Manejo de Trauma Pediatrico

Guidelines for the Management of Pediatric Severe
Traumatic Brain Injury, Third Edition: Update of the

Brain Trauma Foundation Guidelines
Patrick M. Kochanek, MD, MCCM1; Robert C. Tasker, MA, MD, FRCP2; Nancy Carney, PhD3;
Annette M. Totten, PhD4; P. David Adelson, MD, FACS, FAAP, FAANS5;
Nathan R. Selden, MD, PhD, FACS, FAAP6; Cynthia Davis-O’Reilly, BS7; Erica L. Hart, MST8;
Michael J. Bell, MD9; Susan L. Bratton, MD, MPH, FAAP10; Gerald A. Grant, MD11;
Niranjan Kissoon, MD, FRCP(C), FAAP, MCCM, FACPE12;
Karin E. Reuter-Rice, PhD, CPNP-AC, FCCM, FAAN13; Monica S. Vavilala, MD14;
Mark S. Wainwright, MD, PhD15
1
Ake N. Grenvik Professor of Critical Care Medicine, Vice Chair, Depart- Herman and Faye Sarkowsky Endowed Chair, Head, Division of Pediatric Neu-
15
ment of Critical Care Medicine, Professor of Anesthesiology, Pediatrics, rology, University of Washington, Seattle Children's Hospital, Seattle, WA.
Bioengineering, and Clinical and Translational Science, Director, Safar This document was endorsed by the American Association of Neurologi-
Center for Resuscitation Research, University of Pittsburgh School of cal Surgeons/Congress of Neurological Surgeons.
Medicine, UPMC Children’s Hospital of Pittsburgh, Pittsburgh, PA.
Any opinions, findings, and conclusions or recommendations expressed
2
Department of Neurology and Department of Anesthesiology, Critical in this material are those of the authors and do not necessarily reflect the
Care and Pain Medicine, Boston Children's Hospital, Harvard Medical views of the U.S. Army Contracting Command, Aberdeen Proving Ground,
School, Boston, MA. Natick Contracting Division, Stanford University, or the Brain Trauma
3
Professor, Pacific Northwest Evidence-based Practice Center, Depart- Foundation. The information contained in the Guidelines for the Manage-
ment of Medical Informatics and Clinical Epidemiology, Oregon Health & ment of Pediatric Severe Traumatic Brain Injury reflects the current state of
Science University, Portland, OR. knowledge at the time of publication. The Brain Trauma Foundation, Ameri-
4
Associate Professor, Pacific Northwest Evidence-based Practice Cen- can Association of Neurological Surgeons, Congress of Neurological Sur-
ter, Department of Medical Informatics and Clinical Epidemiology, Ore- geons, and other collaborating organizations are not engaged in rendering
gon Health & Science University, Portland, OR. professional medical services and assume no responsibility for patient
outcomes resulting from application of these general recommendations
5
Diane and Bruce Halle Endowed Chair in Pediatric Neurosciences,
in specific patient circumstances. Accordingly, the Brain Trauma Foun-
Chief, Pediatric Neurosurgery, Director, BARROW Neurological Insti-
dation, American Association of Neurological Surgeons, and Congress
tute at Phoenix Children's Hospital, Phoenix, AZ.
of Neurological Surgeons consider adherence to these clinical practice
6
Chair, Department of Neurological Surgery, Oregon Health & Science guidelines will not necessarily assure a successful medical outcome. The
University, Portland, OR. information contained in these guidelines reflects published scientific evi-
7
Research Associate, Pacific Northwest Evidence-based Practice Cen- dence at the time of completion of the guidelines and cannot anticipate
ter, Department of Medical Informatics and Clinical Epidemiology, Oregon subsequent findings and/or additional evidence, and therefore should not
Health & Science University, Portland, OR. be considered inclusive of all proper procedures and tests or exclusive of
8
Research Assistant, Pacific Northwest Evidence-based Practice Center, other procedures and tests that are reasonably directed to obtaining the
Department of Medical Informatics and Clinical Epidemiology, Oregon same result. Medical advice and decisions are appropriately made only
Health & Science University, Portland, OR. by a competent and licensed physician who must make decisions in light
of all the facts and circumstances in each individual and particular case
9
Professor and Chief, Critical Care Medicine, Children’s National Medical and on the basis of availability of resources and expertise. Guidelines are
Center, Washington, DC. not intended to supplant physician judgment with respect to particular
10
Emeritus Professor of Pediatrics, University of Utah, Salt Lake City, UT. patients or special clinical situations and are not a substitute for physician-
11
Department of Neurosurgery, Stanford University, Stanford, CA. patient consultation. Accordingly, the Brain Trauma Foundation, American
Association of Neurological Surgeons, and Congress of Neurological
12
Department of Pediatrics, British Columbia’s Children’s Hospital, Clini-
Surgeons consider adherence to these guidelines to be voluntary, with
cal Investigator, Child and Family Research Institute, University of British
the ultimate determination regarding their application to be made by the
Columbia, Vancouver, BC. Canada.
physician in light of each patient’s individual circumstances.
13
School of Nursing/School of Medicine, Department of Pediatrics, Divi-
sion of Pediatric Critical Care Medicine, Duke University, Durham, NC. Supplemental digital content is available for this article. Direct URL citations
appear in the printed text and are provided in the HTML and PDF versions of
14
Professor & Vice Chair Strategic Affairs, Anesthesiology & Pain Medi- this article on the journal’s website (http://journals.lww.com/pccmjournal).
cine, Professor, Pediatrics, Director, Harborview Injury Prevention and
Research Center (HIPRC), University of Washington, Seattle, WA. Supported, in part, by the U.S. Army Contracting Command, Aberdeen Prov-
ing Ground, and Natick Contracting Division, through a contract awarded
Copyright © 2019 by the Society of Critical Care Medicine and the World to Stanford University (W911 QY-14-C-0086), a subcontract awarded to
Federation of Pediatric Intensive and Critical Care Societies Oregon Health & Science University. Previous editions were supported by
DOI: 10.1097/PCC.0000000000001735 funding from multiple sources through the Brain Trauma Foundation.
Pediatric Critical Care Medicine www.pccmjournal.org S1

Copyright © 2019 by the Society of Critical Care Medicine and the World Federation of Pediatric Intensive and Critical Care Societies.
Unauthorized reproduction of this article is prohibited
Kochanek et al
Dr. Kochanek received funding from the Society of Critical Care Medicine that presents a “Critical Pathway” algorithm of care for both
(Editor-in-Chief of Pediatric Critical Care Medicine), from serving as an
expert witness on cases in pediatric critical care. Drs. Carney and Tot-
first-tier and second-tier (refractory intracranial hypertension)
ten’s, Ms. Davis-O’Reilly’s, and Ms. Hart’s institutions received funding approaches. The algorithm reflects both the evidence-based
from Stanford University. Dr. Selden disclosed that he has stock options recommendations from these guidelines and consensus-based
(current $0 value) in Cerebrotech for scientific advisory board service
(this device is not clinically available and is not referenced in the work).
expert opinion, vetted by the clinical investigators, where evi-
Dr. Reuter-Rice received funding from textbook royalties and curriculum dence was not available. An algorithm was provided in the First
content, and she received support for article research from Robert Wood but not Second Editions of the guidelines, and we believe that
Johnson Foundation funding 2013–2016. Dr. Wainwright received fund-
ing from Sage Therapeutics. The remaining authors have disclosed that
given the new reports available, along with the existing gaps in
they do not have any potential conflicts of interest. evidence, a combination of evidence-based and consensus-based
Hector R. Wong, MD, is a Guest Editor. recommendations provides additional and much-needed guid-
For information regarding this article, E-mail: [email protected]. ance for clinicians at the bedside. The algorithm also addresses
edu (Pediatr Crit Care Med 2019; 20:S1–S82) a number of issues that are important but were not previously
Key Words: critical care; evidence-based medicine; guidelines; covered in the guidelines, given the lack of research and the focus
pediatrics; systematic review; traumatic brain injury on evidence-based recommendations. This includes addressing
issues such as a stepwise approach to elevated ICP, differences in
tempo of therapy in different types of patients, scenarios with a
rapidly escalating need for ICP-directed therapy in the setting of
Severe Traumatic Brain Injury in Infants, Children, impending herniation, integration of multiple monitoring tar-
and Adolescents in 2019: Some Overdue Progress, gets, and other complex issues such as minimal versus optimal
Many Remaining Questions, and Exciting Ongoing therapeutic targets and approaches to weaning therapies. We
Work in the Field of Traumatic Brain Injury Research hope that the readership finds the algorithm document helpful,
In this Supplement to Pediatric Critical Care Medicine, we are recognizing that it represents a challenging albeit important step.
pleased to present the Third Edition of the Guidelines for the Designing and developing this pediatric TBI evidence-based
Management of Pediatric Severe Traumatic Brain Injury (TBI). guidelines document required an expert administrative man-
This body of work updates the Second Edition of the guidelines agement team, and to that end, we are extremely grateful to the
that was published in 2012 (1). It represents a substantial effort staff of the Pacific Northwest Evidence-based Practice Center,
by a multidisciplinary group of individuals assembled to reflect Oregon Health & Science University, for their vital contribu-
the team approach to the treatment of these complex, critically ill tion to this work. We are also grateful to the Brain Trauma
patients that is essential to optimizing critical care and improv- Foundation and the Department of Defense for supporting
ing outcomes. This work also represents the strong and always- the development and publication of these guidelines docu-
evolving partnership between investigators from the medical and ments. We are grateful to the endorsing societies for recogniz-
research communities, forged in Chicago in 2000, from which the ing the importance of this work and for the considerable work
first pediatric TBI guidelines were developed. The mutual trust of the clinical investigators in constructing the final document.
and respect we share have been the foundation of our commit- We are also pleased to have collaborated with the Congress of
ment to bringing evidence-based care to children with TBI. Neurological Surgeons and the journal Neurosurgery that is
Updating these guidelines was particularly exciting to the copublishing the Executive Summary document of these guide-
individuals who have participated in the previous two edi- lines for its readership. We are also grateful to Hector Wong for
tions because several new studies have been published which serving as Guest Editor, along with the external reviewers of this
begin to address a number of major gaps in the pediatric TBI final document. Finally, we thank each of the clinical investiga-
literature—gaps that were specifically identified as targets for tors and coauthors on this project. We believe that the consider-
future research in earlier editions. For example, we are now able uncompensated time and effort devoted to this important
able to include reports on the effects of commonly used seda- project will help to educate clinicians worldwide and enhance
tives and analgesics on intracranial pressure (ICP). Similarly, the outcomes of children with severe TBI. Clinical investigators
initial head-to-head comparisons of the influence of agents in provided Conflict of Interest Disclosures at the beginning of
routine “real world” use such as hypertonic saline (HTS), fen- the process, which were re-reviewed at the time of publication.
tanyl, and others now inform these guidelines (2, 3). A total of No clinical investigator made inclusion decisions or provided
48 new studies were included in this Third Edition. Although assessments on publications for which they were an author.
some progress has been made and should be celebrated, overall Looking forward, it is important to recognize that these guide-
the level of evidence informing these guidelines remains low. lines were written as the Approaches and Decisions in Acute
High-quality randomized studies that could support level I Pediatric TBI Trial (ADAPT) (4–6), one of the most important
recommendations remain absent; the available evidence pro- in the field of pediatric TBI, was coming to a close. The ADAPT
duced only three level II recommendations, whereas most rec- completed enrollment of 1,000 cases of severe pediatric TBI and
ommendations are level III, supported by low-quality evidence. is one example of the recent heightened general interest in TBI as
Based in part on a number of requests from the readership to a disease. This new interest in the importance of TBI has emerged
individual clinical investigators, we have included a companion in part from the recognition of the high prevalence of TBI across
article in the regular pages of Pediatric Critical Care Medicine the injury severity spectrum, particularly concussion, and from
S2 www.pccmjournal.org March 2019 • Volume 20 (Suppl) • Number 3

Supplement
the need for new classification systems and new trial design for
TBI in both children and adults (7, 8). In addition, the emerg-
ing links between TBI and a number of neurodegenerative dis-
eases have broadened the interest in TBI, have led to additional
support of TBI research, and have produced an unprecedented
level of research in TBI and a quest for new therapies (9–11). We
expect that the results of ADAPT, along with those of other ongo-
ing and recently completed research in the field, will help pro-
vide new insight and clarity into the acute medical management
(MM) of infants, children, and adolescents with severe TBI, and
mandate further refinement of the recommendations in these
documents. We know that we speak for the entire team of clinical
investigators in welcoming the opportunity to incorporate addi-
tional high-level evidence into future updates of these guidelines.
METHODS Figure 1. Dynamic process for guidelines, protocols, and future research.
The methods for developing these guidelines were organized The diagram shows the flow of information from available evidence to a
in two phases: a systematic review, assessment, and synthesis guideline. The guideline leads to gaps that identify future research and
consensus-based clinical protocols that fill gaps, both of which lead to a
of the literature; and use of that product as the foundation for generation of new research.
evidence-based recommendations. These guidelines are the
product of the two-phased, evidence-based process. comparators, outcomes, timing, settings, study designs, and
Based on almost 2 decades of collaboration, the team publication types. The criteria for population are as follows:
of clinical investigators and methodologists (Appendix A,
Supplemental Digital Content 1, http://links.lww.com/PCC/ ●● Age 18 years old or younger
A774) is grounded in and adheres to the fundamental princi- ●● TBI
ples of evidence-based medicine to derive recommendations, ●● Glasgow Coma Scale (GCS) score less than 9
and is committed to maintaining the distinction between evi- Included Topics. The team chose to carry forward topics from
dence and consensus. It is important that this distinction is the Second Edition of these guidelines. No new topics were added.
clear to promote transparency and inspire innovative future The topics are organized in three categories that are specific to
research that will expand the evidence base for TBI care. severe TBI in children: monitoring, thresholds, and treatments.
Because these guidelines only provide recommendations
based on available evidence, most often they do not provide Monitoring
direction for all phases of clinical care. Ideally, clinically useful 1. ICP
protocols begin with evidence-based guidelines, and then use 2. Advanced neuromonitoring
clinical experience and consensus to fill the gaps where evi- 3. Neuroimaging
dence is insufficient. The goal is to use the evidence and the
evidence-based recommendations as the backbone to which Thresholds
expertise and consensus can be added to produce protocols
4. ICP
appropriate to specific clinical environments (Fig. 1, “Future
5. Cerebral perfusion pressure (CPP)
Research section”). In a process independent from developing
this Third Edition of the guidelines, the team engaged in a con- Treatments
sensus process and produced the algorithm for treatment of
6. Hyperosmolar therapy
severe TBI in pediatric patients.
The following “Methods section” describes the process we 7. Analgesics, sedatives, and neuromuscular blockade
used to produce the systematic review and evidence-based rec- (NMB)
ommendations. The methods used to develop the algorithm 8. Cerebrospinal fluid (CSF) drainage
are described in that document (12). 9. Seizure prophylaxis
10. Ventilation therapies
11. Temperature control
Phase I: Systematic Evidence Review and 12. Barbiturates
Synthesis 13. Decompressive craniectomy
Scope of the Systematic Review 14. Nutrition
Criteria for Including Publications 15. Corticosteroids
Appendix B (Supplemental Digital Content 1, http://links.
lww.com/PCC/A774) lists the criteria for including studies Major Changes for This Edition. Major changes for this edition
for review using the categories of population, interventions, are summarized here, and details are provided in Appendix C

Kochanek et al
(Supplemental Digital Content 1, http://links.lww.com/PCC/ Use of Indirect Evidence and Intermediate Outcomes
A774). Direct evidence comes from studies that compare important
health outcomes (e.g., mortality, morbidity, function) between
●● The clinical investigators and methods team identified
two or more intervention groups or between an intervention
three primary endpoints considered important health out-
group and a control group that represent the population of
comes for pediatric patients with TBI:
interest, in this case pediatric patients with severe TBI. When
●● To improve overall outcomes (mortality, morbidity, function) direct evidence was limited or not available, indirect evidence
●● To control ICP was used to support a recommendation. Indirect evidence has
●● To prevent posttraumatic seizures (PTSs) been defined in previous work by this methods team (1, 13,
●● Two new meta-analyses were added to the evidence base 14) and other evidence-based methods groups (15, 16). In this
for temperature control. edition, we included two types of indirect evidence.
●● The title of “Hyperventilation” was changed to “Ventilation 1. Evidence That Improvement in an Intermediate Out-
Therapies.” come Is Associated With Important Health Outcomes
●● Recommendations are provided as level I, II, or III. In some cases, there is a lack of direct evidence that uti-
lization of a specific treatment option results in improved
In some cases, publications from the second edition were not patient outcomes such as mortality or morbidity, but there is
included in this 3rd Edition. Our rationale for excluding pre- evidence about changes in an intermediate outcome, which
viously included studies was based on identification of current is then associated with improved mortality or morbidity.
material that superseded our earlier work (See Appendix E, The most notable intermediate outcome for the treatment of
Supplemental Digital Content 1, http://links.lww.com/PCC/ TBI is management of ICP. Multiple studies (cited in the ICP
A774). Similarly, we removed or changed recommendations Monitoring topic of this guideline) consistently demonstrate
from the 2nd Edition when the current literature provided new that patients whose ICP is successfully maintained at or under
and/or more accurate information (see Appendix A, Supple- a maximum threshold have reduced mortality and improved
mental Digital Content 1, http://links.lww.com/PCC/A774). function. As a consequence, the clinical investigators elected
to identify “Control of ICP” as an important intermediate
Study Selection and Compilation of Evidence outcome, and use the available indirect evidence to support
Literature Search Strategies. The research librarian who the recommendations about monitoring ICP and for treat-
worked on the Second Edition reviewed and updated the ments designed to lower ICP.
search strategies for that edition and executed the searches for Intermediate outcomes and indirect evidence of this nature
this Third Edition. Ovid/MEDLINE was searched from 2010 to were used in three topics for this edition of the guidelines: ICP
May of 2015, and an update was performed to include articles Monitoring, Ventilation Therapies, and Temperature Control.
published and indexed through June of 2017. Publications rec- In each of these topics, an intermediate outcome was used as
ommended by peers that were not captured in the search were the endpoint because, although direct evidence was lacking
reviewed, and those meeting inclusion criteria were included in that intervening improves mortality or function, indirect evi-
the final library. The search strategy is in Appendix D (Supple- dence was available associating management of the intermedi-
mental Digital Content 1, http://links.lww.com/PCC/A774). ate outcome with improved mortality or function.
Abstract and Full-Text Review. Abstracts for publications cap- For ICP monitoring, the intermediate outcome was man-
tured in the search were reviewed independently by two members aged ICP; indirect evidence that patients with managed ICP
of the methods team. Articles were retained for full-text review if had better outcomes was used to support the recommenda-
at least one person considered them relevant based on the abstract. tion. For ventilation therapies, the intermediate outcomes were
Two methods team members read each full-text article and deter- prevention of severe hypocarbia (SH). There were no pediat-
mined whether it met the inclusion criteria (Appendix B, Supple- ric studies that directly related hyperventilation to poor out-
mental Digital Content 1, http://links.lww.com/PCC/A774). The comes. However, there was evidence of an association between
included and excluded full-text articles for each topic were also SH and mortality; thus, studies that demonstrated this associa-
reviewed by one or more clinical investigators who took the lead tion were used as indirect evidence. For temperature control,
on each topic, and full-text articles were available for review by all
the intermediate outcomes were mean and peak CSF myelin
authors. The key criteria for inclusion were as follows: the study
basic protein concentrations and phenytoin levels.
population was pediatric patients (age, ≤ 18 yr old) with severe TBI
2. Evidence From Samples With Mixed Ages, Severities, or
(defined as GCS score of 3–8) and the study assessed an included
Pathologies
outcome. Publications with samples that included adults, mod-
In some cases, when direct evidence was lacking, we consid-
erate or mild severities, or pathologies other than TBI (indirect
ered studies that included patients with mixed severities (mild,
evidence) were considered when direct evidence was limited or
moderate, and severe TBI), mixed ages, or mixed pathologies
not available. Discrepancies between reviewers were resolved via
(traumatic and non-TBI) using the following criteria:
consensus or by a third reviewer. A list of studies excluded after
full-text review is in Appendix E (Supplemental Digital Content 1, 1. How relevant to (or different from) our target population
http://links.lww.com/PCC/A774). is the population in the indirect study?

Supplement
2. To what extent does the relevant physiology of the popu- questions. The studies in each topic were reviewed to deter-
lation in the indirect study approximate the relevant mine if quantitative synthesis—meta-analysis—was feasible.
physiology of the population of interest? This involved determining if the patient populations, specif-
3. To what extent are differences in physiology expected to ics of the intervention, and the outcomes were similar enough
influence the outcome? across several studies that the study results could be combined.
4. In what direction would these differences influence the The result of this assessment is included in the Quality of the
observed effect? Body of Evidence table for each subtopic. For this edition, we
did not identify any topics for which quantitative synthesis was
In this edition, indirect evidence from studies with mixed
appropriate according to current standards. For this reason,
severities, ages, or pathologies was included in the topics about
the evidence was synthesized qualitatively.
analgesics, sedatives, and NMB; CSF drainage; and seizure
prophylaxis.
Quality of the Body of Evidence
When indirect evidence was included, it is noted in the table
Assessing the quality of the body of evidence involves four
describing the quality of the body of evidence.
domains: the aggregate quality of the included individual stud-
ies, the consistency of the results across studies, whether the
Quality Assessment of Individual Studies
evidence provided is direct or indirect, and the precision of the
All included studies were assessed for potential for bias, which
estimates of the outcomes. The criteria and ratings are out-
is an approach to assessing the internal validity or quality of an
lined below, and more detailed definitions are given in Appen-
individual study. This assessment is a core component of system-
dix G (Supplemental Digital Content 1, http://links.lww.com/
atic review methods. It is an approach to considering and rating
PCC/A774). In addition, the number of studies and number
studies in terms of how the study design and conduct addressed
of included subjects are considered. Based on these, an overall
issues such as selection bias, confounding, and attrition. The cri-
assessment is made as to whether the quality of the body of
teria used for this edition are described in Appendix F (Supple-
evidence is high, moderate, low, or insufficient. The assessment
mental Digital Content 1, http://links.lww.com/PCC/A774).
of the body of evidence for each subtopic is included in a sum-
Two reviewers independently evaluated each study using
mary table in each section following the recommendations.
the criteria appropriate for the study design (i.e., random-
ized controlled trials [RCTs], observational studies, studies
Criteria
of thresholds) and rated the study as class 1, 2, or 3 evidence
Quality of Individual Studies: This identifies the quality of the
based on the combination of study design and conduct. Class
individual studies. It details how many studies are class 1, class
1 is the highest class and is limited to good-quality RCTs.
2, and class 3.
Class 2 includes moderate-quality RCTs and good-quality
Consistency: Consistency is the extent to which the results
cohort or case-control studies. Class 3 is the lowest class and
and conclusions are similar across studies. It is rated high (all
is given to low-quality RCTs, moderate- to low-quality cohort
are similar), moderate (most are similar), or low (no one con-
or case-control studies, and treatment series and other non-
clusion is more frequent). It is not applicable when the body
comparative designs. Differences in ratings were reconciled via
of evidence consists of a single study.
consensus or the inclusion of a third reviewer as needed.
Directness: We define directness as whether the study pop-
ulation is the same as the population of interest and whether
Data Abstraction
the outcomes are clinical rather than intermediate outcomes.
Data were abstracted from studies by a member of the meth-
Evidence is labeled as direct, indirect, or mixed.
ods team and checked for accuracy by a second member. Infor-
Precision: Precision is the degree of certainty surround-
mation was recorded about the study population, design, and
ing the effect estimate for a given outcome. Precision is rated
results. Key elements of each included study are presented
high, moderate, or low. How this is determined depends on
in the Summary of Evidence tables for each topic. Complete
the type of analysis used in a specific study but may include
abstraction tables are available upon request.
consideration of the width of CIs, other indicators of vari-
ance, or the magnitude of p values used to determine statisti-
Synthesis
cal significance.
The final phase of the evidence review is the synthesis of indi-
Ratings. These criteria are then considered when assigning
vidual studies into information that the clinical investigators
a rating to the body of evidence.
and the methods team use to develop recommendations. This
The ratings are defined as follows:
synthesis is described for each topic in the section titled “Eval-
uation of the Evidence,” following the Recommendations and ●● High: High confidence that the evidence reflects the true
preceding the Evidence Summary. effect. Further research is very unlikely to change the confi-
dence in the estimate of effect.
Identification of Subtopics and Synthesis ●● Moderate: Moderate confidence that the evidence reflects
For each monitoring, thresholds, or treatment topic, the clini- the true effect. Further research may change the confidence
cal investigators identified important subtopics or clinical in the estimate of effect and may change the estimate.

