Journal of Pediatric Surgery xxx (2016) xxx–xxx

Contents lists available at ScienceDirect

Journal of Pediatric Surgery

journal homepage: www.elsevier.com/locate/jpedsurg

Pediatric patients with traumatic epidural hematoma at low risk for

deterioration and need for surgical treatment☆,☆☆
Brian F Flaherty a,⁎, Hannah E Moore b, Jay Riva-Cambrin c, Susan L Bratton a
a
University of Utah, Department of Pediatrics, Division of Critical Care, Salt Lake City, UT 84108
b
University of Southern California, Keck School of Medicine, Los Angeles, CA 90033
c
University of Calgary, Department of Clinical Neurosciences, Section of Neurosurgery, Alberta Children's Hospital, Calgary, AB T3B 6A8

a r t i c l e i n f o a b s t r a c t

Article history: Background: Although some pediatric patients with small traumatic epidural hematoma (EDH) are observed
Received 24 August 2016 without surgical drainage, clinical practice remains variable.
Accepted 6 September 2016 Objectives: Create a prediction rule to identify patients with EDH unlikely to fail hospital observation.
Available online xxxx Methods: Retrospective review at a level I pediatric trauma center between 2003 and 2014. Presenting clinical
and radiographic features were compared between those successfully to failed observation.
Key words:
Results: Two hundred twenty-two patients with EDH whose initial management strategy was observation were
Epidural hematoma
Head injury
analyzed; 196 (88%) were successfully observed. The group failing observation was more likely to present with
Trauma altered mental status (RR 18.8; 95% CI 8.7–49.6), has larger median bleed thickness (observed = 5.6 mm versus
Pediatric failed observation = 10.9 mm, p b 0.01), median bleed volume (observed = 16.7 ml versus failed observation =
Observation 125.5, p b 0.01), and mass effect (RR 3.7; 95% CI 1.8–7.7). No mass effect, EDH volume b 30 ml, and no neurologic
deficits predicted patients at low risk of failing observation with a positive predictive value of 98% (95% CI
93–99%). There was no difference in median discharge Glasgow outcome scores (5 in both groups, p = 0.20).
Conclusion: Patients with no mass effect and EDH volume b30 ml on initial CT scan and no neurologic deficit are at
low risk of failing observation.
Level of evidence: Retrospective cohort level I.
© 2016 Elsevier Inc. All rights reserved.

Traumatic brain injury (TBI) is a leading cause of injury and death in
children, with epidural hematoma (EDH) complicating approximately
3% of patients admitted with TBI [1]. While classic neurosurgical teach-
ing is that all EDHs should be evacuated immediately, there are numer-
ous case reports of both adult and pediatric patients with EDHs
observed without surgical interventions with good outcomes [2–11].
Adult guidelines have been developed which specify a subset of patients
for whom a trial of observation is recommended: Glasgow coma scale
(GCS) 9–15, EDH thickness of b 15 mm, hemorrhage volume b30 ml, and
midline shift b 5 ml [10]. Despite numerous reports of successful observa-
tion of pediatric patients with EDHs, some as large as 2 cm in thickness,
no similar pediatric guidelines have been developed [4–8,11–16].
The primary aim of this study is to define a population of pediatric pa-
tients at low risk for requiring emergency or urgent surgery for EDH by
☆ Funding: this work was supported by the University of Utah Study Design and Biosta-
tistics Center, with funding in part from the National Center for Research Resources and
the National Center for Advancing Translational Sciences [grant no. 8UL1TR000105].
☆☆ Author contributionStudy conception and design: Brian Flaherty, Hannah Moore, Jay Riva-
Cambrin, Susan Bratton.Data acquisition: Brian Flaherty, Hannah Moore.Analysis and data
interpretation: Brian Flaherty, Susan Bratton.Drafting of the manuscript: Brian Flaherty, Susan
Bratton.Critical revision: Brian Flaherty, Hannah Moore, Jay Riva-Cambrin, Susan Bratton.
⁎ Corresponding author.
E-mail address: brian.flaherty@hsc.utah.edu (B.F. Flaherty).
retrospectively evaluating patients with EDH treated at a large, pediatric
level I trauma center. We hypothesize that patients with a GCS 14–15,
and no radiographic signs of mass effect on presentation will be at low
risk for subsequent deterioration and need for surgical or intensive care
resources.
1. Methods
The study was a retrospective cohort study of patients with traumatic
EDH treated between January 1, 2003 and December 31, 2014 at Primary
Children's Hospital (PCH), an American College of Surgery verified level I
pediatric trauma center located in Salt Lake City, UT. The University of
Utah institutional review board approved this study (IRB# 00076164).
1.1. Patient selection and cohort definition
Patients with EDH were initially identified from the hospital's trau-
ma database. The database is maintained by a dedicated trauma regis-
trar who prospectively records data on all patients with a traumatic
injury who present to the PCH ED or are admitted to the hospital. All pa-
tients age 0–18 with a diagnosis of EDH were evaluated. Exclusion
criteria included patients taken for immediate surgery, missing initial
CT imaging or clinical notes, injury deemed so severe that operative

