Validating the Brain Injury Guidelines in a Pediatric Population with Mild Traumatic Brain Injury and Intracranial Injury at a Level I Trauma Center

Abstract

Children with mild traumatic brain injury (mTBI) and intracranial injury (ICI) often receive unnecessary imaging and hospital admission, leading to avoidable burdens on patients and health systems. While most of these patients do not develop critical neurological injuries, identifying those at risk would allow for a more optimal determination of the appropriate level of initial emergency care. The Brain Injury Guidelines (BIG) were developed as a triage tool to identify adult patients with mTBI and ICI who can benefit from repeat imaging, hospital admission, or neurosurgical consultation. Here, we sought to validate BIG in children at a Level I trauma center and determine if the BIG algorithm can accurately identify which patients with mTBI/ICI have critical neurosurgical injuries. We hypothesize that the BIG can identify critical neurological injuries more accurately than the Glasgow Coma Scale (GCS) alone and that more severe injury according to BIG is associated with worse patient outcome. We retrospectively reviewed TBI admissions at a single center (2017–2023) using an institutional registry. Patients included (0–17 years) had an initial head computerized tomography scan with ICI and a GCS of 14–15. Patients were retrospectively classified into the BIG categories (BIG 1, 2, or 3). Medical records were reviewed to identify clinically important TBI (ciTBI): death, neurological deterioration, neurosurgical intervention, intubation >24 h, or hospital admission >48 h due to TBI. Repeat imaging studies obtained were evaluated for progression of injury. The incidence of clinically important TBI (ciTBI) and imaging progression were recorded and compared across BIG categories. Outcomes were evaluated using the Glasgow Outcome Score Extended (GOS-E) 6 months after injury. Univariable and chi-square tests were used to analyze comparisons. Overall, 804 subjects were included in the analysis of which 551 (68.5%) were transfers. Overall, 175 (21.8%) patients had a BIG 1, 402 (50.0%) a BIG 2, and 227 (28.2%) a BIG 3 injury. CiTBI occurred among 64 (8.0%) patients overall, and in 1 (0.6%), 4 (1.0%), and 59 (26.0%) of the BIG 1, 2, and 3 injuries (p < 0.0001). Progression on repeat imaging associated with neurological decline, neurosurgical intervention or resulting in additional evaluation was noted in 0 (0%), 2 (0.5%), and 41 (18.0%) of the BIG 1, 2, and 3 injuries (p < 0.001). Amongst 471 patients (58.6%) with available 6-month patient outcomes, 98% had a GOS-E ≥5 and no outcome difference between BIG categories was observed. Risk stratification of mild TBI using BIG allowed for reasonable identification of children who subsequently develop ciTBI, suggesting that BIG classification can aid in triage and management of patients who might benefit from neurosurgical consultation, repeat imaging, and potentially transfer to a dedicated trauma center. More severe injury according to BIG was not associated with a worse patient outcome.

Introduction

In the United States, traumatic brain injury (TBI) is a leading cause of death and disability in children and results in ∼600,000 emergency department (ED) visits and ∼60,000 hospitalizations each year. Of patients evaluated in EDs with TBI, 75% of cases are classified as mild, with a Glasgow Coma Scale (GCS) of 14 or 15.¹ The high frequency of these mild injuries leads to great burden on the health care system with fiscal estimates of over 1.3 billion USD annually.² Identifying those at high risk of needing hospitalization or neurosurgical intervention is often subjective. In the United States, it is common practice to request neurosurgical consultation for children with mild TBI and evidence of intracranial injury (ICI) or skull fracture on the initial computerized tomography (CT) scan. Many of these children are initially treated at hospitals without neurosurgeons and are thus transferred to regional trauma centers, even though many of these children may not need neurosurgical consultation, neurosurgical intervention, or admission to the hospital. This approach can put significant strain on families, transport teams, providers, and regional medical centers as it may result in unnecessary and costly services including medical transport, repeat imaging, hospital admissions, and specialty consultation.³

To improve the triage of these patients, recent studies have demonstrated that a more measured approach may be possible in the management of some patients with mild TBI and ICI.^4,5 Originally published in 2014, the Brain Injury Guidelines (BIG) outlined a set of criteria to help triage adult patients who can be safely managed by acute care physicians without neurosurgical consultation, hospital admission at a trauma center, or repeat imaging to assess for progression of their injury.⁶ BIG classifies patients with evidence of ICI or skull fracture on their admission CT scan into three injury groups (mild, moderate, severe [described in detail in the Methods section]) utilizing clinical and CT scan findings. Examining 2432 patients of 16 years and older in a multicenter observational prospective study, Joseph et al. demonstrated that applying the BIG criteria can identify patients who are at low risk of experiencing neurological deterioration or requiring neurosurgical intervention and that these events are exceedingly rare in the BIG 1 and 2 injury groups. Based upon these findings, the authors concluded that routine neurosurgical consultation and repeat imaging could be omitted for BIG 1 and 2 patients, optimizing triage and utilization of these resources.

Three, mostly retrospective, studies have examined BIG in pediatric patients (age 0–21), but they are limited to a small number of patients, 502 in total.^7
–9 Two studies exclusively examined the BIG 1 injury population and similarly found that BIG 1 type injury is unlikely to be associated with neurological deterioration or need for neurosurgical intervention. The largest pediatric study reported to date, a retrospective single-institution series, examined 314 subjects across all BIG injury categories and is the only pediatric study to include patients with a BIG 2 injury. Similar to the reported findings in the adult studies of BIG, these authors found very low neurosurgery intervention rates in both the BIG 1 and 2 injury groups and concluded that implementation of the BIG algorithm could reduce utilization of neurosurgical consultation and repeat imaging.

Over many years, we have also observed that children with mild TBI and minimal ICI are often transferred to our trauma center from long distances or hospitalized while few of them require neurosurgical intervention. Recognizing the potential impact of improved triage of these patients, and the limited data available to guide management of this group of patients, we sought to further validate the BIG classification among a larger cohort of pediatric patients admitted to our center and determine if applying the BIG algorithm could improve our triage. Unlike prior BIG studies, which have included patients with severe TBI, we elected to omit patients with lower GCS scores from our study. This approach was chosen to allow us to specifically evaluate the ability of the BIG algorithm to detect critical injuries in a population of children with mild TBI (GCS14-15) and ICI, that is, the population of children most likely eligible for safe redirection of care.

