Selection of children with ultra-severe traumatic brain injury for neurosurgical intervention

Krista Greenan Department of Neurological Surgery, University of California, Davis;
Department of Pediatric Neurosurgery, University of Colorado, Aurora, Colorado; and

Search for other papers by Krista Greenan in
jns
Google Scholar
PubMed
Close
 MD
,
Sandra L. Taylor Division of Biostatistics, Department of Public Health Sciences, School of Medicine, University of California, Davis, Sacramento, California;

Search for other papers by Sandra L. Taylor in
jns
Google Scholar
PubMed
Close
 PhD
,
Daniel Fulkerson Department of Neurological Surgery, Goodman Campbell Brain and Spine, Riley Hospital for Children, Indiana University, Indianapolis, Indiana

Search for other papers by Daniel Fulkerson in
jns
Google Scholar
PubMed
Close
 MD
,
Kiarash Shahlaie Department of Neurological Surgery, University of California, Davis;

Search for other papers by Kiarash Shahlaie in
jns
Google Scholar
PubMed
Close
 MD, PhD
,
Clayton Gerndt Department of Neurological Surgery, University of California, Davis;

Search for other papers by Clayton Gerndt in
jns
Google Scholar
PubMed
Close
 MD
,
Evan M. Krueger Department of Pediatric Neurosurgery, University of Colorado, Aurora, Colorado; and

Search for other papers by Evan M. Krueger in
jns
Google Scholar
PubMed
Close
 DO
, and
Marike Zwienenberg Department of Neurological Surgery, University of California, Davis;

Search for other papers by Marike Zwienenberg in
jns
Google Scholar
PubMed
Close
 MD
Full access

OBJECTIVE

A recent retrospective study of severe traumatic brain injury (TBI) in pediatric patients showed similar outcomes in those with a Glasgow Coma Scale (GCS) score of 3 and those with a score of 4 and reported a favorable long-term outcome in 11.9% of patients. Using decision tree analysis, authors of that study provided criteria to identify patients with a potentially favorable outcome. The authors of the present study sought to validate the previously described decision tree and further inform understanding of the outcomes of children with a GCS score 3 or 4 by using data from multiple institutions and machine learning methods to identify important predictors of outcome.

METHODS

Clinical, radiographic, and outcome data on pediatric TBI patients (age < 18 years) were prospectively collected as part of an institutional TBI registry. Patients with a GCS score of 3 or 4 were selected, and the previously published prediction model was evaluated using this data set. Next, a combined data set that included data from two institutions was used to create a new, more statistically robust model using binomial recursive partitioning to create a decision tree.

RESULTS

Forty-five patients from the institutional TBI registry were included in the present study, as were 67 patients from the previously published data set, for a total of 112 patients in the combined analysis. The previously published prediction model for survival was externally validated and performed only modestly (AUC 0.68, 95% CI 0.47, 0.89). In the combined data set, pupillary response and age were the only predictors retained in the decision tree. Ninety-six percent of patients with bilaterally nonreactive pupils had a poor outcome. If the pupillary response was normal in at least one eye, the outcome subsequently depended on age: 72% of children between 5 months and 6 years old had a favorable outcome, whereas 100% of children younger than 5 months old and 77% of those older than 6 years had poor outcomes. The overall accuracy of the combined prediction model was 90.2% with a sensitivity of 68.4% and specificity of 93.6%.

CONCLUSIONS

A previously published survival model for severe TBI in children with a low GCS score was externally validated. With a larger data set, however, a simplified and more robust model was developed, and the variables most predictive of outcome were age and pupillary response.

ABBREVIATIONS

AUC = area under the curve; ER = Emergency Room; GCS = Glasgow Coma Scale; GOS = Glasgow Outcome Scale; GOS-E = Extended GOS; ICP = intracranial pressure; MVA = motor vehicle accident; NAT = nonaccidental trauma; RH = Riley Hospital; TBI = traumatic brain injury; tSAH = traumatic subarachnoid hemorrhage; UCD = UC Davis.

OBJECTIVE

A recent retrospective study of severe traumatic brain injury (TBI) in pediatric patients showed similar outcomes in those with a Glasgow Coma Scale (GCS) score of 3 and those with a score of 4 and reported a favorable long-term outcome in 11.9% of patients. Using decision tree analysis, authors of that study provided criteria to identify patients with a potentially favorable outcome. The authors of the present study sought to validate the previously described decision tree and further inform understanding of the outcomes of children with a GCS score 3 or 4 by using data from multiple institutions and machine learning methods to identify important predictors of outcome.

METHODS

Clinical, radiographic, and outcome data on pediatric TBI patients (age < 18 years) were prospectively collected as part of an institutional TBI registry. Patients with a GCS score of 3 or 4 were selected, and the previously published prediction model was evaluated using this data set. Next, a combined data set that included data from two institutions was used to create a new, more statistically robust model using binomial recursive partitioning to create a decision tree.

RESULTS

Forty-five patients from the institutional TBI registry were included in the present study, as were 67 patients from the previously published data set, for a total of 112 patients in the combined analysis. The previously published prediction model for survival was externally validated and performed only modestly (AUC 0.68, 95% CI 0.47, 0.89). In the combined data set, pupillary response and age were the only predictors retained in the decision tree. Ninety-six percent of patients with bilaterally nonreactive pupils had a poor outcome. If the pupillary response was normal in at least one eye, the outcome subsequently depended on age: 72% of children between 5 months and 6 years old had a favorable outcome, whereas 100% of children younger than 5 months old and 77% of those older than 6 years had poor outcomes. The overall accuracy of the combined prediction model was 90.2% with a sensitivity of 68.4% and specificity of 93.6%.

CONCLUSIONS

A previously published survival model for severe TBI in children with a low GCS score was externally validated. With a larger data set, however, a simplified and more robust model was developed, and the variables most predictive of outcome were age and pupillary response.

In Brief

The authors used database research to evaluate admission clinical and CT scan characteristics for use as a decision tool to help clinicians caring for children with very severe traumatic brain injury. It may help clinicians identify children who can benefit the most from aggressive medical and surgical intervention.

