clinical article

J Neurosurg Pediatr 17:19–26, 2016

Standardizing ICU management of pediatric

traumatic brain injury is associated with improved
outcomes at discharge
Thomas M. O’Lynnger, MD, MPH,1 Chevis N. Shannon, MPH, MBA, DrPH,1 Truc M. Le, MD,2
Amber Greeno, NP,3 Dai Chung, MD,3 Fred S. Lamb, MD, PhD,2 and John C. Wellons III, MD, MSPH1
Departments of 1Neurological Surgery, 2Pediatrics, Division of Pediatric Critical Care Medicine, and 3Pediatric Surgery, Vanderbilt
University Medical Center, Nashville, Tennessee

Objective  The goal of critical care in treating traumatic brain injury (TBI) is to reduce secondary brain injury by limit-
ing cerebral ischemia and optimizing cerebral blood flow. The authors compared short-term outcomes as defined by
discharge disposition and Glasgow Outcome Scale scores in children with TBI before and after the implementation of a
protocol that standardized decision-making and interventions among neurosurgeons and pediatric intensivists.
Methods  The authors performed a retrospective pre- and postprotocol study of 128 pediatric patients with severe TBI,
as defined by Glasgow Coma Scale (GCS) scores < 8, admitted to a tertiary care center pediatric critical care unit be-
tween April 1, 2008, and May 31, 2014. The preprotocol group included 99 patients, and the postprotocol group included
29 patients. The primary outcome of interest was discharge disposition before and after protocol implementation, which
took place on April 1, 2013. Ordered logistic regression was used to assess outcomes while accounting for injury sever-
ity and clinical parameters. Favorable discharge disposition included discharge home. Unfavorable discharge disposition
included discharge to an inpatient facility or death.
Results  Demographics were similar between the treatment periods, as was injury severity as assessed by GCS
score (mean 5.43 preprotocol, mean 5.28 postprotocol; p = 0.67). The ordered logistic regression model demonstrated
an odds ratio of 4.0 of increasingly favorable outcome in the postprotocol cohort (p = 0.007). Prior to protocol implemen-
tation, 63 patients (64%) had unfavorable discharge disposition and 36 patients (36%) had favorable discharge disposi-
tion. After protocol implementation, 9 patients (31%) had unfavorable disposition, while 20 patients (69%) had favorable
disposition (p = 0.002). In the preprotocol group, 31 patients (31%) died while 6 patients (21%) died after protocol imple-
mentation (p = 0.04).
Conclusions  Discharge disposition and mortality rates in pediatric patients with severe TBI improved after imple-
mentation of a standardized protocol among caregivers based on best-practice guidelines.
http://thejns.org/doi/abs/10.3171/2015.5.PEDS1544
Key Words  critical care; discharge; trauma; traumatic brain injury; protocol

T
raumatic brain injury (TBI) is a heterogeneous con-
dition characterized by marked variability in etiol-
ogy and treatment.4,5,10,26 There have been numerous
studies on the treatment of TBI in the adult population,4,6,
8,9,11,15,16,24,25,29
but less research has been performed on
treating TBI in pediatric patients.2,5,7,18,21,23 These patients’
still-maturing CNS responds differently to injury and the
current treatments available, making it imperative to deter-
mine the best course of action to improve outcomes in this
population.5,16,18,22
The 2003 Brain Trauma Foundation guidelines, 5 which
were most recently updated in 2012,14 summarized prac-
tice standards for treatment of severe TBI in children. The
overarching goal of critical care in treating TBI is to reduce
secondary brain injury by limiting cerebral ischemia and
optimizing cerebral blood flow.5 Despite these evidence-
based guidelines, there is considerable variability in how
different physicians and institutions treat severe TBI, and
the strength of the evidence is low.10
A study by Pineda et al. in 2013 showed significant ben-

Abbreviations  CPP = cerebral perfusion pressure; GCS = Glasgow Coma Scale; GOS = Glasgow Outcome Scale; ICP = intracranial pressure; OR = odds ratio; PICU
= pediatric intensive care unit; TBI = traumatic brain injury.
submitted  January 22, 2015.  accepted  May 11, 2015.
include when citing  Published online October 9, 2015; DOI: 10.3171/2015.5.PEDS1544.