We hypothesize that applying the BIG classification can identify injuries requiring neurosurgical intervention and repeat imaging more accurately than the GCS alone and thus assist in optimizing patient triage and resource utilization of patients with mild TBI. To further our understanding of the BIG classification as it relates to defining injury severity and long-term patient outcome, which has not been explored in previous work, we chose to assess patient outcome, assessed by the Glasgow Outcome Score extended (GOS-E) 6 months after injury as a secondary end-point, anticipating that a more severe injury according to BIG is associated with a greater likelihood of having a worse outcome.

Methods

Study design

This was a single institution retrospective study designed to evaluate the BIG categories on a large pediatric population receiving care at a regional Level I pediatric trauma center. Approval for this study was obtained from the institutional review board (IRB study #2032348).

Patient population

All patient data were collected from our institutional TBI registry between January 2017 and August 2023. Inclusion criteria included patients 17 years and younger at the time of injury, a positive head CT (i.e., any intracranial finding and/or skull fracture), and a postresuscitation GCS score of 14 or higher in accordance with the original classification of mild TBI.¹⁰ We included patients admitted to our hospital, including those who first presented to our ED after injury and those who were transferred to our ED from another hospital. We did not limit inclusion based upon the reason for transfer (i.e., if a transfer was related to a head injury, another systemic injury, or both) but used the abbreviated injury scale (AIS) to assess for the presence of a multisystem injury that may have prompted the patient transfer.¹¹ We defined nonisolated injuries as those with an AIS ≥3 and subsequently used the injury severity score (ISS) to identify patients with nonisolated head injury (polytrauma), defined as an ISS of ≥16.¹² Patients with nonaccidental trauma, a penetrating head injury, known bleeding disorders, a delayed presentation (>24 h after injury, transferred or direct from scene), and those with a normal head CT upon attending neurosurgeon review at the time of consultation were excluded from our analysis.

Patient management

Patients were not managed according to BIG during the timeframe of the study dates. Our regional Level I pediatric trauma center has long had a policy of accepting any patient with mild TBI and ICI from referring hospitals upon request. From 2018 to 2020, any patient with mild TBI and ICI was routinely admitted to the intensive care unit (ICU). Patients >1 year of age with skull fractures were also routinely admitted to the ICU. Following implementation of new institutional pediatric ICU admission guidelines in 2020,¹³ ICU admission has been more selective and has been limited to patients with the following injuries: GCS of 13 or lower, epidural hematoma (EDH), depressed skull fracture, midline shift >5 mm, intraventricular hemorrhage (IVH), cisterns compressed or absent, intraparenchymal, or subarachnoid hemorrhage (SAH) with mass effect on the brain.

During the study dates, the decision to obtain repeat imaging for patients with TBI was made by the neurosurgery consultant and BIG were not explicitly used to support decision making. Any extra-axial hematoma with noted mass effect on the brain, regardless of size, was evaluated with a repeat imaging study. Patients with intracerebral hemorrhages larger than 1 cm and patients with IVH in two or more of the ventricular compartments also typically underwent a repeat imaging study. For isolated traumatic SAH affecting one to two sulci, routine repeat imaging was not obtained. From 2018 to 2019, repeat imaging studies were typically done using CT scan; after 2019, repeat imaging studies were typically done using FAST T2 magnetic resonance imaging or a low dose head CT if the patient required an anesthetic for the repeat imaging study. Vascular imaging studies were obtained for displaced fractures crossing a cerebral venous sinus or skull base fractures close to a major intracranial artery. In the final interpretation of the imaging, differences in techniques were considered and consensus between the neurosurgery consultant and attending neuroradiologist was sought to determine the definitive interpretation.

Data collection

Among patients meeting inclusion criteria, abstracted variables from the registry included age at injury, sex, mechanism of injury, comorbidities, medication use, postresuscitation GCS, pupillary exam, cranial CT findings, neurological exam, routine laboratory parameters (glucose, INR/PTT, hemoglobin) admission to ICU, and hospital length of stay. A toxicology screening was only performed routinely in patients older than 10 years of age. The AIS was used to determine if polytrauma explained the reason for hospital transfer or ICU admission. At the time of ED or hospital discharge, we also abstracted whether patients went home or if they required additional rehabilitation or supervision in a skilled care facility.

Cranial CT findings that were interpreted by the on-call neurosurgeon and radiologist during admission were recorded. In the event of conflicting radiology readings, the neurosurgery assessment was prioritized. For transfers from referring hospitals, the initial cranial CT findings from the referring hospital were entered in our database after formal secondary review of the CT scan at our institution.

Patients were not prospectively managed or assigned according to the BIG classification but retrospectively classified into BIG 1, 2, and 3 injuries based on the clinical and imaging variables from the registry and electronic medical record.⁶ The BIG classification was applied as follows:

The criterion for antiplatelet or anticoagulation medications was omitted from the BIG classification for this study as it is not relevant for the pediatric population.⁷ Patients categorized as BIG 1 had a normal neurological examination finding, minuscule findings on initial head CT scan (any ICH ≤4 mm and no skull fracture). The BIG 2 category included patients with a nondisplaced skull fracture, a localized ICH of 7 mm or less, and a normal neurological exam. Patients could be intoxicated. Patients categorized as BIG 3 could have intoxication, an abnormal neurological exam that included a motor or sensory deficit, but no GCS score below 14. CT scan findings included displaced skull fracture and diffuse or focal ICH ≥8 mm. Patients in the BIG 1 and 2 categories had to meet all clinical and imaging criteria to be included in the category. We did not upgrade patients to a higher injury category if they had two different injuries, that is, a patient with both a skull fracture and a hemorrhage <7 mm was classified as a BIG 2 injury.

