A, Joza S, Kelly K, Panicker C, Rizvydeen S, Niewijk G, Coleman C, Scher BJ, Reed DW, Hockney SM, Buniao G, Stewart T, Trojanowski L, Brogadir C, Price M, Kenny AS, Bradley A, Volpe NJ, Weese-Mayer DE. Pupillometry measures of autonomic nervous system regulation with advancing age in a healthy pediatric cohort. Clin Auton Res. 2020;30(1):43-51. doi: 10.1007/s10286-019-00639-3

19. de Vries LM, Amelynck S, Nyström P, van Esch L, Van Lierde T, Warreyn P, Roeyers H, Noens I, Naulaers G, Boets B, Steyaert J; TIARA Team. Investigating the development of the autonomic nervous system in infancy through pupillometry. J Neural Transm (Vienna). 2023;130(5):723-734. doi: 10.1007/s00702-023-02616-7

20. Schmidt FA, Ruprecht K, Connolly F, Maas MB, Paul F, Hoffmann J, Harms L, Schreiber SJ. B-mode ultrasound assessment of pupillary function: Feasibility, reliability and normal values. PLoS One. 2017;12(12):e0189016. doi: 10.1371/journal.pone.0189016

21. Sargsyan, A.E., Hamilton, D.R., Melton, S.L., Amponsah, D., Marshall, N.E., Dulchavsky, S.A. Ultrasonic evaluation of pupillary light reflex. Crit Ultrasound J; 2009;1:53–57. doi: https://doi.org/10.1007/s13089-009-0012-9

22. Farina F, Vosko MR, Baracchini C, Ermani M, Sommer P, Greisenegger S, Laubichler P,  Struhal W, Kellermair L, Ransmayr G, Brunner C. Ultrasound Examination of the Pupil – A New Tool for the Neuro-Ophthalmological Assessment. Ultraschall Med. 2021;42(1):84-91. doi: 10.1055/a-1208-1482

23. Riggs BJ, Trimboli-Heidler C, Spaeder MC, Miller MM, Dean NP, Cohen JS.. The Use of Ophthalmic Ultrasonography to Identify Retinal Injuries Associated with Abusive Head Trauma. Ann Emerg Med.2016; 67(5):620-4.

24. Barnett SB, Ter Haar GR, Ziskin MC, Rott HD, Duck FA, Maeda K. International recommendations for the use of diagnostic ultrasound in medicine. Ultrasound Med Biol. 2000;26(3):355-66. doi: 10.1016/s0301-5629(00)00204-0

25. Harries A, Shah S, Teismann N, Price D, Nagdev A. Ultrasound assessment of extraocular movements and pupillary light reflex in ocular trauma. Am J Emerg Med. 2010;28(8):956-9. doi: 10.1016/j.ajem.2009.06.026

26. Pierre VA, Smith T, Salerno A. Ocular Ultrasound. Emerg Med Clin North Am. 2024;42(4):891-903. doi: 10.1016/j.emc.2024.05.009

27. Skidmore C, Saurey T, Ferre RM, Rodriguez-Brizuela R, Spaulding J, Lundgreen Mason N. A Narrative Review of Common Uses of Ophthalmic Ultrasound in Emergency Medicine. J Emerg Med. 2021;60(1):80-89. doi: 10.1016/j.jemermed.2020.08.003

28. Abbinante G, Vitiello L, Coppola A, Salerno G, Gagliardi V, Pellegrino A. Optic Nerve Ultrasound Evaluation in Children: A Review. Diagnostics (Basel). 2023;13(3):535. doi: 10.3390/diagnostics13030535

29. Riggs BJ, Cohen JS, Shivakumar B, Trimboli-Heidler C, Patregnani JT, Miller MM, Spaeder MC, Dean NP.  Doppler Ultrasonography of the Central Retinal Vessels in Children with Brain Death. Pediatr Crit Care Med. 2017;18(3):258-264. doi: 10.1097/PCC.0000000000001087

30. Riggs BJ, Hwang M, Cohen JS, Barton K, Thimm MA, Hunt MF, Trimboli-Heidler C, Valvano TJ, Kraus CL, Dean NP, Tekes A. Ophthalmic ultrasonography can identify retinal injury associated with abusive head trauma more quickly and accurately than other neuroimaging modalities. Child Abuse Negl. 2025;165:107474. doi: 10.1016/j.chiabu.2025.107474

Posted in .

Leave a Reply

Your email address will not be published. Required fields are marked *