Traumatic brain injury (TBI) is a leading cause of death and morbidity in pediatric patients.5,27 The Glasgow Coma Scale (GCS) score is a well-accepted measure of injury severity and has been modified to assess children.19,24,26 The most challenging patients are those presenting with post-resuscitation GCS scores of 4 and below. The a priori chance of a poor outcome in these children is high, and many survivors are often left with severe, lifelong disability. While physicians often have pessimistic views about what can be achieved with aggressive intervention, in children, they often institute treatment because there may be a small but unpredictable chance of meaningful survival. This approach saves lives but can simultaneously mean that a substantial number of children survive with severe disability or in a vegetative or minimally conscious state requiring lifelong supportive care.

Currently, clinical and imaging criteria are not robust enough to help physicians definitively select patients who will benefit from intervention. In a recent retrospective study of 67 pediatric patients with a closed head injury and a GCS score of 3 or 4, Fulkerson et al. reported a favorable long-term outcome in 11.9% of patients.11 Using decision tree analysis, these authors provided clinical and imaging criteria present at the time of admission to identify patients with a favorable outcome as well as those who were unlikely to survive or had a high chance of survival in a vegetative state or with severe disability. The presence of an abnormal pupillary response, hypothermia, hypotension, or a history of abuse statistically correlated with a poor outcome. In contrast, a normal pupillary response combined with a certain type of injury (i.e., fall) identified all patients who had a good recovery.

In the present study, we seek to validate the Fulkerson et al. study findings by using our institutional data set and developing a new model utilizing both data sets. Although critical decisions to offer or withhold therapeutic TBI management will likely remain individualized for each patient, additional tools to aid the clinician in making these difficult triage decisions would be extremely valuable. Such tools may prevent or reduce futile intervention, improve the parent counseling process, facilitate the appropriate allocation of resources, and ultimately improve the overall quality of care for children with severe TBI.

Methods

In this retrospective, observational multicenter study, we used the UC Davis (UCD) Medical Center Department of Neurological Surgery TBI Registry. The UCD source data had been prospectively collected from 2008 to 2017. The UCD Institutional Review Board approved the study. We also used the raw data that had been prospectively collected at the Riley Hospital for Children in Indiana from 1988 to 2004 and reported on by Fulkerson et al. (Riley Hospital [RH] Data Registry).11

Patient Selection

The UCD study cohort includes patients younger than 18 years old with TBI, an abnormal head CT, a GCS score of 4 or less, hospitalization between October 2008 and June 2017, and an available 6-month Glasgow Outcome Scale (GOS) score. All patients presenting to the Emergency Room (ER) at UCD Medical Center were entered into the study, including patients with nonsurvivable brain injuries.

An injury was determined to be nonsalvageable by the on-call neurosurgery faculty upon review of the clinical and imaging findings. If nonaccidental trauma (NAT) was suspected, children underwent additional evaluations that included a skeletal survey, eye exam, and assessment by the child abuse team. Criteria for a nonsalvageable injury generally included a post-resuscitation GCS score of 3 (without sedation/paralysis), bilateral nonreactive pupils, and absent brainstem reflexes. In children with the presence of at least one brainstem reflex, head CT findings and time from injury to ER admission were used as additional determinants. In cases of widespread (ischemic) brain injury and/or a prolonged time of absent bilateral pupillary response (> 2 hours), TBI treatment was not instituted, although the majority of patients were medically supported after injury. In patients with large intracranial mass lesions, emergent evacuation was always considered. All patients were re-examined within an hour from ER admission to verify accurate assessment of the initial GCS score and rule out the presence of medications or systemic factors that could have affected assessment.

Patients with brain injuries deemed salvageable were managed according to the established pediatric TBI guidelines, and an intracranial pressure (ICP) monitor was inserted to guide treatment. Patients with mass lesions were taken to the operating room emergently for hematoma evacuation and/or decompressive craniectomy. Most patients with diffuse brain swelling were first medically managed to get ICP control but then underwent emergent decompressive craniectomy (unilateral or bilateral) if the rapidly escalating steps of medical ICP management failed. Institution of a barbiturate coma with blood pressure control and continuous electroencephalography was used in a limited number of patients, usually as a last resort to control ICP in patients after decompressive craniectomy. Similarly, therapeutic hypothermia (35°C) was instituted if other measures of ICP control failed.

Data Collection

Abstracted data variables included age, sex, comorbidities, medication use, vital signs, post-resuscitation GCS score, mechanism of injury, cranial CT findings, neurological exam, and Injury Severity Score (ISS). Cranial CT findings were interpreted by the on-call neurosurgeon and radiologist. At UCD, dedicated unblinded data abstractors entered the admission data into the registry within 24 hours. Prior to analysis, missing data in the UCD Registry were obtained from the electronic medical record when available.

After inspection for accuracy, data from the two registries were merged into a Microsoft Excel file for Mac (version 15.23.2, Microsoft Corp.).

Outcome Measures

We used the GOS score obtained 6 months after injury as the primary outcome measure in the present study. Although both the 6-month GOS and Extended GOS (GOS-E) scores were collected for the UCD Registry, we could not use the GOS-E score for comparison as the RH Registry only collected GOS score as the primary outcome measure. At UCD, the GOS and GOS-E data were collected in a telephone interview using a standardized screening tool by an unblinded provider not involved in the treatment of the children. At RH, the GOS score was abstracted from the surgeon’s follow-up notes or by phone inquiry. It has been reported that the GOS score does not change significantly from 6 months to 1 year after injury in the majority of patients;1,11 therefore, we believed that using the 6-month outcome would serve as a meaningful surrogate of outcome.

For our analysis, the GOS score was dichotomized into favorable outcome (good recovery/moderate disability) versus unfavorable outcome (severe disability/vegetative state/death). A survival analysis was performed separately.

Statistical Analysis

First, the RH Registry data were compared to the UCD Registry data using the chi-square or Fisher exact test for categorical variables and the Wilcoxon rank-sum test for quantitative variables. Chi-square tests were also used to compare categorical variables by surgical intervention. Next, the UCD Registry data were used to externally validate the Fulkerson et al. prediction model for survival. However, because there were very few favorable outcomes in the UCD Registry, it would not provide a good assessment of the performance of the Fulkerson model for favorable versus unfavorable outcomes. The probability of the survival of each subject in the UCD data was predicted using the Fulkerson survival model. Model performance was assessed by comparing observed and predicted survival probabilities and was quantified by calculating the area under the receiver operating characteristic curve.