Kochanek et al
●● Low: Low confidence that the evidence reflects the true development of recommendations. No recommendations were
effect. Further research is likely to change the confidence in made without a basis in evidence.
the estimate of effect and is likely to change the estimate. Once evidence was identified, whether it could be used
●● Insufficient: Evidence is unavailable or does not permit a to inform recommendations was based on the quality of the
conclusion. body of evidence and consideration of applicability. Given
this, there were cases in which evidence was identified, but the
A determination of quality of the body of evidence requires
quality was low and applicability concerns restricted our abil-
a judgment about the relative importance of the criteria, and
ity to translate the evidence into recommendations. Even if a
these may vary across topics and subtopics. The following gen-
recommendation was not made, the evidence was included for
eral examples are provided to illustrate the variations that are
future consideration because in the future, new studies may be
possible but are not intended as exhaustive decision rules. If two
or more class 1 studies demonstrate contradictory findings for a added, resulting in changes in the assessment of the quality of
particular topic, the overall quality of the body of evidence may the body of evidence.
be assessed as low because there is uncertainty about the effect.
Similarly, class 1 or 2 studies that provide indirect evidence may Level of Recommendation
only constitute low-quality evidence overall. In some cases, the Recommendations in this edition are designated as level I, level
body of evidence may be a single study, but the rating may vary. II, or level III. The level of recommendation is determined by
A single study may constitute a high-quality body of evidence if the assessment of the quality of the body of evidence, rather
it is a large, multisite, class 1 RCT; a moderate-quality body of than the class of the included studies. The levels were primarily
evidence if it is a single-site, class 2 study with a sizable sample based on the quality of the body of evidence as follows:
and moderate precision; or insufficient evidence if the sample is ●● Level I recommendations were based on a high-quality
small and the precision of the estimate of effect is low. body of evidence.
●● Level II recommendations were based on a moderate-qual-
Applicability ity body of evidence.
Applicability is the extent to which research findings are useful for ●● Level III recommendations were based on a low-quality
informing recommendations for a broader population (usually body of evidence.
the population that is the target of the recommendations). What
is important to consider when assessing applicability will vary Applicability could result in a level III recommendation
depending on the topic, and the assessment is context specific. (e.g., a “moderate-quality body of evidence” with significant
Consequently, there is currently no generally accepted universal applicability concerns). In this edition, applicability alone was
rating system for applicability. Common considerations focus not used to downgrade a recommendation. However, given the
on the characteristics of the patient population (e.g., to which lack of standards and developed methods in this area, we cited
patients are the results applicable?) and the settings for care deliv- applicability issues that were identified and discussed by the
ery (e.g., where could a similar result be expected?). Even if the clinical investigators.
patient population meets the inclusion criteria established for the “Insufficient” was used in cases where there were no studies
review, there may be specific characteristics that affect applicabil- identified or because the body of evidence had major quality
ity. The characteristics of the setting in which a study was con- limitations. If the evidence was insufficient, no recommenda-
ducted may also be important to consider. For example, a study tions were made.
conducted in a Veterans Administration (VA) Medical Center may
or may not be applicable to other settings, depending on how sim- Recommendation Review and Revision
ilar the Veterans are to the population of interest or how similar Preliminary Topic Reviews. After completion of the literature
the context of the VA is to the care setting of interest. Additional review, identification of new studies, quality assessment, and data
characteristics to be considered may include the geographic loca- abstraction, the methods team sent drafts for each topic to two
tion (e.g., country, state, urban, or rural) and the type of hospital clinical investigators. The clinical investigators read the included
(e.g., level of trauma center). The geographic area and type of hos- studies and the draft recommendations, provided input, and
pital are considered because it is possible that the patients, practice suggested additional studies for consideration. Methods team
patterns, and available services are different across environments. members incorporated the input, acquired and reviewed new
In this edition, we consider the applicability of individual studies studies, and provided the clinical investigators with new pub-
in the “Quality of the Body of Evidence and Applicability section” lications and a revised summary of the evidence for each topic.
immediately following the recommendations. Clinical Investigator Review Meeting. In a day-long meet-
ing in 2016, each topic was presented and discussed by the
group. Based on these discussions, the methods team revised
Phase II: Development of Recommendations
the draft guidelines.
Inclusion of Recommendations Review of Complete Draft. The complete draft of all topics
Class 1, 2, or 3 studies constitute the evidence on which the and the other sections of the guidelines (e.g., Methods; Appen-
recommendations are based. Under our current methods, dices, Supplemental Digital Content 1, http://links.lww.com/
identification of evidence is necessary but not sufficient for the PCC/A774) was sent to all clinical investigators for review and

Supplement
comment. Phone conferences and e-mail exchanges occurred per se, are confounded by the numerous therapeutic interven-
through April 2018 to answer questions, discuss the draft, and tions that have been introduced simultaneously with increased
finalize the document. ICP monitoring and have not been subjected individually to
controlled trials. These confounders include protocol-driven
Peer Review prehospital care, tracheal intubation and oxygenation, aggres-
After revisions were made based on input from the clini- sive treatment of systemic hypotension and hypovolemia,
cal investigators, the complete, revised Third Edition and an osmolar treatment of cerebral edema, rapid cranial CT imag-
Executive Summary were sent to the journal Pediatric Critical ing to detect mass lesions, and improved enteral and parenteral
Care Medicine for peer review. A comprehensive peer review nutrition, among others.
was also conducted by members of the American Association Several studies demonstrate an association between intra-
of Neurological Surgeons/Congress of Neurological Surgeons cranial hypertension and/or systemic hypotension and poor
Joint Guidelines Review Committee, in collaboration with the outcome after severe TBI (25–27). It is less clear, however,
clinical investigators and methods team, to facilitate publica- whether intracranial hypertension or reduced cerebral per-
tion in the journal Neurosurgery. fusion secondary to intracranial hypertension is the primary
mechanism of secondary injury. CPP equals mean arterial
blood pressure (MAP) minus mean ICP (28) and is the most
MONITORING
readily available correlate of global cerebral perfusion (29–32).
ICP Monitoring The relative value of ICP monitoring as a means of evaluating
Recommendations and manipulating CPP, versus avoidance of cerebral herniation
Strength of Recommendations: Weak events, is also unclear (33).
The lack of controlled trials on ICP monitoring limited the
Levels I and II
strength of the recommendations contained in the previous edi-
There was insufficient evidence to support a level I or II recom-
tion of the Guidelines for the Management of Pediatric Severe
mendation for this topic.
TBI (34). This dearth of strong evidence in children is associated
Level III with mixed adoption of guidelines-directed management in the
To Improve Overall Outcomes. III.1. Use of ICP monitoring United States and abroad (35–37). In addition, a single prospective
is suggested. controlled study carried out in South America in predominantly
Changes From Prior Edition. There are no content changes adult patients found no difference in outcome when comparing
from the Second Edition to the recommendations. Three new ICP monitoring-directed therapy or clinical-radiologic–directed
class 3 retrospective observational studies were added to the therapy (38). Although the data in this study are not separable by
evidence base for this topic (17–19). age subgroup, the study did recruit patients more than 12 years
old, and its results have therefore likely informed ongoing debate
Introduction regarding the evidence for ICP monitoring in severe TBI and levels
Secondary injury to the brain after severe TBI is a result of a of adoption at individual centers. A 2007 survey of U.S. neurosur-
pathophysiologic cascade of events that reduces perfusion of geons and nonneurosurgeons caring for such patients found about
surviving neural tissue, oxygen and metabolite delivery, and 60% agreement and conformity with guidelines recommendations
clearance of metabolic waste and toxins. Brain swelling resulting (35). In the United Kingdom in 2006, only 59% of children pre-
from vasogenic and or cytotoxic edema, occurring within the senting with severe TBI underwent ICP monitoring, with only half
closed compartment of the skull, leads to intracranial hyper- of clinical units caring for such children using monitoring technol-
tension, cerebral herniation syndromes, further focal ischemic ogy (36, 37). The use of monitoring in children less than 2 years old
injury, and brainstem compression. Sustained elevation of ICP with severe TBI may be even less likely. Keenan et al (39) observed
thus represents a key pathophysiologic variable in the occur- use of ICP monitoring in only 33% of patients in this young age
rence of secondary brain injury phase following TBI (20–22). group at multiple centers in the state of North Carolina. There is
Since the late 1970s, significant improvements in both also significant variability in the use of various interventions for
survival and functional outcome after severe TBI have been the treatment of intracranial hypertension at different centers (24).
achieved using intensive care management protocols that Because a monitor is required to have an objective measure
center on the measurement of ICP and medical and surgi- of ICP for directed critical care therapies, the outcome benefits
cal treatment of intracranial hypertension (23). Tilford et al of monitoring are considered to be supported inferentially.
(24) demonstrated that a PICU with higher occurrence of ICP
monitoring in severely brain injured children, accompanied by Evaluation of the Evidence
specific ICP-directed medical interventions, resulted in a trend Quality of the Body of Evidence. Studies included for this topic
toward lower mortality than two comparison ICUs. Similarly, address the question about whether the information derived from
Tilford et al (23) demonstrated improved outcomes after severe the ICP monitor to inform treatment decisions improves out-
TBI in an era during which the overall rates of ICP monitoring comes for pediatric patients with TBI. Three large class 3 stud-
in these patients increased. Attempts to evaluate the indepen- ies—two using patients as the unit of measure (17, 19) and one
dent benefit of direct ICP measurement to improve outcomes, using hospitals as the unit of measure (18)—provided low-quality

Kochanek et al
TABLE 1. Intracranial Pressure Monitoring: Quality of the Body of Evidence

Components of Overall Quality: Class 3 Studies
Quality of
Evidence
Consistency Precision (High,
Meta-Analysis Total (High, Directness (High, Moderate,
No. of Studies Possible No. of Moderate, (Direct or Moderate, Low, or
Topic Study Design Recommendation (Yes or Noa) Subjects (n) Low) Indirect) Low) Insufficient)
Use of ICP 2 retrospective III.1. No 6,191 Low Direct Moderate Low

monitoring
(patients as unit of
measure)
Use of ICP 1 retrospective III.1. NA 4,667b NA Direct Moderate Low
monitoring
(hospitals as unit
of measure)
Association of 1 randomized III.1. No 945 Moderate Indirect Low Low
elevated ICP with controlled trial
outcomes 2 prospective
10 retrospective
3 treatment series
ICP = intracranial pressure, NA = not applicable.
Meta-analysis to synthesize results is only possible if several studies address the same criteria/question, and other design criteria are met.
a
Probable overlap of patients across Bennett et al (18, 19).

b
n indicates sample size.
direct evidence to support the recommendation. One RCT (40), than 20 mm Hg at some point during their intensive care
two prospective studies (41, 42), 10 retrospective studies (30, 43– course (52). Intracranial hypertension (ICP > 20 mm Hg) may
51), and three treatment series (52–54) provided indirect evidence also be significantly more prevalent in children with severe
that higher ICP is associated with poorer outcomes. The overall TBI who do not demonstrate spontaneous motor function
quality of the body of evidence is low (Table 1). (80%) than those who do (20%) (42). These studies suggested
Applicability. The studies providing direct evidence (17–19) that children presenting with severe TBI are at notable risk
reported multicenter data from large samples in the United for intracranial hypertension. No specific markers have been
States. The findings were inconsistent, in that two (17, 18) sug- identified which reliably determine the presence or absence of
gested better outcomes for patients who are monitored and the intracranial hypertension without monitoring in this popula-
third (19) suggested no benefit. The small observational studies tion, and thus reliable noninvasive methods to detect intracra-
and treatment series were conducted in the United States, Israel, nial hypertension are not currently available.
United Kingdom, Spain, Lithuania, Switzerland, and Sweden Are ICP Data Useful in Managing Pediatric Severe TBI?.
(30, 40–54). There were no major applicability concerns. Fifteen studies involving 857 pediatric patients demonstrated
Summary of the Evidence. Three class 3 studies provided an association between intracranial hypertension (generally >
direct evidence to support the recommendation (17–19). Six- 20 mm Hg) and poor neurologic outcome or death (30, 40–44,
teen class 3 studies from the Second Edition provided indirect 46–54). Only one small study of 48 patients failed to demon-
evidence that patients with lower ICP have better outcomes strate a clear association between intracranial hypertension
(30, 40–54) (Table 2). and poor outcome (45), but in this study, children with higher
peak ICP were immediately and successfully treated with
Evidence Synthesis decompressive craniectomy. These studies suggest that ICP
Are Children With Severe TBI at Risk of Intracranial Hyper- is an important prognostic variable. It also plays a strong role
tension?. A number of small studies demonstrated a occur- both independently and as a component of CPP in directing
rence of intracranial hypertension in children with severe TBI the management of pediatric severe TBI patients.
(42, 43, 47, 49, 51–54). Some of these studies identified other Does ICP Monitoring and Treatment Improve Outcome?.
clinical factors that, in combination with severe TBI in a child, Three recent retrospective studies using large patient popula-
are indicative of a high occurrence of intracranial hyperten- tions provide direct evidence for the recommendation for this
sion. In these patients, “diffuse cerebral swelling” on CT scan is topic—two using patients as the unit of analysis (17, 19) and
75% specific for the presence of intracranial hypertension (54). one using hospitals as the unit of analysis (18). Alkhoury and
In a study of 56 brain injured patients (39 of whom suffered Kyriakides (17) and Bennett et al (18) suggest that improved
from severe TBI), 32% of children had an initial ICP measure- clinical outcomes were associated with the use of ICP monitor-
ment greater than 20 mm Hg, but 50% had ICP max greater ing for the control of intracranial hypertension. Alkhoury and

Supplement
TABLE 2. Intracranial Pressure Monitoring: Summary of Evidence

Class 3 Studies
Reference
Type of Trauma Study Design
Center n
Geographic Age (yr)
Location Outcomes Data Class Results
Use of information from ICP monitors to inform treatment: recommendation III.1.

Bennett Retrospective registry review Class 3 No ICP vs ICP
et al (19)a n = 3,084 (1,002 with ICP ICP-monitored Composite
Multisite (30 monitoring; 2,082 without) group had greater 484 patients (15.7%) had primary outcome.
children’s Age: mean, 7.03; range, NR treatment intensity 241 (11.6%) vs 243 (24.3%)
hospitals Primary: composite of mortality, than nonmonitored
participating group. Unmeasured Mortality
discharge to hospice or 197 (9.5%) vs 185 (18.5%)
in two national poor functional survival differences between
databases) (placement of both a new the ICP and non- Poor functional survival
United States tracheostomy and a new GT; ICP groups may 55 (5.5%) vs 43 (2.1%)
mortality; poor functional have contributed Mortality and poor functional survival rates were
survival) to the subsequent higher for ICP-monitored group.
treatment intensity.
With propensity matching weights to adjust for
patient-level differences and clustering by
hospital, no significant difference in functional
survival for no ICP monitor group vs ICP monitor
group (OR, 1.31; 95% CI, 0.99–1.74)
ICP monitoring not significantly associated with
hospital mortality but was associated with the
composite outcome including mortality, discharge
to hospice, or either tracheostomy or GT
placement.
The ICP-monitored group had longer hospital LOS,
more mechanical ventilation days, more days
of osmolar therapy, more days of inotropes or
pressors, and more days of pentobarbital.
Alkhoury and Retrospective Class 3 Mortality
Kyriakides n = 3,107 Differential loss to ICP monitoring was associated with a reduction
(17)a Age: follow-up; groups in mortality only for patients with a GCS score
Level I or level II ICP monitor group: mean, different at baseline of 3 (OR, 0.64; 95% CI, 0.43–1.00).
trauma centers 8.8 Monitoring of ICP was performed in only 7.7%
Multiple states No ICP monitor group: of patients who met recommended monitoring
(National mean, 8.4; range, NR criteria.
Trauma Data LOS
Bank) Mortality, hospital LOS, ICU
LOS, ventilator days ICP monitoring group had a longer hospital stay
United States than other groups: 21.0 vs 10.4 d; p < 0.001.
Longer ICU stay: 12.6 vs 6.3 d; p < 0.001
Ventilator days
9.2 vs 4.7; p < 0.001
Bennett et al Retrospective Class 3 Mortality or severe disability
(18)a GCS NR Blinding not specified; Hospitals with higher standardized ICP
Multisite (36 Head Abbreviated Injury Score differential loss monitoring rates had lower rates of mortality
children’s at least 3 to follow-up not or severe disability (p < 0.001 for the slope of
hospitals n = 4,667 specified poor outcomes by hospital monitoring rate).
participating in 55% of patients (2586/4,667) had ICP
the Pediatric Age: mean, NR; range, 0–18
monitoring. ICP monitoring independently
Health Mortality or severe disability associated with ages 1 yr old and older (OR,
Information rates per hospital, ICP 3.1; 95% CI, 2.5–3.8) vs age < 1 yr old.
System Adjusted logistic model indicated that 12.7%
database) (95% CI, 7.7–20.4) of the total variance in ICP
United States monitoring was between-hospital variance not
explained by identified patient factors.
(Continued)

Kochanek et al
TABLE 2. (Continued). Intracranial Pressure Monitoring: Summary of Evidence

Class 3 Studies
Reference
Center n
Geographic Age (yr)
Association of elevated ICP with outcomes (indirect evidence supporting the need for ICP monitoring)
Grinkeviciūte Retrospective Class 3 Survival
et al (45) n = 48 No control for 47 (97.9%) for children admitted to the PICU
PICU Age: mean, 10.6; range, 2.4 confounders GOS
Kaunas, Lithuania mo to 18 yr Indirect evidence 43 (89.6%) favorable outcome
Survival, dichotomized GOS at Association of ICP ICP and CPP
6 mo postinjury, ICP, CPP with outcomes Differences in peak ICP (22.2 vs 24.6 mm Hg,
respectively) in groups with favorable vs
unfavorable outcomes were not statistically
significant; also, no difference was seen
between groups in minimum CPP.
There was no difference in ICP maximum in
groups with good (22.2 mm Hg) vs poor
(24.6 mm Hg) outcomes.
Jagannathan Retrospective Class 3 ICP
et al (46) n = 96 Unclear if analysis of ICP control achieved in 82/96 (85%) overall.
PICU, University Age: mean, 15.1; range, 3–18 ICP monitoring 20/23 (87%) achieved ICP control with external
of Virginia ICP controlled for ventricular drain. Of three not achieving ICP
Health System confounders control, two died and one had craniectomy.
Charlottesville, Indirect evidence Refractory ICP was associated with 100%
VA Association of ICP mortality; the method used to control ICP had
with outcomes no correlation with mortality.
Death was associated with refractory raised
ICP, p < 0.0001, but not with ICP maximum,
irrespective of the surgical or medical
methods(s) used for successful reduction of ICP.
Adelson et al Randomized controlled trial Class 3 Mortality
(40) n = 75 No control for 8 of 48 deaths (17%)
Multisite 48 in multicenter study confounders (class 2 GOS-E
multinational 27 in single-center study for hypothermia trial) ICP of 20 was most sensitive and specific for
hospitals Indirect evidence good outcome.
Age: mean, 6.89; range, 0–13
Pittsburgh, PA Association of ICP The percent time with ICP < 20 mm Hg differed
Mortality, GOS-E at 3 and 6
mo postinjury with outcomes significantly in the good (90.8% ± 10.8%) vs
poor (68.6% ± 35.0%) outcome groups, p <
0.05.
Mean ICP was lower in patients who had a good
outcome versus those with a poor outcome
(good, 11.9 mm Hg; poor, 24.9 mm Hg); p =
0.036.
Wahlström et al Retrospective Class 3 Survival
(50) n = 41 No control for 38 (93%)
Neuro-ICUs Age: median, 8.8; range, 3 mo confounders GOS
at university to 14.2 yr Indirect evidence 80% favorable outcomes
hospitals Survival, dichotomized GOS at Association of ICP ICP
Umea, Sweden median 12 mo postinjury, with outcomes ICP in three nonsurvivors was significantly
ICP higher than in 38 survivors (mean, 43 ± 26 vs
13 ± 4 mm Hg).
Relationship between ICP and outcome in
survivors was not statistically analyzed.
(Continued)

Supplement

Class 3 Studies
Reference
Center n
Geographic Age (yr)
Cruz et al (43) Retrospective Class 3 Mortality

Federal University n = 45 No control for 2 (4.4%)
of São Paulo, Age: confounders GOS
and Clean Favorable outcomes: 6 Indirect evidence 37 favorable
Field Hospital Association of ICP
Unfavorable outcomes: 6.3; 8 unfavorable
São Paulo, Brazil range, 1–12 with outcomes ICP
Mortality, modified ICP peaked on day 4 in both groups.
dichotomized GOS at 6 mo ICP was significantly higher on days 2–5
postinjury, ICP in children with unfavorable vs favorable
outcomes, p = 0.02.
Daily mean ICP values ranged between 15
and 21 mm Hg on days 2–5 in the favorable
outcome group and between 19 and 26 mm Hg
on days 2–5 in the unfavorable outcome group.
Uncontrolled ICP > 40 mm Hg occurred in the
two children who died.
4.4% died; 13.3% had severe disability.
Higher ICP for days 1–5 was significantly
associated with decreased cerebral O2
extraction and worse clinical outcome, p ≤ 0.02.
Pfenninger and Retrospective Class 3 Mortality
Santi (49) n = 51 No control for 14 (27.5%) died
Pediatric Age: mean, 8.1; range, 1 mo confounders GOS
intensive care to 16 yr Indirect evidence 14 (27.5%) dead (GOS 1)
Bern, Switzerland Mortality, GOS at 6 to 12 mo Association of ICP 1 (2%) permanent vegetative state (GOS 2)
postinjury, ICP with outcomes 1 (2%) severe disability (GOS 3)
35 (68.5%) good recovery (GOS 4–5)
ICP
ICP > 40 mm Hg was associated with higher
mortality, p < 0.001.
Thirteen of 16 patients with ICP 20–40 mm Hg
had good outcomes or moderate disability.
Three of three patients with ICP < 20 mm Hg
had good outcomes or moderate disability.
Moderate to severe intracranial hypertension
(mean sustained ICP ≥ 20 mm Hg) was
associated with poor outcome, p < 0.05.
69% of monitored patients had sustained ICP >
20 mm Hg.
Chambers Retrospective Class 3 GOS
et al (30) n = 84 No control for Individual patient data NR.
Neurosurgical Age: median, 10; range, 3 mo confounders; unclear ICP and CPP
Centre at to 16 yr if patient selection Overall, thresholds of 35 mm Hg for ICP and
Newcastle GOS at 6 mo postinjury, ICP, was unbiased 45 mm Hg for CPP were the best predictors
General CPP Indirect evidence of outcome.
Hospital Association of ICP The receiver operating characteristic–defined
Newcastle, with outcomes cutoffs varied depending on the Marshall CT
United classification and ranged from 21 to 59 mm Hg.
Kingdom
ICP maximum predictive of poor outcome
was > 35 mm Hg.
(Continued)

Kochanek et al

Class 3 Studies
Reference
Center n
Geographic Age (yr)
White et al Retrospective Class 3 Survival

(51) n = 136 No control for 104 (76%) survived
Division of n = 37 with ICP monitoring confounders for ICP ICP
Pediatric Age: analysis 14% of survivors and 41% of nonsurvivors had
Critical Care Survivors: median, 6.8 Indirect evidence ICP > 20 mm Hg in the first 72 hr.
Medicine Association of ICP Those with lower mean ICP were more likely to
Nonsurvivors: median, 7.7;
Washington, DC range, 0–17 with outcomes be survivors, p < 0.005.
Survival at discharge, ICP ICP maximum and ICP measured 6, 12, and
24 hr after admission were all significantly
lower in survivors.
Downard et al Retrospective Class 3 Mortality
(44) n = 118 Retrospective review 33 (28%) died
Neurosurgery and Age: mean, 7.4; range, 0–15 Indirect evidence GOS
Emergency Mortality, GOS at last recorded Association of ICP 33 (28%) dead (GOS 1)
Medicine, patient interaction, ICP with outcomes 13 (11%) permanent vegetative state or severe
Oregon Health disability (GOS 2–3)
& Science
University and 25 (21%) moderate disability (GOS 4)
Department 47 (40%) good recovery (GOS 5)
of Pediatrics, ICP
Emanuel In a stepwise logistic regression analysis, mean ICP
Hospital and > 20 mm Hg in the initial 48 hr was significantly
Health Center associated with an increased risk of death.
Portland, OR
Michaud et al Retrospective Class 3 Mortality
(48) n = 75 Retrospective review 25 (33%) died
Level 1 trauma n = 51 with ICP monitoring Indirect evidence GOS
center, Age: mean, 8.2; range, 3 mo Association of ICP 25 (33%) dead
Harborview to 16 yr with outcomes 4 (5%) vegetative state
Medical Center Mortality, GOS at hospital 14 (19%) severe disability
Seattle, WA discharge, ICP 9 (12%) moderate disability
23 (31%) good recovery
ICP
94% of children with ICP maximum < 20 mm
Hg vs 59% with ICP maximum > 20 mm Hg
survived, p = 0.02.
48% of children with ICP elevation > 1 hr
survived compared with 89% of children with
ICP elevated for < 1 hr.
Outcome was also better in children with ICP
elevation for < 1 hr.
No statistically signiﬁcant relationship was found
between peak ICP and degree of disability.
Barzilay et al Treatment series Class 3 Mortality
(52) n = 56 Uncontrolled series 15 (27%) died
PICU, The Chaim n = 41 TBI Indirect evidence GOS
Sheba Medical Age: mean, 6.2; range, NR Association of ICP 15 (27%) died
Center Mortality, dichotomized GOS at with outcomes 17 (30%) poor recovery
Tel Aviv, Israel hospital discharge, ICP Mixed pathologies 24 (43%) good recovery
ICP
For children with severe TBI, ICP maximum
was 16.9 ± 3.1 in survivors (n = 32) and
53.7 ± 10.8 in nonsurvivors (n = 9); p < 0.01.
(Continued)
Supplement

Class 3 Studies
Reference
Center n
Geographic Age (yr)
Kasoff et al Retrospective Class 3 Mortality