In patients younger than 10 years of age, in whom a toxicology screening was not obtained, their screening was recorded as negative for the purposes of BIG classification. In any patient who underwent repeat imaging with CT or MRI, we explored the reasons why the study was obtained, including interval change in the hemorrhage and vascular injury screening. We also reviewed whether repeat imaging was associated with worsening clinical status or led to surgical intervention, ICU admission, or endotracheal intubation. If repeat imaging was obtained for evaluation of vascular imaging only, it was documented and analyzed separately. If the scan was obtained to assess for hematoma expansion, it was classified as follows: 0 = no change, 1 = change within the same BIG category, 2 = change to next BIG category (e.g., BIG 1 to BIG 2 or BIG 3, BIG 2 to BIG 3, progression within BIG 3). If the hematoma was smaller on repeat imaging, it was classified as no change.

We used the previously published and well-validated Pediatric Emergency Care Research Network (PECARN) criteria to define clinically important TBI (ciTBI)¹⁴ as ICI resulting in death, neurosurgical intervention, intubation for >24 h, or hospital admission for at least two nights due to head-related injury. The GOS-E was used to assess the outcome 6 months following injury.¹⁵ The GOS-E was abstracted from the institutional TBI registry and collected by personnel specifically trained to obtain outcome data utilizing a structured telephone interview prior to this retrospective study. The study personnel entering outcome data into the registry were trained with standardized patient scenarios to achieve consistency in the outcome assessment, but interobserver agreement was not formally evaluated. Informed consent to perform the outcome assessment was not specifically obtained. Per our institutional policy, informed consent was not deemed necessary for obtaining and including GOS-E outcome data to the registry.

Statistical analyses

Univariable statistics, including frequency, arithmetic mean, and standard deviation, were calculated for continuous variables. Frequency and percentage were calculated for categorical variables. Chi-square analyses were used to compare the observed values to the expected values for outcomes among different BIG categories, including the frequency of ciTBI and the frequency of injury progression on repeat imaging. No alternative methods were used to adjust for multiple comparisons. All statistical analyses were conducted using Stata Statistical Software, version 15.

Results

Between January 2017 and August 2023, 1259 pediatric patients were evaluated in the ED for TBI with a positive CT scan. Of this group, 804 patients met inclusion criteria (a postresuscitation GCS of 14–15 and were admitted to the hospital) and remained after applying exclusion criteria. Among patients included in the analyses, most (551, 68.5%) were transferred to our institution’s ED from another hospital’s ED prior to hospital admission. Additional baseline patient characteristics and BIG classifications are enumerated in Table 1. The CT documented injuries of patients in the different BIG categories are outlined in Table 2.

Table 1.

Baseline Patient Characteristics

BIG category	1	2	3	All categories
Patient count, n (% of category)	175 (21.8)	402 (50.0)	227 (28.2)	804
Age at injury in years
Median	7	6	7	6
Interquartile range	8	7	8	8
Range	1–19	1–23	1–17	1–23
Sex, n (% of category)
Female	73 (41.7)	148 (36.8)	93 (41.0)	314 (39.0)
Male	102 (58.3)	254 (63.2)	134 (59.0)	490 (61.0)
Race, n (% of category)
African American	18 (10.3)	32 (8.0)	13 (5.7)	63 (7.8)
Asian	12 (6.9)	33 (8.2)	16 (7.1)	61 (7.6)
White	99 (56.6)	234 (58.2)	129 (56.8)	462 (57.5)
Other	46 (26.3)	103 (25.6)	69 (30.4)	218 (27.1)
Ethnicity, n (% of category)
Hispanic or Latino	63 (36.0)	106 (26.4)	63 (27.7)	232 (28.8)
Not Hispanic or Latino	110 (62.9)	292 (72.6)	160 (70.5)	562 (70.0)
Other	2 (1.1)	4 (1.0)	4 (1.8)	10 (1.2)
GCS, n (% of category)
Score 14	18 (10.3)	35 (8.7)	44 (19.4)	97 (12.1)
Score 15	157 (89.7)	367 (91.3)	183 (80.6)	707 (87.9)
Mechanism of Injury, n (% of category)
Assault	2 (1.2)	5 (1.2)	3 (1.3)	10 (1.3)
Auto vs. Pedestrian	6 (3.4)	24 (6.0)	7 (3.1)	37 (4.6)
Fall	96 (54.9)	289 (71.9)	132 (58.1)	517 (64.3)
MVA	35 (20.0)	38 (9.5)	35 (15.4)	108 (13.4)
Object to the head	6 (3.4)	3 (0.7)	11 (4.9)	20 (2.5)
Recreational activity^a	21 (12.0)	25 (6.2)	17 (7.5)	63 (7.8)
Other/unknown	9 (5.1)	18 (4.5)	22 (9.7)	49 (6.1)

Recreational activity includes all-terrain vehicle, bicycle, horse, and sport.

BIG, Brain Injury Guidelines; GCS, Glasgow Coma Scale; MVA, motor vehicle accident.

Table 2.

Distribution of Initial Computerized Tomography Findings Across Brain Injury Guidelines Groups

	BIG 1 (n = 175)	BIG 2 (n = 402)	BIG 3 (n = 227)
Skull fracture (n, %)	0 (0.0)	251 (62.4)	31 (13.7)
EDH (n, %)	9 (5.1)	28 (7.0)	68 (30.0)
SDH (n, %)	74 (42.3)	95 (23.6)	73 (32.1)
IPH (n, %)	5 (2.9)	6 (1.5)	5 (2.2)
IVH (n, %)	0 (0.0)	0 (0.0)	0 (0.0)
SAH (n, %)	57 (32.6)	50 (12.4)	50 (22.0)

BIG, Brain Injury Guidelines; EDH, epidural hematoma; IPH, intraparenchymal hemorrhage; IVH, intraventricular hemorrhage; SAH, subarachnoid hemorrhage; SDH, subdural hematoma.