We subsequently combined the RH and UCD data sets and developed prediction models for survival and dichotomized GOS outcome. Using the combined data set, we constructed decision trees through recursive partitioning using the function Rpart in R Statistical Computing Software with a minimum node size of 1, cross-validation of 10, and complexity parameter of 0. Variables evaluated in building the decision trees were age, sex, pupillary response (bilateral reactive, unilateral nonreactive, bilateral nonreactive), mechanism of injury, midline shift (> 5 mm), epidural hematoma, subdural hematoma, intraparenchymal hematoma, intraventricular hemorrhage, traumatic subarachnoid hemorrhage (tSAH), compression of cisterns, hypoxia, hypothermia (< 36°C), and hypotension (> 2 SD for age). For mechanism of injury, “fall from bike” and “fall from steps” were combined with “fall,” and “struck” was combined with “assault.” A p value < 0.05 was considered statistically significant.

Results

Patient Demographics and Clinical Characteristics

A summary of demographic and clinical characteristics and outcomes for the two patient cohorts is presented in Tables 13 and Fig. 1. The RH cohort was significantly younger and more likely to have suffered NAT as the mechanism of injury. Patients in the UCD cohort were more likely to be involved in a motor vehicle accident (MVA). Hypotension and hypothermia were more common in the UCD cohort; however, data regarding these two disorders were unknown for many patients in the RH cohort.

TABLE 1.

Demographics, imaging characteristics, and outcome measures in the combined cohort of pediatric patients with severe TBI

VariableValue
Clinical characteristic
 Age in yrs (range)5.94 (0–17.99)
 Sex: F/M/unknown41/70/1
 Transfer from outside hospital58 (52%)
 Mechanism of injury
  MVA29 (26%)
  Fall13 (12%)
  Auto vs pedestrian26 (23%)
  NAT39 (35%)
  Assault2 (2%)
  Other3 (3%)
 Hypotension (before or w/in 24 hrs of admit)57 (51%)
 Hypoxia60 (54%)
 Hypothermia (<36°C)70 (63%)
 Temperature range in °C29.1–40
 Pupillary response: BNR/UNR/BR76 (68%)/7 (6%)/29 (26%)
 GCS score 381 (72%)
 GCS score 431 (28%)
Imaging characteristic
 Midline shift >5 mm39 (35%)
 IVH18 (16%)
 ICH22 (20%)
 ASDH58 (52%)
 EDH6 (5%)
 Basal cisterns (compressed/absent)52 (46%)
Intervention
 Surgical intervention70 (63%)
 Craniotomy42 (38%)
 ICP monitor70 (63%)
GOS at 6 mos
 Death69 (62%)
 Vegetative10 (9%)
 Severe disability14 (12%)
 Moderate disability15 (13%)
 Good recovery4 (4%)

ASDH = acute subdural hematoma; BNR = bilateral nonreactive; BR = bilateral reactive; EDH = epidural hematoma; ICH = intracerebral hemorrhage; IVH = intraventricular hemorrhage; UNR = unilateral nonreactive.

TABLE 2.

Comparison of characteristics between UCD and RH patients

VariableRH Cohort (n = 67)UCD Cohort (n = 45)p Value
No. of males (%)40 (59.7%)30 (66.7%)0.733
Age in yrs (range)3.1 (1.2–6.8)6.5 (1.2–14.8)0.009
Age in yrs, no. (%)
 0–114 (20.9%)7 (15.6%)
 >1–427 (40.3%)10 (22.2%)
 5–917 (25.4%)10 (22.2%)
 10–146 (8.9%)3 (6.7%)
 15–173 (4.5%)10 (22.2%)
Transfer from outside hospital40 (59.7%)18 (40.0%)0.064
Mechanism of injury0.008
 Assault2 (3%)2 (4%)
 Auto vs pedestrian14 (21%)12 (27%)
 Fall10 (15%)3 (7%)
 Jumping from vehicle0 (0%)1 (2%)
 MVA11 (16%)18 (40%)
 NAT30 (45%)9 (20%)
 Other0 (0%)1 (2.2%)
Hypotension*0.001
 Yes30 (45%)18 (60%)
 No23 (34%)27 (40%)
 Unknown14 (21%)0 (0%)
Hypoxia<0.001
 Yes38 (57%)22 (49%)
 No12 (18%)23 (51%)
 Unknown17 (25%)0 (0%)
Hypothermia<0.001
 Yes38 (46%)39 (87%)
 No24 (36%)6 (13%)
 Unknown12 (18%)0 (0%)
Pupil response0.267
 BNR42 (63%)34 (75%)
 UNR4 (6%)3 (7%)
 BR21 (31%)8 (18%)
Midline shift26 (38.9%)13 (28.9%)0.380
GCS score 3 (vs 4)44 (38.9%)37 (82.2%)0.088
IVH11 (16.4%)7 (15.6%)0.999
ICH12 (17.9%)10 (22.2%)0.749
ASDH44 (65.7%)27 (60.0%)0.681
EDH2 (3.0%)5 (11.1%)0.115
tSAH46 (68.7%)23 (51.1%)0.094
Cisterns compressed/absent51 (76.1%)28 (62.2%)0.171

n = number

Mean blood pressure > 2 SD below normal for age.

SaO2 < 92%.

Temperature < 36°C.

TABLE 3.

Outcomes in two pediatric cohorts

OutcomeRH Group (n = 67)UCD Group (n = 45)
GOS(-E)
 Good recovery
  Upper8 (11.9%)0 (0%)
  Lower2 (3%)0 (0%)
 Moderate disability
  Upper4 (6.0%)0 (0%)
  Lower5 (11.1%) 
 Severe disability
  Upper7 (10.4%)1 (2.2%)
  Lower4 (8.9%)
 Vegetative3 (4.5%)4 (8.9%)
 Death38 (56.7%)31 (68.9%)
Dichotomized GOS
 Favorable outcome14 (20.9%)5 (11.1%)
 Unfavorable outcome48 (71.6%)40 (88.9%)
 Lost to follow-up5 (7.5%)0 (0%)
FIG. 1.
FIG. 1.

