The RAMER Reviews: Brain Point of Care Ultrasound in Young Children Receiving Computed Tomography in the Emergency Department: A Proof of Concept Study

Written By: Kyle Soldevilla, DO; Edited by: Timothy Khowong, MD

BACKGROUND

Ultrasound as a means of detecting emergent intracranial hemorrhage (ICH) in pediatric emergency departments remains largely unstudied. The widely accepted imaging modality of choice for quickly investigating signs or symptoms suspicious for brain injury is unenhanced computerized tomography (CT) of the head, however it is not without its risks and limitations. First, there is a known increased risk of malignancy from irradiating young children. Secondly, lesser resourced emergency departments may not have qualified radiologists readily available to interpret pediatric imaging; and finally, some children require sedation to undergo imaging further exposing them to the risk of sedation. On the contrary, point-of-care ultrasound (POCUS) is generally readily available to most ED physicians without delay and confers multiple advantages such as no radiation, no sedation requirement, and is easily repeatable at bedside. 

In neonatal intensive care, brain POCUS is the modality of choice to identify intracranial pathology for those same reasons. It is repeatable, easy to use, and portable–which is the most optimal for those infants whose transport to alternative imaging may prove to be unnecessarily risky and dangerous.

The purpose of this study was to examine the ability of brain POCUS to identify clinically significant traumatic brain injuries in children with head injuries and/or abnormal neurologic exams (as defined by PECARN). Additionally, this study sought to determine the sensitivity and specificity of brain POCUS in a single-center tertiary pediatric hospital emergency department for children under 15 months of age compared to head CT. 

THE STUDY/METHODS

This was a single-center prospective cohort feasibility study designed to investigate whether or not pediatric emergency medicine (PEM) faculty can adequately identify clinically significant brain injury using brain POCUS as a primary outcome. “Clinically significant” meaning traumatic brain injuries that may or may not require neurosurgical intervention, intubation, hospital admission or ultimately result in death. As a secondary outcome, the investigators also sought to determine sensitivity and specificity values for their patient population.

Overall, the participants involved were limited to the primary research institution which was a pediatric tertiary care center. These included ultrasound-trained PEM physicians and any child presenting to their ED with traumatic head injury requiring head CT. All POCUS and PEM trained were invited to a 2-hour POCUS brain workshop designed in collaboration with a NICU staff physician that was formally trained in ultrasound. Patient participants were essentially any child who presented to the ED under 15 months of age who were deemed to require head CT at the discretion of the on-shift physician. Exclusion criteria was not strictly elucidated, however adult EM physicians were naturally excluded due to the setting of this study. Patients with head trauma over the age of 15 months were also excluded but not explicitly stated. No specific exclusion criteria were listed in regards to comorbid conditions or genetic abnormalities. 

Sensitivity and specificity was calculated on the basis of positive and negative findings reported on POCUS versus findings reported formally on CT. Positive predictive value (PPV), negative predictive value (NPV), Accuracy, and Kappa coefficient were obtained similarly.

CHARACTERISTICS/RESULTS

Out of the total 21 patients enrolled in the study, 18 patients had positive CT findings–approximately 86%. Five (24%) of these patients had skull fractures noted on head CT. Another 5 (24%) had CTs positive for intracranial hemorrhage. The remainder of patients had positive CTs notable for incidental and non-emergent findings i.e. ethmoid sinus opacification, inflammatory changes, asymmetrical ventricles, and prominent extra-axial spaces. 

Of the 5 patients with positive head CTs for intracranial hemorrhage, 3 also had a positive brain POCUS. The remaining two had false negative brain POCUS where CT ultimately revealed small bleeds– one being a small occipital extra-axial hemorrhage and the other a small parietal subdural hemorrhage. 

Derived from the positive CTs, POCUS was found to have a sensitivity of 60% (95% confidence interval [CI] 15%-95%) and a specificity 94% (CI 70% - 100%). PPV and NPV were 75% (CI 19% - 99%) and 88% (CI 64% - 99%), respectively. Accuracy was approximately 86% (CI 64% - 97%). Notably, inter-rater reliability was the only statistically significant finding of 35% (p = 0.05).

DISCUSSION

This appears to be the first study of its kind within a largely under-studied knowledge base looking at the utility of POCUS in the evaluation of clinically significant traumatic brain injury. Notably, it does appear that the primary investigators were relatively aware of the large limitations of this study, commenting on their unusually high positive CT rate within this cohort at about 24% when compared to previously published rates of 3-5%. Some additional limitations worthwhile to address was that this was a study performed primarily at a pediatric tertiary care center with only PEM physicians and with the primary investigator performing a vast majority of the imaging. The unusually high rate of positive CTs and non-blinded nature of the study not only impacts the internal validity of the study but the overall external validity making it difficult to generalize this to Emergency Medicine at large. Additionally, the study was small and all scans were performed within a 2 hour time-frame. It begs the question of whether not the statistical analysis would be drastically different should the cohort size increase or even the time frame increase.

That being said, again, it is a feasibility study so its level of evidence is relatively low and limitations are expected. At the very least, it contributes to the limited body of knowledge we have on the utility of this modality in pathology that can often be frustrating to investigate.. Studies in the past have looked at results retrospectively and have found similar sensitivity and specificity value. For example, Elkhunovich et al. (2018) performed a retrospective study over 5-years worth of intracranial cranial bleeds finding ultrasound to have a sensitivity and specificity of 67% (confidence interval [CI], 50%-81%) and 99% (CI, 97%-100%) respectively. McCormick et al. (2015) found an even more impressive sensitivity and specificity rate amongst two ultrasound-trained observers (upwards of 100% sensitivity and 100% specificity for a single observer) in an earlier retrospective study, however this study was also markedly limited by size. Fundamentally, this study was the first of its kind to pursue POCUS first with prospective comparison to CT, but I ultimately would not say this is practice changing. It does, on the other hand, prompt further inquiry and desire for more robust investigations into the utility of cranial ultrasound for infants with traumatic head injury. 

References: 

Elkhunovich M, Sirody J, McCormick T, Goodarzian F, Claudius I (2018). The utility of cranial ultrasound for detection of intracranial hemorrhage in infants.  Pediatric Emergency Care 34(2): 96-101

Davenport, S., Ben Fadel , N., Davila, J., Barrowman, N., Bijelic , V., & Shefrin, A. (2023). Brain Point of Care Ultrasound in Young Children Receiving Computed Tomography in the Emergency Department: A Proof of Concept Study. POCUS Journal, 8(2), 165–169. https://doi.org/10.24908/pocus.v8i2.16435

McCormick T, Chilstrom M, Childs J, McGarry R, Seif D, Mailhot T, et al. Point-of-Care Ultrasound for the Detection of Traumatic Intracranial Hemorrhage in Infants: A Pilot Study. Pediatr Emerg Care [Internet].2015;00(00):1–3.Available from: http://www.ncbi.nlm.nih.gov/pubmed/26308609