http://dx.doi.org/10.1016/j.jpedsurg.2016.09.005
0022-3468/© 2016 Elsevier Inc. All rights reserved.

Please cite this article as: Flaherty BF, et al, Pediatric patients with traumatic epidural hematoma at low risk for deterioration and need for surgical
treatment, J Pediatr Surg (2016), http://dx.doi.org/10.1016/j.jpedsurg.2016.09.005
2 B.F. Flaherty et al. / Journal of Pediatric Surgery xxx (2016) xxx–xxx
therapy was not offered because of fatal injury, and patients undergoing
a craniotomy for an indication other than their EDH, such as elevation of
a depressed skull fracture.
Patients who required no surgical intervention were considered the
“observation” group. Patients whose admitting neurosurgery and trau-
ma surgery notes stated the initial EDH management plan was observa-
tion, but subsequently the patients underwent a surgical intervention
for the EDH were considered the “failed observation” group.
1.2. Data extraction
Initial presenting clinical data were abstracted from the emergency
department trauma flowsheet, trauma intake note, and neurosurgery
consult notes. All initial head CTs were reviewed (BF or HM) with
greatest EDH thickness, height, and length measured and volume calcu-
lated by using the formula for an ellipsoid (4/3 × π × thickness × length
× height) as previously described [11,13]. To insure consistency in mea-
sures, BF and HM both reviewed the first 20 scans together and an addi-
tional 20 scans, reviewed separately, were assessed for inter-rater
reliability by calculating intraclass correlation coefficients (ICC) with
95% confidence intervals and by constructing Bland–Altman plots
(Appendix Fig. 1) [17,18]. The ICC for thickness was 0.95 (0.89–0.98),
for height was 0.82 (0.6–0.92), for length was 0.98 (0.94–0.99), and
for volume was 0.96 (0.89–0.98). The presence of mass effect, midline
shift, herniation, and additional injuries (skull fractures, subdural hema-
toma, intraparenchymal bleed, contusion, axonal injury) was based on
the initial pediatric radiologist's clinical reading.
Additional data abstracted from the chart included duration of pedi-
atric intensive care unit (PICU) and hospital stay, discharge neurologic
status, and discharge location. Glasgow outcome score (GOS) was ob-
tained from the trauma registry. For patients failing observation, the in-
dication for surgery (worsening CT, change in neurologic status,
increased ICP) was determined from the neurosurgery operative note.
1.3. Statistical analysis
Summary data are presented as median values with interquartile
range (IQR) and as percentages. Observed patients were compared to
those whose management plan changed from observation to surgical
management after admission to the trauma service. Categorical data
were compared with the chi-squared and Fischer's exact test when
the expected number per cell was ≤5. Rates were compared using the
relative risk (RR) estimate with 95% confidence intervals (CI). For mul-
tiple comparisons, p values were adjusted using the Bonferroni method.
Continuous variables were compared using the Mann–Whitney U test.
All comparisons were 2 sided with a p value of 0.05 used for significance.
Factors that significantly differed in the univariate comparisons were
considered for the prediction rule. The prediction rule was created using
recursive partitioning by selection of the parameter that had the greatest
chi squared value [19]. Each subset of patients was then tested again for
all parameters that differed significantly in the univariate analysis and
the parameter with the greatest chi squared value was again chosen so
long as the differences remained significantly different (p b 0.05) for
each new branch. For continuous variables, the cut-off point used for
the prediction rule was determined from receiver-operator curve (ROC)
analysis. The analyses were performed with R version 3.2.4 (Vienna,
Austria) and SPSS version 24 (IBM, Armonk, NY) [20].
2. Results
2.1. Patient population
Review of the trauma database identified 392 patient with EDH of
which a majority, 292 (75%), underwent initial observation. As outlined