Comparison of mechanisms of injury between UCD cohort and Fulkerson et al. study. Asterisks indicate statistical significance at p < 0.008. Peds = pedestrian.

Characteristics and Outcome of Patients Not Offered ICP-Directed TBI Management

In the UCD cohort that did not receive ICP-directed TBI management (19 patients [42% of cohort]), 26% were injured from NAT, 37% from an MVA, 26% from an auto versus pedestrian injury, and 11% from a fall. Patients shared the following common characteristics: GCS score 3, bilaterally nonreactive pupils, hypothermic upon admission, and midline shift < 5 mm. Additionally, 89% of these patients were hypotensive, 69% had hypoxia, 63% had compressed cisterns, 79% had tSAH, and 1 patient had an epidural hematoma. Patients received general supportive care after admission to allow the family to be with their child. All patients died after supportive measures were held.

In the RH cohort that did not receive ICP-directed TBI management (23 patients [34% of cohort]), the majority of patients also had a GCS score of 3 and bilaterally nonreactive pupils, but overall the group was more heterogeneous than the UCD cohort. The majority of patients had NAT (74%). Other injuries included MVA (9%), auto versus pedestrian injury (9%), and other (8%). A few patients had a GCS score of 4 (13%), reactive pupils (17%), and a temperature > 36°C (17%). Hypotension was present in 48% of patients and hypoxia in 61%. On head CT, 74% of patients had compressed cisterns, 70% had tSAH, and 87% had a midline shift < 5 mm. One patient in this group made a full recovery, 1 patient survived in a vegetative state, and 3 patients had severe disability. The remaining 18 patients (78%) died.

Outcome of ICP-Directed TBI Management

Neurosurgical intervention and ICP-directed TBI management were offered to the majority of patients in both cohorts. The proportion of patients who underwent any type of neurosurgical intervention (ICP monitor and/or craniotomy) appeared higher in the RH cohort (66% vs 58%, not significant); however, craniotomy rates (37% vs 38%) in the two cohorts were similar. At the RH, craniotomy (25 cases) was performed only when a patient had a mass lesion, and in most patients a decompressive craniectomy was performed. At UCD, decompressive craniectomy was performed for brain swelling in the absence of a mass lesion: of the 17 craniotomy patients, 5 (29%) underwent bilateral decompressive craniectomy and 2 (12%) underwent craniectomy for brain swelling only.

Compared to the nonintervention cohort, the patients who underwent TBI treatment had nonreactive pupils less frequently (57% vs 90%, p < 0.001) and fewer instances of hypotension (36% vs 67%, p < 0.001) and hypoxia (43% vs 64%, p < 0.02). The occurrence of hypothermia was not different (47% vs 55%, p < 0.4). Midline shift > 5 mm (51% vs 7%, p < 0.0001) was more common and tSAH (54% vs 86%, p < 0.0002) and compressed cisterns (71% vs 93% p < 0.006) were less common in the patients receiving intervention.

In the UCD cohort, overall survival was 31% (14/45) and 18% (3/17) of craniotomy patients had a favorable outcome. In the RH cohort, overall survival was 43% (29/67), and 5 craniotomy patients (20%) had a favorable outcome. Overall, 40% (17/42) survived and 21% (9/42) had a favorable outcome after craniotomy.

Validation of the Fulkerson et al. Model

The Fulkerson study included two decision trees, one for survival and one for favorable outcome. Of the 45 UCD patients, 31 (68.9%) died and 40 (88.9%) had an unfavorable outcome. Because only a few patients had a good outcome, applying the Fulkerson outcome model to the UCD data would not provide a good assessment of the model’s performance. Therefore, only the survival model was assessed for the validation analysis.

The Fulkerson survival model performed modestly. The area under the curve (AUC) for the model applied to the UCD data was 0.68 with a wide 95% CI of 0.47, 0.89. The overall predicted survival of 13.5% for the UCD group was very close to the observed survival for the cohort (13%), but when the expected versus observed survival was compared, the proportions of patients surviving with similar initial clinical and imaging characteristics were quite variable between the RH and UCD cohorts. In a group of patients with an expected low probability of survival (0%–4%) based on the Fulkerson prediction tree, a higher percentage of UCD patients survived but all with lower severe disability or in a vegetative state (data not shown). Conversely, UCD patients in a group with an expected high survival rate (56%–100%) had lower overall survival rates than the Fulkerson patients.

Decision Tree for Favorable Outcomes Based on the Combined Data Set

Data from both registries were combined, and new outcome and survival decision trees were developed. Table 4 summarizes the patient characteristics used to construct the decision trees.

TABLE 4.

Predictor variables used to construct decision trees

Variable
Sex
Age
Transfer outside hospital
Mechanism of injury
Hypotension
Hypothermia
Hypoxia
Pupillary response
Midline shift
Basal cisterns compressed
tSAH present
EDH present
SDH present
IVH present

Using all clinical and imaging data (UCD and RH cohorts), pupillary response and age were the only predictors retained in the outcome model. The resultant decision tree is shown in Fig. 2. A bilaterally nonreactive pupillary response led to an unfavorable outcome for 73 (96%) of 76 patients. If the pupillary response was normal in at least one eye (i.e., bilaterally or unilaterally reactive), the outcome subsequently depended on age. An age ≤ 0.38 years (4.5 months) always yielded an unfavorable outcome (100%). In children between 0.38 and 6.2 years, the proportion of children with an unfavorable outcome was much lower (5 of 18 [28%]), whereas this proportion would rise again in children 6.2 years or older (77%). The overall accuracy of the combined prediction model was 90.2% with a sensitivity of 68.4% and specificity of 93.6%.

FIG. 2.
FIG. 2.

Decision tree for favorable outcomes in the combined data set. BR = bilateral reactive; n = number of patients; UNR = unilateral nonreactive.