(47) n = 25 Selection not specified 5 (20%) died
Department of Age: mean, 8.8; range, 3 mo (25 cases selected ICP
Neurosurgery to 17 yr over a 3-yr period) Mean of peak ICP in patients who died (n = 5)
New York Mortality, ICP Indirect evidence was 81 mm Hg (range, 55–120 mm Hg).
Association of ICP Mean of peak ICP was 18.7 mm Hg (range,
with outcomes 10–30 mm Hg) in patients who did not require
additional treatment for ICP; no deaths; no
statistical analysis presented.
Children with elevated ICP had a lower survival
rate than children with normal ICP; no
statistical analysis presented.
Alberico et al Prospective Class 3 Mortality
(41) n = 100 No control for 24 (24%) died
Medical College Age: mean, 13.39; range, confounders GOS
of Virginia 0–19 Indirect evidence 43 (43%) good outcome
Hospital Mortality, dichotomized GOS at Association of ICP
pediatric ICP
3 mo and 1 yr, ICP with outcomes 70% good outcome in children with ICP < 20 mm
service
Richmond, VA Hg with treatment vs 8% good outcome in
children with ICP refractory to treatment
(> 20 mm Hg), p < 0.05
Reducible ICP was significantly associated with
better outcome than nonreducible ICP.
Esparza et al Treatment series Class 3 Mortality
(53) n = 56 Uncontrolled series 18 (32%) died
Pediatric Age: mean, 7.6; range, 3 mo Indirect evidence GOS
Neurosurgery to 14 yr Association of ICP 18 (32%) died
Madrid, Spain Mortality, GOS (timing unclear), with outcomes 0 vegetative state
ICP 1 (1.8%) severe disability
2 (3.6%) moderate disability
35 (62.5%) good recovery
ICP
Thirteen of 13 patients (100%) with ICP
> 40 mm Hg had poor outcome (severe
disability, vegetative, or dead), and all the
patients with poor outcome died.
Four of 14 patients (≈28%) with ICP > 20–
40 mm Hg had poor outcome.
Two of 29 patients (≈7%) with ICP 0–20 mm Hg
had poor outcome.
Outcomes:
93% good, 7% poor for patients with ICP
maximum ≤ 20 mm Hg
maximum > 20 to 40 mm Hg
maximum > 40 to 60 mm Hg
maximum > 60 mm Hg (no significance test
reported)
(Continued)

Kochanek et al

Class 3 Studies
Reference
Center n
Geographic Age (yr)
Shapiro and Treatment series Class 3 Mortality

Marmarou (54) n = 22 Uncontrolled series 5 (23%) died
Albert Einstein Age: range, 3 mo to 15 yr Indirect evidence Outcome
College of Mortality, outcomes (not Association of ICP Four of 17 survivors were severely disabled.
Medicine specified), PVI, ICP with outcomes Thirteen of 17 had a good outcome or were
New York moderately disabled.
PVI and ICP
Two of the five deaths were due to uncontrolled
ICP.
Sixteen of 22 patients had PVI measured before
and after therapy.
Drainage increased PVI and decreased ICP in
14 of 16.
86% of children had ICPs exceeding 20 mm
Hg. ICP could be controlled in 14 of the 16
children whose pressure-volume index was
measured, and in those patients, there were
no deaths.
Bruce et al (42) Prospective Class 3 Mortality
Children’s n = 85 No control for 8 (9%) died
Hospital of n = 40 with ICP monitoring confounders GOS
Philadelphia Age: mean, 7.1; range, 4 mo Indirect evidence 8 (9%) died
Philadelphia, PA to 18 yr Association of ICP 3 (3.5%) persistent vegetative state
Mortality, dichotomized GOS at with outcomes 74 (87.5%) good recovery or moderate disability
6 mo, ICP ICP
Of those who had ICP monitoring (n = 40): level
of ICP related to outcome: ICP < 20 (n = 9):
67% good recovery/moderate disability; 11%
severe disability/persistent vegetative state;
22% died
ICP > 20 ≤ 40 (n = 17): 88% good recovery/
moderate disability; 6% severe disability/
persistent vegetative state; 6% died
ICP > 40 (n = 14): 57% good recovery/
moderate disability; 7% severe disability/
persistent vegetative state; 36% died
CPP = cerebral perfusion pressure, GCS = Glasgow Coma Scale, GOS = Glasgow Outcome Scale, GOS-E = Glasgow Outcome Scale Extended, GT =
gastrostomy tube, ICP = intracranial pressure, LOS = length of stay, NR = not reported, OR = odds ratio, PVI = pressure-volume index, TBI = traumatic brain injury.
New study.
a
Kyriakides (17) found that the use of ICP monitoring versus with lower rates of mortality or severe disability. However, a sub-
no ICP monitoring was associated with a reduction in mortal- sequent study by Bennett et al (19), using a propensity-weighted
ity in more severely injured patients but also showed monitor- effectiveness analysis that linked two national databases (n =
ing of ICP was performed in only 7.7% of patients who met 3,084; 1,002 with ICP monitoring and 2,082 without), reported
recommended monitoring criteria (17). Interestingly, the ICP no significant difference in functional survival between groups,
monitoring group had a longer hospital stay, longer ICU stay, no significant association between monitoring and hospital
and more ventilator days. In another retrospective study of 36 mortality, but an association between monitoring and higher
institutions, using the Pediatric Health Information Systems mortality, discharge to hospice, or either tracheostomy or gas-
database, Bennett et al (18) showed that hospitals with higher trostomy tube placement (19). The ICP-monitored group had
standardized ICP monitoring rates had better patient outcomes greater treatment intensity than the nonmonitored group, and

Supplement
authors caution that the findings could be due to unmeasured Note 2. Use of advanced neuromonitoring (brain oxygen-
differences between the groups that may have contributed to ation) should only be for patients with no contraindications
the subsequent treatment intensity. to invasive neuromonitoring such as coagulopathy and for
Multiple studies contribute indirect evidence to support the patients who do not have a diagnosis of brain death.
recommendation for this topic. For example, two studies of Changes From the Prior Edition. There are no content
combined treatment strategies suggest that improved clinical changes from the Second Edition to the recommendations.
outcomes are associated with successful control of intracranial The notes are new to this edition. Two new class 3 treatment
hypertension (41, 46). A prospective observational study of 100 series were added to the evidence base for this topic (55, 56).
children with severe TBI treated with varying combinations of
hyperventilation, diuretics, CSF drainage, sedation, pharmaco- Introduction
logic paralysis, and barbiturates reported that children whose Advanced monitoring systems provide information about
ICP was successfully lowered had better 1-year outcomes than cerebrovascular and metabolic function. In children with
children whose ICP was uncontrollable (but worse than those severe TBI, the addition to ICP monitoring of advanced neu-
without intracranial hypertension) (41). A retrospective review romonitoring techniques such as microdialysis, electrophysiol-
of a prospectively acquired TBI database showed that reduced ogy assessments, and examination of cerebral autoregulation
survival and worsened outcomes in children with severe TBI were may help identify patients needing particular treatments (57).
associated with intracranial hypertension refractory to treatment, If treatment then prevents unwanted cerebral pathophysiologic
rather than peak ICP per se (46). In this study, successful control processes and is shown to improve function and outcome, the
of ICP, irrespective of treatment modality (osmolar therapy, CSF use of these advanced monitoring systems may be warranted as
drainage, decompression, etc), was deemed to be important. part of optimal critical care (58, 59).
The decision to insert and use any monitoring device
depends on understanding the data and information derived Evaluation of the Evidence
from the monitor that permits targeted evidence-based care. Quality of the Body of Evidence. Studies included for this topic
Because there are no imaging or other biomarkers that indicate addressed the use of advanced neuromonitoring methods to
a patient with intracranial hypertension, it is recommended improve outcomes for children with severe TBI, and what thresh-
that ICP is measured to determine if intracranial hypertension old value should be targeted for measures of cerebrovascular and
is present. Given that much of present care is predicated on metabolic function. No studies meeting inclusion criteria were
prevention and treatment of elevated ICP, detection of elevated identified that evaluated the use of Pbro2 monitoring and linked
ICP with monitoring is considered to be more capable of allow- their use to improvements in outcomes. Four studies—three treat-
ing for timely delivery and accurate titration of treatment than ment series (55, 56, 60) and one prospective cohort (61)—consti-
without the use of an ICP monitor. Although they represent tuted the evidence for the recommendation about a threshold, if
only class 3 evidence for long-term outcomes related to ICP Pbro2 monitoring was used. The studies were small, with moder-
monitoring, these studies support the association of successful ate consistency and low precision, reporting direct evidence. The
ICP monitor–based management of intracranial hypertension overall quality of the body of evidence is low (Table 3).
with improved survival and neurologic outcome. Applicability. The included studies were small and con-
ducted at single sites. They included a range of ages for pedi-
Indications From Adult Guidelines atric patients. Two were conducted in the United States (56,
Consistent with the recommendations in this edition, the 60) and two in South Africa (55, 61). The applicability of the
Fourth Edition of the adult guidelines provides a level III rec- evidence is limited.
ommendation to monitor ICP (14).
Summary of Evidence
Four class 3 studies (55, 56, 60, 61), two new (55, 56) and two
Advanced Neuromonitoring
from the Second Edition, provided evidence to support the rec-
Recommendations ommendation (60, 61). One prospective cohort (61) included
Strength of Recommendation: Weak patients from one of the treatment series (55) (Table 4).
Levels I and II
Evidence Synthesis.
What Threshold Value Should Be Targeted for Measures of
Cerebrovascular and Metabolic Function?
Level III The four studies that focused on this question looked for an
To Improve Overall Outcomes. III.1. If brain tissue oxygen- association between Pbro2 levels and favorable or unfavorable
ation (Pbro2) monitoring is used, maintaining a level greater outcome (55, 56, 60, 61).
than 10 mm Hg is suggested. Stippler et al (56) analyzed over 8,000 hours of monitoring
Note 1. There was insufficient evidence to support a rec- of 46 children with severe TBI who were treated according to a
ommendation for the use of a monitor of Pbro2 to improve protocol targeting Pbro2 at 25 mm Hg. Overall levels were high
outcomes. and 30 mm Hg represented the highest combined sensitivity and

Kochanek et al
TABLE 3. Advanced Neuromonitoring: Quality of the Body of Evidence

Quality of
Evidence
Total Consisten- Precision (High,
No. of Meta-Analysis No. of cy (High, Directness (High, Moderate,
Studies Recommen Possible Subjects Moderate, (Direct or Moderate, Low, or
Topic Study Design dations (Yes or Noa) (n) Low) Indirect) Low) Insufficient)
Brain tissue O2 1 prospective III.1 No, different 114b Moderate Direct Low Low
monitoring 3 treatment designs
thresholds series
a
The two articles by Figaji et al (55, 61) are assumed to include the same patients.
b
TABLE 4. Advanced Neuromonitoring: Summary of Evidence

Class 3 Studies
Reference Study Design

Type of Trauma Center n
Geographic Age (yr)
Pbro2 monitoring thresholds: recommendation III.1.

Stippler et al (56)a Treatment series Class 3 Mortality
Pediatric n = 46 Uncontrolled 2 deaths (4.3%) during acute hospitalization
Neurotrauma Age: mean, 9.4; range, series GOS
Center, University 0.1–16.5 70% of patients had favorable outcomes
of Pittsburgh Mortality, dichotomized GOS at ICP
Pittsburgh, PA 6 mo postinjury, ICP, CPP There was no significant difference in ICP during the
first 6 d postinjury between patients with favorable vs
unfavorable outcomes.
CPP (with protocol targeting maintenance of Pbro2 >
25 mm Hg)
A Pbro2 of 30 mm Hg was associated with the highest
combined sensitivity (20.0%)/specificity (80.7%) for
favorable neurologic outcome at 6 mo.
Pbro2 alone was not independently associated with outcome.
High, rather than the expected low, values of PbrO2
were observed in some patients with refractory
intracranial hypertension and low CPP.
Figaji et al (55)a Treatment series Class 3 GOS
Red Cross War n = 28 (appear to be part of Uncontrolled A greater delta Pbro2/delta Pao2 was associated with a
Memorial the cohort for Figaji et al series lower probability of favorable outcome.
Children’s [61]) Estimate: –1.839; 95% CI, 0.03–0.78; p = 0.02
Hospital Age: mean, 5.8; range, 9 mo GOS outcomes:
Cape Town, South to 11 yr GOS 1 (died) (n = 3, 12.5%)
Africa GOS at 6 mo postdischarge, GOS 2 (n = 0)
Fio2, Pao2, Pbro2 GOS 3 (n = 4, 16.7%)
GOS 4 (n = 7, 29.2%)
GOS 5 (n = 10, 41.7%)
Fio2, Pao2, Pbro2
Effect of increased Fio2 on Pao2 and Pbro2:
Induced hyperoxia significantly increased both Pao2, p
< 0.0001, and Pbro2, p < 0.0001.
When Fio2 was increased, Pao2 significantly increased
(p < 0.0001) and Pbro2 significantly increased (p <
0.0001).
Normobaric hyperoxia increased Pbro2, but the
responses varied in patients.
(Continued)
Supplement
TABLE 4. (Continued). Advanced Neuromonitoring: Summary of Evidence

Class 3 Studies

Geographic Age (yr)
Figaji et al (61) Prospective Class 3 Pbro2 < 5 mm Hg for >1 hr or

Red Cross n = 52 Unclear if Pbro2 < 10 mm Hg for >2 hr
Children’s Age: mean, 6.5; range, 9 mo outcome Mortality
Hospital to 14 yr assessment Adjusted OR, 26.8; 95% CI, 2.7–265; p = 0.005
Cape Town, South Mortality at 6 mo postinjury was unbiased
PCPC
Africa PCPC Unfavorable: severe disability or death
Adjusted OR, 27.4; 95% CI, 1.9–391; p = 0.015
Independent of other significant factors such as ICP, CT,
low Pao2, and CPP mortality
Narotam Treatment series Class 3 Mortality
et al (60) n = 16 Uncontrolled Normal initial Pbro2 (≥ 10 mm Hg)
Level II trauma Age: mean, 14; range, 1.5–18 series No patients died
center Mortality at 3 mo postinjury Mean initial Pbro2
Creighton University 10 survivors vs 6 deaths
Medical Center
16.07 ± 18.7 vs 6.76 + 6.69 mm Hg; p = 0.247
Omaha, NE
Mean final Pbro2
Survivors vs nonsurvivors
Pbro2, 25.0 ± 11.57 vs 8.53 ± 11.0 mm Hg; p = 0.01
CPP = cerebral perfusion pressure, GOS = Glasgow Outcome Score, ICP = intracranial pressure monitoring, OR = odds ratio, Pbro2 = brain tissue
oxygen, PCPC = pediatric cerebral performance category.
New study.
a
Different abbreviations such as pBto2/Pbto2 and Ptio2 are used to denote brain tissue oxygen and brain tissue oxygen tension; we use PbrO2 for consistency,
which may differ from what was used by the authors of the cited studies.
specificity for favorable outcomes. However, the sensitivity was Neuroimaging

low (20%), and the unexpected observations of high rather than
Recommendations
low Pbro2 in patients with intracranial hypertension and com-
promised CPP suggest the need for studies designed to under- Strength of Recommendation: Weak
stand what may be a complex relationship between Pbro2 and
Levels I and II
outcome. Figaji et al (61) used a treatment threshold of 20 mm
Hg and reported an association of poor outcome with Pbro2 less
than 10 mm Hg, which was even stronger for Pbro2 less than 5
mm Hg. In another analysis of a subgroup of the studied patients, Level III
Figaji et al (55) reported that in patients whose Pbro2 changed To Improve Overall Outcomes. III.1. Excluding the possibility
more in response to changes in Pao2, outcomes were worse. of elevated ICP on the basis of a normal initial (0–6 hr after
Narotam et al (60) also reported an association between unfavor- injury) CT examination of the brain is not suggested in coma-
able outcome and Pbro2 less than 10 mm Hg. Combined, this evi- tose pediatric patients.
dence suggests that if this advanced monitoring modality is used, III.2. Routinely obtaining a repeat CT scan greater than 24
it would be prudent to target the more conservative threshold of hours after the admission and initial follow-up is not suggested
greater than 10 mm Hg. Although all these studies suggest that for decisions about neurosurgical intervention, unless there is
higher Pbro2 levels are associated with better outcomes to some either evidence of neurologic deterioration or increasing ICP.
degree, the relationship appears complex and the studies were Changes From the Prior Edition. Recommendation III.1. is
not designed to isolate the effect of treating to specific thresholds new to this edition. Two new class 3 studies—one retrospective
or to compare the utility of Pbro2 to other measures. observational study (62) and one treatment series (63)—were
added to the evidence base for this topic.
Indications From Adult Guidelines
Due to insufficient evidence, there are no recommendations Introduction
in the Fourth Edition of the adult guidelines regarding Pbro2 CT of the head is the current preferred imaging technique for
monitoring or thresholds (14). rapid detection of intracranial injury, signs of mass effect, and/

Kochanek et al
or cerebral edema for trauma victims with severe TBI. Acute used to rule out intracranial hypertension, and whether rou-
CT imaging is universally performed in high-income countries tine repeated CT scans, after the initial scan, should be used
during initial evaluation, and detection of intracranial injury to make decisions about neurosurgical intervention. The over-
is common (62–75%) among these severely injured patients all quality of the body of evidence is low and is based on one
(64, 65). new class 3 treatment series (63), one new class 3 retrospective
The Rotterdam head CT score grades severity of injury and study (62), and one class 3 retrospective study from the Second
increasing scores are associated with greater mortality (66). Edition of these guidelines (72) (Table 5).
Risk of mortality increases with Rotterdam scores in children Applicability. The recommendations are supported by
in a similar pattern to adults (67). three small, uncontrolled studies (62, 63, 72). Applicability is
Although an abnormal head CT has been an indication to considered limited.
monitor ICP, a question remains regarding the risk of intra- Summary of Evidence. Three class 3 studies—two new (62,
cranial hypertension in patients with a normal head CT. In the 63) and one from the Second Edition—provide evidence to
Second and Third Editions of the adult TBI guidelines, ICP support the recommendations (72) (Table 6).
monitoring was indicated for patients with a normal initial
head CT if two or more features were present: age more than Evidence Synthesis
40 years old, unilateral or bilateral motor posturing, or systolic Use of CT to Rule Out Intracranial Hypertension. Bailey et
blood pressure (SBP) less than 90 mm Hg (68, 69). The First and al (63) conducted a retrospective chart review of moderate to
Second Editions of the Pediatric TBI Guidelines referenced this severe pediatric TBI cases to evaluate factors associated with
recommendation (1, 70). However, for the Fourth Edition of the ICP monitoring and to determine to what extent normal ICP
adult TBI guidelines, this recommendation was removed due to (< 20 mm Hg) can be predicted by normal initial head CT
lack of evidence meeting current standards (14). No decision scans (n = 299). ICP monitors were placed in 13% of children
rule regarding CT and indications for ICP monitoring has been (9/68) with normal initial head CTs (Rotterdam and Marshall
endorsed for infants and children with severe TBI. 1) and 31% (30/98) of those with Marshall 2 (diffuse injury,
An initial CT is obtained acutely to evaluate intracranial open cisterns, and midline shift 0 mm to 5) scores. Seven of
injury, need for neurosurgical procedures, and for signs of nine children (78%) with an initial normal head CT but who
intracranial hypertension; however, routine repeat CT imag- were unable to localize pain when examined by a pediatric
ing is more controversial. Repeating a CT scan in children with neurosurgeon after initial resuscitation developed ICP greater
severe TBI is usually considered when there is 1) no evidence than 20 mm Hg during the first 24 hours of pressure monitor-
of neurologic improvement; 2) persistent or increasing ICP; or ing. This series of nine patients constitutes a highly selected
3) an inability to assess neurologic status (e.g., sedation, para- and uncontrolled sample, providing low-quality evidence to
lytic agents) (71). support the recommendation to not exclude the possibility of
intracranial hypertension based on a normal initial head CT.
Evaluation of the Evidence Use of Repeated CT Scans to Make Decisions About Neu-
Quality of the Body of Evidence. Studies included for this topic rosurgical Interventions. Bata and Yung (62) conducted a
addressed the questions about whether initial CTs should be retrospective review of children with mild, moderate, and
TABLE 5. Neuroimaging: Quality of the Body of Evidence

Quality of
Meta- Consisten- Direct- Precision Evidence
Analysis Total No. cy (High, ness (High, (High, Moderate,
No. of Studies Recommen- Possible of Subjects Moderate, (Direct or Moderate, Low, or
Topic Study Design dation (Yes or Noa) (n) Low) Indirect) Low) Insufficient)
Use of initial CT 1 treatment series III.1. NA 9 NA Direct NA Low

to rule out (subset of a
intracranial retrospective
hypertension review)
Use of repeated 2 retrospective III.2. No 73 severe Moderate Direct Low Low
CT scans
to make
decisions about
neurosurgical
interventions
NA = not applicable.
a

Supplement
TABLE 6. Neuroimaging: Summary of Evidence

Class 3 Studies
Reference Study Design n

Type of Trauma Center Age (yr)
Geographic Location Outcomes Data Class Results
Use of initial CT to rule out intracranial hypertension: recommendation III.1.

Bailey et al (63)a Treatment series subset of a Class 3 Early intracranial hypertension
Primary Children’s retrospective review Extrapolated a subset Of nine patients with a Marshall
Medical Center n=9 of nine patients, score of I who had ICP monitoring,
Salt Lake City, UT Age: resulting in a highly seven had early intracranial
ICP monitor: median, 7 selected sample hypertension. This was not
without control for significantly different from those
No ICP monitor: median, 6.7; range, NR confounders with radiographic pathology
Early intracranial hypertension with meeting Marshall 2 criteria.
normal initial head CT (Marshall I)
Use of repeated CT scans to make decisions about neurosurgical interventions: recommendation III.2.
Bata and Yung (62)a Retrospective Class 3 36 had routine repeated CTs.
Women’s and n = 71 No control for potential 35 had clinically indicated repeat CTs.
Children’s 33 severe confounders Neurosurgical procedures
Hospital Age: median, 10 yr; range, NR Five patients required surgical
North Adelaide, SA, Neurosurgical procedures resulting intervention (two severe had
Australia from repeat CT, change in routine repeat CT; two severe
management or need for further and one moderate had clinically
imaging resulting from repeat CT indicated repeat CT).
Medical management
Of the 33 severe cases, 10 had
change in medical management
following repeat CT.
OR for delayed ICP monitoring or
EVD was not significantly different
between patients with worse repeat
head CT vs those with stable
radiologic lesions:
ICP monitoring
OR, 2.48; 95% CI, 0.38–16.29;
p = 0.31
EVD insertion
OR, 3.6; 95% CI, 0.21–61; p = 0.40
Figg et al (72) Retrospective Class 3 115 repeat CT scans conducted
Level 1 trauma n = 40 No control for potential 87 routine (76%)
center Age: mean, 9.6; range, 2 mo to 17 yr confounders 24 for increased ICP (21%)
Grand Rapids, MI Urgent neurosurgical procedures 4 for neurologic change 3%)
resulting from serial CT scans Neurosurgical procedures
Five patients (4.3%) had a surgical
intervention based on findings
from the serial CT scans; however,
all five scans were ordered due
to clinical indicators (ICP or
neurologic status), not as routine
follow-up.
Findings from repeat scans after the
admission and initial follow-up study
No change: 61 (53%)
Improvement: 39 (34%)
Worsening: 15 (13%)
EVD = external ventricular drain, ICP = intracranial pressure, NR = not reported, OR = odds ratio.
New study.
a

Kochanek et al
severe TBI treated over a decade. They evaluated whether Introduction

routine head CT scans altered surgical or MM. Among 1,675 In children with severe TBI, mortality is often due to refractory
admissions with head trauma, 71 met criteria for the analysis sustained increases in ICP, defined as intracranial hypertension.
(33 severe). Of five patients who had progression of intracra- Because management of severe TBI in the PICU is largely focused
nial injury demonstrated on repeat imaging, two occurred on the management of raised ICP and preservation of perfusion
in patients who were reimaged without worsening signs as represented by CPP (MAP minus the mean ICP [28]), at pres-
and symptoms of intracranial injury. Both had severe TBI. ent, preventing intracranial hypertension is central to current
The other three repeat CTs were obtained because of signs neurocritical care of these children. Prevention of severe intracra-
and symptoms of worsening injury. The likelihood of hav- nial hypertension is also thought to be important to avoid cere-
ing delayed ICP monitoring or an external ventricular drain bral herniation events leading to a cascade of often fatal sequelae.
(EVD) after initial trauma evaluation was not significantly Brief increases in ICP that return to normal in less than 5 minutes
greater in subjects with progression of intracranial injury ver- are believed to be insignificant, although some have challenged
sus those with stable radiologic lesions. that belief in adult patients (75). However, sustained increases of
Figg et al (72) retrospectively reviewed 40 pediatric patients greater than or equal to 20 mm Hg for greater than or equal to
treated from January 1990 to December 2003 for severe TBI, 5 minutes likely warrant treatment (76). Based in large part on
and examined whether serial CT scans led to urgent neuro- adult studies, an ICP treatment threshold of 20 mm Hg has been
surgical operative intervention. One hundred fifteen serial CT used in most centers for decades.
scans were ordered (76% routine follow-up, 21% increased Normal values for MAP and hence CPP are lower in chil-
ICP; 3% neurologic change). Results of these scans showed dren, particularly in infants and young children, but optimal
no change (53%), improvement (34%), and worsening (13%). ICP in the postinjury period is undefined. It has been shown in
Five patients (4.3%) had a surgical intervention supported by anesthetized children without TBI that the lower CPP limit of
results of the serial CT scan (one evacuation of epidural hema- autoregulation of cerebral blood flow (CBF) is similar in young
toma; one evacuation of a subdural hematoma; one burr hole; children versus older children—and does not decrease below
and three for additional EVDs). All five scans were ordered ≈60 mm Hg (77). Thus, young children have less autoregulatory
based on a clinical indicator (ICP or neurologic status), not as reserve than older children. For example, the difference in CPP
routine follow-up. between normal and the lower limit of autoregulation is smaller
Both reports by Bata and Yung (62) and Figg et al (72) were in infants and young children than it is in older children. This
studies conducted at a single center. They lacked consistent cri- suggests the possible need to set a lower ICP therapeutic target
teria for routine reimaging; time elapsed after injury or initial to maintain adequate perfusion for infants and young children
CT imaging also varied. Thus, the quality of evidence is low to when compared with older children or adults with TBI.
support the recommendation against use of serial CT scans to Although an ICP threshold of 20 mm Hg is generally used,
make decisions about neurosurgical interventions. an even lower threshold may be physiologically appropriate for
infants and young children. It should also be recognized that some
Indications From Adult Guidelines of the studies defining the ICP threshold used therapies that are
The adult guidelines do not include neuroimaging as a no longer routinely recommended such as aggressive hyperventi-
topic (14). lation or moderate hypothermia. Finally, in light of the heteroge-
neity of the pathology and pathophysiology in pediatric TBI, ICP
management may need to be individualized in some cases.
THRESHOLDS
Evaluation of the Evidence
Thresholds for Treatment of Intracranial Hypertension
Quality of the Body of Evidence. Studies included for this topic
Recommendations addressed the question about the target threshold for treating
Strength of Recommendation: Weak intracranial hypertension to produce the best outcomes. The
evidence consists of 10 small (30, 40, 43, 45, 47, 49, 51, 73, 74, 78)
Levels I and II
and two somewhat larger class 3 observational studies providing
direct evidence (41, 44). The consistency is moderate and preci-
sion low, rendering a low overall quality of evidence (Table 7).
Level III Applicability. The included studies were conducted in
To Improve Overall Outcomes Spain, Switzerland, United States, United Kingdom, Brazil,
III.1. Treatment of ICP targeting a threshold of less than 20 mm and Lithuania. Ten (30, 40, 43, 45, 47, 49, 51, 73, 74, 78) of
Hg is suggested. the twelve (30, 40, 41, 43–45, 47, 49, 51, 73, 74, 78) have small
Changes From Prior Edition. There are no content changes samples and most have other design flaws that taken together,
from the Second Edition to the recommendations. Two new call into question their applicability.
class 3 retrospective observational studies were added to the Summary of Evidence. Twelve class 3 studies, two new (73,
evidence base for this topic (73, 74), and one class 3 study from 74) and 10 from the Second Edition (30, 40, 41, 43–45, 47, 49, 51,
the Second Edition was removed (53). 78), provide evidence to support the recommendations (Table 8).