in Fig. 1, after applying exclusion criteria, 222 patients underwent anal-
ysis. One hundred ninety-six patients (88%) were successfully observed,
Please cite this article as: Flaherty BF, et al, Pediatric patients with traumatic epidural hematoma at low risk for deterioration and need for surgical
treatment, J Pediatr Surg (2016), http://dx.doi.org/10.1016/j.jpedsurg.2016.09.005
while 26 (12%) of the patients failed initial observation management and
required surgery at a median time of 33 h (IQR 14.9–49.2) from injury.
Based on neurosurgery operatives notes, 17 (65%) of the patients were
taken to surgery for combined worsening CT changes and neurologic
symptoms, 6 (23%) patients were taken based on neurologic symptoms
only, 2 (8%) patients based on CT changes only, and 1 (4%) patient
based on CT changes and increasing intracranial pressure reading. Basic
demographics of the cohort are summarized in Table 1. Those who failed
observation had a significantly older median age and had a higher medi-
an injury severity scores (ISS) compared to those successfully observed.
The median time to hospital arrival was similar in both groups (observa-
tion = 5.5 h versus failed observation = 4.4 h, p value = 0.44).
2.2. Clinical characteristics
Table 2 compares presenting clinical features of the patients who
were successfully observed to those who failed observation. Those
who failed observation were significantly more likely to have altered
mental status when arriving in the ED (RR 18.8, 95% CI 8.7–49.6).
2.3. Radiographic characteristics
The median time to first CT scan was also similar in both groups
(observation = 3.6 h versus failed observation = 2.2 h, p = 0.39).
Table 3 compares the two groups by radiographic features. Factors relat-
ed to EDH size were significantly greater in the failed observation
groups. Mass effect was also significantly more common among those
who failed observation (RR 3.7, 95% CI 1.8–7.7).
2.4. Outcomes
A majority of patients were discharged with Glasgow outcome
scores of 5 (observation = 93% versus failed observation = 84%, p
value = 0.2) and no neurologic deficits (observation = 3% versus failed
observation = 4%, p value = 0.70). Table 4 compares hospital and dis-
charge outcomes by study group. The PICU and total hospital length of
stay was significantly longer for patients in the failed observation
group. The only death was in the failed observation group. The lone
death was a pedestrian hit by a car who presented with a GCS of 4
after a prolonged on scene cardiac arrest. The patient was taken to the
OR after increased ICP was noted and repeat CT imaging showed an
epidural of 15 mm in thickness. Because of diffuse anoxic brain injury,
the family later elected to withdraw care.
2.5. Prediction of patients at low risk for failing observation
Using recursive partitioning, we found that the absence of mass
effect and EDH volume b30 ml on initial CT scan and the absence of neu-
rologic deficits identified a group of patients at low risk for deterioration
and need for surgery. Fig. 2 summarizes this prediction rule. When an-
alyzing the ability of this rule to detect successful observation, it yielded
a positive predictive value of 98% (95% CI = 93–99%), negative predic-
tive value of 25% (95% CI = 16–35%), specificity of 88% (95% CI =
68–97%) and sensitivity of 64% (95% CI = 57–70%).
3. Discussion
Our goal was to identify early clinical and/or radiographic features to
identify patients with EDH at low risk for deteriorating and failing ob-
servation. We found that the presence of altered mental status, EDH
size, the presence of mass effect, and the presence of midline shift
were significantly different between the group that was successfully
observed and the group that failed. Based on our recursive
partitioning analysis, a combination of the absence of mass effect,
EDH volume b30 mL and no neurologic deficits predicted those at
low risk for deterioration and observation failure.
 B.F. Flaherty et al. / Journal of Pediatric Surgery xxx (2016) xxx–xxx 3