Decision Tree for Survival Based on the Combined Data Set

A decision tree for survival was created using the same predictors as those used for the outcome tree. Pupillary response, hypotension, and mechanism of injury were identified as significant predictors (Fig. 3). If patients had at least one reactive pupil, the likelihood of survival was 81%. In contrast, if patients had bilateral nonreactive pupils, the odds were reversed, and the likelihood of death was 82%. If the patients with nonreactive pupils were hypotensive as well, the likelihood of death increased to 91%. In patients without hypotension but injured when hit by a car or from an assault, mortality was 29%.

FIG. 3.
FIG. 3.

Decision tree for survival in the combined data set. BNR = bilateral nonreactive.

Discussion

The results of this study showed that patient age and pupillary response are the strongest predictors of outcome in children with a post-resuscitation GCS score of 3 or 4. Using combined registry data, we found that the outcome of patients with bilateral nonreactive pupils was almost universally unfavorable, with an 82% mortality rate, an overall unfavorable outcome in 96%, and a favorable outcome in only 3 of the surviving patients. Subsequent hypotension or injury by NAT, MVA, or fall was associated with increased mortality.

Patients with at least one reactive pupil did much better. The overall survival rate in this group was high (81%), and the outcome in this group subsequently depended on age. Most children between 5 months and 6 years old had a favorable outcome (72%), whereas all children younger than 4.5 months old (100%) and most children older than 6 years old (77%) had an unfavorable outcome.

GCS score, pupillary response, and age have been consistently shown to predict morbidity and mortality in both adult and pediatric populations with severe TBI.1,2,6,8–10,16,17,20,22,29 There are only a few studies that have focused exclusively on patients with a GCS score of 3 or 4, all except the Fulkerson et al. study conducted in adult patients. The results in these studies reflect what we found: mortality is exceedingly high and significantly worse in patients with fixed, dilated pupils. For example, Tien et al. reported on 179 patients of 15 years and older and found 100% mortality for the group with fixed pupils and 42% mortality for the patients with reactive pupils.28 In a study including 51,425 patients from the German Trauma Registry, only 8% of patients with bilateral fixed pupils on presentation had a favorable outcome.15

In a recent publication on adult severe TBI, a new score that includes GCS score and pupillary status (GCS-P) was found to correlate well with prognosis and performed as well as more complex scoring systems.4 In a preliminary look at our data set (data not shown), a GCS-P of 1 (i.e., GCS score 3 with nonreactive pupils) was associated with an 88% mortality rate and a 96% unfavorable outcome rate, and higher scores correlated with improved outcomes. We plan to report our findings in a separate study.

Children who present with severe TBI tend to have lower mortality and a better long-term outcome than adults who present with similar TBI severity, although this effect is not as robust in children with very severe injuries.1,10 Furthermore, the relationship between age and outcome does not appear to be linear in children; that is, younger patients with TBI (age < 1 year) are reported to have worse outcomes. This is perhaps related to the vulnerability of the brain in earlier stages of development. In our study, we found the poorest outcomes in children younger than 5 months of age, a more favorable outcome in children up to 6 years old, and a worsening outcome at ages thereafter. Almost all of the very young children suffered from NAT, which may represent a very different mechanism than other types of injury, with its potential for repetitive injury or secondary insult to the brain. Additionally, many of these children had evidence of hypoxic injury upon admission, which may worsen the overall insult to the brain and therefore prognosis.

The relatively superior outcome in children between 5 months and 6 years of age and the subsequently worse outcome in older children were unexpected findings. The older children in our study appeared to have had a high-speed mechanism of injury (i.e., MVA or auto vs pedestrian injury) more often and may have been more severely injured as a result. Of the children between 5 months and 6 years of age, a substantial number had NAT, but concurrent hypoxic injury was noted less frequently than in the younger children, which may explain a superior outcome. Contrary to our results, in a recent publication on 188 pediatric patients with TBI (45 with severe TBI), children ages 7–9 years had the worst outcomes compared to those in the other age groups.7

In this study, we sought to confirm the clinical scenarios from the Fulkerson et al. study that could drive decision-making by strongly predicting outcome. For example, in the decision tree reported by Fulkerson and colleagues, all patients with reactive pupils and injured from a fall had a favorable long-term recovery. In this patient group, aggressive and sustained intervention to treat elevated ICP would be indicated. In contrast, only 4% of patients with pupillary abnormalities (unilateral or bilateral), hypothermia, and a mechanism of injury other than MVA were found to survive, and none of these patients had a favorable outcome. In these patients, initial treatment could be offered with the understanding that the upfront chance of survival with a favorable outcome would be exceedingly small. If, in the subsequent 24 hours, no clinical improvement is seen, transition to comfort care should be considered. We did not find similar associations for these clinical scenarios using our data set, and as a result we found a modest AUC. This is most likely because the number of subjects in each node was very small; therefore, individual patients with extreme results, favorable or unfavorable, have had profound impact on the observed overall outcome in the node. Additionally, there were significant differences between the two cohorts in terms of the types of patients included, their clinical presentations and interventions, and the data collection, and the modest AUC may indicate that they are not fully comparable.

We did find that, similar to the Fulkerson et al. study, pupillary response was a primary predictor for both survival and favorable outcomes in the models published here. Mechanism of injury and hypotension were also included in both the Fulkerson survival model and the survival model published with the current analysis. Thus, there were commonalities between our models in the discriminatory predictors, although the specific clinical scenarios varied. Differences in patient characteristics could contribute to the discrepancies, and larger studies with diverse patients will be needed to develop a broadly applicable decision-support tool.13 Powerful validated tools such as the International Mission on Prognosis in Traumatic Brain Injury (IMPACT) calculator are available for the adult TBI population.12,14,18,21,23,25 Applying these validation tools to the pediatric population in future studies may yield additional and more robust instruments.30

Study Limitations

In addition to the retrospective nature of this work, data were collected over a prolonged period of time, and overall care for pediatric brain injury has changed over that same period.

Additionally, this study includes patients for whom care was withdrawn, which introduces some complicating factors. The time at which care was withdrawn was not a variable in our analysis. This means that there were patients who were included and intervened on and who were then deemed nonsalvageable shortly thereafter. Lastly, there were likely institutional differences in patient selection for intervention, type of intervention chosen, or the withholding of care if it was deemed futile. We believed that it was important to include these patients in our analysis because excluding them would have created a significant selection bias.