Clinically important traumatic brain injury

In the final cohort of 804 patients, 64 (8.0%) had ciTBI. Following categorization using the BIG criteria, ciTBI most often occurred in the BIG 3 category (59 of 227 patients, 26.0%) and was rarely observed among those with BIG 2 category injuries (4 of 402, 1.0%) (p < 0.0001). The negative predictive value of experiencing a ciTBI among patients with a BIG 1 category was 99.4%, with only 1 patient of 175 (0.6%) experiencing a ciTBI related to a hospital stay >48 h for TBI symptoms.

Characteristics of BIG 1 patients

Of the 804 patients, 175 patients (21.8%) had a BIG 1 type injury (Table 3). A total of 112 patients (64.0%) were transferred from an outside hospital ED and 32 patients (18.2%) had an AIS >16 indicating polytrauma. Forty-three patients (24.6%) were admitted to the ICU, 12 (27.9%) of those with polytrauma. No patients categorized with a BIG 1 injury underwent neurosurgical intervention. Twelve patients (6.9%) underwent repeat imaging to assess for hematoma stability at the discretion of the neurosurgery service. Progression on imaging was noted in two patients (1.1%). Expansion was not associated with clinical symptoms and further evaluation with imaging was not deemed necessary. Hematoma progression on repeat imaging was found in one patient (0.6% of BIG 1). Notably, there were no clinical symptoms or need for subsequent neurosurgical intervention.

Table 3.

Clinical Course and Repeat Imaging According to Brain Injury Guidelines Categories

	BIG 1 (n = 175)	BIG 2 (n = 402)	BIG 3 (n = 227)
Presentation to primary ED (%)	63 (36.0)	111 (27.6)	79 (34.8)
Transfer from referring ED (%)	112 (64.0)	291 (72.4)	148 (65.2)
Hospital course
LOS >48 h due to TBI, n (%)	1 (0.6)	3 (0.8)	15 (7.1)
Admitted to ICU, n (%)	43 (24.6)	122 (30.3)	167 (73.6)
Intubation >24 h, n (%)	0 (0.0)	0 (0.0)	0 (0.0)
Neurosurgical intervention, n (%)	0 (0.0)	2 (0.5)	35 (15.5)
Patients with ciTBI, n (%)	1 (0.6)	4 (1.0)	59 (26.0)
Repeat imaging
Patients with repeat imaging, n (%)	12 (6.9)	114 (28.4)	146 (64.3)
Patients with stable repeat imaging, n (%)	11 (6.3)	98 (24.4)	114 (50.2)
Patients with interval change on repeat imaging, n (%)^a	0 (0.0)	13 (3.2)	N/A
Patients with progression on imaging, n (%)^b	1 (0.6)	3 (0.7)	32 (14.1)
Patients with progression on imaging and neurosurgical intervention, n (%)	0 (0.0)	2 (0.5)	41 (18.0)

Repeat imaging studies that demonstrate interval change but not sufficient to advance to a higher BIG category.

Repeat imaging studies that demonstrate progression to higher BIG category or progression within BIG 3.

BIG, Brain Injury Guidelines; ciTBI, clinically important TBI; ED, emergency department; ICU, intensive care unit; LOS, length of stay; TBI, traumatic brain injury.

Characteristics of BIG 2 patients

There were 402 patients (50.0%) with an injury resulting in a BIG 2 categorization (Table 3). Two patients (0.5%) in this cohort underwent neurosurgical intervention, one for a lumbar drain for a cerebrospinal fluid leak and another for craniotomy to evacuate a hematoma. Repeat imaging was obtained in 114 (28.4%) of patients with BIG 2 injuries. CT angiography, CT venography, magnetic resonance angiography, and/or magnetic resonance venography were performed in 21 cases (5.2%) to assess for vascular injury (i.e., sinus injury, thrombosis, or arterial injury). No vascular injury was found and no subsequent intervention or medical treatment other than maintaining adequate hydration was instituted. Three patients (0.7%) demonstrated hematoma progression to BIG 3 upon repeat imaging, one of which required surgical intervention. Two expanding hematomas were epidural, but in only one patient (0.6%), this was clinically notable and led to neurosurgical intervention (i.e., craniotomy). Of the 28 patients with BIG 2 EDHs, one patient required surgical intervention.

Characteristics of BIG 3 patients

There were 227 patients (28.2%) with injuries classified in the BIG 3 category (Table 3). Neurosurgical intervention was required for 41 (18.0%) of these patients with procedures including craniotomy, hematoma evacuation, and elevation of depressed skull fractures. Fifteen (42%) craniotomies were performed for evacuation of EDHs. Sixty-seven patients were noted to have an EDH and 15 of the 67 (22%) patients underwent craniotomy for EDH evacuation. Repeat imaging was performed in 146 patients (64.3%) and additional imaging to assess vascular and sinus lesions were completed in 29 BIG 3 patients (12.8%).

Hospital discharge and clinical outcomes

Seven hundred and seventy-five (96.3%) patients were discharged home from the ED or hospital. Twenty-nine patients (3.6%) were discharged to the care of a foster home or relative. Six-month clinical outcomes were collected in 470 (58.5%) patients, limited to those who could be reached and participate in the GOS-E structured interview (Table 4). Therefore, 334 (41.5%) patients were lost to follow-up. The loss to follow-up was distributed proportionally across groups, with no apparent pattern of missingness. Of the patients in whom the GOS-E was collected, 99.1% of 107 patients with a BIG 1 injury, 98.7% of 224 patients with a BIG 2 injury, and 98.6% of 139 patients with a BIG 3 injury had a favorable outcome, defined as a GOS-E score of lower moderate disability to upper good recovery (GOS-E 5–8, Table 4). With the majority of patients experiencing a good recovery, we further analyzed the outcome by assigning the patients to upper good recovery, lower good recovery, and below good recovery categories (Table 4). We did not detect differences in outcome between injury groups (p = 0.28). Thirty-four patients overall (7.2%) had a less than good recovery. Six patients (1.3%) had severe disability, and the medical records of these patients were explored separately to better understand this outcome. In all six patients, this score was assigned because they required full assistance with grooming and dressing or continuous supervision. In two patients, this was found to be related to a pre-existing behavioral condition and in one patient due to an extremity fracture.