Fig. 1. Assignment of patients with traumatic EDH to “observed” and “failed observation” groups.

The makeup of our cohort is similar to prior reports regarding the
care of acute EDH in children with respect to age, mechanism of injury,
and presenting clinical symptoms. Similar to other reports of injured
children, we found falls to be the most common mechanism of injury
and those patients undergoing observation generally presented with
GCS of 14 or higher [7,9,11,13,21]. The proportion of patients undergo-
ing trial of observation in our study was also similar to prior reports,
with studies in the last decade noting initial observation rates ranging
from 60 to 83%. Our rate of observation failure of 12% was also similar
to prior studies with failure rate ranging from 5 to 12%
[5,11,12,15,21,22]. This similarity to numerous other cohorts of pediat-
ric EDH speaks to the generalizability of our results.
With respect to developing treatment guidelines for pediatric EDH,
two prior studies have attempted to develop standards to guide the
management of pediatric EDH. Both studies compared patients who
were successfully observed to those who underwent immediate surgery
[11,13]. Likely because of small sample size and no further validation
studies, the findings of these studies have not altered EDH management.
Further, both studies compared observed versus initial surgery groups, a

Table 1
Patient demographics and mechanism of injury. Table 2
Comparison of presenting clinical symptoms between observed and failed observation
Failed groups.
Observed observation
N = 196 N = 26 P-value Observed Failed observation
N = 196 N = 26 P-value
Median age in years (IQR) 6 (2–10) 8 (6–10) 0.03
Male, n (%) 108 (55) 16 (62) 0.60 Loss of consciousness, n (%) 0.52
Median ISS (IQR) 16 (9–17) 25 (16–25) b0.01 ≥1 min 14 (7) 3 (12)
Median time to presentation, hours b1 min 12 (6) 1 (4)
(IQR) 5.5 (2.7–20) 4.4 (2.6–13.2) 0.31 Unknown duration 26 (13) 4 (15)
Mechanism of injury, n (%) 0.78 None 110 (56) 11 (42)
Fall from standing height or higher 111 (57) 14 (54) Unknown 34 (17) 7 (27)
Fall from less than standing height 16 (8) 3 (12) Median arrival GCS (IQR) 15 (14–15) 15 (15–15) 0.09
Fall of unknown height 8 (4) 0 (0) Neurologic deficit, n (%) 22 (1) 10 (38) b0.01
Bicycle fall 20 (10) 4 (15) Altered mental status, n (%) 19 (10) 9 (35) b0.01
Bicycle versus MV 1 (b1) 0 (0) Motor deficit, n (%) 1 (1) 0 (0) N0.99
Pedestrian versus MV 6 (3) 2 (8) Sensory deficit, n (%) 1 (1) 0 (0) N0.99
MVA less than 60 mph 1 (b1) 0 (0) Pupillary deficit, n (%) 4 (2) 2 (4) 0.15
MVA 60 mph or faster 5 (3) 3 (12) Cranial nerve deficit, n (%) 1 (1) 0 (0) N0.99
MVA unknown speed 5 (3) 0 (0) Vomiting, n (%) 80 (41) 16 (62) 0.45
Sports related 3 (2) 3 (2) Headache, n (%) 35 (18) 9 (35) 0.44
Assault 10 (5) 0 (0) Irritability, n (%) 29 (15) 0 (0) 0.03
Unknown 10 (5) 0 (0) Vision changes, n (%) 4 (2) 0 (0) N0.99
Seizures, n (%) 8 (4) 8 (4) N0.99
IQR = interquartile range, ISS = injury severity score, MV = motor vehicle,
MVA = motor vehicle accident. GCS = Glasgow coma score, IQR = interquartile range.