The data collection at UCD has been rigorous and systematic; as a result, there were very few missing values in the registry. For example, to avoid inaccurate GCS scores, neurosurgical exam data were cross-checked with nursing notes and documentation by other nonneurosurgical providers at admission. However, clinical events such as subclinical posttraumatic seizures could have interfered with assessment of the GCS score but were not excluded from analysis.

For accurate GOS scores, all patients in the UCD Registry underwent a formal outcome assessment by a dedicated, nonblinded provider at set times after injury. In contrast, outcome assessment for the RH data set relied on the surgeon in office evaluation, which may not have been as precise. The RH Registry also contained substantially more missing values, which may have affected the reported associations. The outcomes were recorded many years after injury, whereas UCD patients were followed only 6 months after injury. In most reported series, including the Fulkerson series itself, the most improvement in outcome occurs in the first 6 months after injury. Clinical improvement beyond that tends to slow down, especially in patients with very severe injuries, but neurological recovery does not stop completely, and there is still improvement that occurs between 6 and 12 months after injury and, occasionally, even beyond that. Thus, the overall favorable outcome of the UCD cohort may have been underestimated and not entirely comparable to the RH cohort.

Finally, a number of assessment tools exist to evaluate the outcome of pediatric head injury. As the GOS was the only outcome measure used in the Fulkerson study, we could only use the GOS to compare the two cohorts. The GOS is well validated and has been in use for 40 years in the outcome assessment of TBI. In future studies, however, the pediatric GOS-E or developmental assessments should be included in the design if high-impact triage decisions are to be made based on the reported results.3

Conclusions

We externally validated previous survival and outcome models with only modest success. With a combined data set, however, a more robust model was developed, and the variables most predictive of outcome were patient age and pupillary response. The presence of fixed and dilated pupils portends a very poor prognosis in this pediatric patient group, especially those with a GCS score of 3, and aggressive intervention may be warranted in only select patients. Further studies using larger data sets and more refined outcome tools are needed to validate these findings and perhaps derive more detailed clinical findings to support decision-making.

Acknowledgments

We would like to acknowledge Nancy Rudisill, MscN, RN, and Karen Smith, RN, CNRN, for their assistance in compiling and maintaining the UC Davis TBI Registry, which was critical to this work.

Disclosures

This work was supported by a financial grant from the UC Davis Department of Neurological Surgery and the UC Davis Clinical and Translational Science Center supported by the National Institutes of Health through grant no. UL1 TR001860.

Author Contributions

Conception and design: Zwienenberg. Acquisition of data: Zwienenberg, Fulkerson. Analysis and interpretation of data: Zwienenberg, Greenan, Taylor. Drafting the article: Zwienenberg, Greenan. Critically revising the article: Zwienenberg, Greenan, Taylor, Fulkerson, Shahlaie, Gerndt. Reviewed submitted version of manuscript: Zwienenberg, Greenan, Taylor, Fulkerson, Shahlaie, Gerndt. Approved the final version of the manuscript on behalf of all authors: Zwienenberg. Statistical analysis: Taylor. Study supervision: Zwienenberg. Literature search and bibliography: Krueger.

Supplemental Information

Previous Presentations

Portions of this work were presented in abstract form as proceedings at the 46th Annual Meeting of the AANS/CNS Section on Pediatric Neurological Surgery, in Houston, TX, on November 30, 2017.

References

  • 1

    Anderson V, Godfrey C, Rosenfeld JV, Catroppa C: Predictors of cognitive function and recovery 10 years after traumatic brain injury in young children. Pediatrics 129:e254e261, 2012

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 2

    Arnett AB, Peterson RL, Kirkwood MW, Taylor HG, Stancin T, Brown TM, et al.: Behavioral and cognitive predictors of educational outcomes in pediatric traumatic brain injury. J Int Neuropsychol Soc 19:881889, 2013

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 3

    Beers SR, Wisniewski SR, Garcia-Filion P, Tian Y, Hahner T, Berger RP, et al.: Validity of a pediatric version of the Glasgow Outcome Scale–Extended. J Neurotrauma 29:11261139, 2012

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 4

    Brennan PM, Murray GD, Teasdale GM: Simplifying the use of prognostic information in traumatic brain injury. Part 1: The GCS-Pupils score: an extended index of clinical severity. J Neurosurg 128:16121620, 2018

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 5

    Centers for Disease Control and Prevention: Ten leading causes of death by age group, United States, 2016. CDC.gov (https://www.cdc.gov/injury/wisqars/LeadingCauses.html) [Accessed February 15, 2019]

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 6

    Chiaretti A, Piastra M, Pulitanò S, Pietrini D, De Rosa G, Barbaro R, et al.: Prognostic factors and outcome of children with severe head injury: an 8-year experience. Childs Nerv Syst 18:129136, 2002

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 7

    Crowe LM, Catroppa C, Babl FE, Rosenfeld JV, Anderson V: Timing of traumatic brain injury in childhood and intellectual outcome. J Pediatr Psychol 37:745754, 2012

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 8

    Di Battista A, Soo C, Catroppa C, Anderson V: Quality of life in children and adolescents post-TBI: a systematic review and meta-analysis. J Neurotrauma 29:17171727, 2012

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 9

    Ducrocq SC, Meyer PG, Orliaguet GA, Blanot S, Laurent-Vannier A, Renier D, et al.: Epidemiology and early predictive factors of mortality and outcome in children with traumatic severe brain injury: experience of a French pediatric trauma center. Pediatr Crit Care Med 7:461467, 2006

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 10

    Emami P, Czorlich P, Fritzsche FS, Westphal M, Rueger JM, Lefering R, et al.: Impact of Glasgow Coma Scale score and pupil parameters on mortality rate and outcome in pediatric and adult severe traumatic brain injury: a retrospective, multicenter cohort study. J Neurosurg 126:760767, 2017

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 11

    Fulkerson DH, White IK, Rees JM, Baumanis MM, Smith JL, Ackerman LL, et al.: Analysis of long-term (median 10.5 years) outcomes in children presenting with traumatic brain injury and an initial Glasgow Coma Scale score of 3 or 4. J Neurosurg Pediatr 16:410419, 2015