Supplement
TABLE 7. Threshold for Treatment of Intracranial Hypertension: Quality of the Body of

Evidence
Quality of
Evidence
No. of Meta- Consistency Precision (High,
Studies Analysis Total (High, Directness (High, Moderate,
Study Recommen- Possible No. of Moderate, (Direct or Moderate, Low, or
Topic Design dation (Yes or Noa) Subjects (n) Low) Indirect) Low) Insufficient)
Target intracranial 3 prospective III.1. No 649b Moderate Direct Low Low

pressure threshold to
9 retrospective
improve outcomes
a
Patients in Mehta et al (73) are a subset of the sample in Miller Ferguson et al (74).
b
TABLE 8. Threshold for Treatment of Intracranial Hypertension: Summary of Evidence

Class 3 Studies
Study Design
Reference n
Target ICP threshold to improve outcomes: recommendation III.1.

Miller Ferguson Retrospective Class 3 Outcome by threshold (> 14, > 20, > 30 mm
et al (74)a n = 85 Selection of Hg).
PICU of a tertiary Age: mean, 5.1; range, not reported patients for ICP GOS
children’s hospital Dichotomized GOS at 6 mo monitoring at No significant difference in outcomes
Pittsburgh, PA postinjury (1–2 unfavorable, 3–5 discretion of across three estimates utilizing three
favorable) neurosurgery different thresholds
Note: article (p. 446) indicates
“favorable 1–2, unfavorable 3–5”
Mehta et al (73)a Retrospective Class 3 Outcome by threshold (< 15 mm Hg, < 20 mm
Pediatric Neurotrauma n = 22 Small sample size Hg)
Registry Age: median, 6.1 mo; range, 0–2 yr GOS
Pittsburgh, PA Dichotomized GOS 6 mo postinjury No significant difference between outcome
(favorable 3–5; unfavorable 1–2); groups in daily mean ICP over first 7 d.
ICP; CPP; physiologic variables No significant difference between number
Subset of 85 patients in Miller of hourly readings of ICP > 20 mm Hg
Ferguson et al (74) between outcome groups.
Note: article (p. 415) indicates
“favorable [GOS: 1–2] and
unfavorable [GOS: 3–5]
outcomes.” This is transposed
from the generally accepted
order (e.g., GOS 1 = dead).
Grinkeviciūte et al (45) Prospective Class 3 Survival
Kaunas University of n = 48 No control for 97.9% for children admitted to the PICU
Medicine Age: mean, 10.6; range, 2.4 mo to confounders. GOS
Kaunas, Lithuania 18 yr Insufficient No significant difference in peak ICP in
Survival, dichotomized GOS at 6 mo power to detect groups with favorable vs unfavorable
postinjury outcome outcomes (22.2 vs 24.6 mm Hg,
respectively)
Five patients were described as having poor
outcomes.
(Continued)

Kochanek et al
TABLE 8. (Continued). Threshold for Treatment of Intracranial Hypertension: Summary of

Evidence
Class 3 Studies
Study Design
Reference n
Adelson et al (40) Randomized controlled trial of Class 3 GOS

Multicenter study moderate hypothermia vs No control for Mean ICP was lower in children with
(n = 6)b normothermia plus medical confounders in good (11.9 ± 4.7 mm Hg) vs poor
management ICP analysis (24.9 ± 26.3 mm Hg; p < 0.05) outcome.
Post hoc analysis of relationship Percent time with ICP < 20 mm Hg differed
between ICP and outcome significantly in the good (90.8% ± 10.8%)
n = 47 vs poor (68.6% ± 35.0%; p < 0.05)
Age: mean, 6.89; range, 0–13 outcome groups.
Dichotomized GOS at 3 and 6 mo ICP > 20 mm Hg was the most sensitive and
specific for poor outcome.
Cruz et al (43) Prospective Class 3 GOS
Federal University n = 45 No control for Daily mean ICP values ranged between
of São Paulo, and Children with ICP < 15 mm Hg confounders 15 and 21 mm Hg on days 2–5 in the
Clean Field Hospital were excluded favorable outcome group and between
São Paulo, Brazil Age: 19 and 26 mm Hg on days 2–5 in the
unfavorable outcome group.
Favorable Outcome: mean, 6
ICP was significantly higher (p ≤ 0.02) on
Unfavorable Outcome: mean, 6.3; days 2–5 in children with unfavorable vs
range, 1–12 favorable outcomes.
Dichotomized GOS at 6 mo 82% of the patients had favorable outcome.
postinjury
ICP peaked on day 4 in both groups.
Uncontrolled ICP > 40 mm Hg occurred in
the two children who died.
Pfenninger and Santi Retrospective Class 3 GOS
(49) n = 26 with ICP monitoring and No control for Mean sustained ICP ≥ 20 mm Hg was
University Children’s critical care management; 51 confounders, associated with poor outcome (p < 0.05).
Hospital total potential
Bern, Switzerland Age: mean, 8.1; range, 1 mo to 16 selection bias
yr in children who
Dichotomized GOS at 6 to 12 mo received ICP
postinjury monitoring
Chambers et al (30) Retrospective Class 3 GOS

Neurosurgical Centre n = 84 No control for Threshold of ICP over 35 mm Hg was the
at Newcastle Age: median, 10; range, 3 mo to confounders; best predictor of poor outcome.
General Hospital 16 yr unclear if patient The receiver operating characteristic–
Newcastle, United Dichotomized GOS at 6 mo selection was defined cutoffs varied depending on the
Kingdom postinjury unbiased Marshall CT classification and ranged
from 21 to 59 mm Hg.
White, 2001 (51) Retrospective Class 3 Mortality
Johns Hopkins n = 37 with ICP monitoring; 136 No control for 14% of survivors and 41% of nonsurvivors
Hospital total confounders for had ICP > 20 mm Hg in the first 72 hr.
PICU database and Age: ICP analysis, Patients with lower mean ICPs more likely
the Pediatric Trauma Survivors: median, 6.8 potential to survive (p < 0.005).
Registry selection bias No other threshold was specifically
Nonsurvivors: median, 7.7; range, in patients who
0–17 examined.
received ICP
Mortality monitoring ICP maximum and ICP measured at 6,
12, and 24 hr after admission were
significantly lower in survivors.
(Continued)

Supplement
TABLE 8. (Continued). Threshold for Treatment of Intracranial Hypertension: Summary of

Evidence
Class 3 Studies
Study Design
Reference n
Downard et al (44) Retrospective Class 3 Mortality

Oregon Health n = 118 Primary objective In a stepwise logistic regression analysis,
Sciences University Age: mean, 7.4; range, 0–15 was CPP mean ICP > 20 mm Hg in the initial 48 hr
trauma registry and Mortality thresholds; was significantly associated with an
Legacy Emanuel treatments increased risk of death (p = 0.001).
Hospital and Health varied according
Center to physician
Portland, OR preference.
Kasoff et al (47) Retrospective Class 3 Mortality

Westchester County n = 25 No control for Mean of peak ICP in patients who died (n
Medical Center Age: mean, 8.8; range, confounders, = 5) was 81 mm Hg (range, 55–120 mm
New York 3 mo to 17 yr selection bias Hg). Mean of peak ICP was 18.7 mm Hg
Mortality unclear (range, 10–30 mm Hg) in patients who did
not require additional treatment for ICP
and there were no deaths; no statistical
analysis was presented.
Alberico et al (41) Prospective Class 3 GOS
Medical College of n = 100 No control for 70% good outcome in children with ICP <
Virginia Hospital Age: mean, 13.39; range, 0–19 confounders 20 mm Hg with treatment vs 8% good
pediatric service GOS at 1-yr postinjury outcome in children with ICP refractory to
Richmond, VA treatment (> 20 mm Hg), p < 0.05
Pfenninger et al (78) Retrospective Class 3 Mortality

University Children’s n = 24 No control for ICP > 40 mm Hg was associated with
Hospital Age: mean, 6.6; range, 3 mo to 14 confounders higher mortality (p < 0.001)
Bern, Switzerland yr GOS
Mortality; GOS (follow-up time not 13 of 16 patients with ICP 20–40 mm Hg
specified) had good outcome or moderate disability.
Three of three patients with ICP < 20 mm
Hg had good outcome or moderate
disability.
CPP = cerebral perfusion pressure, GOS = Glasgow Outcome Scale, ICP = intracranial pressure.
New study.
a
1) Children’s Hospital of Pittsburgh, Pittsburgh, PA; 2) University of California, Davis, Sacramento, CA; 3) Jackson Memorial Hospital and Children’s Hospital
b
of Miami, Miami, FL; 4) Primary Children’s Hospital, Salt Lake City, UT; 5) Pennsylvania State University Hershey Medical Center, Hershey, PA; and 6)
Harborview Medical Center, Seattle, WA.
Evidence Synthesis 18 mm Hg for 25–96 mo; ≥ 20 mm Hg for 97–156 mo). They
What Is the Target Threshold to Produce the Best Outcomes?. reported significantly lower mean ICP in children with good
Two retrospective studies assessed patient outcomes across dif- versus poor outcomes but did not report information about
ferent thresholds for ICP (73, 74). The 22 patients in Mehta et the effect of different thresholds by age on outcome.
al (73), who were under the age of 2 years old, were a subset of Seven studies used the ICP threshold of 20 mm Hg as the
the 85 pediatric patients in Miller Ferguson et al (74). Miller indicator of intracranial hypertension (41, 44, 45, 47, 49, 51,
Ferguson et al (74) associated thresholds of greater than 14, 78). One prospective study reported no significant differ-
20, and 30 mm Hg with outcome, and Mehta et al (73) used ence in 6-month postinjury GOS based on ICP levels for 48
thresholds of less than 15 or 20 mm Hg (74). No significant patients (45). Two studies, one retrospective (49) (n = 26) and
difference was observed across thresholds in dichotomized one prospective (n = 100) (41), reported significantly better
Glasgow Outcome Scale (GOS) at 6 months postinjury in GOS scores at 6 and 12 months postinjury for patients with
either study. lower ICP. The remaining four retrospective studies, including
In a subanalysis of a RCT of hypothermia (n = 47), Adelson a total of 204 patients, reported higher mortality in patients
et al (40) instituted treatment for increased ICP at different with higher ICP. Methods to identify ICP and initiate treat-
thresholds according to age (≥ 15 mm Hg for 0–24 mo; ≥ ment varied across studies.

Kochanek et al
In a prospective study of 45 pediatric patients, Cruz et al (86). If autoregulation is disrupted after TBI, then a decrease in
(43) used the ICP threshold of 15 mm Hg as the indicator of CPP may induce cerebral ischemia. With continuous monitor-
intracranial hypertension, and reported lower daily mean ICP ing of MAP and ICP, CPP can be followed and manipulated by
values (days 2–5) in the favorable versus unfavorable outcome interventions that attempt to avoid both regional and global
group using the GOS at 6 months postinjury. In a retrospective ischemia. The optimal CPP for therapy remains unknown.
study of 84 pediatric patients, Chambers et al (30) reported There are age-related differences in MAP, CBF, and CMRO2
that an ICP threshold of 35 mm Hg was the best indicator of from infancy through adulthood. Because pediatric values are in
poor outcome using the GOS at 6 months postinjury. general lower than adult values, we need to know whether there
A methodologic concern with studies of ICP thresholds is are age-specific thresholds or targets for CPP that should be used
whether the studies were conducted with an inherent bias for ICP during critical care management of pediatric severe TBI.
less than 20 mm Hg as the a priori therapeutic target for some or There are three main limitations in comparing CPP data
all patients. In addition, variable statistical approaches were used from various studies for the purpose of identifying whether
to adjust for confounding variables in examining the association low CPP is harmful, or whether there is an age-related “critical
between ICP and outcome. Another important limitation was threshold” that should be targeted in treatment. First, there may
that there was no consistent approach to assessing the relation- be a problem with the measurement of CPP, which relates to
ship between outcome and either the timing or duration of intra- position of zero calibration for ICP and MAP, ICP device used,
cranial hypertension after TBI. Generally, mean or peak values, and the practice of head elevation. Together, these differences
or ICP values within a given epoch were used. Consequently, the may lead to a measurement difference of 5–10 mm Hg (related
available evidence provides a weak level III recommendation. to the vertical distance between the two zeroing points). Second,
the real-time numerical value of CPP not only reflects intracra-
Indications From Adult Guidelines nial tissue and fluid dynamics but also the CPP level that is being
The Fourth Edition of the adult TBI guidelines provides a level targeted by those at the bedside. There is considerable variance
III recommendation that suggests a combination of ICP values in the referenced studies, with respect to their description of any
and clinical and brain CT findings may be used to make man- ICP- or CPP-directed strategy (Table 11). Third, the CPP sum-
agement decisions (14). This recommendation is also consid- mary statistic that is used in the analysis is different in many of
ered clinically relevant in pediatric patients. the studies (Table 11) and only one report describes excluding
preterminal data; the rest of the reports do not discuss whether
Thresholds for CPP these data are included or excluded.
Recommendations Taken together, caution should be applied when interpret-
Strength of Recommendations: Weak ing the results from the pediatric TBI CPP studies and apply-
ing the information to treatment strategies for TBI.
Levels I and II
There was insufficient evidence to support a level I or II recom- Evaluation of the Evidence
mendation for this topic. Quality of the Body of Evidence. Studies included for this
topic addressed the questions about what are the minimum
Level III thresholds and target ranges for managing CPP; are ranges
To Improve Overall Outcomes. III.1. Treatment to maintain a age-specific, and what is the target threshold for infants? Mul-
CPP at a minimum of 40 mm Hg is suggested. tiple class 3 studies provided low-quality evidence supporting
III.2. A CPP target between 40 and 50 mm Hg is suggested to a minimum target of 40 mm Hg and use of age-specific ranges
ensure that the minimum value of 40 mm Hg is not breached. (Table 11). Although one class 2 study provided data support-
There may be age-specific thresholds with infants at the lower ing use of age-specific ranges, it was not considered sufficient
end and adolescents at or above the upper end of this range. to make a level II recommendation (79). Evidence from two
Changes From Prior Edition. There are no content changes small class 3 studies was insufficient to make a recommenda-
from the Second Edition to the recommendations. Of the 15 tion specific to infants (73, 80) (Table 9).
included studies (30, 40, 44, 52, 60, 61, 73, 74, 79–85), four are Applicability. Twelve of the 15 studies were published since
new to this edition. One new class 2 (79) and three new class 3 2000 (30, 40, 44, 60, 61, 73, 74, 79, 81, 83–85). The body of evi-
retrospective observational studies were added to the evidence dence included multisite studies and use of registry data from
base for this topic (73, 74, 85). multiple sites. Countries included the United States, United King-
dom, Israel, Germany, South Africa, and Switzerland. Although
Introduction two studies focused on infants (73, 80), the remainder contained
CPP—as defined by MAP minus the mean ICP (28)—is the a range of ages. There are no major applicability concerns.
pressure gradient driving CBF, which, in turn, in the normal Summary of Evidence. Of the 15 studies summarized in the
state, is autoregulated and coupled with cerebral metabolic rate Evidence Tables, 13 provide evidence to support the recom-
for oxygen (CMRO2). Autoregulation refers to the mechanisms mendations for this topic: one new class 2 (79), two new class
of changes in cerebral vascular resistance by which CBF is 3 (74, 85), and 10 class 3 studies from the Second Edition (30,
maintained over a wide range of increases or decreased in CPP 40, 44, 52, 60, 61, 81–84) (Tables 10 and 11).

Supplement
TABLE 9. Cerebral Perfusion Pressure: Quality of the Body of Evidence

Quality of
Evidence
Meta- Total Consistency Precision (High,
Analysis Pos- No. of (High, Directness (High, Moderate,
No. of Studies Recommen- sible Subjects Moderate, (Direct or Moderate, Low, or
Components of overall quality: class 2 study

CPP threshold 1 retrospective III.2. NA 317 NA Direct Low Low
Mixed ages
Components of overall quality: class 3 studies
CPP threshold 1 prospective III.1. No; different 836 Low Direct Low Low
Mixed ages 2 retrospective III.2. thresholds;
9 treatment series different
populations
CPP threshold for 1 retrospective None No 39 Moderate Direct Low Insufficient
infants < 2 yr old
1 treatment series
CPP = cerebral perfusion pressure, NA = not applicable.
a
TABLE 10. Cerebral Perfusion Pressure: Summary of Evidence

Study
Reference Design
Geographic Age (yr)
Class 2 study
CPP age-specific ranges: recommendation III.2.
Allen et al Retrospective Class 2 Predefined CPP thresholds by age group
(79)a n = 317 pediatric Relevant criteria 40 and 30 mm Hg for 0–5 yr
22 (of 46 Age: for minimizing 50 and 35 mm Hg for 6–11 yr
designated) 0–5: 55 bias were met 60 and 50 for 12 yr old and older
trauma centers
in New York: 6–11: 65 CPP categories
20 are level I 12–17: 197 CPP-H: events above high threshold
centers and two Mortality at 14 d postinjury, CPP-B: events between high and low threshold
level II CPP, hypotension, elevated CPP-L: events below low threshold
New York ICP Mortality
Rates of survival and relative risk of mortality at 14 d
postinjury:
Regarding age (0–11 vs 12–17 yr), there was
a difference between survivors (n = 101) and
nonsurvivors (n = 19) for CPP-B.
For CPP-L, there was a significant difference in the 0-
to 11-yr-olds, not present in 12- to 17-yr-olds.
Refer to Tables 3 and 4 in the publication for specific
data about relationship between mortality and CPP
thresholds by age.
Data suggest that CPP targets should be age specific:
Above 50 mm Hg in 6- to 17-yr-olds
Above 44 mm Hg in 0- to 5-yr-olds
(Continued)

Kochanek et al
TABLE 10. (Continued). Cerebral Perfusion Pressure: Summary of Evidence

Reference
Type of Trauma Center Study
Geographic Design n Age (yr)
Class 3 studies
Minimum threshold and target ranges for managing CPP: recommendations III.1. and III.2.
Miller Retrospective Class 3 Outcome by threshold (< 40, < 45, < 50, < 55, < 60 mm
Ferguson et al n = 85 Selection of Hg during first 5 d after ICP monitor placement)
(74)a Age: mean, 5.1; range, NR patients for ICP GOS
PICU of a tertiary Dichotomized GOS monitoring at No significant difference in outcomes across five
children’s at 6 mo postinjury discretion of estimates utilizing five different thresholds.
hospital (1–2 unfavorable, 3–5 neurosurgery
Pittsburgh, PA favorable)
Note: article (p. 446)
indicates “favorable 1 to 2,
unfavorable 3 to 5.”
Vavilala et al Retrospective Class 3 Discharge GOS (survivors only)
(85)a n = 236 Outcome Adjusted risk ratio (95% CI)
Five Pediatric Age: mean, 8.0; range, NR assessment not Reference: any CPP ≤ 40 (no surgery)
Trauma Centers Dichotomized discharge GOS blinded, unclear Operating room
Seattle, WA; if groups are
similar, some, All CPP > 40 mm Hg 0.61 (0.58–0.64)
Pittsburg PA;
Chicago, IL; but not all key ICU
Torrance, CA; confounders All CPP > 40 mm Hg 0.73 (0.63–0.84)
Columbus, OH controlled for CPP > 40 mm Hg was associated with favorable
discharge GOS
Kapapa et al Treatment series Class 3 GOS
(83) Analysis of CPP in relation to Uncontrolled Patients with CPP values below the age-specific lower
(Location not age-specific lower limit (up series limit for just a single occurrence had a significantly
cited) to 1 mo, > 40 mm Hg; 2 mo worse outcome (p = 0.013).
up to 1 yr, > 45 mm Hg; 1
yr up to 7 yr, > 50 mm Hg;
> 7 yr, 55–60 mm Hg)
n = 36
Age: mean, NR; range, 0–16
Dichotomized GOS at varied
timepoints
Chaiwat et al Treatment series Class 3 GOS
(81) n = 36 patients (two inflicted Uncontrolled On univariate analysis, CPP < 40 mm Hg during the first
Level I pediatric TBI) series 72 hr had no association with poor outcome.
trauma center Age: mean, 9.1; range, 0.8–16 When logistic regression was performed, using a number
Seattle, WA Dichotomized GOS of factors, only impaired autoregulatory index remained
dichotomized at 6 mo an independent predictor of poor outcome.
postinjury
Figaji et al Treatment series Class 3 GOS
(61) n = 52 Uncontrolled Median (interquartile range) for lowest CPP was
Cape Town, South Age: mean, NR; range, 9 mo series significantly lower in patients with unfavorable
Africa to 14 yr outcome: 29 mm Hg (20–45 mm Hg) vs 44 mm Hg
Dichotomized GOS at ≥ 6 mo (35–51 mm Hg), p = 0.023
postinjury Patients with unfavorable outcome had more episodes of
CPP < 40 mm Hg: 3 (0–10) vs 0 (0–1), p = 0.03
There was no difference in the number of episodes of
CPP < 50 mm Hg.
(Continued)

Supplement

Geographic Age (yr)
Narotam et al Treatment series Class 3 Mortality

(60) n = 16 Uncontrolled Mean CPP was 81.52 ± 16.1 mm Hg for survivors vs
Level II trauma Age: mean, 14; range, series 50.33 ± 31.7 mm Hg for nonsurvivors (p < 0.033).
center 1.5–18 GOS
Omaha, NE Mortality, GOS at 3 mo All survivors had good outcome.
postinjury
Stiefel et al Treatment series Class 3 Mortality
(84) n=6 Uncontrolled 1 in 6 died
Level I trauma Age: mean, 12; range, 6–16 series Mean daily CPP in survivors was 75.63 ± 11.73 mm Hg.
center Mortality, discharge GOS GOS
Philadelphia, PA 4 of 6: 5
1 of 6: 3
1 of 6: 1
Adelson et al Randomized controlled trial Class 3 GOS
(40) of hypothermia therapy No control for Average CPP was 69.19 ± 11.96 mm Hg for favorable
Multicenter: Analysis of average CPP confounders vs 56.37 ± 20.82 mm Hg for unfavorable (p = 0.0004)
Pittsburgh, PA; over the first 5 d of care in CPP outcome groups.
Sacramento, n = 102 analysis (for Percent time with CPP > 50 mm Hg was 94.2%
CA; Miami, FL; Age: mean age in two part hypothermia, ± 16.9% for favorable vs 87.3% ± 29.5% for
Salt Lake City, study 6.89 and 6.95 yr this is a class 2 unfavorable (p = 0.0001).
UT; Hershey, study) Mean CPP on day 1 was higher in the hypothermia
PA; Seattle, WA Range: 0–13
Dichotomized GOS at 6 mo group (70.75 mm Hg) than the normothermia group
(level I pediatric (64.84 mm Hg), p = 0.037.
trauma center) postinjury
No significant differences between groups on days 2–5,
and GOS was not assessed in relation to differences
in CPP on day 1.
Chambers et Treatment series Class 3 GOS
al (30) n = 84 Uncontrolled Poor outcome in all eight cases with CPP < 40 mm Hg;
Neurosurgical Age: median, 10 yr; range, 3 series more patients had good outcome than poor outcome
Centre at mo to 16 yr when mean CPP was > 40 mm Hg.
Newcastle GOS dichotomized at 6 mo
General Hospital postinjury
Newcastle, United
Kingdom
Downard et al Treatment series Class 3 Mortality
(44) n = 118 Uncontrolled All children with mean CPP < 40 mm Hg died.
OHSU and Legacy Age: mean, 7.4; range, 0–15 series GOS
Emmanuel Mortality; last recorded GOS No significant difference in GOS when mean CPP was
Hospital and in records at 3 mo or later, divided into deciles from 40 to > 70 mm Hg.
Health Center dichotomized
(both are level I More patients had a good outcome than poor outcome
trauma centers when mean CPP was > 50 mm Hg, but no statistical
with immediately analysis.
available
neurosurgical
services)
Portland, OR
(Continued)