©AANS, 2016 J Neurosurg Pediatr  Volume 17 • January 2016 19

T. M. O’Lynnger et al.
efit to the implementation of a neurocritical care program
with standardized treatment of severe TBI using a proto-
col based on the 2003 guidelines.5,21 The authors analyzed
discharge disposition in pediatric patients with severe TBI
at St. Louis Children’s Hospital before and after imple-
mentation of a pediatric neurocritical care program. The
protocol was designed to facilitate communication among
specialists and to define a plan for monitoring and treat-
ment of children with severe TBI. The protocol was in-
stituted in 2005, and Pineda et al.’s retrospective cohort
study looked at short-term outcomes in patients from 1999
to 2012, comparing preprotocol and postprotocol peri-
ods. The authors found that, after protocol implementa-
tion, patients had a 67% favorable disposition, defined as
home with or without therapy, compared with 48% before
protocol implementation. An ordinal regression model
indicated that outcomes improved across the spectrum
of discharge disposition status and Glasgow Coma Scale
(GCS) scores after protocol implementation. Notably, a
controlled trial by Chesnut et al. from 2012 demonstrated
no significant survival benefit when intracranial pres-
sure (ICP)–focused critical care management was used
in adolescent and adult patients compared with imaging
and clinical examination–based management alone.6,17,20,27
Given these conflicting results, further research into criti-
cal care protocols in TBI is necessary.
We conducted a study to evaluate short-term outcomes
in children with TBI after the implementation of a pro-
tocol that standardized decision-making and intervention
among neurosurgeons and intensivists (Fig. 1). Our study
had 2 a priori aims: 1) compare pre- and postprotocol dis-
charge disposition, and 2) determine how Glasgow Out-
come Scale (GOS) scores differed between the pre- and
poststandardization cohorts.
Methods
For this retrospective cohort study, we used data from
the prospectively maintained Vanderbilt University Medi-
cal Center pediatric trauma registry. We included patients
less than 18 years of age presenting with TBI with a GCS
score of less than 8 between April 1, 2008, and May 31,
2014. All patient electronic medical records with radio-
graphic evidence of TBI were reviewed for this study, and
the GCS score used for determination of enrollment was
based on the examination by the neurosurgery team after
resuscitation. This process helped to limit those patients
classified as having an artificially low GCS score upon
initial emergency department assessment secondary to
sedating medications. Variables not maintained prospec-
tively were extracted from the electronic medical record,
including ICP monitoring and hyperosmolar therapy. Pa-
tients were followed for the length of their hospitalization.
Those who died in the emergency department were ex-
cluded from the study. The Vanderbilt institutional review
board approved the study protocol.
Chart review was used to extract parameters, includ-
ing age, sex, race, GCS score after resuscitation, need for
surgery, injury type and mechanism, discharge disposi-
tion, GOS score at discharge, pediatric intensive care unit
(PICU) length of stay, total length of stay, use of hyper-
20 J Neurosurg Pediatr  Volume 17 • January 2016
osmolar therapy, use of barbiturates, and ICP monitoring.
Due to the complexity of each patient’s clinical narrative
and associated multiplicity of variables generated, pro-
tocol adherence was challenging to assess. However, use
of 3% hypertonic saline instead of mannitol was identi-
fied as a surrogate measure of adherence given the 2012
guidelines’ focus on 3% NaCl as treatment for elevated
ICP and its consistent availability in the medical record.
Any usage of mannitol was considered a protocol devia-
tion. Strict usage of solely 3% NaCl for elevated ICP was
considered consistent with protocol requirements. Due to
the real-time and rapid nature of patient care, we found
that compliance for cerebral perfusion pressure (CPP) was
challenging to adequately categorize during patients’ hos-
pitalization, and it was not clear whether these data were
an accurate representation of the actual clinic course.
Therefore, this parameter was not used as a surrogate
measure of adherence.
Protocol Implementation
The TBI protocol was implemented on April 1, 2013.
Prior to 2013, no specific multidisciplinary protocols
were used at the institution in the management of pedi-
atric brain injury. Variation in the overall management
paradigm for patients with severe TBI was common. The
evidence-based protocol was developed by a multidisci-
plinary group composed of local experts from pediatric
services, including neurosurgery, critical care medicine,
trauma surgery, and emergency medicine. Source mate-
rial that was reviewed included the 2003 Brain Trauma
Foundation guidelines5 and the 2012 update14 and insti-
tutional protocols from St. Louis Children’s Hospital and
Children’s of Alabama. The standardized clinical protocol
was devised to guide medical therapy in a stepwise fash-
ion, with an emphasis on maintaining CPP and ICP within
strict parameters to reduce secondary brain injury by op-
timizing cerebral blood flow. The tiered approach to ther-
apy focuses first on optimization of oxygen delivery and
cerebral perfusion, secondly on CSF diversion, thirdly on
maintenance of adequate sedation/analgesia, and finally
on maximization of hyperosmolar therapy prior to pro-
gression to second-tier therapies, as defined by the 2012
Brain Trauma Foundation guidelines.14 Each intervention
step is followed by immediate reevaluation to determine
efficacy and need for further escalation. When developing
the protocol, it was determined that simplifying the algo-
rithm to a single hyperosmolar therapy would decrease
variation in care. The protocol used hypertonic saline, so
this was used as a surrogate adherence measure. In addi-
tion, the 2012 Brain Trauma Foundation guidelines note
Level II and III evidence for the use of hypertonic saline.
There were no studies about the use of mannitol that met
inclusion criteria.
Outcomes of Interest
The primary outcome was discharge disposition, cat-
egorized as discharge home, discharge to rehabilitation,
or death. A single patient was transferred to another acute
care facility per the family’s request. This patient was
considered to have a “rehabilitation” discharge disposition
to maintain the model with only 3 discharge categories.
 Standardizing ICU management of pediatric traumatic brain injury