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 12

    Gómez PA, de-la-Cruz J, Lora D, Jiménez-Roldán L, Rodríguez-Boto G, Sarabia R, et al.: Validation of a prognostic score for early mortality in severe head injury cases. J Neurosurg 121:13141322, 2014

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 13

    Haider AH, Crompton JG, Oyetunji T, Risucci D, DiRusso S, Basdag H, et al.: Mechanism of injury predicts case fatality and functional outcomes in pediatric trauma patients: the case for its use in trauma outcomes studies. J Pediatr Surg 46:15571563, 2011

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 14

    Han JX, See AAQ, Gandhi M, King NKK: Models of mortality and morbidity in severe traumatic brain injury: an analysis of a Singapore neurotrauma database. World Neurosurg 108:885893, 893.e1, 2017

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 15

    Hoffmann M, Lefering R, Rueger JM, Kolb JP, Izbicki JR, Ruecker AH, et al.: Pupil evaluation in addition to Glasgow Coma Scale components in prediction of traumatic brain injury and mortality. Br J Surg 99 (Suppl 1):122130, 2012

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 16

    Hukkelhoven CW, Steyerberg EW, Rampen AJ, Farace E, Habbema JD, Marshall LF, et al.: Patient age and outcome following severe traumatic brain injury: an analysis of 5600 patients. J Neurosurg 99:666673, 2003

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 17

    Jagannathan J, Okonkwo DO, Yeoh HK, Dumont AS, Saulle D, Haizlip J, et al.: Long-term outcomes and prognostic factors in pediatric patients with severe traumatic brain injury and elevated intracranial pressure. J Neurosurg Pediatr 2:240249, 2008

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 18

    Kesmarky K, Delhumeau C, Zenobi M, Walder B: Comparison of two predictive models for short-term mortality in patients after severe traumatic brain injury. J Neurotrauma 34:22352242, 2017

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 19

    Kirkham FJ, Newton CR, Whitehouse W: Paediatric coma scales. Dev Med Child Neurol 50:267274, 2008

  • 20

    Michaud LJ, Rivara FP, Grady MS, Reay DT: Predictors of survival and severity of disability after severe brain injury in children. Neurosurgery 31:254264, 1992

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 21

    Perel P, Arango M, Clayton T, Edwards P, Komolafe E, Poccock S, et al.: Predicting outcome after traumatic brain injury: practical prognostic models based on large cohort of international patients. BMJ 336:425429, 2008

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 22

    Pillai S, Praharaj SS, Mohanty A, Kolluri VR: Prognostic factors in children with severe diffuse brain injuries: a study of 74 patients. Pediatr Neurosurg 34:98103, 2001

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 23

    Rizoli S, Petersen A, Bulger E, Coimbra R, Kerby JD, Minei J, et al.: Early prediction of outcome after severe traumatic brain injury: a simple and practical model. BMC Emerg Med 16:32, 2016

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 24

    Simpson D, Reilly P: Pediatric coma scale. Lancet 2:450, 1982

  • 25

    Steyerberg EW, Mushkudiani N, Perel P, Butcher I, Lu J, McHugh GS, et al.: Predicting outcome after traumatic brain injury: development and international validation of prognostic scores based on admission characteristics. PLoS Med 5:e165, 2008

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 26

    Tatman A, Warren A, Williams A, Powell JE, Whitehouse W: Development of a modified paediatric coma scale in intensive care clinical practice. Arch Dis Child 77:519521, 1997

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 27

    Thurman DJ, Alverson C, Dunn KA, Guerrero J, Sniezek JE: Traumatic brain injury in the United States: a public health perspective. J Head Trauma Rehabil 14:602615, 1999

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 28

    Tien HC, Cunha JR, Wu SN, Chughtai T, Tremblay LN, Brenneman FD, et al.: Do trauma patients with a Glasgow Coma Scale score of 3 and bilateral fixed and dilated pupils have any chance of survival? J Trauma 60:274278, 2006

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 29

    Tude Melo JR, Di Rocco F, Blanot S, Oliveira-Filho J, Roujeau T, Sainte-Rose C, et al.: Mortality in children with severe head trauma: predictive factors and proposal for a new predictive scale. Neurosurgery 67:15421547, 2010

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 30

    Young AM, Guilfoyle MR, Fernandes H, Garnett MR, Agrawal S, Hutchinson PJ: The application of adult traumatic brain injury models in a pediatric cohort. J Neurosurg Pediatr 18:558564, 2016

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • Collapse
  • Expand
  • Comparison of mechanisms of injury between UCD cohort and Fulkerson et al. study. Asterisks indicate statistical significance at p < 0.008. Peds = pedestrian.

  • Decision tree for favorable outcomes in the combined data set. BR = bilateral reactive; n = number of patients; UNR = unilateral nonreactive.

  • Decision tree for survival in the combined data set. BNR = bilateral nonreactive.

  • 1

    Anderson V, Godfrey C, Rosenfeld JV, Catroppa C: Predictors of cognitive function and recovery 10 years after traumatic brain injury in young children. Pediatrics 129:e254e261, 2012

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 2

    Arnett AB, Peterson RL, Kirkwood MW, Taylor HG, Stancin T, Brown TM, et al.: Behavioral and cognitive predictors of educational outcomes in pediatric traumatic brain injury. J Int Neuropsychol Soc 19:881889, 2013

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 3

    Beers SR, Wisniewski SR, Garcia-Filion P, Tian Y, Hahner T, Berger RP, et al.: Validity of a pediatric version of the Glasgow Outcome Scale–Extended. J Neurotrauma 29:11261139, 2012

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 4

    Brennan PM, Murray GD, Teasdale GM: Simplifying the use of prognostic information in traumatic brain injury. Part 1: The GCS-Pupils score: an extended index of clinical severity. J Neurosurg 128:16121620, 2018

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 5

    Centers for Disease Control and Prevention: Ten leading causes of death by age group, United States, 2016. CDC.gov (https://www.cdc.gov/injury/wisqars/LeadingCauses.html) [Accessed February 15, 2019]