Kochanek et al

Geographic Age (yr)
Barzilay et al Treatment series Class 3 Survival (TBI group)

(52) n = 56 total Uncontrolled CPP was 65.5 ± 8.5 mm Hg for survivors vs 6.0 ± 3.9 mm
Chaim Sheba n = 41 TBI series Hg for nonsurvivors (p < 0.01).
Medical Center Age: mean, 6.2; range, NR 32 patients (78%) survived (mean ICP maximum
with PICU 16.9 ± 3.1 and CPP minimum 65.5 ± 8.5 torr)
Survival at hospital discharge
Tel Aviv, Israel compared with nine patients (22%) who died (mean
ICP maximum 53.7 ± 10.8 and CPP minimum 6 ± 3.9
torr) (p < 0.01).
Kaiser and Treatment series Class 3 Survival
Pfenninger n = 24 Uncontrolled All survivors (n = 19) had minimum CPP > 50 mm Hg.
(82) Age: mean, 6.3 yr; range, series Three of the five children who died also had CPP >
ICU, University 1–14 50 mm Hg.
Children’s Survival; GOS at 6–12 mo GOS
Hospital postinjury 1, n = 5
Bern, Switzerland 2, n = 0
3, n = 3
4, n = 3
5, n = 13
CPP threshold for infants < 2 yr old: no recommendation
Mehta et al Retrospective Class 3 Outcome by threshold (< 40, < 45, < 50 mm Hg during the
(73)a n = 22 Small sample size first 7 d)
Pediatric Age: mean, 6.1 mo; range, GOS
Neurotrauma 0–2 yr No difference in daily mean CPP between outcome
Registry Dichotomized GOS at 6 mo groups
Pittsburgh, PA postinjury (favorable 3–5, Patients in favorable outcome group had significantly
unfavorable 1–2), ICP, fewer hourly readings of CPP < 45 mm Hg than
CPP, physiologic variables unfavorable group (p = 0.046).
Subset of 85 patients in No difference between number of hourly readings
Miller Ferguson et al (74). of CPP < 40 or < 50 in patients with favorable or
Note: article (p. 415) unfavorable outcomes
indicates “favorable [GOS:
1–2] and unfavorable
[GOS: 3–5] outcomes.”
This is transposed from
the generally accepted
order (e.g., GOS 1 = dead).
Barlow and Treatment series Class 3 Outcome scale
Minns (80) n = 17 all inflicted TBI Uncontrolled Lowest CPP correlated with poor outcome (p < 0.0047)
Edinburgh, Age: mean, 5.1 mo; range, series
Scotland, United 1–20 mo
Kingdom A 6-point outcome scale
assessed 3–122 mo
postinjury (mean, 33 mo).
CPP = cerebral perfusion pressure, GOS = Glasgow Outcome Scale, ICP = intracranial pressure monitoring, NR = not reported, TBI = traumatic brain injury.
New study.
a
Evidence Synthesis the total 5 days of care (40). The authors used dichotomized
What Are the Minimum Threshold and Target Ranges for GOS outcome (good in 28 cases, 14 hypothermia patients and
Managing CPP, and Are Ranges Age Specific?. A RCT of hypo- 14 normothermia patients; poor in 40 cases, 18 hypothermia
thermia (32° to 33°) therapy (a class 2 study for the evidence patients and 22 normothermia patients) assessed at 6 months
about hypothermia, but class 3 for the evidence about CPP) after injury to examine differences in CPP. The average CPP for
reported average CPP over the first 5 days of care as well as for all 5 days was higher in the good outcome group (good outcome

Supplement
Summary of Treatments, Cerebral Perfusion Pressure Target, and Cerebral

TABLE 11.
Perfusion Pressure Statistics Used in Studies
Variations in Treatment and Measurement of CPP
Treatments Used
Induced Barbiturate Decompressive CPP Target CPP Statistic Used

Reference Hyperventilation Hypothermia Coma Craniectomy Strategy Time Period
Adelson NO YES YES YES Age-related Mean CPP

et al (40)
45 mm Hg Days 1–5
Allen et al (79)a Threshold Threshold Threshold Threshold Threshold Number of events
below minimum,
above maximum, or
between thresholds
Any time during ICU stay
Barlow and Threshold Threshold Threshold Threshold Threshold Lowest CPP
Minns (80) Timing NR
Barzilay et al (52) YES YES YES NO — Lowest CPP
Timing NR
Chaiwat et al (81) — — — — — Lowest CPP
First 72 hr
Chambers Threshold Threshold Threshold Threshold Threshold Minimum CPP
et al (30) Range: 6.3–173 hr;
median: 41.2
Downard YES NO NO YES Age-related Mean CPP
et al (44) 45 mm Hg First 48 hr
Figaji et al (61) YES YES YES YES Age-related Initial and lowest CPP
45 mm Hg Mean duration
123.7 ± 67.1 hr
Kaiser and YES YES YES NO — Mean CPP
Pfenninger Mean time of 9 d; range
(82) 1–15 d
Kapapa et al (83) YES YES YES YES Age-related Lowest CPP
40 mm Hg Any point during
hospitalization with
monitor
Mehta et al (73)a NO YES YES YES ICP of 20 Mean CPP
and CPP Daily mean for first 7 d
50 mm Hg
Miller Ferguson NO YES YES YES ICP of 20 Mean CPP daily mean
et al (74)a and CPP for first 5 d
50 mm Hg
Narotam YES NO NO NO ICP-related Mean CPP
et al (60) Timing NR
Stiefel et al (84) — — — — Age-related Mean CPP
40 mm Hg From admission for at
least 72 hr
Vavilala et al (85)a YES — YES YES CPP > Below or above
40 mm Hg threshold
At any time in operating
room or ICU
CPP = cerebral perfusion pressure, ICP = intracranial pressure monitoring, NR = not reported.
New study.
a
In the studies that describe therapy, “YES” denotes use of therapy and “NO” denotes where treatment is not used. “Threshold” denotes where the study is
aimed at defining a threshold about burden from CPP insult and outcome, rather than it being an intervention study. Dash indicates where no information is
given in the report.

Kochanek et al
69.19 ± 11.96 mm Hg vs poor outcome 56.37 ± 20.82 mm Hg; (p < 0.0001, Fisher exact test). Chaiwat et al (81) analyzed 36
p = 0.0004). In addition, the percent time with CPP greater cases of TBI for predictors of poor outcome. ICP of greater than
than 50 mm Hg was higher in the good outcome group (good 20 mm Hg and CPP less than 40 mm Hg during the first 72 hours
outcome 94.2% ± 16.9% vs poor outcome 87.3% ± 29.5%; were not associated with outcome. However, on logistic regres-
p = 0.0001). In contrast, a recent retrospective study of 85 pedi- sion, an estimate of impaired CBF autoregulation using Doppler
atric patients with severe TBI (mean age, 5.1 ± 0.8 yr) analyzed ultrasonography—the autoregulatory index (ARI)—was an
outcome in five threshold groups (< 40, 45, 50, 55, and 60 mm independent predictor of poor outcome (adjusted odds ratio
Hg). No significant difference across groups was found in the [OR], 23.1; 95% CI, 1.9–279.0). Impaired ARI was an indepen-
dichotomized GOS at 6 months postinjury. dent risk factor when the authors entered CPP less than 40 mm
Five treatment series found higher CPP associated with better Hg, SBP less than fifth percentile for age and gender during the
outcomes and their findings were as follows (52, 60, 61, 80, 84). first 72 hours after TBI, low middle cerebral artery (MCA) veloc-
Barzilay et al (52) studied 41 consecutive TBI admissions to their ity, and impaired ARI into the model (adjusted OR, 29.8; 95%
PICU with coma for at least 6 hours before admission. Survivors CI, 1.7–521.4). Because ARI is calculated as the percent change in
had higher minimum CPP than nonsurvivors (65.5 ± 8.5 vs cerebrovascular resistance (CVR) per percent change in CPP, and
6.0 ± 3.9 mm Hg; p < 0.01). All patients with head trauma were CVR is defined as the ratio of CPP to MCA velocity, it is impos-
treated for 5 days with dexamethasone, and neither the tim- sible to disentangle the relationship between outcome and CPP.
ing nor duration of ICP and CPP derangements was specified. ARI represents a research tool.
“Preterminal” data were not removed, resulting in the low aver- Five class 3 studies contain data concerning CPP threshold
age CPP for nonsurvivors. Figaji et al (61) studied prospectively above 40 mm Hg (3, 5, 10, 12, 14). Two retrospective treatment
52 children with TBI and found median lowest CPP experienced series support the idea that there may be an age-related CPP thresh-
during the course of monitoring was higher in those with better old above 40 mm Hg. Kapapa et al (83) analyzed 16 children under
outcome. By using dichotomized GOS outcome assessed at least 6 16 years old and reported dichotomized GOS in relation to age-
months after injury, those with favorable outcome had higher low- specific lower limits in CPP (i.e., > 40 mm Hg, infants up to 1 mo
est CPP median (interquartile range) of 44 mm Hg (35–51 mm old; > 45 mm Hg, infants 2 mo to 1 yr old; > 50 mm Hg, children
Hg) versus 29 mm Hg (20–45 mm Hg) in comparison with those between 1 and 7 yr old; 55–60 mm Hg, children > 7 yr old). The
with unfavorable outcome (p = 0.023). Narotam et al (60) ana- authors found that patients with CPP value below the age-specific
lyzed data from 16 children 1.5–18 years old (mean, 14 yr), 15 of lower limit for just a single occurrence had a significantly worse
whom had GCS less than or equal to 8. All 10 survivors had GOS 5 outcome (p = 0.013). Kaiser and Pfenninger (82) reported findings
at 3 months. Mean CPP was higher in survivors (81.52 ± 16.1 mm in 24 consecutive admissions to their PICU of patients with a GCS
Hg) than nonsurvivors (50.33 ± 31.7 mm Hg; p = 0.033). Stiefel et less than 8, average age 6.3 years (10 patients between 1 and 5 yr
al (84) studied brain tissue oxygen monitoring in six patients (age, old), and showed that all survivors had CPP greater than 50 mm
6–14 yr old; GCS = 3–7) and found that mean daily CPP in the Hg (p < 0.005, Fisher exact test). Two studies did not observe a
five survivors was 75.63 ± 11.73 mm Hg. Last, in a sample of TBI threshold above 40 mm Hg (44, 61). In the study reported by Figaji
cases restricted to 17 young children with inflicted injury (age, et al (61) (see above), the authors also reported outcome in relation
1–20 mo old; mean, 5.1 mo), Barlow and Minns (80) reported to the number of episodes during monitoring that CPP was less
that lowest CPP during intensive care was associated with poorer than 50 mm Hg: there was no difference in the number of episodes
outcomes in a 6-point scale 3–122 months (mean, 33 mo) after in those with unfavorable (8 [2–18.5]) versus favorable (3 [0–8.8])
injury (p = 0.0047). outcome (p = 0.137). Of note, two thirds of the children in this
Four studies reported findings in relation to a threshold in series were younger than 8 years old. As discussed above, in the
CPP of 40 mm Hg (30, 44, 61, 81). In the study reported by Figaji study reported by Downard et al (44), 100% of children with mean
et al (61), the authors found that more episodes of CPP less than CPP less than 40 mm Hg died when compared with only 25% of
40 mm Hg were observed in those with unfavorable (3 [0–10]) children who had a CPP greater than 40 mm Hg. The difference in
versus favorable (0 [0–1]) outcome (p = 0.0.03); a more complex mortality was statistically significant (p < 0.0001, Fisher exact test).
relationship between CPP and outcome involved data from auto- Last, in the study of young children with abusive TBI reported by
regulation of CBF. Chambers et al (30) analyzed 84 children age 3 Barlow and Minns (80) (see above), only one infant in the series of
months to 16 years old (median, 10 yr) and examined minimum 17 had lowest CPP greater than 50 mm Hg.
CPP in relation to dichotomized GOS at 6 months. Sixty-three These studies, in aggregate, suggest that in the pediatric age
out of 76 cases with CPP greater than 40 mm Hg had good out- range, there may be an age-related threshold between 40 and
come, and all eight cases with CPP less than 40 mm Hg had poor 50 mm Hg, with infants at the lower end and adolescents at the
outcome (p < 0.0001; Fisher exact test). Downard et al (44) ana- upper end of this range. Finally, studies specifically focused on
lyzed 118 pediatric TBI cases of children up to 15 years old (mean assessment of the optimal upper limit for CPP management in
age, 7.4 yr; 99 cases with GCS = 3–8) and reported dichotomized pediatric TBI were lacking.
GOS at 3 months or later in relation to CPP thresholds. Seventy-
two out of 96 patients with CPP greater than 40 mm Hg had Indications From Adult Guidelines
good outcome, whereas all 22 cases with CPP less than 40 mm The recent Fourth Edition of the adult guidelines does not fur-
Hg died. The difference in mortality was statistically significant ther inform the pediatric guidelines for this topic (14).

Supplement
TREATMENTS assessing fluid balance, central venous pressure, urine output,

blood urea nitrogen, serum creatinine, and clinical examination.
Hyperosmolar Therapy
The placement of a Foley catheter is also routinely used to quan-
Recommendations tify urine output and to avoid potential bladder rupture.
Strength of Recommendations: Weak Mannitol. Mannitol is commonly used to manage raised
ICP in pediatric TBI (96). Despite this fact, mannitol has not
Level I
been subjected to contemporary controlled clinical trials ver-
There was insufficient evidence to support a level I recommen-
sus placebo, other osmolar agents, or other therapies in chil-
dation for this topic.
dren. Most of the investigations on the use of mannitol were
Level II carried out in the 1970s on mixed disease populations in both
For ICP Control. II.1. Bolus HTS (3%) is recommended in children and adults (93).
patients with intracranial hypertension. Recommended effec- Mannitol can reduce ICP by two distinct mechanisms.
tive doses for acute use range between 2 and 5 mL/kg over Mannitol can reduce ICP by reducing blood viscosity. This
10–20 minutes. effect is immediate and results from a viscosity-mediated reflex
Level III vasoconstriction (intact autoregulation) which allows CBF to
For ICP Control. III.1. Continuous infusion HTS is suggested be maintained despite a reduced level of cerebral blood volume
in patients with intracranial hypertension. Suggested effective (CBV) (97, 98). The effect of mannitol administration on blood
doses as a continuous infusion of 3% saline range between 0.1 viscosity is transient (< 75 min) (97). Mannitol administration
and 1.0 mL/kg of body weight per hour, administered on a slid- also reduces ICP by an osmotic effect, which develops more slowly
ing scale. The minimum dose needed to maintain ICP less than (over 15–30 min), due to the gradual movement of water from the
20 mm Hg is suggested. brain parenchyma into the systemic circulation. The effect persists
III.2. Bolus of 23.4% HTS is suggested for refractory ICP. up to 6 hours and requires an intact blood-brain barrier (99, 100).
The suggested dose is 0.5 mL/kg with a maximum of 30 mL. Mannitol may accumulate in injured brain regions (101), where a
Safety Recommendation (applies to all recommendations reverse osmotic shift may occur—with fluid moving from the vas-
for this topic). In the context of multiple ICP-related therapies, cular compartment into the brain parenchyma, possibly increas-
avoiding sustained (> 72 hr) serum sodium greater than 170 ing ICP. This phenomenon has been suggested to occur when
mEq/L is suggested to avoid complications of thrombocytope- mannitol is used for extended periods of time (102, 103).
nia and anemia, whereas avoiding a sustained serum sodium Mannitol is excreted unchanged in urine, and a risk of the
greater than 160 mEq/L is suggested to avoid the complication development of acute tubular necrosis and renal failure has
of deep vein thrombosis (DVT). been suggested with mannitol administration with serum
Note. Although mannitol is commonly used in the manage- osmolarity levels greater than 320 mOsm in adults (104–106).
ment of raised ICP in pediatric TBI, no studies meeting inclu- However, the literature supporting this finding is limited in
sion criteria were identified for use as evidence for this topic. scope and was generated at a time when dehydration rather
Changes From Prior Edition. Recommendation II.1. is new than euvolemia was the therapeutic target.
to this Third Edition. It replaces the level II recommendation HTS. In the initial description in 1919 of the reduction in
from the Second Edition suggesting the use of HTS in general, ICP by IV administration of hyperosmolar agents, HTS was
with doses ranging between 6.5 and 10 mL/kg. Recommenda- the agent used (91). Its use in the treatment of increased ICP,
tions III.2. is new to this Third Edition. The recommendation however, failed to gain clinical acceptance. Resurgence in inter-
from the Second Edition to maintain serum osmolarity less est in this treatment emerged in the late 1980s (107), leading to
than 360 mOsm/L was removed from this edition. The Safety the studies providing current evidence.
Recommendation is new to this edition. One new class 2 pro- Like mannitol, the penetration of sodium across the blood-
spective observational study (2) and five new class 3 studies— brain barrier is low (108). Sodium thus shares both the favor-
four retrospective observational (85, 87–89) and one treatment able rheologic and osmolar gradient effects involved in the
series (90)—were added to the evidence base for this topic. reduction in ICP. HTS has other theoretical beneficial effects
including restoration of normal cellular resting membrane
Introduction potential and cell volume (109, 110), stimulation of arterial
Hyperosmolar Therapy for Intracranial Hypertension. IV natriuretic peptide release (111), inhibition of inflammation
administration of hyperosmolar agents was shown to reduce (108), and enhancement of cardiac output (112). Possible side
ICP in 1919 (91). Mannitol was introduced into clinical use in effects of HTS include rebound in ICP, central pontine myelin-
1961 (92). Despite widespread use of a number of osmolar agents olysis, renal impairment, subarachnoid hemorrhage, natriure-
(mannitol, urea, and glycerol) up until the late 1970s (93, 94), sis, high urinary water losses, hyperchloremic acidosis, and
mannitol gradually became the agent of choice to manage ICP masking of the development of diabetes insipidus (108).
(93). Subsequently, HTS was introduced in the 1990s (95), and Much higher levels of serum osmolarity (≈360 to 370
although both are currently used, HTS use has increased while mOsm) may be tolerated in children when induced with HTS
mannitol use has decreased (96). Additionally, a goal of euvolemia (87, 113) versus mannitol although this point remains contro-
rather than dehydration as a therapeutic target is achieved by versial (35).

Kochanek et al
In 14 adults with severe TBI, Lescot et al (114) suggested outcomes and control ICP, and the subquestions about HTS
important differences in the response of contused versus non- versus other agents, mode of administration, effective dose
contused brain tissue to HTS, with reductions in the volume of ranges, and the risk of DVT and other complications. Two class
noncontused brain but increases in the volume of contusions 2 RCTs (95, 115), one class 2 prospective study (2), three class
after treatment. Studies of regional effects of HTS or mannitol 3 retrospective studies (85, 88, 113), and one class 3 treatment
have not been carried out in pediatric TBI. series (90) provide evidence for treatment effects on intracra-
A second use of HTS is to treat hyponatremia due to cere- nial hypertension for this topic. Two class 3 retrospective stud-
bral salt wasting if it develops in pediatric patients after TBI ies provide evidence for the Safety Recommendation (87, 89).
(1). However, the focus of this guideline is on the use of hyper- There was insufficient evidence to support a recommendation
osmolar agents in the treatment of raised ICP. about use of HTS to improve outcomes, or about the compara-
tive effectiveness of agents. The overall quality of the body of
Evaluation of the Evidence evidence is moderate to low (Table 12).
Quality of the Body of Evidence. Studies included for this topic Applicability. The evidence includes current studies; how-
addressed the question of the effectiveness of HTS to improve ever, they are observational, class 3. The two class 2 RCTs were
TABLE 12. Hyperosmolar Therapy: Quality of the Body of Evidence

Quality of Evi-
No. of Meta- Total Consistency Precision dence (High,
Studies Analysis No. of (High, Directness (High, Moderate,
Study Recommen- Possible Subjects Moderate, (Direct or Moderate, Low, or
Topic Design dations (Yes or Noa) (n) Low) Indirect) Low) Insufficient)

Bolus hypertonic 1 RCT II.1. No, 34 Moderate Direct Moderate Moderate
saline to control different
ICP monitoring 1 prospective comparators
Continuous infusion 1 RCT III.1. NA 35 NA Direct Low Low
hypertonic saline
to control ICP
monitoring
Continuous 1 retrospective III.1. NA 68 NA Direct Low Low to
infusion moderate
hypertonic saline
to control ICP
monitoring
Bolus hypertonic 1 treatment III.2. NA 32 NA Direct Low Low
saline to control series
refractory ICP
monitoring
Bolus hypertonic 1 retrospective No NA 16 NA Direct Low Insufficient
saline vs recommen
mannitol to dation
control refractory
ICP monitoring
Hypertonic saline 1 retrospective No NA 236 NA Direct Low Insufficient
to improve recommen
overall outcomes dation
Risk of deep vein 1 retrospective Safety NA 58 NA Direct Moderate Low
thrombosis with recommen
hypertonic saline dation
Hypertonic saline 1 retrospective Safety NA 48 NA Direct Low Low
complications/ recommen
upper limits dation
ICP = intracranial pressure, NA = not applicable, RCT = randomized controlled trial.
a

Supplement
from the 1990s (95, 115). With the exception of one (85), the study comparing IV administration of 3% saline (513 mEq/L,
studies were conducted at single centers with sample sizes less 1,027 mOsm/L) and 0.9% saline (154 mEq/L, 308 mOsm/L)
than 100. Six were conducted in the United States (2, 85, 87, 89, in 18 children with severe TBI. Bolus doses of each agent
95, 113), one in Australia (90), one in Switzerland (115), and one were equal and ranged between 6.5 and 10 mL/kg. During the
in Canada (88). The studies included a range of ages. The small 2-hour trial, HTS use was associated with an approximately
sample sizes and single-center designs limit the applicability. 7 mEq/L increase in serum sodium concentration, lower
ICP, and reduced need for other interventions. Concomitant
Summary of the Evidence therapies used for patient management in this study included
Of the nine studies summarized in the evidence table, two class thiopental, dopamine, mannitol, and hyperventilation. CSF
2 studies (2, 95) provided evidence to support recommendation drainage was not used.
II.1. One class 2 study (115) and one class 3 study (113) provided More recently, Shein et al (2) carried out a prospective
evidence to support recommendation III.1. One class 3 study cohort study comparing the effects of bolus IV administration
(90) provided evidence to support recommendation III.2. There of 3% saline (3 mL/kg [range, 2–5 mL/kg] over 10–20 min),
was insufficient evidence to support a recommendation for the fentanyl, pentobarbital, or mannitol, on ICP and CPP in 16
use of HTS to improve outcomes. Two class 3 studies (87, 89) children with severe TBI. The study featured use of a data
supported the safety recommendations for this topic (Table 13). acquisition system sampling every 5 seconds, and over 2.7
million timepoints were analyzed for values of ICP, MAP, and
Evidence Synthesis CPP. The response to these therapies (collectively, 362 doses)
Bolus Administration of HTS to Control ICP. Fisher et al (95) was assessed in real world use rather than a randomized com-
carried out a double-blind randomized controlled crossover parison to treat ICP greater than 20 mm Hg for greater than
TABLE 13. Hyperosmolar Therapy: Summary of Evidence

Class 2 studies
Bolus hypertonic saline to control ICP: recommendation II.1.
Shein et al (2)a Prospective Class 2 Hypertonic saline vs other drugs for
Children’s Hospital n = 16 in analysis Differences in patients by intracranial hypertension
University of Age: mean, 44 mo; treatment, limited control for Decrease in ICP; increase in CPP
Pittsburgh range, 34–124 mo confounding Associated with a two-fold faster resolution
Pittsburgh, PA ICP and CPP of intracranial hypertension than either
fentanyl or pentobarbital
Adjusted hazard ratio, 2.171 (95% CI,
1.062–4.439)
Hypertonic saline may be first-line
treatment given favorable hemodynamics
and resolution of intracranial
hypertension.
Comparison drugs: beneficial effects on
ICP only
Fentanyl: ICP decreased; CPP decreased.
Highest rate of treatment failure rate
Pentobarbital: ICP decreased; CPP no
significant change
Note: Mannitol not included due to limited
use (seven doses out of 362 total; four
out 196 analyzed)
Fisher et al (95) RCT Class 2 3% saline vs 0.9%
San Diego Children’s n = 18 Randomization and allocation During the 2-hr trial, hypertonic saline was
Hospital Age: mean, 8.3; range, concealment methods not associated with a lower ICP and reduced
San Diego, CA 0.6–14.5 reported; crossover study need for additional interventions (thiopental
ICP lacking reporting on first- and hyperventilation) to control ICP.
period comparison of baseline Serum sodium concentration increased ≈7
characteristics; small sample size mEq/L after 3% saline.
(Continued)

Kochanek et al
TABLE 13. (Continued). Hyperosmolar Therapy: Summary of Evidence

Continuous infusion hypertonic saline to control ICP: recommendation III.1.