Fig. 1. Traumatic brain injury protocol. CVP = central venous pressure; EEG = electroencephalogram; ETCO2 = end-tidal CO2;
EVD = external ventricular drain; Hct = hematocrit; HOB = head of bed; MAP = mean arterial pressure; Na = sodium; NMB =
neuromuscular blockade; NS = normal saline; NSGY = neurosurgery; pCO2 = partial pressure of CO2; TP = transpyloric.

Our secondary outcome was GOS score at discharge.1 We
did not use the extended scale, because we did not believe
that the retrospective nature of this analysis was sensitive
enough to adequately reproduce the scale in a meaningful
way.
Statistical Analysis
No trends in outcomes were detected in the 5 years of
the preprotocol cohort; as such, this group was analyzed
as a whole versus the postprotocol cohort. Study outcomes
measured before and after TBI protocol implementation
were compared. Mean age, GCS score, and length of stay
between cohorts were compared using the Wilcoxon rank-
sum test. Need for surgery, the various injury types and
mechanisms, ICP monitor placement, barbiturate use, and
hyperosmolar use were compared using the chi-square test
and Fisher’s exact test where appropriate. Discharge dis-
position and GOS scores were compared using the Krus-
kal-Wallis test. Individual groups within discharge dispo-
sition were compared using the Wilcoxon rank-sum test.
To compare study outcomes before and after TBI pro-
tocol implementation while accounting for potential con-