Table 4.

Discharge Status and Outcome According to Brain Injury Guidelines Categories

	BIG 1 (n = 175)	BIG 2 (n = 402)	BIG 3 (n = 227)
Discharge status
Home	172	392	211
Other^a	3	10	16
GOS-E @ 6 months (n, %)
UGR (n, %)	80 (45.7)	189 (47.0)	116 (51.1)
LGR (n, %)	17 (9.7)	21 (5.2)	13 (5.7)
UMD (n, %)	6 (3.4)	9 (2.2)	8 (3.5)
LMD (n, %)	3 (1.7)	2 (0.5)	0 (0.0)
USD (n, %)	1 (0.6)	2 (0.5)	2 (0.9)
LSD (n, %)	0 (0.0)	1 (0.2)	0 (0.0)
V (n, %)	0 (0.0)	0 (0.0)	0 (0.0)
D (n, %)	0 (0.0)	0 (0.0)	0 (0.0)
Unknown (n, %)	68 (38.9)	178 (44.3)	88 (38.8)

Other includes foster care and nonprimary guardian homes (e.g., relative).

BIG, Brain Injury Guidelines; D, dead; GOS-E, Glasgow Outcome Score Extended; LGR, lower good recovery; LMD, lower moderate disability; LSD, lower severe disability; UGR, upper good recovery; UMD, upper moderate disability; USD, upper severe disability; V, vegetative state.

Resource utilization

Implementing the BIG protocol could have potentially prevented the transfer for neurosurgical consultation of 112 patients with BIG 1 injuries, provided all of their transfers were initiated for TBI, which was not definitively determined by this study. To account for transfers that may have been initiated for polytrauma, 23 patients with both a BIG 1 injury and an AIS >16 were identified that could be excluded. This more conservative estimate of interhospital transfers that could have been avoided would be 89 of 551 (16.1%), affecting ∼15 children per year. Of the 43 children with a BIG 1 injury admitted to the ICU, 12 (27.9%) had an AIS >16. Therefore, we estimated that 31 children (17.7%) were admitted to the ICU for an isolated mild TBI, affecting 5 children with a BIG 1 TBI annually. Updating the imaging protocol to limit routine repeat imaging to patients with BIG 3 injuries could reduce utilization by 49% (126 of 262), affecting up to 21 children each year.

Discussion

In this study, we retrospectively examined the BIG classification in 804 pediatric patients with mild TBI and ICI to assess its ability to detect critical injury in this patient population. We found a low incidence (<1%) of ciTBI in patients with BIG 1/2 injuries and a significantly higher incidence in patients with a BIG 3 type injury (26%). Compared with the entire population studied, in which an overall incidence of ciTBI of 8% was noted, application of BIG further stratified and more definitively identified the patients at risk of developing neurological complications or ciTBI (i.e., they mostly are patients with a BIG 3 injury). More severe injuries identified by BIG did not appear to be associated with a worse clinical outcome. In our care setting, findings indicated that future application of the BIG classification could reduce interhospital patient transfers for neurosurgical consultation by 16%, ICU admission in 17%, and utilization of repeat imaging by 49%.

The original 2014 BIG,⁵ validated in 2022 in a multi-institutional prospective study of patients 16 years and older admitted to 10 major trauma centers, showed all neurosurgical interventions occurred in patients with BIG 3 injuries. Only two patients (0.7%) with a BIG 2 injury developed neurological deterioration, but no patient with a BIG 1 or 2 injury required neurosurgical intervention. In our much smaller series of children and young adults, the findings are very similar. Moreover, we excluded patients with moderate and severe TBI from our study, that is, patients who would be classified as a BIG 3 by the algorithm, and who because of the strong association between GCS and severity of injury,¹⁶ would be more likely to have ciTBI. These findings further support utilization of BIG to identify critical neurological injuries and indicate that the BIG algorithm can be applied to find the critical injuries amongst patients with mild TBI: that is, out of the small percentage of patients who will have ciTBI after their mild initial injury (8% in our study), applying the BIG algorithm subsequently identifies the patient who is most likely to have that critical injury or ciTBI.

As alluded to in the introduction, the literature to examine the applicability of BIG in the pediatric population has been limited to three small pediatric studies primarily focused on BIG 1 injuries.^7
–9 Our study adds many pediatric patients with BIG 2 injuries, providing a more robust dataset to examine this injury subtype. In the preceding adult and pediatric studies, it has been reported that management of patients with BIG 1 injuries without a neurosurgical consultant would reduce the number of repeat imaging studies. Azim et al. addressed this question in a prospective cohort of 80 pediatric patients with BIG 1 injury managed by acute care surgery (ACS) without involvement of a neurosurgery consultant and found a reduction of 44%.^7
–9 In our study, which examined a larger cohort of patients with BIG 1 injury, our observed reimaging rate was similar to that in their described ACS-managed cohort. However, all of our patients were managed by a neurosurgical consultant, and we did not find that involvement of a neurosurgery consultant was associated with a high repeat imaging rate. Formulated institutional repeat imaging rules, although not as specific as BIG, a cohesive, well-resourced, and experienced multidisciplinary trauma team may explain our findings.

Another pediatric study examined BIG 1 injuries in 28 children over a 4-year period and included nine patients with minor skull fractures. Similarly, none of their patients underwent neurosurgical intervention. The authors suggested that re-classification of simple skull fractures to BIG 1 injury could be considered but would require validation in a larger study. We separately assessed 251 patients with simple skull fractures in our study and found no evidence of ciTBI in this group (data not presented), indicating that reclassification may be reasonable and safe. We will review this in detail in a future publication.