Simma et al RCT Class 2 Hypertonic saline (1.7%) vs lactated Ringer’s
(115) n = 35 Not blinded, insufficient power solution
Children’s Hospital of Age: mean, 87 mo; No difference between groups in survival
Zurich range, 12–173 mo rate and length of hospital stay
Zurich, Switzerland ICP, CPP, need for Patients treated with hypertonic saline
other interventions, required fewer interventions than those
fluid requirements, treated with lactated Ringer’s to maintain
ICU stay, and survival ICP control (p < 0.01).
rate The hypertonic saline treatment group had
shorter length of ICU stay (p = 0.04),
shorter duration of mechanical ventilation
(p = 0.10 not significant), and fewer
complications than the lactated Ringer’s-
treated group (p = 0.09) for two or more
complications, not significant, without p
value reported for one complication).
Class 3 studies
Continuous infusion hypertonic saline to control ICP: recommendation III.1.
Peterson et al Retrospective Class 3 3% hypertonic saline, continuous infusion
(113) n = 68 No control for confounders Survival rate was higher than expected
San Diego Children’s Age: mean, 7.8; range, based on Trauma and Injury Severity
Hospital NR Score (41 predicted, 58 actual).
San Diego, CA ICP, 6-mo GOS score 53% had good outcome, 20.5% moderate,
10% severe, 1.5% vegetative, and 15%
died; three died of uncontrolled ICP.
No patients developed renal failure.
Central pontine myelinolysis, subarachnoid
hemorrhage, or rebound increases in ICP
were not observed.
Bolus hypertonic saline to control refractory ICP: recommendation III.2.
Piper and Treatment series Class 3 All received 23.4% hypertonic saline infused
Harrigan (90)a n = 32 (4 [87.5%] with No control for confounding; over 10 min, maximum dose of 30 mL
John Hunter Hospital GCS ≥ 9) differences across patients All-cause mortality
Newcastle, NSW, Age: mean, 14; range, 6% (2/32) at 7 d
Australia 8 mo to 17 yr No deaths after 7 d
ICP, mortality, GOS, ICU GOS
LOS 48% 5 (normal life activities)
26% 4 (disabled but independent)
19% 3 (conscious but severely disabled)
0% 2 (persistent vegetative state)
6% 1 (deceased)
ICP
Mean ICP response to HTS (pre HTS ICP-
ICP 60 min post)
10 mm Hg (range, 1–30; sd, 8)
ICU LOS
Mean 10 d (range, 2–25; sd, 6)
(Continued)

Supplement

Study Design
Reference n
Bolus hypertonic saline vs mannitol to control refractory ICP: no recommendation

Roumeliotis et al Retrospective Class 3 ICP
(88)a n = 16 Limited control for confounding; Both HTS and Mannitol produced a
Pediatric tertiary care Age: mean, 13 11 of 16 patients had nonsignificant decrease in ICP in first 4 hr.
and trauma center Interquartile range: cointerventions CPP
Montreal, QC, Canada 10–15 No change in CPP postbolus
ICP, CPP, and serum Serum sodium
sodium No significant change in serum sodium
Hypertonic saline to improve overall outcomes: no recommendation
Vavilala et al Retrospective Class 3 Discharge survival
(85)a n = 236 Limited control for confounding, Hypertonic saline or mannitol used for high
Five pediatric trauma Age: mean, 8.0; range, outcome assessors not ICP in operating room (adjusted relative
centers NR blinded, impact of baseline risk, 0.62; 95% CI, 0.42–0.93)
Seattle, WA; Pittsburg In-hospital mortality, characteristics, and missing data Reference: neither used with high ICP
PA; Chicago, IL; discharge GOS unclear (measured in prehospital, emergency
Torrance, CA; department, operating room, ICU.
Columbus, OH Combined not significant)
Note: frequency of each agent or results
not reported separately
Risk of DVT with hypertonic saline: safety recommendation
Webster et al Retrospective Class 3 DVT
(89)a DVT compared with no Potential confounding; Cumulative total bolus volume of HTS (mL/
Cincinnati Children’s DVT; all given HTS measurement and detection kg) associated with DVT (OR, 1.6; 95%
Hospital the exposure is the concerns for DVT CI, 1.2–2.4; p = 0.01)
Cincinnati, OH bolus volume and Peak sodium level and 72-hr sustained sodium
sodium level levels: associated with DVT (p = 0.05)
n = 58 Sustained sodium level of at least 160
Age: median mmol/L: associated with DVT (p = 0.02)
DVT = 8 Mortality
No DVT = 4.5 69% alive at 30 d postinjury
Range: NR Total bolus volume of HTS during the
DVT, survival at 30 d, hospital stay was not significantly
and GOS associated with survival (p = 0.62)
GOS
A favorable GOS was not associated with a
higher cumulative total bolus volume of
HTS administered.
Odds of a favorable GOS of 4 or 5 were
less for those subjects who maintained a
sustained sodium of at least 160 mmol/L
(OR, 0.1; 95% CI, 0.03–0.38; p = 0.008).
The sample-average sustained sodium for
the subjects with a discharge GOS of
1–3 was 160.1 mmol/L, and the sample-
averaged sustained sodium for the
subjects with a discharge GOS of 4–5
was 144.8 mmol/L, a difference of 15.3
mmol (95% CI, 9.3–21.3; p ≤ 0.01).
(Continued)

Kochanek et al

Study Design
Reference n
Hypertonic saline complications/upper limits: safety recommendation
Gonda et al (87)a Retrospective Class 3 Complications with sustained peak serum sodium
Rady Children’s n = 48 TBI Control for some identified level (mEq/L) ≥ 170 Reference: < 170
Hospital Out of 88 total confounders ORs
San Diego, CA Age: mean, 8; range, Multivariate (95% CI)
3 wk to 19 yr Thrombocytopenia: 31.5 (7.0–194)/71.6
Complications (6.8–1,807), p < 0.001
associated with Neutropenia: 26.7 (4.1–530.5)/195 (0.8–
hypernatremia for 64,053,150), p = 0.146
TBI group RBC transfusion: 33.6 (5.7–650.7)/22.8
Mortality, GOS at (1.7–935.4), p < 0.001
discharge, ICP Fresh frozen plasma transfusion: 9.5 (2.5–
monitor duration for 41.6)/3.9 (0.6–27.1), p < 0.001
entire group (indirect Renal failure: 10.2 (2.1–76.3)/17.5 (1.0–
evidence) 540.7), p < 0.009
Complications for Acute respiratory distress syndrome: 12.5
subset of TBI (1.8–253.2)/7.7 (0.3–354.4), p < 0.028
patients (direct
evidence) Overall mortalityb
15.9% (14/88), 7/88 due to cerebral
herniation
Overall Glasgow Outcome Scoreb
< 3: 10 (17.2%) vs 14 (46.7%), p = 0.005
Overall ICP monitor durationb
Mean 10.9 d (sd, ± 4.7) vs 19 d (sd, ±
12.1), p < 0.001
CPP = cerebral perfusion pressure, DVT = deep vein thrombosis, GOS = Glasgow Outcome Scale, HTS = hypertonic saline solution, ICP = intracranial
pressure, LOS = length of stay, NR = not reported, OR = odds ratio, RCT = randomized controlled trial, TBI = traumatic brain injury.
New study.
a
Rates based on total sample size including non-TBI patients.

b
5 minutes. HTS administration was associated with a two- in favor of continuous infusion of HTS to control ICP: 1) the
fold faster resolution of intracranial hypertension than either control group received infusion of lactated Ringer’s, and 2) the
fentanyl or pentobarbital and was the only agent that also HTS concentration that was used (1.7%) has limited clinical
improved CPP. Mannitol use could not be assessed given that use in pediatric TBI.
only seven doses were administered. Peterson et al (113) reported a retrospective class 3 study
These two class 2 studies provide evidence to support the on the use of a continuous infusion of 3% saline (sodium 513
level II recommendation for bolus administration of HTS (3% mEq/L, 1,027 mOsm/L) titrated to reduce ICP to less than or
solution) to treat intracranial hypertension (2, 95). equal to 20 mm Hg in 68 infants and children with TBI. The
Continuous Infusion Administration of HTS to Con- mean daily doses of HTS over a 7-day period ranged between
trol ICP. Simma et al (115) carried out an RCT of 1.7% HTS 11.76 and 26.94 mL/kg/d. There was no control group. Three
(sodium 268 mmol/L, 598 mOsm/L) versus lactated Ringer’s patients died of uncontrolled ICP, and mortality rate was lower
solution (sodium 131 mmol/L, 277 mOsm/L) administered IV than expected based on Trauma and Injury Severity Score (ISS)
over the initial 3 days in 35 children with severe TBI. Patients categorization. No patient with a serum sodium concentration
treated with HTS required fewer interventions (including greater than 180 mEq/L had a good outcome. No patients devel-
mannitol use) to control ICP than those treated with lactated oped renal failure. Concomitant therapies included sedation,
Ringer’s solution. Patients in the HTS treatment group also NMB, mannitol, hyperventilation, and barbiturates. CSF drain-
had shorter length of PICU stay (p = 0.04) and fewer compli- age was used in three children. The mean daily dose of mannitol
cations than the lactated Ringer’s-treated group (p = 0.09 for was between 0.83 and 2.16 g/kg/d. Rebound in ICP, central pon-
two or more complications, not significant for one complica- tine myelinolysis, and subarachnoid hemorrhage were not seen.
tion). Due to design flaws and insufficient power, the evidence These two studies provided the evidence to support the level
from this study is class 2. Two facets of this study precluded the III recommendation for continuous infusion of HTS (3%) to
clinical investigators from making a level II recommendation treat intracranial hypertension (113, 115).

Supplement
Bolus Administration of HTS for Refractory ICP. Piper and ICP greater than 25 mm Hg, but no differences in mortal-
and Harrigan (90) reported a treatment series on the use of ity or long-term outcome were noted (14, 116, 117).
23.4% saline to treat refractory intracranial hypertension in 32
infants and children with severe TBI. Refractory intracranial Analgesics, Sedatives, and NMB
hypertension was defined as an ICP greater than 20 mm Hg
Recommendations
for greater than 5 minutes and not responding to a stepwise
Strength of Recommendations: Weak
protocol that included sedation, analgesia, head elevation, mild
hyperventilation, mild hypothermia (36.5°C), NMB, and use of
Levels I and II
inotropic support to support age-appropriate CPP. Contrain-
dication to the administration of 23.4% saline in this study was
a serum sodium level of greater than 155 mmol/L. The dose of
23.4% saline was 0.5 mL/kg administered over a period of 10
Level III
minutes, with a maximum dose of 30 mL. The mean reduction
For ICP Control. III.1. With use of multiple ICP-related ther-
in ICP with treatment was 10 mm Hg (range, 1–30 mm Hg),
apies, as well as appropriate use of analgesia and sedation in
and the highest serum sodium level observed in any patient
routine ICU care, avoiding bolus administration of midazolam
was 161 mmol/L. GOS greater than 3 was achieved in 74% of
the patients. This study provided the evidence to support the and/or fentanyl during ICP crises is suggested due to risks of
level III recommendation for the use of 23.4% saline to treat cerebral hypoperfusion.
refractory intracranial hypertension. Note 1. In the absence of outcome data, the specific indi-
Use of Hyperosmolar Therapy to Improve Outcomes. There cations, choice, and dosing of analgesics, sedatives, and neu-
was insufficient evidence to support a recommendation for the romuscular blocking agents should be left to the treating
use of hyperosmolar therapy to improve overall outcomes. physician.
A single study, Vavilala et al (85) assessed acute care clinical Note 2. Based on guidance from the U.S. Food and Drug
indicators associated with in-hospital mortality and discharge Administration, prolonged continuous infusion of propofol
GOS score in a retrospective cohort study of 236 infants and for either sedation or the management of refractory intracra-
children with severe TBI across five medical centers. The use nial hypertension is not recommended.
of HTS or mannitol for high ICP in the operating room was Changes From Prior Edition. Recommendation III.1. is
associated with favorable outcome versus no hyperosmolar new to this Third Edition. The recommendation about the
therapy. However, when the comparison included the emer- use of thiopental from the Second Edition has been removed
gency department (ED), operating room, and ICU, a signifi- and the study by de Bray et al (118) has been removed. Five
cant effect was not observed. new class 3 studies—two prospective (2, 119), two retrospec-
tive (85, 120), and one treatment series (3)—were added to the
Safety Recommendations evidence base for this topic.
Two class 3 studies contributed evidence supporting the safety
recommendations for this topic (87, 89). Webster et al (89) car- Introduction
ried out a single-center case-control study of 58 infants and In management of pediatric severe TBI requiring tracheal
children with severe TBI treated with bolus and/or continuous intubation and mechanical ventilation, analgesics and seda-
infusion of HTS. Eight patients developed DVTs. There was no tives are needed for comfort and tolerance (121). Neuromus-
association between the volume of HTS administered during cular blocking agents are not needed routinely, except at the
the initial 72 hours and DVT; however, a serum sodium level time of rapid sequence intubation or when severe acute lung
greater than or equal to 160 mEq/L sustained for 72 hours was injury mechanics pose a problem with supportive ventilation
significantly associated with DVT (p = 0.02). Gonda et al (87) (122). This topic evaluates the use of these agents during ICU
reported a retrospective cohort study of 48 infants and chil- treatment in patients who are appropriately sedated, when
dren with severe TBI in which bolus and infusion treatment needed specifically for management of ICP, and for optimizing
with 3% saline were used. A sustained level (defined as ≥ 72 hr) cerebral perfusion.
of serum sodium greater than or equal to 170 mEq/L was sig- In a mechanically ventilated patient, appropriate use
nificantly associated with thrombocytopenia and the need for of analgesia and sedation will treat any pain or distress and
erythrocyte transfusion. In addition, although the association mitigate patient-ventilator dyssynchrony, both of which may
with renal failure was not significant (p = 0.064), this could result in episodic rise in CBV and raise ICP (123). Routine care
indicate a trend toward harm that merits further study. (e.g., oral hygiene, tracheal tube suctioning) while the patient
is awake or aware may also lead to unanticipated rises in ICP
Indications From Adult Guidelines (124). Pediatric practice is currently limited in the choice of
The Fourth Edition of the adult TBI guidelines inform the sedative/analgesics that are available. Practice is markedly
pediatric guidelines in that class 2 and 3 studies comparing different in adult than in pediatric neurocritical care; espe-
HTS to mannitol have been carried out and suggest that HTS cially in the frequent use of propofol, which in pediatrics
may be more effective that mannitol with regard to ICP burden has a U.S. Food and Drug Administration warning (125). As

Kochanek et al
a consequence, given the relatively long half-life of the drugs who received high-dose fentanyl, low-dose midazolam, or
that are administered, frequently the neurologic examination high-dose fentanyl plus low-dose midazolam, there was an
can be obscured. increase in ICP for all treatment conditions (3). It is unclear
Neuromuscular blocking agents may reduce ICP by either if the finding was due to other patient factors, and the study
optimizing patient-ventilator interactions or by prevention of did not address use of sedation beyond the context of ICP. In
shivering. a prospective observational study of 16 pediatric patients, fen-
tanyl decreased ICP, but less effectively than HTS or pentobar-
Evaluation of the Evidence bital (2). Furthermore, it decreased CPP, and had the highest
Quality of the Body of Evidence. Studies included for this topic treatment failure rate. This study suggests that in the setting
addressed the questions of the effectiveness of combinations of of adequate analgesia/sedation, HTS might be preferable over
NMB, ketamine, fentanyl, midazolam, etomidate, and pento- fentanyl. Taken together, these studies provide the evidence to
barbital to control ICP for children with severe TBI. There was support the recommendation for this topic.
insufficient evidence to derive a recommendation about mixed The findings from the studies by Welch et al (3) and Shein
NMB, ketamine, etomidate, or pentobarbital. The overall qual- et al (2) should be interpreted cautiously given they included
ity of the body of evidence is low (Table 14). routine use of analgesia and sedation in ICU care. An ICP crisis
Applicability. For the question of the use of NMB, the two could be caused by pain or anxiety in the absence of appropri-
studies were conducted in multiple sites and sample sizes were ate analgesia and sedation. In that setting, additional fentanyl
moderate, indicating cautious confidence in their applicabil- and/or midazolam may reduce ICP. However, these two stud-
ity. The remaining studies enrolled small samples and were ies may indicate that these agents are ineffective at reducing
conducted at single sites, indicating limited confidence in their ICP in patients receiving adequate analgesia and sedation. In
applicability. that setting, they confer either lesser efficacy than other tier 1
choices such as HTS and/or reduce MAP, negating benefit on
Summary of the Evidence CPP. Alternatively, a tier 2 intervention such as pentobarbital
Two new class 3 studies provided evidence to support the rec- may be required at that juncture of care (2).
ommendation (2, 3) (Table 15). Use of NMB for ICP Control. The NMB studies suggest
differences between subjects who received a NMB agent and
Evidence Synthesis those who did not (85, 120). However, for reasons discussed
Use of Fentanyl and/or Midazolam for ICP Control. One in the “Introduction section,” the studies evaluated also did not
small retrospective study found that in 31 pediatric patients include the indication for a given dose of NMB.
Analgesics, Sedatives, and Neuromuscular Blockade: Quality of the Body of

TABLE 14.
Evidence
Quality of
Meta- Evidence
Analysis Total Consistency Precision (High,
No. of Possible No. of (High, Directness (High, Moderate,
Studies Study Recommen- (Yes Subjects Moderate, (Direct or Moderate, Low, or
Topic Design dations or Noa) (n) Low) Indirect) Low) Insufficient)
Use of 2 retrospective No NA 326 Moderate Direct Low Insufficient

neuromuscular recommen
blockade (mixed) dation
Use of ketamine 1 prospective No NA 30 NA Indirect Low Insufficient
recommen
dation
Use of fentanyl 1 treatment series III.1. NA 31 NA Direct Low Low
and/or
midazolam
Use of etomidate 1 treatment series No NA 8 NA Direct Low Insufficient
recommen
dation
Use of fentanyl or 1 prospective III.1. NA 16 NA Direct Low Low
pentobarbital
a

Supplement
TABLE 15. Analgesics, Sedatives, and Neuromuscular Blockade: Summary of Evidence

Class 3 Studies

Use of NMB (mixed): no recommendation

Chin et al (120)a Retrospective (secondary analysis Class 3 GOS-E
17 hospitals of data from an randomized Unclear allocation No significant difference
United States, controlled trial) concealment; not blinded ICP
Australia, and New n = 90 Increased number of daily ICP
Zealand Group 1 NMB every day readings > 20 mm Hg (4.4 ± 1.1 vs
n = 31 2.4 ± 0.5; p = 0.015)
Group 2 none or intermittent NMB CPP
n = 59 No difference in frequency of low
Age: mean, 9.5; range, NR CPP
GOS-E at discharge and 3, 6, and Complications
12 mo postinjury; ICP; CPP; No significant difference
complications; ICU LOS ICU LOS
Longer ICU and hospital LOS (p =
0.003 and 0.07) for group 1
Consistent administration of NMB was
associated with ICH and increased ICU
LOS. However, it was not associated
with an increase in complications or
improvement in GOS-E.
Vavilala et al (85)a Retrospective Class 3 Discharge GOS
Five pediatric trauma n = 236 Selection; blinding; Adjusted risk ratio (95% CI)
centers Age: mean, 8; range, NR differential loss to All locations
Seattle, WA; Discharge GOS follow-up; baseline NMB monitored 1.13 (0.85–1.51) not
Pittsburg PA; differences unclear; did significant
Chicago, IL; not control for all relevant
confounders No NMB given 1.37 (1.24–1.53)
Torrance, CA;
Columbus, OH Reference: given, not monitored
In OR
NMB monitored 1.33 (0.93–1.9) not
significant
No NMB given 1.42 (1.28–1.58)
Reference: given, not monitored
In OR and ICU
Adherence = NMB monitoring (all
112/121 OR patients received)
Odds ratio: 18.8% overall; range,
0–32.1
ICU 3% (0–4.8)
NMB monitoring does not have a
significant effect on outcomes
compared with using NMB without
monitoring
(Continued)

Kochanek et al
(Continued). Analgesics, Sedatives, and Neuromuscular Blockade: Summary of

TABLE 15.
Evidence
Class 3 Studies
Study Design
Reference n
Use of ketamine: no recommendation

Bar-Joseph et al Prospective Class 3 ICP and CPP
(119)a Pre/post Selection bias; no control for Post ketamine administration
Meyer Children’s n = 30 confounders Overall in both groups
Hospital, Rambam 5 (17%) not TBI Indirect evidence ICP decreased by 30% (from
Medical Center Mixed severities; GCS 25.8 ± 8.4 to 18.0 ± 8.5 mm Hg) (p
Age: mean, NR; range, 6 mo to 18 yr
Israel NR; mixed pathologies < 0.001)
ICP; CPP
All elevated ICP (> 18 mm Hg) After 65 ketamine administrations
resistant to first-tier therapies 61 ICP decreased
received a single ketamine dose 3 increase < 2 mm Hg
(1–1.5 mg/kg) 1 increase > 2 mm Hg
Group 1: to prevent further ICP CPP increased from 54.4 ± 11.7 to
increase during a potentially 58.3 ± 13.4 mm Hg (p < 0.005).
distressing intervention Group 1
Group 2: additional measure to lower Mean ICP decreased from 25.2 ± 5.4
ICP to 17.9 ± 5.5 mm Hg within the first
2 min of ketamine administration (p <
0.001) and during distressing activity
increased slightly up to 19.6 ± 6.7 at
minute 7 and then decreased again.
Group 2
ICP decreased by 33% (from
26.0 ± 9.1 to 17.5 ± 9.1 mm Hg) (p
< 0.0001) within 2 min following
ketamine administration.
Result contradicts prior concerns that
ketamine increases ICP.
Use of fentanyl and/or midazolam: recommendation III.1.
Welch et al (3)a Treatment series Class 3 AUC used to represent cumulative ICH
Trauma PICU n = 31 Uncontrolled series exposure
St. Louis Children’s 8 (26%) abusive head trauma The mean change AUC, ICP increased
Hospital Age: mean, 8; range, 0–18 after drug administration
St. Louis, MO ICP, CPP, and MAP, difference Overall
between predrug and postdrug Increase in ICP of 17 mm Hg; p =
administration 0.04
Mean change in AUC for ICP
All significant
High-dose fentanyl: p = 0.02
Low-dose midazolam; p = 0.006
High-dose fentanyl plus low-dose
midazolam; p = 0.007
Dosing of fentanyl and midazolam did
not reduce the ICP.
Age and postresuscitation GCS did
not correlate with change AUC-
ICH, time after injury significantly
correlated with change AUC-ICH r
= –0.12, p = 0.02
(Continued)

Supplement
(Continued). Analgesics, Sedatives, and Neuromuscular Blockade: Summary of

TABLE 15.
Evidence
Class 3 Studies
Study Design
Reference n
Use of etomidate: no recommendation
Bramwell et al Treatment series Class 3 Etomidate administration resulted in
(126) n=8 Uncontrolled series a decrease in ICP vs baseline (p <
Primary Children’s Age: mean, NR; range, NR 0.05) without change in MAP; thereby
Medical Center, increasing CPP at each 5 min interval.
ICP At 6 hr after etomidate administration,
PICU
Salt Lake City, UT adrenocorticotropic hormone
stimulation tests showed adrenal
suppression in four of the eight
patients; however, no patient required
treatment with steroids.
Use of fentanyl or pentobarbital: recommendation III.1.
Shein et al (2)a Prospective Class 3 Hypertonic saline vs other drugs for ICP
University of n = 16 in analysis Inadequate control for Decrease in ICP; increase in CPP
Pittsburgh Age: mean, 44 mo; range, 34–124 confounders; small Associated with a two-fold faster
Pittsburgh, PA mo sample resolution of ICH than either fentanyl
Outcomes: ICP and CPP or pentobarbital.
Adjusted hazard ratio, 2.171 (1.062–
4.439)
Fentanyl
ICP decreased but had highest rate of
treatment failure rate
Pentobarbital
ICP decreased; CPP no significant
change
AUC = area under the curve, CPP = cerebral perfusion pressure, GCS = Glasgow Coma Scale, GOS = Glasgow Outcome Scale, GOS-E = Glasgow
Outcome Scale Extended, ICH = intracranial hypertension, ICP = intracranial pressure, LOS = length of stay, MAP = mean arterial pressure, NMB =
neuromuscular blockade, NR = not reported, OR = operating room.
New study.
a
Use of Ketamine for ICP Control. Traditionally, ketamine CSF Drainage

has been considered contraindicated in patients with raised
ICP, with the concern that it increases CBF and might increase Recommendations
ICP further. This reasoning has been questioned in a variety of Strength of Recommendation: Weak
studies and systematic reviews, and the report by Bar-Joseph Levels I and II
et al (119) adds to this evidence. However, a level III recom- There was insufficient evidence to support a level I or II recom-
mendation on ketamine was not made because the GCS score mendation for this topic.
for patients was not indicated, precluding the ability to con-
firm that the inclusion criteria were met for the publication. Level III
Use of Etomidate for ICP Control. Use of etomidate for ICP For ICP Control. III.1. CSF drainage through an EVD is sug-
control is largely of historical interest because we now have gested to manage increased ICP.
other sedatives and analgesics available that do not suppress Changes From Prior Edition. The recommendation from
adrenal function (126). A single dose of etomidate is, however, the Second Edition about use of lumbar drain (LD) was elimi-
commonly used for intubation in pediatric TBI (122). nated. One new class 3 treatment series was added to the evi-
dence base for this topic (127).
Indications From Adult Guidelines
The clinical investigators do not believe that the recom- Introduction
mendations about analgesics, sedatives, and NMB from the The EVD can be used not only to measure ICP in children fol-
adult guidelines can be used to guide treatment decisions in lowing TBI but may also provide added therapeutic benefit of
children. CSF drainage. CSF drainage through an EVD has frequently

Kochanek et al
TABLE 16. Cerebrospinal Fluid Drainage: Quality of the Body of Evidence