J Neurosurg Pediatr  Volume 17 • January 2016 21

T. M. O’Lynnger et al.

founders, we used multivariate ordered logistic regres- TABLE 1. Patient characteristics: pre- and postprotocol
sion. Variables were determined a priori based on clinical comparison*
significance and perceived importance. These variables p
were pre- and postprotocol status, GCS score after resus- Variable Preprotocol Postprotocol Value
citation, age, ICP monitor placement, and PICU length of
stay. The number of parameters was limited to 5 to pre- No. of patients 99 29
vent overfitting the model. In the model, odds ratios (ORs) Demographics
greater than 1 were associated with increasingly favorable   Mean age in yrs (SD) 6.54 (5.41) 5.89 (6.03) 0.239
discharge disposition. Ordered logistic regression was also   Sex, male 52 (53) 16 (55) 0.802
used to predict discharge disposition based on GCS score
  Race, white 78 (79) 25 (86) 0.375
after resuscitation across the spectrum of TBI severity and
to create a plot comparing trends. Statistical significance Severity of injury
was set a priori at p < 0.05, and the analysis was conducted   Mean GCS score (SD) 5.43 (1.73) 5.28 (1.85) 0.671
using Stata statistical software (version 13, StataCorp).   Neurosurgical operation 20 (20) 7 (32) 0.648
  Subdural hemorrhage 33 (33) 17 (61) 0.009
Results   Epidural hemorrhage 6 (6) 3 (11) 0.412
A total of 128 patients (preprotocol n = 99, postpro- Injury mechanism
tocol n = 29) were included in the study. Table 1 shows  MVC 35 (35) 5 (17) 0.064
demographics, injury severity, injury mechanism, length  Pedestrian 7 (7) 1 (3) 0.682
of stay, and ICP treatment parameters. Baseline demo-  Fall 14 (14) 4 (14) 0.999
graphics were not significantly different in the pre- and   Abusive head trauma 20 (20) 14 (48) 0.003
postprotocol groups. Injury severity as assessed by initial
GCS was similar, with a mean of 5.43 in the preprotocol  Other 23 (23) 5 (17) 0.493
cohort versus 5.28 in the postprotocol group (p = 0.671). Length of stay in days
Rates of patients requiring a neurosurgical operation were   Mean PICU stay (SD) 7.0 (6.5) 6.1 (4.9) 0.986
not significantly different (20% preprotocol vs 32% post-   Mean hospital stay (SD) 12.3 (14.6) 10.6 (9.5) 0.871
protocol; p = 0.648). A significant difference existed in the ICP Treatment
number of patients presenting with subdural hemorrhage   ICP monitor placement 46 (46) 8 (28) 0.070
as the predominant radiographic finding (33% preprotocol
vs 61% postprotocol; p = 0.009). The injury mechanism   Hyperosmolar (only 3% NaCl) 22 (22) 12 (41) 0.040
was similar, except there was significantly more abusive   Mannitol w/wo 3% NaCl 48 (48) 4 (14) 0.001
head trauma in the postprotocol group (48% vs 20%, p  Barbiturates 16 (16) 6 (21) 0.680
= 0.003) and a trend toward fewer motor vehicle colli- MVC = motor vehicle collision.
sions (35% vs 17%, p = 0.064). Length of stay was not *  Data are shown as number and percentage (%) unless otherwise indicated.
significantly different between the pre- and postprotocol
cohorts. Mean PICU length of stay and overall length of
stay did not differ between cohorts (p = 0.986 and p = ity placement) and 36 patients (36%) had favorable dis-
0.871, respectively). Before protocol implementation, ICP charge disposition (classified as discharge home). After
monitors were placed in 46% of patients compared with protocol implementation, 9 patients (31%) had unfavorable
28% of patients after protocol implementation (p = 0.07). disposition while 20 patients (69%) had favorable disposi-
Use of barbiturates was similar before and after protocol tion (p = 0.002). The number of deaths was significantly
implementation (16% vs 21%; p = 0.68). decreased, as were overall unfavorable outcomes. In the
preprotocol group, 31 patients (31%) died while 6 patients
Protocol Adherence (21%) died after protocol implementation (p = 0.041). GOS
Protocol adherence was difficult to assess based on ex- scores were generally improved after the protocol was ini-
isting electronic medical records, but use of 3% hyperton- tiated, but this difference was not statistically significant
ic saline over mannitol was used as a surrogate measure. (p = 0.124).
The protocol calls for preferential use of 3% hypertonic
saline for treatment of elevated ICP. After protocol imple- Regression Models
mentation, hypertonic saline was given to a greater per- A proportional odds ordered logistic regression model
centage of patients (22% vs 41%, p = 0.04) and mannitol of discharge disposition revealed that an improvement in
was administered to a smaller percentage (48% vs. 14%, outcomes was associated with being in the postprotocol
p = 0.001). implementation group and increasing GCS score (Table
3). Treatment in the postprotocol implementation group
Short-Term Outcomes was associated with an OR of 4.046 (p = 0.007) of increas-
Table 2 shows pre- and postprotocol outcomes. In unad- ingly favorable outcomes. GCS score was associated with
justed bivariate analysis, discharge disposition improved an OR of 1.844 (p < 0.001). ICP monitor placement itself
significantly after protocol implementation. Prior to pro- was associated with worsening categorical outcome, with
tocol implementation, 63 patients (64%) had unfavorable an OR of 0.206 (p < 0.001). Increasing PICU length of stay
discharge disposition (classified as death or inpatient facil- was associated with increasingly favorable outcomes, but