McNickle et al. published the largest pediatric series examining BIG to date and further modified the pediatric BIG classification using the GCS as a distinguishing factor. They re-classified BIG 1 injuries to only include a GCS of 15, BIG 2 injuries to include a GCS of 13–14, and BIG 3 injuries to include a GCS of 12 or below.⁹ Imaging criteria were as originally defined, except for EDH <4 mm which was upgraded to a BIG 2 injury. Other criteria such as abnormal anticoagulation status, presence of intoxication, and transfer from an outside hospital were removed. They evaluated 314 patients and the primary end-point of their study was the need for neurosurgical intervention. In line with their major role as a regional referral center, patient transfer rates were very similar to those observed in our study (68%). The occurrence of neurosurgical intervention was also similar to our study in patients with BIG1 and 2 injuries, but substantially higher in their BIG 3 group, in part because their BIG 3 group included patients with a lower GCS and therefore higher initial injury severity. We classified EDH as originally defined by BIG in our study. We did not observe clinically significant hematoma expansion in the 15 patients with EDH in the BIG 1 injury group but noted clinically significant expansion requiring craniotomy in one of the 29 patients with BIG 2 EDH (3.4%). The majority of the craniotomies performed for EDH (94%); however, were in patients with BIG 3 head injuries, revealing the discriminatory power of BIG for identifying injuries requiring intervention.

While we did not specifically analyze the association between EDH size and clinical deterioration or intervention in this study, and the overall number of patients with EDH in our study is small, we can infer from our data that a clinically significant or expanding EDH may be found frequently enough in patients with a BIG 2 head injury such that routine neurosurgical consultation may remain desirable instead of omission as originally put forward by the BIG study group. Therefore, we recommend reclassifying any EDH to the BIG 3 injury group, so that repeat imaging and neurosurgical consultation is automatically obtained in this group of patients in whom hematoma expansion may require emergent and lifesaving neurosurgical intervention. This approach is also consistent with Schwartz et al. in their multicenter review of adult patients stratified by the original BIG classification, where they found EDH was an independent predictor (OR 6.73) of progression in BIG 1 and 2 patients.⁸

In the PECARN study examining the decision to obtain CT imaging in pediatric patients with mild TBI, a low risk of ciTBI was defined as a predicted risk of 0.05–0.06%, intermediate risk as 0.8–0.9%, and high risk as 4.3% or higher.⁴ These risk assessments informed subsequent decision making for obtaining a head CT to assess the injury. Utilizing similar thresholds to define low, intermediate and high risk for ciTBI in the patients who have ICI or a skull fracture, we advocate to classify BIG 1 patients as low to intermediate risk, BIG 2 patients as intermediate risk, and BIG 3 patients as high risk for ciTBI. In this setting, our proposed management of these patients and recommendations for when to engage a neurosurgical consultant (Table 5) and consider repeat imaging align with these risk assessments.

Table 5.

Proposed Brain Injury Guidelines Classification

Variable	UCD-BIG 1 (estimated ciTBI risk 0.6%)	UCD-BIG 2 (estimated ciTBI risk 0.99%)	UCD-BIG 3 (estimated ciTBI risk 26%)
Neurological exam	Normal	Normal	Abnormal
Intoxication	No	No/Yes	No/Yes
Skull fracture	Simple linear not crossing sutures	Other nondisplaced	Displaced/depressed
EDH	No	No	Yes
SDH	≤4 mm	5–7 mm	≥8 mm
IPH	≤4 mm, 1 location	3–7 mm, 2 locations	≥8 mm, or multiple locations
IVH	No	No	Yes
SAH	Trace, 1–2 sulci	Localized, 3 or more sulci	Scattered
Therapeutic plan
Hospitalization	No, ED observation 6–8 h	Yes	Yes
Routine repeat imaging	No	No	Yes
NSG consultation	Yes/no^a	Yes	Yes

Determined by clinical context, training institution, expertise, and available resources.

BIG, Brain Injury Guidelines; EDH, epidural hematoma; IPH, intraparenchymal hemorrhage; IVH, intraventricular hemorrhage; SAH, subarachnoid hemorrhage; SDH, subdural hematoma.

Our study is the first to observe long-term outcomes (6 months) following discharge with the GOS-E but is hampered by a large loss to follow-up rate. The proportion of patients lost to follow-up was similar in all BIG injury categories and we did not observe a worse outcome with increasing injury severity defined by BIG. Our patients had similar favorable discharge outcomes compared with prior studies (i.e., they went home after their injuries) across BIG injury categories, in line with the expected outcome after mild TBI.^7
–9,17 We suspect we did not find an association between injury severity and GOS-E because all patients had a mild TBI with a GCS of 14–15 and were adequately managed for their injuries. Moreover, the GOS-E may not be able to pick up subtle outcome differences in these patients. To explore this further, we did comparisons of upper good recovery versus other and good recovery versus other, but this also did not yield any differences across BIG injury categories. Future larger prospective studies with sufficient patient follow-up may be able to determine if the BIG classification is able to predict more subtle outcome differences, recovery differences that can significantly affect the lives of the patients.

Hospital admissions and repeat imaging studies

According to the original BIG, management of those with BIG 2 injuries includes hospital admission and exclusion of both repeat imaging and neurosurgical consultation. Aligned with our historical institutional practice, we obtained repeat imaging studies in a large proportion of patients with evidence of intracranial hemorrhage but found few clinically significant injuries in the BIG 2 injury group. This low rate of clinically important injury indicates that routine repeat imaging should not be pursued in this group and that it may be feasible to rely more on clinical examination rather than repeat imaging to assess these patients during hospitalization. To support this approach, we advocate for continued neurosurgical consultation for patients categorized as BIG 2 and involvement in both the initial neurological assessment and subsequent evaluation.

Utility of BIG protocol: Resource allocation

Implementation of the BIG may be significant at the individual patient level and affect multiple parts of their health care experience. First, the logistics involved in transferring the patient and the aftermath of a visit to a hospital can be a burden for patients and their families who must travel long distances and may have to negotiate health insurance costs affiliated with medical transport, hospital stay, and care at a higher-level center. Additionally, while higher-level hospitals may have specialized services available for these patients, bed availability for patient access is finite. Patients with injuries that may not require specialty services compete with the larger population needing access to a tertiary care facility.