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 6

    Chiaretti A, Piastra M, Pulitanò S, Pietrini D, De Rosa G, Barbaro R, et al.: Prognostic factors and outcome of children with severe head injury: an 8-year experience. Childs Nerv Syst 18:129136, 2002

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 7

    Crowe LM, Catroppa C, Babl FE, Rosenfeld JV, Anderson V: Timing of traumatic brain injury in childhood and intellectual outcome. J Pediatr Psychol 37:745754, 2012

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 8

    Di Battista A, Soo C, Catroppa C, Anderson V: Quality of life in children and adolescents post-TBI: a systematic review and meta-analysis. J Neurotrauma 29:17171727, 2012

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 9

    Ducrocq SC, Meyer PG, Orliaguet GA, Blanot S, Laurent-Vannier A, Renier D, et al.: Epidemiology and early predictive factors of mortality and outcome in children with traumatic severe brain injury: experience of a French pediatric trauma center. Pediatr Crit Care Med 7:461467, 2006

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 10

    Emami P, Czorlich P, Fritzsche FS, Westphal M, Rueger JM, Lefering R, et al.: Impact of Glasgow Coma Scale score and pupil parameters on mortality rate and outcome in pediatric and adult severe traumatic brain injury: a retrospective, multicenter cohort study. J Neurosurg 126:760767, 2017

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 11

    Fulkerson DH, White IK, Rees JM, Baumanis MM, Smith JL, Ackerman LL, et al.: Analysis of long-term (median 10.5 years) outcomes in children presenting with traumatic brain injury and an initial Glasgow Coma Scale score of 3 or 4. J Neurosurg Pediatr 16:410419, 2015

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 12

    Gómez PA, de-la-Cruz J, Lora D, Jiménez-Roldán L, Rodríguez-Boto G, Sarabia R, et al.: Validation of a prognostic score for early mortality in severe head injury cases. J Neurosurg 121:13141322, 2014

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 13

    Haider AH, Crompton JG, Oyetunji T, Risucci D, DiRusso S, Basdag H, et al.: Mechanism of injury predicts case fatality and functional outcomes in pediatric trauma patients: the case for its use in trauma outcomes studies. J Pediatr Surg 46:15571563, 2011

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 14

    Han JX, See AAQ, Gandhi M, King NKK: Models of mortality and morbidity in severe traumatic brain injury: an analysis of a Singapore neurotrauma database. World Neurosurg 108:885893, 893.e1, 2017

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 15

    Hoffmann M, Lefering R, Rueger JM, Kolb JP, Izbicki JR, Ruecker AH, et al.: Pupil evaluation in addition to Glasgow Coma Scale components in prediction of traumatic brain injury and mortality. Br J Surg 99 (Suppl 1):122130, 2012

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 16

    Hukkelhoven CW, Steyerberg EW, Rampen AJ, Farace E, Habbema JD, Marshall LF, et al.: Patient age and outcome following severe traumatic brain injury: an analysis of 5600 patients. J Neurosurg 99:666673, 2003

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 17

    Jagannathan J, Okonkwo DO, Yeoh HK, Dumont AS, Saulle D, Haizlip J, et al.: Long-term outcomes and prognostic factors in pediatric patients with severe traumatic brain injury and elevated intracranial pressure. J Neurosurg Pediatr 2:240249, 2008

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 18

    Kesmarky K, Delhumeau C, Zenobi M, Walder B: Comparison of two predictive models for short-term mortality in patients after severe traumatic brain injury. J Neurotrauma 34:22352242, 2017

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 19

    Kirkham FJ, Newton CR, Whitehouse W: Paediatric coma scales. Dev Med Child Neurol 50:267274, 2008

  • 20

    Michaud LJ, Rivara FP, Grady MS, Reay DT: Predictors of survival and severity of disability after severe brain injury in children. Neurosurgery 31:254264, 1992

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 21

    Perel P, Arango M, Clayton T, Edwards P, Komolafe E, Poccock S, et al.: Predicting outcome after traumatic brain injury: practical prognostic models based on large cohort of international patients. BMJ 336:425429, 2008

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 22

    Pillai S, Praharaj SS, Mohanty A, Kolluri VR: Prognostic factors in children with severe diffuse brain injuries: a study of 74 patients. Pediatr Neurosurg 34:98103, 2001

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 23

    Rizoli S, Petersen A, Bulger E, Coimbra R, Kerby JD, Minei J, et al.: Early prediction of outcome after severe traumatic brain injury: a simple and practical model. BMC Emerg Med 16:32, 2016

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 24

    Simpson D, Reilly P: Pediatric coma scale. Lancet 2:450, 1982

  • 25

    Steyerberg EW, Mushkudiani N, Perel P, Butcher I, Lu J, McHugh GS, et al.: Predicting outcome after traumatic brain injury: development and international validation of prognostic scores based on admission characteristics. PLoS Med 5:e165, 2008

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 26

    Tatman A, Warren A, Williams A, Powell JE, Whitehouse W: Development of a modified paediatric coma scale in intensive care clinical practice. Arch Dis Child 77:519521, 1997

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 27

    Thurman DJ, Alverson C, Dunn KA, Guerrero J, Sniezek JE: Traumatic brain injury in the United States: a public health perspective. J Head Trauma Rehabil 14:602615, 1999

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 28

    Tien HC, Cunha JR, Wu SN, Chughtai T, Tremblay LN, Brenneman FD, et al.: Do trauma patients with a Glasgow Coma Scale score of 3 and bilateral fixed and dilated pupils have any chance of survival? J Trauma 60:274278, 2006

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 29

    Tude Melo JR, Di Rocco F, Blanot S, Oliveira-Filho J, Roujeau T, Sainte-Rose C, et al.: Mortality in children with severe head trauma: predictive factors and proposal for a new predictive scale. Neurosurgery 67:15421547, 2010

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 30

    Young AM, Guilfoyle MR, Fernandes H, Garnett MR, Agrawal S, Hutchinson PJ: The application of adult traumatic brain injury models in a pediatric cohort. J Neurosurg Pediatr 18:558564, 2016

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation

Metrics

All Time Past Year Past 30 Days
Abstract Views 1845 144 0
Full Text Views 305 58 11
PDF Downloads 278 59 12
EPUB Downloads 0 0 0