22 J Neurosurg Pediatr  Volume 17 • January 2016

 Standardizing ICU management of pediatric traumatic brain injury

TABLE 2. Pre- and postprotocol outcomes* TABLE 3. Ordered logistic regression

Variable Preprotocol Postprotocol p Value Variable OR 95% CI p Value
Discharge Disposition Group (pre- vs postprotocol) 4.046 1.476–11.087 0.007
 Death 31 (31) 6 (21) 0.041† GCS score 1.844 1.469–2.315 < 0.001
 Rehabilitation 32 (32) 3 (10) Age (yrs) 1.042 0.974–1.114 0.232
 Home 36 (36) 20 (69) 0.017‡ ICP monitor placement 0.206 0.084–0.501 < 0.001
  Unfavorable (death/rehab) 63 (64) 9 (31) PICU length of stay (days) 1.080 1.001–1.158 0.032
  Favorable (home) 36 (36) 20 (69) 0.002 CI = confidence interval.
GOS (score)
  1 (death) 31 (31) 6 (21) Our results corroborate the findings of Pineda et al.
  2 (vegetative) 3 (3) 0 (0) and other groups that instituted standardized ICU care
  3 (severe disability) 10 (10) 3 (10) based on guidelines.11,20,21,29 Cuschieri et al. showed in
  4 (moderate disability) 19 (19) 5 (17) 2012 that disease-focused implementation of standard op-
  5 (good recovery) 36 (36) 15 (52) 0.124§
erating procedures improves outcomes.9 Additionally, a
2014 study by Vavilala et al. demonstrated improvement
  Unfavorable (1–3) 44 (44) 9 (31) in mortality and discharge GOS score with increasing ad-
  Favorable (4–5) 55 (56) 20 (69) 0.197 herence to clinical indicators derived from the 2012 Brain
*All data shown as number (%) unless otherwise indicated. Trauma Foundation guidelines,31 including maintenance
†  Difference between death and home. of CPP greater than 40 mm Hg and early start of nutrition.
‡  Difference among home, rehab, and death. Our results build upon these findings, further reinforcing
§  Difference among all GOS scores. the evidence that guideline-based care can improve out-
comes in pediatric patients with severe TBI.
the OR close to 1 reveals this association to be clinically Our primary outcome of interest was discharge disposi-
immaterial. Age was not associated with outcome in the tion, which was significantly improved in the postprotocol
model. Given that GOS scores were not statistically sig- cohort. With regard to discharge status, the neurosurgery,
nificantly improved in the postprotocol group, an ordered critical care, and trauma surgery teams worked with the
logistic regression model did not reveal a significant posi- physical therapists and case management teams to deter-
tive relationship with postprotocol status. mine the optimal environment for discharge. There was
Predicted outcomes improved across the range of GCS no intentional influence by providers to have patients dis-
scores based on an ordered logistic regression model. charged home versus to rehabilitation. In addition, because
Figure 2 demonstrates a dramatic shift in the probability cohorts were roughly similar, we anticipate that unmea-
of discharge home in the postprotocol group, as well as sured factors such as family preference and availability of
a marked decline in the probability of death postproto- outpatient therapy services would look similar between
col. This remained consistent across the spectrum of GCS cohorts, as well. One major benefit of using a 5-year pre-
scores studied. protocol for comparison was that there was no general-
ized trend toward discharge disposition status noted prior
to protocol implementation. There was a trend toward
Discussion improvement in the secondary outcome GOS scores, but
Treatment in the postprotocol implementation group with 5 distinct categories and low overall sample size, this
was associated with favorable discharge disposition and improvement did not reach the level of statistical signifi-
decreased mortality. Our ordered logistic regression model cance. Discharge disposition is a better measure of overall
demonstrated improved discharge disposition in the post- well-being at discharge compared with GOS, which more
protocol group with increasing GCS score, as one would narrowly represents neurological recovery. However, we
expect with decreasing injury severity. In the model, the chose to additionally analyze the GOS score because of
OR for postprotocol status can be interpreted as follows: its widespread use and focus on neurological outcome.18
in the postprotocol group, patients are 4 times as likely to We were unable to use the extended scale due to the retro-
be discharged home versus the combined disposition of spective nature of the analysis, which could not capture the
rehabilitation or death, and patients are 4 times as likely finer variation in disability required of this scale.
to be discharged home or to rehabilitation versus suffer The pre- and postprotocol cohorts were similar except
death during hospitalization. Prior to protocol implemen- the postprotocol group had a higher rate of abusive head
tation, the mortality rate for severe TBI was 31%, and the trauma as well as a higher rate of subdural hemorrhage
rate of discharge home was 36%. This mortality rate is as the predominant radiographic finding. Studies have
consistent with that in large cohorts in the pediatric trau- shown that mortality and outcomes are worse in children
matic literature.3,12,30,32 After protocol implementation, the exposed to abusive head trauma,33 which could potentially
mortality rate dropped to 21%, and the rate of discharge lead to an underestimation of the overall improvement in
home increased dramatically to 69%. This is a significant outcomes seen in our postprotocol cohort. The finding
improvement and demonstrates the possibility of improv- that the rate of subdural hemorrhage in the postprotocol
ing short-term outcomes by standardizing PICU care for cohort was increased is significant given the association
pediatric patients with severe TBI. between subdural hemorrhage and more severe underly-