Quality of
Evidence
No. of Meta- Total Consistency Precision (High,
Study Recommen- Possible Subjectsb Moderate, (Direct or Moderate, Low, or
Topic Design dations (Yes or Noa) (n) Low) Indirect) Low) Insufficient)
External ventricular 1 retrospective III.1. No, study 56 Moderate Mixed Low Low
drainage to designs
reduce ICP 2 treatment differ
and improve series
outcomes
Lumbar drain to 2 treatment No No, samples 21 Moderate Direct Low Insufficient
reduce ICP series recommen differ
and improve dation
outcomes
ICP = intracranial pressure.
Meta-analysis to synthesize results is only possible if several studies address the same criteria/question and other design criteria are met.
a
The total number of subjects 18 yr old or younger when the studies included mixed ages.
b
been used in patients with intraventricular hemorrhage and Evidence Synthesis

hydrocephalus. With its ability to potentially decrease ICP EVD to Reduce ICP and Improve Outcomes. Refractory ICP
with CSF drainage, an EVD has been used as a therapeutic contributes to mortality, therefore, controlling elevated ICP is
device following TBI. Different techniques for CSF drainage, an important factor in patient survival following severe pedi-
both intermittent and continuous, as well as route of drain- atric TBI. In one study, Shapiro and Marmarou (54) retrospec-
age, EVD, or LD, have been reported in the pediatric literature tively studied 22 children with severe TBI, defined as a GCS
as potentially associated with increased risk of complications score of less than or equal to 8, treated with an EVD and CSF
from hemorrhage and malposition (46, 54, 127–129). drainage. Parameters measured included ICP, pressure-volume
index (PVI), mortality, and outcome. Draining CSF increased
Evaluation of the Evidence (improved) PVI and decreased ICP. Two neurologic deaths
Quality of the Body of Evidence. Studies included for this topic occurred in patients with refractory intracranial hypertension;
addressed questions about the use of EVD and LD to reduce however, the study did not report the ICP of the other three
ICP and improve outcomes for children with severe TBI. One patients who died or the four survivors with severe disability.
class 3 observational study (46) and two treatment series (54, Consequently, the absolute influence of CSF drainage and ICP
127) provided evidence to support the recommendation about in this sample cannot be determined. In contrast, Jagannathan
EVD. Two class 3 treatment series were insufficient to support a et al (46) retrospectively studied 96 children with severe TBI
recommendation about LD (128, 130). The studies were small comparing management of ICP alone versus ICP along with
with moderate consistency in the results and low precision of surgery using either an EVD or operative treatment (evacua-
effect estimates. One study (127) provided indirect evidence tion of hematoma or decompressive craniectomy). ICP con-
because it included patients with mixed severities and ages. trol was achieved in 82 patients (85%). Methods employed to
The overall quality of the body of evidence is low (Table 16). achieve ICP control included maximal medical therapy (seda-
Applicability. Three studies were conducted 20 years ago tion, hyperosmolar therapy, and NMB) in 34 patients (35%),
or more (54, 128, 130) and two were more recent (2008, 2011) EVD in 23 patients (24%), and surgery in 39 patients (41%).
(46, 127). Four of the five studies are treatment series (54, In this study, refractory ICP resulted in 100% mortality but the
127, 128, 130), and three of them had sample sizes greater authors concluded that it was unclear if controlling elevated
than 25 (54, 128, 130). All were conducted at single sites. ICP using CSF drainage was important in patient survival fol-
Four were conducted in the United States (46, 54, 128, 130) lowing severe pediatric TBI. More recently, Andrade et al (127)
and one (the most recent) in Brazil (127). Age ranges varied. reported on 58 patients who were treated with CSF drainage
The study designs and conduct in single sites limited their using an EVD. Eleven (19%) were pediatric, and 44 (76%) were
applicability. severe. Given the mixed ages and severity levels, the evidence
is indirect, and the sample size is small, limiting the level of
Summary of the Evidence evidence about ICP control or outcomes. Complications of
Three class 3 studies—one new (127) and two from the Second ventriculitis were seen in 8.3% of patients. The evidence from
Edition (46, 54)—provided evidence to support the recom- these studies supports recommendation III.1. about ICP con-
mendation (Table 17). trol. The data presented about patient outcomes reported in

Supplement
TABLE 17. Cerebrospinal Fluid Drainage: Summary of Evidence

Class 3 Studies
Study Design
Reference n
External ventricular drain to reduce ICP and improve outcomes: recommendation III.1.
Andrade et al (127)a Treatment series Class 3 All patients treated with ventricular CSF
Clinicas Hospital n = 58 No control for drainage
University of São n = 11, < 17 years confounders, Mortality
Paulo Medical n = 44 Glasgow Coma Scale enrollment criteria 3 of 11
School 4–8 unclear Neurologic assessment
São Paulo, Brazil Age: mean, 29; range, 4–65 Indirect evidence Among patients < 17 yr old
Mortality, neurologic assessment, Mixed ages and Favorable outcome 6 (54.5%)
complications severities
Unfavorable outcome 5 (45.5%) (not
separated by severity)
No significant difference in outcome,
between groups separated by age
(adults vs children), p > 0.05
Complications: not separated by age or
severity
Overall rate of infection (ventriculitis): 8.3%
Infection did not contribute to clinical
worsening or death, p > 0.05
Jagannathan et al Retrospective Class 3 Mortality
(46) n = 96 Control for confounders 3 of 23
University of Virginia n = 23 treated with CSF unclear for ICP, only GOS
Health System using EVD (other groups: patients with 2-yr No significant difference in mean or median
Charlottesville, VA craniectomy and medical follow-up included in GOS across treatment groups
management) analysis
ICP
Age: mean, 5.1; range, 3–18 Overall 20/23 (87%) achieved ICP control
Mortality, GOS at mean 2 yr, ICP, with EVD. Of three not achieving ICP
complications control, two died, one had a craniectomy
and then died.
Refractory ICP was associated with 100%
mortality; the method used to control ICP
had no correlation with mortality.
Complications
Higher rate of meningitis (p < 0.05)
in patients with drain (5/23, 22%)
compared with patients treated with
craniectomy (3/40, 7%)
Shapiro and Treatment series Class 3 Mortality
Marmarou (54) n = 22 Uncontrolled series 5/22
Albert Einstein College Age: mean, NR; range, 3 mo to Outcome
of Medicine 15 yr 4/17 severe disabilities 13/17 good
New York Mortality, outcome (scale not outcome or moderately disabled
specified), ICP ICP
Decreased 14/16
16 of 22 patients had PVI measured before
and after therapy.
Drainage increased PVI. Two of the five
deaths were due to uncontrolled ICP.
(Continued)

Kochanek et al
TABLE 17. (Continued). Cerebrospinal Fluid Drainage: Summary of Evidence

Class 3 Studies
Study Design
Reference n
LD to reduce ICP and improve outcomes: no recommendation

Baldwin and Rekate Treatment series Class 3 Mortality
(128) n=5 Uncontrolled series 2/5
Children’s Hospital and Age: mean, NR; range, 8–14 Outcome
Medical Centers Mortality, outcome (scale One moderate disability, two good recovery
Arizona undefined), ICP ICP
Decreased in 5/5
LD was added to an aggressive ICP control
protocol that included EVD.
Levy et al (130) Treatment series Class 3 Mortality
St. Joseph’s Hospital n = 16 Uncontrolled series 2/16
and Phoenix Age: mean, NR; range, 1–15 GOS
Children’s Hospital Mortality, GOS at 6 mo Eight good recovery; three moderate
Arizona postinjury, ICP disability, three severe disabilities
ICP
Decreased in 14/16
Lumbar drainage was added to an
aggressive ICP control protocol that
included EVD.
CSF = cerebrospinal fluid, EVD = external ventricular drainage, GOS = Glasgow Outcome Scale, ICP = intracranial pressure, LD = lumbar drain, NR = not
reported, PVI = pressure-volume index.
New study.
a
these studies were insufficient to determine the influence of Note. At the present time, there is insufficient evidence to
EVD on outcomes. recommend levetiracetam over phenytoin based on either effi-
LD to Reduce ICP and Improve Outcomes. The evidence cacy in preventing early PTS (EPTS) or toxicity.
was insufficient to support a recommendation about LD. Changes From Prior Edition. Recommendation III.1. is
modified from the Second Edition of these guidelines, with
Indications From Adult Guidelines phenytoin removed. The note regarding levetiracetam is new
In the adult guidelines, the level III recommendations for CSF to this Third Edition. Three new class 3 studies—one prospec-
drainage that might be additive to the pediatric recommendations tive observational (131), one retrospective observational (132),
include the following: the EVD system zeroed to the midbrain and one treatment series (133)—have been added to the evi-
with continuous drainage of CSF may be more effective in lower- dence base for this topic.
ing the ICP burden than intermittent drainage; and the use of CSF
drainage to lower ICP in patients with an initial GCS less than 6 Introduction
during the first 12 hours after injury may be considered (14). PTSs are defined as occurring early, within 7 days of injury, or
late, beyond 8 days of recovery (134). Risk factors associated
Seizure Prophylaxis with the occurrence of PTS include location of the lesion, cere-
Recommendations bral contusions, retained bone and metal fragments, depressed
Strength of Recommendation: Weak skull fracture, focal neurologic deficits, loss of consciousness,
GCS greater than 10, severity of injury, length of posttraumatic
Levels I and II amnesia, subdural or epidural hematoma, penetrating injury,
and age. Infants and children have lower seizure thresholds
(135), adding to the challenge of recognition of subtle clini-
Level III cal seizures (132) in critically ill children. The occurrence of
For Seizure Prevention (Clinical and Subclinical). III.1. Pro- electrographic seizures (seizures detected by continuous elec-
phylactic treatment is suggested to reduce the occurrence of troencephalogram recording) following severe TBI is higher in
early (within 7 d) PTSs. children than adults, occurring in up to 70% of cases (136).

Supplement
TABLE 18. Seizure Prophylaxis: Quality of the Body of Evidence

Quality of
Meta- Evidence
No. of Analysis Total Consistency Precision (High,
Studies Possible No. of (High, Directness (High, Moderate,
Study Recommen- (Yes or Subjects Moderate, (Direct or Moderate, Low, or
Topic Design dation Noa) (n) Low) Indirect) Low) Insufficient)
New topic: 1 prospective None NA 74 Low Indirect Low Insufficient

levetiracetam for
1 treatment
the prevention
series
of early
posttraumatic
seizures
Prophylactic 2 III.1. NA 469 (252 Moderate Mixed Low Low
treatment retrospective with severe
to reduce traumatic
posttraumatic brain injury)
seizures
a
Evaluation of the Evidence indirect (132) evidence to support the level III recommenda-
Quality of the Body of Evidence. Studies included for this tion for this topic.
topic addressed the questions about the use of levetiracetam Use of Levetiracetam to Prevent PTSs. Two studies assessed
or phenytoin to reduce PTSs. Indirect evidence from one treat- the use of levetiracetam; one small treatment series (n = 34)
ment series (133) and one phase II trial (131) was insufficient to prevent EPTS (133) and one phase II trial (n = 40) to pre-
to support a recommendation about levetiracetam. Two class vent long-term seizures measured at 2 years postinjury (131).
3 retrospective studies provide a low-quality body of evidence Chung and O’Brien (133) found that 17% of treated patients
to support the recommendation about phenytoin (132, 137) had EPTS. Pearl et al (131) reported one case of posttraumatic
(Table 18). epilepsy at 2 years postinjury. The small samples included
Applicability. Both studies supporting the recommenda- mixed severities, providing indirect evidence considered insuf-
tion were retrospective and conducted in single sites (132, 137). ficient to support a recommendation for this topic.
The studies addressing the use of levetiracetam, one single cen-
ter (133) and the other in two centers (131), used small sample Indications From Adult Guidelines
sizes and provided indirect evidence. Applicability is limited. The clinical investigators do not think that the recommenda-
tions about seizure prophylaxis from the adult guidelines can
Summary of the Evidence be used to guide treatment decisions in children.
Of the four class 3 studies included in the evidence tables for
this topic (131–133, 137), two provided evidence to support Ventilation Therapies
the recommendation (132, 137) (Table 19). Recommendations
Strength of Recommendations: Weak
Evidence Synthesis
Use of Phenytoin to Prevent PTSs. Two single-center, class 3 Levels I and II
retrospective studies reported use of prophylactic phenytoin There was insufficient evidence to support a level I or II recom-
to prevent EPTSs (132, 137). In a retrospective study with a mendation for this topic.
sample of 275 patients (221 with severe TBI), for the 133 who Level III
received an antiepileptic drug, 126 received phenytoin or fos- To Improve Overall Outcomes. III.1. Prophylactic severe
phenytoin (95%) (132). Of those, 2.4% had early seizures and hyperventilation to a Paco2 less than 30 mm Hg in the initial 48
97.5% did not. Both clinical and electrographic seizures were hours after injury is not suggested.
included in this analysis. An older retrospective review includ- III.2. If hyperventilation is used in the management of
ing 194 patients (31 severe) found in the severe TBI group a refractory intracranial hypertension, advanced neuromonitor-
significantly lower rate of PTSs in patients treated prophy- ing for evaluation of cerebral ischemia is suggested.
lactically with phenytoin than in those who were not treated Changes From Prior Edition. There are no content changes
prophylactically (137). These studies provide direct (137) and from the Second Edition to the recommendations. The title was

Kochanek et al
TABLE 19. Seizure Prophylaxis: Summary of Evidence

Class 3 Studies

Geographic Age (yr)
Levetiracetam for the prevention of EPTSs: no recommendation

Chung et al (133)a Treatment series Class 3 Early seizures
Nationwide Children’s n = 34 Uncontrolled series 6/34 patients (17%) had clinical seizures
Hospital, PICU Age: median, 6 mo; Fifteen of the 69 patients despite levetiracetam prophylaxis.
Columbus, OH range, 5 d to 16 yr underwent continuous Note: This is higher than rate reported in the
Effect of levetiracetam electroencephalogram literature for patients who receive phenytoin
on EPTSs, GCS, ICP monitoring, which was prophylaxis (2–15%).
monitoring not a standard practice. EPTSs occurred in younger patients (median
Indirect evidence age, 4 mo; p < 0.001) and abusive head
Mixed severities trauma patients (p < 0.0004).
Initial GCS score for patients with seizures
slightly higher (median, 9) than patients
without seizures (median, 7), p = 0.04.
GCS score for patients with and without
seizures were 9 (mean, 7–12) and 7 (mean,
3–12), p = 0.04.
Elevated ICP monitoring (of those monitored)
for patients with and without seizures was
0/1 (0%) and 11/14 (79%), respectively.
Pearl et al (131)a (note: Prospective (phase II Class 3 Posttraumatic epilepsy
these are the results trial) Unclear that selection was 1 of 40 subjects (2.5%) developed
for children included in n = 40 unbiased; no control for posttraumatic epilepsy (defined as seizures
a larger study by Klein Age: mean, NR; range, confounders > 7 d after trauma).
et al [138]) 6–17 Indirect evidence In the adult subjects in the same study (Klein
Children’s National Medical Effect of levetiracetam Mixed severities et al [138]), there were higher rates.
Center, Washington, on posttraumatic Acute and chronic treatment with
DC; and Mid-Atlantic epilepsy at 2 yr levetiracetam (55/kg/d) was safe and
Epilepsy and Sleep postinjury well tolerated in children 6–17 yr old
Center, Bethesda, MD with traumatic brain injury. No children
discontinued treatment because of side
effects.
Prophylactic phenytoin to reduce posttraumatic seizures: recommendation III.1.
Liesemer et al (132)a Retrospective Class 3 Early seizures
Primary Children’s Medical n = 54 moderate Selection process not No seizures or impact seizure only vs EPTS:
Center n = 221 severe specified; blinding Received AED before seizure
Salt Lake City, UT Median age: 7.4 w/o and qualifications of Yes 125 (52%) vs 8 (24%)
seizures; 1.4 w/ outcome assessors not
specified. No 116 (48%) vs 26 (76%), p < 0.01
seizures AED treatment is protective
Range: NR Indirect evidence
Mixed severities Odds ratio, 0.2 (95% CI, 0.07–0.5)
Effect of phenytoin on Completed AED course (7 d of treatment)
EPTSs (within first
7 d) 78 (32%) vs 8 (24%), p < 0.01
AED used
Fosphenytoin or phenytoin: 123 (51%) vs 3
(9%)
Phenobarbital: 2 (1%) vs 4 (12%)
Both: 0 (0) vs 1 (3)
23/34 (68%) developed EPTS in the first
12 hr postinjury
(Continued)

Supplement
TABLE 19. (Continued). Seizure Prophylaxis: Summary of Evidence

Class 3 Studies
Study Design
Reference n
Lewis et al (137) Retrospective Class 3 EPTS

Harbor University of n = 31 severe Control for confounders For children with GCS 3–8, treatment with
California Pediatric n = 194 total only in analysis of prophylactic phenytoin was associated with
Trauma Center Age: median, 6 yr; predictors of seizure, a reduced rate of seizures (2/13, 15%)
Torrance, CA range, 3 mo to 15 yr not for comparison of compared with patients not treated with
groups based on seizure prophylactic medication (9/17, 53%), p = 0.04
GCS: 3–8 (31–16%); prophylaxis one-tailed Fisher; p = 0.057, two-tailed.
9–15 (163–84%)
Rate of seizures in total group of 194 was
Effect of prophylactic 9.3%.
phenytoin on EPTS
AED = antiepileptic drug, EPTS = early posttraumatic seizure, GCS = Glasgow Coma Scale, ICP = intracranial pressure, NR = not reported.
New study.
a
changed from “Hyperventilation” to “Ventilation Therapies.” No is supported by studies in adults (145, 146). The use of hyper-
new studies have been added to the evidence base for this topic. ventilation is addressed in the TBI treatment algorithm article
which is a companion to these guidelines.
Introduction
Patients with severe TBI are comatose and may lack both air- Evaluation of the Evidence
way protective reflexes and normal ventilatory drive. Thus, Quality of the Body of Evidence. Studies included for this topic
airway protection and controlled mechanical ventilation and addressed the use of hyperventilation to manage pediatric
oxygenation are necessary. Hyperventilation has been used in patients with severe TBI. One class 3 retrospective study (141)
the management of severe pediatric TBI for the rapid reduc- contributed indirect evidence, and one class 3 treatment series
tion of ICP since the 1970s (139). It reduces ICP by producing (139) contributed direct evidence, and together they provided
hypocapnia-induced cerebral vasoconstriction with a reduc- a low-quality body of evidence to support the recommenda-
tion in CBF and CBV. The use of hyperventilation was based tions for this topic (Table 20).
on the assumption that hyperemia was common after pediatric Applicability. The studies supporting the recommendation
TBI and it was thought to reduce ICP by reducing luxury perfu- about use of hyperventilation—one large (141) and one small
sion (20). Subsequent pediatric studies, however, showed that (139)—were both conducted at single sites, which limited their
hyperemia is uncommon (58, 140), that low (rather than high) applicability.
CBF is associated with unfavorable outcome (58, 140), and that
hyperventilation can produce hypoperfusion or ischemia (58, Summary of the Evidence
140). Concerns have thus been raised about the safety of hyper- Two class 3 studies from the Second Edition provided evidence
ventilation therapy (141, 142). After TBI, the CBF response to to support the recommendations (139, 141) (Table 21).
changes in Paco2 can also be unpredictable (142). Hypocarbia
has also been shown experimentally to reduce the buffering Evidence Synthesis
capacity of CSF, an effect which may increase vulnerability of Use of Hyperventilation to Manage Severe TBI in Children.
the brain to abrupt increases in ICP in response to increases in Of the two class 3 studies included as evidence for this topic,
Paco2 (143). Thus, ventilation targeting normal arterial levels neither represented a comparison of hyperventilation to nor-
of Co2 (35–45 mm Hg) is currently recommended. mal ventilation, or to any other therapy targeting control of
Despite a prior recommendation in the 2003 guidelines ICP (139, 141). Similarly, there were no reports in children spe-
against prophylactic hyperventilation, several subsequent cifically addressing the effects of varying levels or duration of
reports suggested that it was still a commonly used therapy in hyperventilation on ICP or outcome, or studies of the transient
children (36, 144). This trend, however, may be reversing based application of hyperventilation in the setting of impending
on a report of the pretrial survey of the ADAPT study (4). herniation or ICP crisis. Last, neither study had a standardized
Finally, although clinical studies are lacking on the topic protocol to assess Paco2, measuring it only intermittently.
of reversal of transtentorial herniation in children after TBI, One report described the effects of hyperventilation on CBF
titrating the use of hyperventilation to effect (i.e., reversal of and brain physiology, and reported GOS at 6 mo (139). Skippen
pupillary dilation and resolution of Cushing’s triad) is recom- et al (139) carried out a selected treatment series of 23 children
mended as an integral component of the approach to the emer- (3 mo to 16 yr old) with isolated severe TBI. CBF was measured
gent treatment of transtentorial herniation, and this approach by Xenon-enhanced CT during Paco2 adjustments to greater

Kochanek et al
TABLE 20. Ventilation: Quality of the Body of Evidence

Quality of
Evidence
No. of Meta- Total Consistency Precision (High,
Study Recommen Possible Subjects Moderate, (Direct or Moderate, Low, or
Topic Design dation (Yes or Noa) (n) Low) Indirect) Low) Insufficient)
General 1 retrospective III.1. No, different 487 Low Mixed Low Low
use of designs
1 treatment III.2.
hyperventilation
series
a
TABLE 21. Ventilation: Summary of Evidence

Class 3 Studies
Reference Study
Type of Trauma Center Design n
Geographic Age (yr)
General use of hyperventilation: recommendations III.1. and III.2.

Curry et al (141) Retrospective Class 3 SH
Harborview Level I n = 464 Unclear if outcome SH on initial measurement was more common in
Pediatric Trauma Age: mean, 8; range, 0–14 assessment infants (≤ 2 yr) vs older children 30.8% vs 19.3%,
Center Prevalence of SH (SH, Paco2 methods unbiased respectively, p = 0.02.
Seattle, WA < 30 mm Hg) during Indirect evidence Prevalence of SH in the first 48 hr was similar
the initial 48 hr and risk Although SH between age groups, 58.9% for infants vs 58.0%
of inpatient mortality; is associated for older children; p = 0.91.
association between SH with mortality, Mortality
and morality it is unclear to Mortality-adjusted odds ratio,
what extent SH 1.44; 95% CI, 0.56–3.73 for one episode of SH
was caused 4.18; 95% CI, 1.58–11.03 for two episodes of SH
by intentional
hyperventilation. 3.9; 95% CI, 1.61–9.62 for ≥ 3 episodes
Skippen et al Treatment series Class 3 CBF measured at three levels of Paco2: > 35, 25–35,
(139) n = 23 Uncontrolled series < 25 mm Hg
British Columbia’s Age: mean, 11; range, 3 mo Areas of CBF below ischemic threshold 28.9%,
Children’s to 16 yr 59.4%, and 73.1%, respectively (not compared
Hospital PICU Ischemic threshold defined statistically)
Vancouver, BC, as < 18 mL/100 g/min Mean vasoreactivity 2.7% change in CBF per mm
Canada CBF; GOS score at 6 mo Hg change in Paco2 (range, –2.3% to 7.1%)
(the association between Frequency of regional cerebral ischemia increased
hyperventilation and GOS significantly with hyperventilation.
was not assessed) GOS
Ten (52.2%) had good or moderate outcome; 10
(43.5%) were severe or vegetative; one (4.3%)
died (no analysis).
CBF = cerebral blood flow, GOS = Glasgow Outcome Scale, SH = severe hypocarbia.
than 35, 25–35, and less than 25 mm Hg. The ischemic thresh- ICP and increased CPP. A relationship between the level of
old was defined as CBF less than 18 mL/100 g/min. However, the hypocarbia and frequency of cerebral ischemia was observed.
ischemic threshold in children is not known and may vary with The frequency of regional ischemia was 28.9% during normo-
the severity of tissue injury and patient age. Co2 reactivity of capnia and increased to 59.4% and 73.1% for Paco2 25–35 mm
CBF was also assessed. Management included CSF drainage and Hg and less than 25 mm Hg, respectively. However, no statistical
hyperosmolar therapy. As Paco2 was reduced with hyperventi- analysis was done. Fifty-two percent had good or moderate out-
lation, CBF decreased in almost all patients despite decreased come, 43.5% were severely disabled or vegetative, and 4.3% died.