With regards to obtaining repeat imaging, small changes can be significant. Especially in young children (under 6 years of age), imaging studies can often only be obtained under sedation or general anesthesia which puts a demand on institutional resources and clinical coordination of care. In addition, in very young children (under 3 years of age), administering general anesthesia may have long-term effects if the anesthetic is prolonged.¹⁸ Moreover, if repeat imaging is done with CT, it is associated with radiation exposure, which may have long-term exposure risks.¹⁹

A potentially cost-effective workflow that preserves access to subspecialty patient care and quality without interhospital transfer is utilization of telehealth. Despite limited literature available on this topic, there are reports that indicate promise in utilizing telehealth for evaluating pediatric trauma.^20,21 These studies indicate that with proper technology infrastructure and human resources for logistic coordination, patients can be adequately and safely assessed remotely. Future studies may help clarify where and how telehealth can be used to advance assessment of mild TBI.

Limitations and strengths

This study is subject to limitations consistent with retrospective single-institution studies. That is, there is limited generalizability, thus multi-institutional studies may assist in widening the application of this guideline. A strength of our study is the sample size: this is the largest pediatric study reported to date, examining patients in all BIG injury categories. Furthermore, our data were meticulously collected in an institutional database within 24 h of patient admission, after medical record and imaging review by dedicated study personnel. Systematic data audits were completed to ensure integrity of the data set and reduce data collection errors. Additionally, three individuals independently determined the BIG categorization and ciTBI classification and met consensus, although no formal inter-rater reliability assessments were conducted (e.g., Cohen’s Kappa).²²

Another limitation is that a toxicology screen was not performed on all children which could have affected classification of the injuries and misclassified some of the BIG 1 injuries. Additionally, this is a retrospective study that focuses predominantly on patients with GCS 14–15 in accordance with the original definition of mild TBI, while other clinical studies have included patients with a wider GCS range. Notably, this study included polytrauma patients. While we considered this during data review and attempted to identify the patients with multisystem injuries by AIS in our study, transfer ratios and outcomes may be skewed due to this patient population. It is also possible that race/ethnicity/socioeconomic status and age may be effect modifiers as they may affect patient outcome and components of ciTBI such as the length of hospital stay.^23,24 Given our sample size, we could not analyze these categories separately. Socioeconomic data were not collected. The outcome data collected in this study were incomplete in a large proportion of patients, making it possible to miss outcome differences between BIG injury groups. Additionally, this study captured data through a time period when our institutional policy for ICU admission changed, and this could have affected patient outcome, even though ICU admission was not a specific end-point of our study. Moreover, only a small proportion of our patients were admitted to the ICU, precluding meaningful analysis of outcome differences, that may have been very subtle to begin with. This may be an area for future research. Lastly, this study was conducted at a major Level I regional trauma center with an experienced trauma team, expert radiology support, and neurosurgery on site. It therefore remains to be determined if our findings can be extrapolated and implemented in a lower-resource environment.

Conclusion

Our study indicates that utilizing the BIG improves risk stratification for children with mild TBI by predicting clinically important TBI. With BIG implementation, we may be able to reduce transfers for neurosurgical consultation to our hospital, ICU admissions, and utilization of repeat imaging studies, while the likelihood of missing a critical neurosurgical injury in those not evaluated by a neurosurgical consultant would remain very low. Our results support prior work that has concluded that patients with a BIG 1 injury can be managed safely without involvement of a neurosurgeon in an ED with appropriate trauma expertise and skilled radiology support. However, we suggest that patients with EDHs and BIG 2 injuries may need continued involvement of a neurosurgeon to obtain a careful neurological examination, guide the decision for obtaining repeat cranial imaging in select patients, and manage any injuries that are detected in follow-up. A prospective study evaluating these adaptations may be necessary to fully understand the implications and impact on patient care. Our interim recommendations are summarized in Table 5.

Transparency, Rigor, and Reproducibility Statement

The retrospective study was approved by the IRB (#2032348-1). The analysis plan was not formally preregistered. The data of 2000 pediatric patients with TBI enrolled in our institutional Trauma Registry were initially screened to identify admissions with mild TBI to our institute and 1259 TBI admissions were identified over 5 years. After excluding patients with a GCS of 13 and below and following the other exclusion criteria outlined in our methods, 804 patients were selected for this retrospective study. No experiments were performed on subjects, and randomization was not required for this study. Data were labeled using codes not linked to participant identifying information.

Data were collected from our institutional TBI registry, an IRB-approved longitudinal database project. Study personnel not involved in the direct clinical care of the patients were trained to chart review pediatric patients who were treated for TBI and collect and store predefined variables in the registry using REDCap electronic data capture. A waiver of consent was used to enter all clinical and imaging data obtained in the acute care phase into this registry. Other study personnel, blinded to the acute phase of patient care, were trained to contact the care provider (i.e., parent, legal guardian) of each patient 6 months after injury. These individuals were trained separately by providing written case scenarios on how to determine GOS-E. These data were also stored in the registry using the REDCap electronic data capture tool and maintained separately from the acute care phase data. All equipment and software used to perform acquisition and analysis are widely available from Microsoft (i.e., Microsoft Word, Microsoft Excel) and Stata. Due to the retrospective nature of the study, investigators were not blinded. The study was developed by investigators with decades of medically licensed practice managing pediatric TBIs in the hospital setting, in addition to experience in designing clinical trials and leadership in Quality Improvement initiatives. Statistical analysis was performed by J.P.M. with qualifications including a Master of Public Health with a concentration in Biostatistics. No replication or external validation studies have been performed or are planned/ongoing to our knowledge. De-identified data from this study are not available in a public archive. There is not an analytic code associated with this study. Materials used to conduct the study are not publicly available. The authors agree to provide the full content of the article on request by contacting M.Z.

Footnotes

Authors’ Contributions

N.Y.: Data curation, writing—original draft, and writing—review and editing. J.C.: Data curation and writing—review and editing. J.E.K.: Writing—review and editing. J.P.M.: Writing—review and editing. D.K.N.: Writing—review and editing. J.M.: Data curation. K.S.: Writing—review and editing. L.K.: Writing—original draft and writing—review and editing. M.Z.: Conceptualization, data curation, formal analysis, methodology, supervision, writing—original draft, and writing—review and editing.