J Neurosurg Pediatr  Volume 17 • January 2016 23

T. M. O’Lynnger et al.
Fig. 2. Predictive model of discharge disposition probability before and
after protocol implementation.
ing intracranial injury. One would expect outcomes to be
worse in the more severely injured group. However, the
postprotocol cohort had improved outcomes, so we may
conclude that our results are an underestimate of the true
impact of the protocol on outcomes.
Our ordered logistic regression model demonstrated im-
proved outcomes in the postprotocol group and increasing
GCS score, as one would expect with decreasing injury se-
verity. Interestingly, ICP monitor placement was a negative
predictive factor despite the protocol’s focus on ICP man-
agement. We postulated this might be due to bias in initial
GCS score reporting, as some patients may initially dem-
onstrate a lower GCS score secondary to sedation. When
reevaluated prior to ICP monitor placement, some patients’
examinations may improve, resulting in no ICP monitor
placement. Those patients who did receive monitors were
those with the worst examinations and hence the greatest
injury severity. This would decrease expected outcomes in
those who ended up receiving an ICP monitor. There still
remains controversy within the literature regarding the ef-
fectiveness of ICP-based treatment of TBI on outcomes.
Although the protocol indicated patients with GCS scores
less than 8 should receive an ICP monitor, this intervention
was not strictly enforced at the time of the initial examina-
tion, as evidenced by an ICP monitor rate of 28% postpro-
tocol. The protocol includes the overall management strat-
egy, in which placement of an ICP monitor was only a part,
albeit an important one. We hypothesize that the protocol,
in particular neurosurgical presence and the use of short-
acting sedation, enabled a second examination in which the
patients may have been found to be improving and not in
need of monitor placement.
While our study demonstrates that standardizing care
among neurosurgeons and intensivists can improve short-
term outcomes, it does not allow us to draw any conclu-
sions on the particular effectiveness of ICP-focused care.
Our study was not intended to measure adherence to the
protocol, although we noted significantly increased use
of hypertonic saline and decreased use of mannitol per
guideline recommendations for the treatment of elevated
ICP. Interestingly, there was a trend toward fewer ICP
24 J Neurosurg Pediatr  Volume 17 • January 2016
monitor placements in the postprotocol cohort. While it
could be concluded that the overall proportion of patients
receiving ICP monitors being lower in the postprotocol
group signals a lack of adherence, the overall goal of this
study was not to evaluate the utility of ICP monitors, but
rather a focused attention on the evaluation and manage-
ment of these complex patients across multiple services,
with a standardized approach to assessment and decision-
making. A patient receiving an ICP monitor and elevated
care in the past for a poor GCS score on the first assess-
ment may now be more likely to be placed on the appropri-
ate sedating medication for rapid neurological assessment
and receive serial examinations over the earliest portions
of their hospital course, which may show improvement
beyond the need for ICP monitoring. It may be that the
attention to a protocol (i.e., the Hawthorne effect) may
be at play here, but the overall goal of improved care is
achieved. Whether the effect is sustained in this cohort in
follow-up assessments or in future patients is a source of
ongoing study and remains to be seen. Our intent was to
determine how a multidisciplinary approach with com-
mitment from various groups to care for these challeng-
ing patients impacted care. As evidenced by our results,
we believe this did improve outcomes. Those who did not
require a monitor were still treated under the protocol with
goals to maintain adequate mean arterial pressure, nor-
monatremia, adequate hematocrit, sufficient sedation and
analgesia, appropriate oxygenation, and glucose control. In
addition, the initial neurosurgical examination was used to
guide treatment of patients rather than presenting exami-
nations in the emergency department, which can often be
clouded by patient sedation.
Additionally, our study did not show significant differ-
ences in length of stay despite the improvement in out-
comes. ICP monitoring–focused care may be associated
with longer PICU stays and overall increased length of
stay,6,13 but this did not hold true based on the results in
this study. Although costs were not measured in this study,
length of stay has been validated as a surrogate measure for
cost.19 The results of this study could thus imply that out-
comes were improved without increasing costs, a signifi-
cant benefit in a cost-conscious environment.11,28 However,
it is premature to draw conclusions in this regard, because
complexity and intensity of therapy may be increased with
protocol-based care, as demonstrated in a study by Palmer
et al.20 This could ultimately increase overall costs despite
similar lengths of stay. However, improving outcomes in
pediatric patients would be justified even if costs were in-
creased.
There were some limitations to the study, including the
retrospective portion of the design, small sample size, and
single-center involvement. However, our study attempted
to account for these limitations by analyzing outcomes
over several years to increase sample size and ensuring
our ordered logistic regression model was not overfit with
excessive parameters. Ultimately, we were able to demon-
strate that short-term outcomes in pediatric patients with
severe TBI were improved after a standardized protocol
was implemented among caregivers. Further studies are
needed that include multiple pediatric critical care centers
and assess long-term outcomes.