Please cite this article as: Flaherty BF, et al, Pediatric patients with traumatic epidural hematoma at low risk for deterioration and need for surgical
treatment, J Pediatr Surg (2016), http://dx.doi.org/10.1016/j.jpedsurg.2016.09.005
4 B.F. Flaherty et al. / Journal of Pediatric Surgery xxx (2016) xxx–xxx

Table 3
Comparison of initial CT findings between observed and failed observation groups.

Observed Failed observation

N = 196 N = 26 P-value

Time to initial CT scan, hours

(IQR) 3.6 (1.6–17.6) 2.2 (1.6–11.7) 0.39
EDH location, n (%) 0.34
Frontal 49 (25) 7 (27)
Parietal 104 (53) 14 (54)
Temporal 38 (19) 12 (46)
Occipital 14 (7) 2 (8)
Posterior fossa 15 (8) 3 (12)
Median EDH thickness, mm (IQR) 5.6 (4.1–8.5) 10.9 (7.2–13.9) b0.01
Median EDH volume, mL (IQR) 16.7 (5.2–43.5) 125.5 (22.5–207.7) b0.01
Presence of mass effect, n (%) 51 (26) 16 (62) b0.01
Median midline shift, mm (IQR) 0 (0) 0 (0–1) b0.01
Herniation, n (%) 3 (2) 2 (8) 0.11
Skull fracture, n (%) 175 (89) 22 (85) 0.51
Additional intracranial injury, n
(%) 64 (33) 14 (54) 0.33
Skull fracture, n (%) 175 (89) 22 (85) 0.76

IQR = interquartile range.

decision which not only reflects the patient signs and symptoms and
initial radiology findings, but also the subjective surgical opinion. As
such, there were likely many patients undergoing surgery who may
have done well with just observation. In contrast to these studies, we
purposely developed a larger cohort and compared patients who were
successfully observed to those that failed observation based on more Fig. 2. Prediction rule for patients with EDH at low risk for failing observation.