Author Disclosure Statement

The authors have no competing interests to disclose.

Funding Information

There was no funding provided for this research.

References

Keane

, Escobar

Jr , Neff

, et al. Pediatric mild traumatic brain injury: Who can be managed at a non-pediatric trauma center hospital? A systematic review of the literature. Am Surg, 2022; 88(3):447–454; doi: 10.1177/00031348211050804

Yue

, Krishnan

, Andrews

, et al. Update on pediatric mild traumatic brain injury in rural and underserved regions: A global perspective. J Clin Med, 2023; 12(9); doi: 10.3390/jcm12093309

Joseph

, Pandit

, Haider

, et al. Improving hospital quality and costs in nonoperative traumatic brain injury: The role of acute care surgeons. JAMA Surg, 2015; 150(9):866–872; doi: 10.1001/jamasurg.2015.1134

Kuppermann

, Holmes

, Dayan

, et al. Pediatric Emergency Care Applied Research Network (PECARN). Identification of children at very low risk of clinically-important brain injuries after head trauma: A Prospective Cohort Study. Lancet, 2009; 374(9696):1160–1170; doi: 10.1016/s0140-6736(09)61558-0

Joseph

, Friese

, Sadoun

, et al. The BIG (Brain Injury Guidelines) project: Defining the management of traumatic brain injury by acute care surgeons. J Trauma Acute Care Surg, 2014; 76(4):965–969; doi: 10.1097/ta.0000000000000161

Joseph

, Aziz

, Pandit

, et al. Prospective validation of the brain injury guidelines: Managing traumatic brain injury without neurosurgical consultation. J Trauma Acute Care Surg, 2014; 77(6):984–988; doi: 10.1097/ta.0000000000000428

Azim

, Jehan

, Rhee

, et al. Big for small: Validating brain injury guidelines in pediatric traumatic brain injury. J Trauma Acute Care Surg, 2017; 83(6):1200–1204; doi: 10.1097/ta.0000000000001611

Schwartz

, Crandall

, Hsu

, et al. Applying pediatric brain injury guidelines at a level I adult/pediatric safety-net trauma center. J Surg Res, 2020; 255:106–110; doi: 10.1016/j.jss.2020.05.042

McNickle

, Jones

, Yacoub

, et al. BIG Kids: Application of a modified brain injury guideline in a pediatric trauma center. J Pediatr Surg, 2023; 58(3):552–557; doi: 10.1016/j.jpedsurg.2022.07.020

10.

Jennett

, Bond

. Assessment of outcome after severe brain damage. Lancet, 1975; 1(7905):480–484; doi: 10.1016/s0140-6736(75)92830-5

11.

Copes

, Champion

, Sacco

, et al. The injury severity score revisited. J Trauma, 1988; 28(1):69–77; doi: 10.1097/00005373-198801000-00010

12.

Management AfCIIoTaI. Injury Scoring. Available from: https://aci.health.nsw.gov.au/networks/trauma/data/injury-scoring [Last accessed: March 15, 2024 ].

13.

Ament

, Greenan

, Tertulien

, et al. Medical necessity of routine admission of children with mild traumatic brain injury to the intensive care unit. J Neurosurg Pediatr, 2017; 19(6):668–674; doi: 10.3171/2017.2.Peds16419

14.

Nigrovic

, Lee

, Hoyle

, et al. Traumatic Brain Injury (TBI) Working Group of Pediatric Emergency Care Applied Research Network (PECARN). Prevalence of clinically important traumatic brain injuries in children with minor blunt head trauma and isolated severe injury mechanisms. Arch Pediatr Adolesc Med, 2012; 166(4):356–361; doi: 10.1001/archpediatrics.2011.1156

15.

Beers

, Wisniewski

, Garcia-Filion

, et al. Validity of a pediatric version of the Glasgow Outcome Scale-Extended. J Neurotrauma, 2012; 29(6):1126–1139; doi: 10.1089/neu.2011.2272

16.

Teasdale

, Jennett

. Assessment of coma and impaired consciousness. A practical scale. Lancet, 1974; 2(7872):81–84; doi: 10.1016/s0140-6736(74)91639-0

17.

Rivara

, Koepsell

, Wang

, et al. Disability 3, 12, and 24 months after traumatic brain injury among children and adolescents. Pediatrics, 2011; 128(5):e1129-38–e1138; doi: 10.1542/peds.2011-0840

18.

Administration USFD. FDA Drug Safety Communication: FDA Review Results in New Warnings About Using General Anesthetics and Sedation Drugs in Young Children and Pregnant Women. 2018. Available from: https://www.fda.gov/drugs/drug-safety-and-availability/fda-drug-safety-communication-fda-review-results-new-warnings-about-using-general-anesthetics-and?source%3Dgovdelivery&utm_medium%3Demail&utm_source%3Dgovdelivery

19.

Ward

, Eckenhoff

. Neurocognitive adverse effects of anesthesia in adults and children: Gaps in knowledge. Drug Saf, 2016; 39(7):613–626; doi: 10.1007/s40264-016-0415-z

20.

Taylor

, Knochel

, Proctor

, et al. Pediatric trauma telemedicine in a rural state: Lessons learned from a 1-year experience. J Pediatr Surg, 2021; 56(2):385–389; doi: 10.1016/j.jpedsurg.2020.10.020

21.

Kim

, Falcone

RA.

Jr ., The use of telemedicine in the care of the pediatric trauma patient. Semin Pediatr Surg, 2017; 26(1):47–53; doi: 10.1053/j.sempedsurg.2017.01.008

22.

McHugh

. Interrater reliability: The kappa statistic. Biochem Med (Zagreb), 2012; 22(3):276–282.

23.

Haider

, Chang

, Efron

, et al. Race and insurance status as risk factors for trauma mortality. Arch Surg, 2008; 143(10):945–949; doi: 10.1001/archsurg.143.10.945

24.

Nakayama

, Copes

, Sacco

. The effect of patient age upon survival in pediatric trauma. J Trauma, 1991; 31(11):1521–1526; doi: 10.1097/00005373-199111000-00010