Conclusions
Discharge disposition and mortality rates in pediatric
patients with severe TBI improved after implementation
among caregivers of a standardized protocol based on
best-practice guidelines. This improvement occurred de-
spite a higher rate of abusive head trauma in the postproto-
col group. Those patients ultimately undergoing ICP mon-
itor placement had worse outcomes as a subgroup. Length
of stay was not increased in the postprotocol cohort.
Acknowledgments
The initial protocol that was modified was based in part on
that designed and implemented by José Pineda, MD, and David
Limbrick, MD, PhD, at Washington University in St. Louis, and
James Johnston, MD, at the University of Alabama–Birmingham.
We are grateful for their feedback and willingness to collaborate.
In addition, the physicians and caregivers in the Vanderbilt Divi-
sion of Pediatric Critical Care and the Department of Neurosurgery
are to be acknowledged for their willingness to standardize care
for these complex children. The project described was supported
by CTSA award no. UL1TR000445 from the National Center for
Advancing Translational Sciences, as well as the Vanderbilt Sur-
gical Outcomes Center for Kids (SOCKs). Its contents are solely
the responsibility of the authors and do not necessarily represent
official views of the National Center for Advancing Translational
Sciences or the National Institutes of Health.
References
  1. 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
29:1126–1139, 2012
  2. Bell MJ, Carpenter J, Au AK, Keating RF, Myseros JS, Yaun
A, et al: Development of a pediatric neurocritical care ser-
vice. Neurocrit Care 10:4–10, 2009
  3. Brown AW, Leibson CL, Malec JF, Perkins PK, Diehl NN,
Larson DR: Long-term survival after traumatic brain injury:
a population-based analysis. NeuroRehabilitation 19:37–43,
2004
  4. Bulger EM, Nathens AB, Rivara FP, Moore M, MacKenzie
EJ, Jurkovich GJ: Management of severe head injury: insti-
tutional variations in care and effect on outcome. Crit Care
Med 30:1870–1876, 2002
  5. Carney NA, Chesnut R, Kochanek PM: Guidelines for the
acute medical management of severe traumatic brain injury
in infants, children, and adolescents. Crit Care Med 31 (6
Suppl):S407–S491, 2003
  6. Chesnut RM, Temkin N, Carney N, Dikmen S, Rondina C,
Videtta W, et al: A trial of intracranial-pressure monitoring
in traumatic brain injury. N Engl J Med 367:2471–2481,
2012
  7. 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 18:129–136, 2002
  8. Cremer OL, van Dijk GW, van Wensen E, Brekelmans GJ,
Moons KG, Leenen LP, et al: Effect of intracranial pressure
monitoring and targeted intensive care on functional outcome
after severe head injury. Crit Care Med 33:2207–2213, 2005
  9. Cuschieri J, Johnson JL, Sperry J, West MA, Moore EE, Mi-
nei JP, et al: Benchmarking outcomes in the critically injured
trauma patient and the effect of implementing standard oper-
ating procedures. Ann Surg 255:993–999, 2012
10. Dean NP, Boslaugh S, Adelson PD, Pineda JA, Leonard JR:
Physician agreement with evidence-based recommendations
Standardizing ICU management of pediatric traumatic brain injury
for the treatment of severe traumatic brain injury in children.
J Neurosurg 107 (5 Suppl):387–391, 2007
11. Fakhry SM, Trask AL, Waller MA, Watts DD: Management
of brain-injured patients by an evidence-based medicine pro-
tocol improves outcomes and decreases hospital charges. J
Trauma 56:492–500, 2004
12. Greenwald BD, Hammond F, Harrison-Felix CL, Nakase-
Richardson R, Howe LL, Kreider S: Mortality following
traumatic brain injury among individuals unable to follow
commands at the time of rehabilitation admission: a NIDRR
TBI model systems study. J Neurotrauma [epub ahead of
print], 2015
13. Haddad S, Aldawood AS, Alferayan A, Russell NA, Tamim
HM, Arabi YM: Relationship between intracranial pressure
monitoring and outcomes in severe traumatic brain injury
patients. Anaesth Intensive Care 39:1043–1050, 2011
14. Kochanek PM, Carney N, Adelson PD, Ashwal S, Bell MJ,
Bratton S, 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 13
(Suppl 1):S1–S82, 2012
15. Lane PL, Skoretz TG, Doig G, Girotti MJ: Intracranial pres-
sure monitoring and outcomes after traumatic brain injury.
Can J Surg 43:442–448, 2000
16. Maas AI, Harrison-Felix CL, Menon D, Adelson PD, Balkin
T, Bullock R, et al: Standardizing data collection in traumatic
brain injury. J Neurotrauma 28:177–187, 2011
17. Mauritz W, Steltzer H, Bauer P, Dolanski-Aghamanoukjan