objective changes in status. This provides a comparison of patients

who could be observed versus those who more likely needed surgery.
Importantly, we were also able to describe the outcomes of a large need surgery. As indicated by the negative predictive value, a number of
cohort of patients with EDH who failed observation. We found that patients not meeting all 3 low risk criteria still succeeded in observation.
most patients failed observation because of worsening neurologic and As such, not meeting the low risk criteria should not be considered an
CT findings in a median time of 33 h. Despite failing observation, how- indication for surgery.
ever, these patients have good clinical outcomes similar to that of the Our study has some limitations. Because of its retrospective nature,
successfully observed group. When considering observation therapy there is the potential that some clinical features were not captured in
for patients with EDH it is critical to know that potentially delaying sur- the medical record and, therefore, not analyzed. If these features could
gery in those that fail observation does not lead to worse neurologic out- predict treatment strategy success, this would have an impact our pre-
comes. As the other studies to comment on the outcomes of failed diction rule. Approximately 20% of the cohort was excluded because of
observation involved only 1–3 patients, our study provides more com- missing CT scans. We defined the presence of “mass effect” on CT scan
pelling evidence that failing observation in a monitored level I trauma based on the initial radiologist's clinical reading. At our institution, CT
center with immediate access to pediatric radiology and neurosurgical scans were read by pediatric radiologists and in recent years by pediat-
capabilities is unlikely to lead to poor outcomes [11,12,15]. ric neuroradiologist and centers with less experience may not read mass
Taken as a whole, our findings have important implications for the effect with similar accuracy. Additionally, we were limited in our ability
treatment of pediatric EDH. Our risk stratification and outcome data to collect follow-up data to ensure no late presentation of worsening
identify a group of patients who can safely have observation in the hos- EDH symptoms and need for late intervention. However, as the only pe-
pital with 24 h neurosurgical availability as their initial treatment strat- diatric trauma and neurosurgery center for a several hundred mile radi-
egy. Given their low risk of deterioration, early de-escalation of the level us, it is extremely unlikely that any patient requiring late intervention
of care or avoidance of the PICU may be appropriate and worthy of fu- would have been cared for at another institution. Finally, the prediction
ture study. It is important to note that the study was designed to identify rule has not yet been validated on an external cohort, which is a point of
patients at low risk of deterioration and not to identify patients likely to future study. Recursive partitioning can overfit data making a validation
group all the more important.
Table 4
Comparison of lengths of stay and outcomes between observed and failed observation
groups. 4. Conclusions
Observed Failed observation
N = 196 N = 26 P-value A majority of patients with EDH can be safely observed. A lack of
mass effect and EDH volume b 30 ml on initial CT scan and no neurologic
Median ICU LOS, days (IQR) 0.7 (0.6–0.9) 1.7 (1.4–2.5) b0.01
Median LOS (IQR) 1.7 (0.9–2.8) 3.7 (2.6–5) b0.01 deficits on presentation predicts a group of patients at low risk for dete-
Median discharge GCS (IQR) 15 (15–15) 15 (15–15) 0.99 riorating and requiring surgery. Patients who fail observation appear to
Glasgow outcome score, n (%) 0.23 have similar, good outcomes compared to those who do not fail.
5 138 (93) 16 (84)
4 8 (5) 3 (16)
3 2 (1) 0
b3 0 0 Acknowledgements
Residual neurologic deficit, n (%) 5 (3) 1 (4) 0.53
Death 0 1 0.12 Katrina Blacker FNP, Kathryn Swenson PA-C, Jennifer Harvey PNP for
LOS = length of stay, GCS = Glasgow coma score, IQR = interquartile range. assistance in data entry.

Please cite this article as: Flaherty BF, et al, Pediatric patients with traumatic epidural hematoma at low risk for deterioration and need for surgical
treatment, J Pediatr Surg (2016), http://dx.doi.org/10.1016/j.jpedsurg.2016.09.005
 B.F. Flaherty et al. / Journal of Pediatric Surgery xxx (2016) xxx–xxx
Appendix A. Bland–Altman Plots of EDH Measures.
References
[1] Faul M, Xu L, Wald MM, et al. Traumatic brain injury in the United States: emergency
department visit, hospitalizations and deaths 2002–2006. Atlanta (GA): Centers for Dis-
ease Control and Prevention, National Center for Injury Prevention and Control; 2010.
[2] Munro D, Maltby GL. Extradural hemorrhage: a study of forty-four cases. Ann Surg
1941;113:192–203.
[3] Bullock R, Smith RM, van Dellen JR. Nonoperative management of extradural hema-
toma. Neurosurgery 1985;16:602–6.
[4] Duthie G, Reaper J, Tyagi A, et al. Extradural haematomas in children: a 10-year review.
Br J Neurosurg 2009;23:596–600. http://dx.doi.org/10.3109/02688690902978157.
[5] Irie F, Le Brocque R, Kenardy J, et al. Epidemiology of traumatic epidural hematoma in
young age. J Trauma 2011;71:847–53. http://dx.doi.org/10.1097/TA.0b013e3182032c9a.
[6] Knuckey NW, Gelbard S, Epstein MH. The management of “asymptomatic” epidural
hematomas. A prospective study. J Neurosurg 1989;70:392–6. http://dx.doi.org/10.
3171/jns.1989.70.3.0392.
[7] Paiva WS, de Andrade AF, Mathias Júnior L, et al. Management of supratentorial epidu-
ral hematoma in children: report on 49 patients. Arq Neuropsiquiatr 2010;68:888–92.
Please cite this article as: Flaherty BF, et al, Pediatric patients with traumatic epidural hematoma at low risk for deterioration and need for surgical
treatment, J Pediatr Surg (2016), http://dx.doi.org/10.1016/j.jpedsurg.2016.09.005
5
[8] Pang D, Horton JA, Herron JM, et al. Nonsurgical management of extradural hematomas in
children. J Neurosurg 1983;59:958–71. http://dx.doi.org/10.3171/jns.1983.59.6.0958.
[9] Rocchi G, Caroli E, Raco A, et al. Traumatic epidural hematoma in children. J Child
Neurol 2005;20:569–72.
[10] Bullock MR, Chesnut R, Ghajar J, et al. Surgical management of acute epidural hemato-
mas. Neurosurgery 2006;58:S7–S15. http://dx.doi.org/10.1227/01.NEU.0000210363.
91172.A8.
[11] Flaherty BF, Loya J, Alexander MD, et al. Utility of clinical and radiographic findings
in the management of traumatic epidural hematoma. Pediatr Neurosurg 2014.
http://dx.doi.org/10.1159/000363143.
[12] Balmer B, Boltshauser E, Altermatt S, et al. Conservative management of significant
epidural haematomas in children. Childs Nerv Syst 2006;22:363–7. http://dx.doi.
org/10.1007/s00381-005-1254-x.
[13] Bejjani GK, Donahue DJ, Rusin J, et al. Radiological and clinical criteria for the
management of epidural hematomas in children. Pediatr Neurosurg 1996;25:
302–8.
[14] Jamous MA, Abdel Aziz H, Al Kaisy F, et al. Conservative management of acute epidu-
ral hematoma in a pediatric age group. Pediatr Neurosurg 2009;45:181–4. http://dx.
doi.org/10.1159/000218200.
6 B.F. Flaherty et al. / Journal of Pediatric Surgery xxx (2016) xxx–xxx

[15] Khan MB, Riaz M, Javed G. Conservative management of significant supratentorial
epidural hematomas in pediatric patients. Childs Nerv Syst 2014;30:1249–53.
http://dx.doi.org/10.1007/s00381-014-2391-x.
[16] Kochanek PM, Carney N, Adelson PD, et al. Guidelines for the acute medical manage-
ment of severe traumatic brain injury in infants, children, and adolescents – second
edition. Pediatr Crit Care Med 2012;13(Suppl. 1):S1–82. http://dx.doi.org/10.1097/
PCC.0b013e31823f435c.
[17] Rankin G, Stokes M. Reliability of assessment tools in rehabilitation: an illustration of
appropriate statistical analyses. Clin Rehabil 1998;12:187–99.
[18] Giavarina D. Understanding Bland Altman analysis. Biochem Med 2015;25:141–51.
http://dx.doi.org/10.11613/BM.2015.015.
[19] Strobl C, Malley J, Tutz G. An introduction to recursive partitioning: rationale, appli-
cation, and characteristics of classification and regression trees, bagging, and ran-
dom forests. Psychol Methods 2009;14:323–48. http://dx.doi.org/10.1037/
a0016973.
[20] R Core Team. R: a language and environment for statistical computing. R Foundation
for Statistical Computing; 2016[http://www.r-project.org].
[21] Hardtke Teichert J, Rosales PR, Blanco Lopes P, et al. Extradural hematoma in chil-
dren: case series of 33 patients. Pediatr Neurosurg 2013;48:216–20. http://dx.doi.
org/10.1159/000345849.
[22] Schutzman SA, Barnes PD, Mantello M, et al. Epidural hematomas in children. Ann
Emerg Med 1993;22:535–41.

Please cite this article as: Flaherty BF, et al, Pediatric patients with traumatic epidural hematoma at low risk for deterioration and need for surgical
treatment, J Pediatr Surg (2016), http://dx.doi.org/10.1016/j.jpedsurg.2016.09.005