L, Metnitz P: Monitoring of intracranial pressure in patients
with severe traumatic brain injury: an Austrian prospective
multicenter study. Intensive Care Med 34:1208–1215, 2008
18. McCauley SR, Wilde EA, Anderson VA, Bedell G, Beers SR,
Campbell TF, et al: Recommendations for the use of com-
mon outcome measures in pediatric traumatic brain injury
research. J Neurotrauma 29:678–705, 2012
19. Newgard CD, Fleischman R, Choo E, Ma OJ, Hedges JR,
McConnell KJ: Validation of length of hospital stay as a sur-
rogate measure for injury severity and resource use among
injury survivors. Acad Emerg Med 17:142–150, 2010
20. Palmer S, Bader MK, Qureshi A, Palmer J, Shaver T, Borzat-
ta M, et al: The impact on outcomes in a community hospital
setting of using the AANS traumatic brain injury guidelines.
J Trauma 50:657–664, 2001
21. Pineda JA, Leonard JR, Mazotas IG, Noetzel M, Limbrick
DD, Keller MS, et al: Effect of implementation of a paediat-
ric neurocritical care programme on outcomes after severe
traumatic brain injury: a retrospective cohort study. Lancet
Neurol 12:45–52, 2013
22. Potoka DA, Schall LC, Gardner MJ, Stafford PW, Peitzman
AB, Ford HR: Impact of pediatric trauma centers on mortal-
ity in a statewide system. J Trauma 49:237–245, 2000
23. Salorio CF, Slomine BS, Guerguerian AM, Christensen JR,
White JR, Natale JE, et al: Intensive care unit variables and
outcome after pediatric traumatic brain injury: a retrospec-
tive study of survivors. Pediatr Crit Care Med 9:47–53,
2008
24. Saul TG, Ducker TB: Effect of intracranial pressure moni-
toring and aggressive treatment on mortality in severe head
injury. J Neurosurg 56:498–503, 1982
25. Saul TG, Ducker TB: Intracranial pressure monitoring in
patients with severe head injury. Am Surg 48:477–480, 1982
26. Scaife ER, Statler KD: Traumatic brain injury: preferred
methods and targets for resuscitation. Curr Opin Pediatr
22:339–345, 2010
27. Shafi S, Diaz-Arrastia R, Madden C, Gentilello L: Intracrani-
al pressure monitoring in brain-injured patients is associated
with worsening of survival. J Trauma 64:335–340, 2008
28. Shi J, Xiang H, Wheeler K, Smith GA, Stallones L, Groner J,
et al: Costs, mortality likelihood and outcomes of hospital-
J Neurosurg Pediatr  Volume 17 • January 2016 25
T. M. O’Lynnger et al.
ized US children with traumatic brain injuries. Brain Inj
23:602–611, 2009
29. Stein SC, Georgoff P, Meghan S, Mirza KL, El Falaky OM:
Relationship of aggressive monitoring and treatment to im-
proved outcomes in severe traumatic brain injury. J Neuro-
surg 112:1105–1112, 2010
30. Tude Melo JR, Di Rocco F, Blanot S, Oliveira-Filho J, Rou-
jeau T, Sainte-Rose C, et al: Mortality in children with severe
head trauma: predictive factors and proposal for a new pre-
dictive scale. Neurosurgery 67:1542–1547, 2010
31. Vavilala MS, Kernic MA, Wang J, Kannan N, Mink RB,
Wainwright MS, et al: Acute care clinical indicators associat-
ed with discharge outcomes in children with severe traumatic
brain injury. Crit Care Med 42:2258–2266, 2014
32. White JR, Farukhi Z, Bull C, Christensen J, Gordon T,
Paidas C, et al: Predictors of outcome in severely head-
injured children. Crit Care Med 29:534–540, 2001
33. Xiang J, Shi J, Wheeler KK, Yeates KO, Taylor HG, Smith
GA: Paediatric patients with abusive head trauma treated
in US Emergency Departments, 2006–2009. Brain Inj
27:1555–1561, 2013
Disclosure
The authors report no conflict of interest concerning the materi-
26 J Neurosurg Pediatr  Volume 17 • January 2016
als or methods used in this study or the findings specified in this
paper.
Author Contributions
Conception and design: O’Lynnger, Shannon, Lamb, Wellons.
Acquisition of data: O’Lynnger, Greeno. Analysis and inter-
pretation of data: all authors. Drafting the article: O’Lynnger,
Shannon, Le, Wellons. Critically revising the article: O’Lynnger,
Shannon, Le, Wellons. Reviewed submitted version of manu-
script: all authors. Approved the final version of the manu-
script on behalf of all authors: O’Lynnger. Statistical analysis:
O’Lynnger, Shannon. Administrative/technical/material support:
Greeno. Study supervision: Chung, Lamb, Wellons.
Supplemental Information
Previous Presentation
Portions of this work were presented in poster form at the AANS/
CNS Section on Pediatric Neurosurgery, Amelia Island, Florida,
December 3, 2014.
Correspondence
Thomas M. O’Lynnger, Pediatric Neurosurgery, Vanderbilt
University Medical Center, 9222 Doctors’ Office Tower, 2200
Children’s Way, Nashville, TN 37232-9557. email: thomas.
olynnger@vanderbilt.edu.