Supplement
Again, no statistical comparison of outcomes was conducted, Note: See the last paragraph of the “Evidence Synthesis sec-
and outcomes were not reported by hyperventilation treatment. tion” for a discussion that explains the seemingly contradictory
A second report examined the association between hypo- recommendations.
carbia and outcome at hospital discharge in a large pediatric Changes From Prior Edition. Recommendations II.1. is
series of severe TBI victims who were all mechanically venti- modified with regard to timing from the Second Edition of
lated (141). Curry et al (141) carried out a retrospective cohort these guidelines. Recommendation III.1 on the use of hypother-
study of 464 patients less than 15 years old with an admis- mia for ICP control is now at a Level III (rather than a level II)
sion GCS score less than 9 and a head Abbreviated Injury given that none of the studies of higher class had ICP control as a
Score greater than or equal to 3, and with a Paco2 recorded primary outcome and given the safety concerns that were identi-
in the first 48 hours of admission for the years 2000 to 2005. fied. A level II recommendation about rewarming rate from the
The authors examined the prevalence of SH (Paco2, < 30 mm Second Edition has been removed and replaced with the more
Hg) and its relationship with neurologic outcome before (375 specific safety recommendation. A level III recommendation
patients) and after (89 patients) the publication of the 2003 about use of moderate hypothermia from the Second Edition
Pediatric TBI Guidelines (147). They found a nonsignificant has been removed. The safety recommendation about phenyt-
change in the prevalence of SH from 60% of patients before oin is new. One class 3 treatment series from the Second Edition
to 52% after (p = 0.19). Patients with one documented epi- has been removed (148). Two new meta-analyses (149–152),
sode of SH, controlling for ED GCS score, lowest ED SBP, ISS, three new RCTs—one class 1 (153), one class 2 (154), and one
Paco2 sampling frequency, and year of admission had adjusted class 3 (155)—and three new secondary analyses of RCTs—one
ORs for mortality of 1.44 (95% CI, 0.56–3.73) for one episode class 2 (156) and two class 3 (157, 158)—have been added to the
of SH, 4.18 (95% CI, 1.58–11.03) for two episodes, and 3.93 evidence base for this topic.
(95% CI, 1.61–9.62) for greater than or equal to three epi-
sodes, compared with patients with mild or no hypocarbia. Introduction
These findings, although retrospective, show an association of The definitions of hypothermia and hyperthermia vary.
SH with poor outcomes. However, there might be other con- Posttraumatic hypothermia is often classified as a core body
tributors to hypocarbia such as marked reduction in metabolic temperature less than 35°C, whereas a temperature greater
rates or acidosis from systemic shock. Furthermore, although than 38.0–38.5°C represents fever/pyrexia if it results from
SH was associated with mortality, it is unclear to what extent an altered thermoregulatory set point, and represents hyper-
SH was caused by intentional hyperventilation. Thus, the evi- thermia if it is imposed upon a normal set point. For sim-
dence from this study is indirect, and the exact contribution of plicity, the term hyperthermia is used to reflect an elevated
induced hyperventilation to poor outcome cannot be clearly core body temperature throughout this topic. Experimental
defined from this study. studies in animal models and clinical studies in children
Indications From Adult Guidelines. The recent Fourth Edi- demonstrated that hyperthermia correlates with poor out-
tion of the adult guidelines does not further inform the pediat- comes and it has been recommended that hyperthermia
ric guidelines for this topic (14). following TBI in children should be prevented (159, 160).
However, no studies of the influence of hyperthermia on
Temperature Control/Hypothermia outcomes in children were identified for inclusion in this
guideline.
Recommendations There are compelling reasons to explore the rationale for
Strength of Recommendation: Moderate the use of therapeutic hypothermia to limit secondary brain
Level I injury based on its role in decreasing cerebral metabolic
There was insufficient evidence to support a level I recommen- demands, inflammation, lipid peroxidation, excitotoxicity, cell
dation for this topic. death, and acute seizures (161, 162). Clinical studies reviewed
on temperature regulation for these guidelines addressed
Level II
global functional outcome and its effect on ICP. The influence
To Improve Overall Outcomes. II.1. Prophylactic moderate
on outcomes of reduction of ICP following severe TBI in chil-
(32–33°C) hypothermia is not recommended over normother-
dren remains to be determined. As discussed in other topics,
mia to improve overall outcomes.
the lowering of severely elevated ICP with respect to the treat-
Level III ment threshold may be a desirable outcome.
For ICP Control. III.1. Moderate (32–33°C) hypothermia is
suggested for ICP control. Evaluation of the Evidence
Safety Recommendation 1. If hypothermia is used and Quality of the Body of Evidence. Studies included for this topic
rewarming is initiated, it should be carried out at a rate of compared hypothermia with normothermia in the treatment
0.5–1.0°C every 12–24 hours or slower to avoid complications. of severe TBI in pediatric patients, and considered its influ-
Safety Recommendation 2. If phenytoin is used during hypo- ence on mortality, neurologic outcome, and control of ICP.
thermia, monitoring and dosing adjusted to minimize toxicity, Two meta-analyses (149–152); five RCTs—one class 1 (153),
especially during the rewarming period, are suggested. three class 2 (40, 144, 154), and one class 3 (155); and three

Kochanek et al
secondary analyses—one class 2 (156) and two class 3 (157, across the meta-analyses and class 1 and 2 RCTs. The overall
158) provide evidence for this topic. Findings are consistent quality of the body of evidence is moderate (Table 22).
TABLE 22. Temperature Control/Hypothermia: Quality of the Body of Evidence

Quality of
Meta- Evidence
Analysis Consistency Precision (High,
Possible (High, Directness (High, Moderate,
No. of Studies Study Recommen (Yes or Total No. of Moderate, (Direct or Moderate, Low, or
Topic Design dation Noa) Subjects (n) Low) Indirect) Low) Insufficient)
Components of overall quality: meta-analyses

Comparison of Two meta- II.1. NA 504b High Direct High Moderate
hypothermia analyses
and
normothermia Tasker et al (150)
Crompton et al (149,
Tasker and Akhondi-
Asl [151], and
Crompton and
Sharma [152])
Comparison of 1 RCT II.1. & Safety NA 77 NA Direct Moderate High
hypothermia Rec. 1.
and Adelson et al (153)
normothermia
Comparison of 4 RCTs No 350 High Direct Moderate Moderate
hypothermia
and Beca et al (154) II.1. NA 50 NA Direct NA NA
normothermia III.1.
Adelson et al (40) II.1. NA 75 NA Direct NA NA
III.1.
Hutchison II.1. NA 225 NA Direct NA NA
et al (144)
III.1.
Hutchison II.1. NA 225 (2008 NA Direct NA NA
et al (156) sample)
Comparison 1 RCT Safety No 119 High for Mixed Low Insufficient
of recommen Empey et al
hypothermia 2 retrospective dation 2 (155) and
and Bourdages
normothermia et al (157)
Su et al (158) No NA 84 NA Indirect NA NA
recommen
dation
Empey et al (155) Safety NA 19 NA Indirect NA NA
recommen
dation 2
Bourdages et al (157) No NA 16 NA Direct NA NA
recommen
dation
NA = not applicable, RCT = randomized controlled trial.
a
b
The total number of subjects for two meta-analyses was calculated with sample sizes of included studies in Tasker et al (150), except for Adelson et al (40), for
which the sample size from the original publication was used, and Salonia et al (164), which was included in Crompton et al (149) and not Tasker et al (150).

Supplement
Applicability. The meta-analyses and three of the RCTs are 152]), three new RCTs (one class 1 [153], one class 2 [154],
recent (149–155). Study sites were located in the United States, New and one class 3 [155]), and three new secondary analysis (one
Zealand, Australia, Canada, France, and the United Kingdom. Sam- class 2 [156] and two class 3 [157, 158]) were added to two
ple sizes for individual studies ranged from 16 to 225 patients, and
class 2 RCTs (40, 144) from the Second Edition. Together they
ages ranged from 0 to 18 years. Applicability concerns are minimal.
provided evidence to support the recommendations for this
Summary of the Evidence topic. Two new studies were reviewed and included in the evi-
Of the 10 studies included in the evidence tables, two new dence table (157, 158) but not used to support a recommen-
meta-analyses (one fair [150] and one poor quality [149, 151, dation (Tables 23 and 24).
TABLE 23. Temperature Control/Hypothermia: Summary of Evidence

Meta-Analyses
Study Design n
Reference Age (yr) Outcomes
Type of Trauma Center Hypothermia
Geographic Location Protocol Data Class Results
Comparison of hypothermia and normothermia: recommendations II.1. and III.1.

Tasker et al (150)a MA Fair-quality MA Mortality
Recommendation II.1. 7 RCTs Duplicate review not With conventional MA, no difference
Adelson et al (153) n = 470 specified; search in mortality.
Adelson et al (40) Age: 0–17 not comprehensive; With Bayesian MA, probability of
exclusions reducing mortality with hypothermia
Beca et al (154) Mortality not specified; compared with normothermia is
Biswas et al (165) Protocol varied across studies publication bias not 0.40 (with RR < 1 or RRR > 0).
Hutchison et al (144) assessed; conflict Probability of RRR of death >
Li et al (166) of interest NR 20%, with hypothermia rather than
normothermia, 0.28
Crompton et al (149, MA Poor-quality MA Mortality
Tasker and Akhondi-Asl Eight pediatric studies Duplicate review No significant difference (RR, 1.53;
(151), and Crompton and 7 RCTs; 1 observational not specified; 95% CI, 0.92–2.54; p = 0.10)
Sharma (152)a extent of literature Neurologic outcome
n = 454
Recommendation II.1. search unclear; No significant difference; 10%
Age: 3 mo to 18 yr exclusions not
Adelson et al (153) decrease in favorable neurologic
Mortality, neurologic outcome, specified; quality of
Adelson et al (40) GOS outcomes (RR, 0.90; 95% CI,
individual studies 0.80–1.01; p = 0.06)
Beca et al (154) Protocol varied across studies not assessed
Biswas et al (165) or reported; GOS
Hutchison et al publication bias not No significant difference; GOS scores
Salonia et al (164) assessed; conflict decreased by 0.17 points (mean
of interest NR difference, –0.17; 95% CI, –0.64 to
0.31; p = 0.50)
Class 1 and 2 Studies
Adelson et al (153)a RCT Class 1 Hypothermia vs normothermia
15 sites Cool Kids Trial (Evidence for Safety Mortality within 3 mo
United States, New Zealand, n = 77 Rec. #1.) 6 (15%) vs 2 (5%), p = 0.15
and Australia Age: median, 10.9 GOS and GOS-E
Recommendation II.1. and IQR: 3.4–14.6 No significant difference for GOS
safety recommendation 1 Hypothermia: median, 9.7; IQR, (p = 0.90) or GOS-E (p = 0.73)
4.2–14.5 Complications (acute nonserious
Normothermia: median, 12.5; IQR, infection and late nonserious
3.3–14.8 infection): no difference, p =
Mortality, GOS and GOS-E at 3 0.0622.
mo, complications Decompressive craniotomy: 17 (45%)
Hypothermia 32–33°C for vs 7 (18%), p = 0.0220
48–72 hr with slow rewarming; Trial terminated early for futility
0.5–1.0°C every 12–24 hr
(Continued)

Kochanek et al
TABLE 23. (Continued). Temperature Control/Hypothermia: Summary of Evidence

Study Design n
Reference Age (yr) Outcomes
Beca et al (154) a
RCT Class 2 Hypothermia vs normothermia
8 PICUs Phase 2 trial Criteria met except Mortality
Australia: Adelaide, Brisbane, n = 50 blinding which 3 (13%) vs 1 (4%), p = 0.34
Melbourne, Perth, Sydney Hypothermia: n = 24 was not possible Poor outcome (included death)
New Zealand: Auckland Normothermia: n = 26 or unclear; sample
size likely not 3 (12%) vs 4 (17%)
Canada: Vancouver, BC Age: adequate PCPC at 12 mo
Recommendations II.1. and Hypothermia: mean, 11.0; IQR, 4 (17%) vs 3 (12%), p = 0.70
III.1. 6.9–14.2 PICU length of stay (d) in survivors
Normothermia: mean, 9.5; IQR, 12 vs 11, p = 0.87
5.2–13.8
Hospital length of stay (d) in survivors
Mortality, PCPC, PICU length of
stay, hospital length of stay, 48 vs 40, p = 0.70
and mechanical ventilation; ICP, Mechanical ventilation (d) in survivors
MAP, CPP; complications 8 vs 10, p = 0.77
Early hypothermia (32–33°C) No predefined covariables were
for 72 hr with slow based on significant in the multivariable model.
individual patient to maintain ICP, MAP, CPP
ICP and CPP (no > 0.5°C every ICP was lower by 1.8 mm during
3 hr) cooling (95% CI, 0.3–3.4; p = 0.02)
No significant group difference in
MAP or CPP during cooling, p =
0.44 and p = 0.77
No significant group difference in MAP,
ICP, or CPP during rewarming, p =
0.68, p = 0.59, p = 0.07
Complications
Net effect of hypothermia during
cooling:
Heart rate drop: 23.4 beats/min; 95%
CI, 15.9–30.9; p < 0.001
Adelson et al (40) 2 RCTs Class 2 Mortality and GOS
Multicenter Total n = 75 Unclear reporting No difference between groups in
United States 2 samples/studies of randomization mortality or 3 and 6 mo GOS.
Recommendations II.1. and n = 48 (patients who met methods, allocation ICP
III.1. inclusion criteria, multicenter) concealment Overall, there was no statistical
methods, and difference in mean ICP between
n = 27 (patients who did not attrition
meet inclusion criteria, single the groups during the 5-d period, p
center) = 0.37 except within the first 24 hr,
Age: when the ICP was reduced in the
hypothermia group, p = 0.024.
Sample 1: mean, 6.89; range,
NR Complications
Hypothermia: mean, 6.92 No difference between groups in
complication rates
Normothermia: mean, 6.86
Sample 2: mean, 6.95; range, NR
Hypothermia: mean, 7.17
Normothermia: mean, 5.6
Mortality, 3 and 6 mo GOS, ICP,
complications
Hypothermia 32–33°C for 48 hr;
slow rewarming 1°C every
3–4 hr, with slower rates for
individual patients based on ICP
(Continued)

Supplement
TABLE 23. (Continued). Temperature Control/Hypothermia: Summary of Evidence

Study Design
n
Age (yr)
Reference Outcomes
Hutchison et al (144) RCT Class 2 Mortality

Multicenter n = 225 Some differences There were 23 deaths (21%) in
Canada, France, and United Age: between groups on the hypothermia group and 14
Kingdom Hypothermia: mean, 9.8; range, baseline prognostic deaths (12%) in the normothermia
Recommendations NR factors group (relative risk, 1.40; 95% CI,
II.1 and III.1. 0.90–2.27; p = 0.06). Although
Normothermia: mean, 10.2; this is not significant, it suggests
range, NR that additional studies to rule out
Mortality, PCPC at 4–6 mo or confirm this potential harm are
postinjury, ICP, complications needed.
Hypothermia 32.5°C for 24 hr; PCPC
rewarming 0.5°C every 2 hr No difference between groups on
functional outcomes at 6 mo
(p = 0.14)
ICP
ICP was lower during cooling and
higher during warming in the
hypothermia group at 16 hr
(p < 0.02), 24 hr (p < 0.01), 48 hr
(p = 0.01), and 72 hr (p = 0.03).
Complications
Significantly more episodes of
hypotension (p = 0.047), and lower
mean blood pressures and CPPs
(p < 0.001) in hypothermia group
CPP = cerebral perfusion pressure, GOS = Glasgow Outcome Scale, GOS-E = Glasgow Coma Scale Extended, ICP = intracranial pressure, IQR =
interquartile range, MA = meta-analysis, MAP = mean arterial pressure, NR = not reported, PCPC = Pediatric Cerebral Performance Category, RCT =
randomized controlled trial, RR = risk reduction, RRR = relative risk reduction.
New study.
a
Evidence Synthesis Five RCTs—one class 1 (153) and four class 2 (40, 144, 154,
Influence of Hypothermia on Mortality and Outcome. Two 156)—were also used as evidence to support recommenda-
meta-analyses, one moderate quality (150) and one poor qual- tion II.1. Both the publications by Hutchison et al (144, 156)
ity (149, 151, 152), contributed to the evidence to support rec- reported on the same set of patients, with the earlier focusing
ommendation II.1. against the use of hypothermia to improve on mortality and outcomes and the latter on the association
outcomes. Together they included seven RCTs (n = 469) (40: between hypotension and outcomes. The five studies included
samples 1 and 2, 153, 154, 163–166). The review by Crompton 427 patients, randomized to either hypothermia or normo-
et al (149) included one study (164) that reported an analy- thermia. Adelson et al (40), Beca et al (154), and Adelson et al
sis using patients from another study included in the review (153) reported no difference between groups in mortality or
outcomes (40, 153, 154). Hutchison et al (144) reported no dif-
(40). The Tasker et al (150) analysis of these studies found no
ference between groups on outcomes overall and no significant
significant difference in mortality between hypothermia and
difference in morality (p = 0.06), but characterized this as a non-
normothermia groups, using conventional methods. With
significant trend toward increased mortality in the hypothermia
Bayesian analysis, the probability of reducing mortality with group. This is one way to acknowledge the potential for harm
hypothermia versus normothermia was 0.40. The analysis by that merits further study.
Crompton et al (149) originally reported significantly higher Both publications by Adelson et al (40, 153) reported no dif-
mortality (p = 0.03), no difference in neurologic outcomes ference between groups in complications. Beca et al (154) found
(p = 0.06), and no difference in GOS scores (p = 0.50) for decreased heart rate for hypothermia patients with cooling.
hypothermia versus normothermia. However, after repeating Hutchison et al (144) reported significantly more hypotension,
the analysis to correct for “double counting” in patient num- pressor requirements, and lower blood pressures and CPPs in
bers across the studies, there was no significant difference in the hypothermia group; the 2011 analysis of that sample identi-
mortality (p = 0.10) (151, 152). fied in the hypothermia group a significant association between

Kochanek et al
TABLE 24. Temperature Control/Hypothermia Summary of Evidence: Secondary Analyses

Study Topic Study Design
Reference n
Class 2 Study
Secondary analysis of Hutchison et al (144)
Hutchison et al Retrospective (secondary Class 2 Hypothermia patients: association of ≥ 1
(156) analysis of 2008 RCT) RCT was class 2. episodes of hypotension and unfavorable
17 trauma centers n = 225 Blinding NR. Original outcome from 25 to 72 hr, p = 0.04, but
Canada, France, and Age: publication (2008) not from 0 to 24 hr after injury, p = 0.24.
United Kingdom Hypothermia: mean, 9.8; reported baseline Hypothermia vs normothermia
Recommendation II.1. range, NR differences between Hypotension: low systolic pressure
groups on prognostic Odds Ratio 95% CI p
Normothermia: mean, 10.2; factors.
range, NR 0–24 hr 1.25 86–1.83 0.24
Low systolic pressure, low mean 3.76 1.29–11.01 0.02
arterial pressure, low CPP, 25–72 hr 1.13 1.00–1.27 0.04
and episodes of hypotension 1.69 1.03–2.76 0.04
and unfavorable outcomes
0–72 hr 1.11 1.00–1.23 0.04
1.18 1.02–1.37 0.03
Hypotension: low mean arterial pressure
Odds Ratio 95% CI p
0–24 hr 1.18 0.81–1.74 0.39
2.20 1.32–3.68 0.003
25–72 hr 1.10 0.98–1.23 0.12
1.14 0.97–1.33 0.10
0–72 hr 1.08 0.98–1.20 0.12
1.18 1.02–1.36 0.02
Low CPP
Odds Ratio 95% CI p
0–24 hr 1.08 0.83–1.42 0.57
1.37 1.00–1.86 0.05
25–72 hr 1.13 1.02–1.26 0.03
1.16 1.02–1.32 0.03
0–72 hr 1.09 1.00–1.19 0.05
1.13 1.02–1.26 0.02
Class 3 Studies
Secondary analysis of Hutchison et al (144)
Bourdages et al Retrospective (ancillary study of Class 3 Hypothermia vs normothermia
(157)a 2008 RCT) RCT was class 2; this Mortality: 3 (43%) vs 1 (11%)
Multicenter and n = 16 study was conducted ARDS: 2 (29%) vs 0
university affiliated Age: mean, 12.7; range, in a single center; no Pneumonia: 3 (43%) vs 4 (44%)
PICU in a level III 7.2–17.0 power analysis; of 23
trauma center eligible patients, 16 Septic shock: 2 (29%) vs 0
Hypothermia: mean, 13.5; Brain herniation: 2 (29%) vs 1 (11%)
Quebec, Canada range, 8.8–16 were included.
No recommendation (Statistical significance for parameters above NR.)
Normothermia: mean, 12.2;
range, 7.2–17.0 PICU LOS (d): 11 (5–21) vs 15 (4–32)
Mortality, ARDS, pneumonia, ICP, CPP
septic shock, brain herniation, No significant difference
PICU LOS, ICP, CPP; Complications
complications Ventilator-free at 28 d: 17 (0–24) vs 13 (0–23)
Holter results:
-Arrhythmias: 5 (71%) vs 2 (22%), p = 0.13
-Minimum heart rate: 58 (51–83) vs 83
(64–104), p < 0.01
-Maximum heart rate: 111 (84–121) vs 137
(104–147), p < 0.01
(Continued)
Supplement
(Continued). Temperature Control/Hypothermia Summary of Evidence:

TABLE 24.
Secondary Analyses
Study Topic Study Design
Reference n
Secondary analysis of Adelson et al (40)

Su et al (158)a Retrospective (analysis of data Class 3 Mean and peak CSF MBP concentrations
Children’s Hospital of from a phase II RCT) RCT was class 2; this Hypothermia vs normothermia
Pittsburgh n = 27 w/TBI study selected 27 No significant difference between groups
Pittsburgh, PA 57 non-TBI controls patients from the trial Mean 45.27 ± 7.48 vs 56.70 ± 12.44
No recommendation Age: (hypothermia n = 14;
normothermia n = 13) Peak 69.38 ± 22.76 vs 90.33 ± 32.37
Hypothermia: mean, 6.78 and compared CSF No significant difference between groups
Normothermia: mean, 6.96 MBP to 57 controls. on days 1 or 2 before rewarming
Range for TBI patients: 7 wk to Indirect evidence TBI vs non-TBI controls
16 yr Intermediate outcome Concentrations in TBI patients were
Controls: mean, 0.73; range, 8 generally 400- to 500-fold greater than
d to 11 yr concentrations in controls (50.49 ± 6.97
Mean and peak CSF MBP vs 0.11 ± 0.01 ng/mL)
Empey et al (155)a RCT Class 3 Phenytoin levels

Children’s Hospital of n = 19 (Evidence for Safety Rec. Elevated free phenytoin concentrations in
Pittsburgh ICU Age: #2.) the hypothermia group in the rewarming
Pittsburgh, PA Hypothermia: median, 11.1; RCT was class 2; this and posttreatment periods (temperature
Safety range, 2.1–14.7 study selected a effect: p = 0.051; study period effect:
recommendation 2 subset of patients who p = 0.023; interaction: p = 0.633).
Normothermia: median, 13.6;
range, 2.5–16.2 received fosphenytoin The cumulative dose of fosphenytoin
or phenytoin. administered to each patient was
Phenytoin levels; hypothermia not different between the groups
measurements (Evidence for Safety Rec.
#2) (temperature effect: p = 0.853; study
Moderate hypothermia period effect: p = 0.249; interaction:
(32 33°C for 48 hr; Indirect evidence
p = 0.660).
rewarming 1°C every Intermediate outcome
Hypothermia vs normothermia
12–24 hr)
Albumin mean (sd):
Day 1: 3.0 (0.7) vs 3.2 (0.5)
Day 3: 2.4 (0.5) vs 2.4 (0.2)
Day 7: 2.7 (0.4) vs 2.2 (0.3)
Aspartate transaminase median (range):
41 (31–111) vs 60 (38–1303)
Alanine transaminase median (range):
30 (23–88) vs 26 (11–530)
Alkaline phosphatase median (range):
168 (66–288) 145 (70–382)
Total bilirubin mean (sd):
0.6 (0.4) vs 0.7 (0.8)
Serum creatinine mean (sd):
0.5 (0.2) vs 0.5 (0.2)
ARDS = acute respiratory distress syndrome, CPP = cerebral perfusion pressure, CSF = cerebrospinal fluid, ICP = intracranial pressure, LOS = length of
stay, MBP = myelin basic protein, NR = not reported, RCT = randomized controlled trial, TBI = traumatic brain injury.
New study.
a
1 or more episodes of hypotension and unfavorable outcomes Influence of Hypothermia on Intracranial Hypertension.
from 25 to 72 hours postinjury (p = 0.04), but not from 0 to Three class 2 RCTs contributed evidence to support recom-
24 hours postinjury (p = 0.24) (156). The rewarming rate for mendation III.1. for the use of hypothermia to decrease ICP (40,
patients in Hutchison et al (144) was faster than that of patients 144, 154). Beca et al (154) reported significantly lower ICP dur-
in Adelson et al (153) (0.5°C every 2 hr vs 0.5–1°C every 12– ing cooling for the hypothermia group (p = 0.02). Adelson et al
24 hr, respectively). (40) found that ICP decreased within the first 24 hours in the

Kochanek et al
hypothermia group, but no difference between groups over 5 Changes From Prior Edition .There are no content changes
days. Hutchison et al (144) found that ICP was significantly lower from the Second Edition to the recommendations (1). Two
during cooling, but then higher during warming in the hypother- new class 3 studies—one retrospective observational (85) and
mia group at 16 hours (p < 0.02), 24 hours (p < 0.01), 48 hours one treatment series (167)—were added to the evidence base
(p = 0.01), and 72 hours (p = 0.03). Although the evidence does for this topic.
not suggest a long-term benefit for ICP control with hypother-
mia, it does suggest that hypothermia produces an immediate Introduction
decrease in ICP. High-dose barbiturates lower ICP by suppression of metab-
Safety Recommendations. Safety recommendation 1 cautions olism and alteration of vascular tone (168–170) causing
against rapid rewarming that may be a source of complications improved coupling of regional blood flow to metabolic
seen in previous work in which temperature was increased at a demands resulting in higher brain oxygenation (171), with
rate of 0.5°C every 2 hours (144). The current recommended lower CBF and decreased ICP from decreased CBV. Barbi-
parameters are based on the protocol used in the study by Adel- turates may lower ICP when first-tier medical and surgical
son et al (153), which found no significant difference between management have not resulted in adequate control. However,
hypothermia and normothermia groups for adverse events.
cardiorespiratory side effects are very common and potentially
Safety recommendation 2 is based on the RCT by Empey
harmful, including decreased cardiac output, hypotension, and
et al (155) which demonstrated elevated free phenytoin lev-
increased intrapulmonary shunt, resulting in lower CPP and
els on rewarming from hypothermia (rate 1°C per 12–24 hr)
hypoxia. Thus high-dose barbiturate therapy has been reserved
although the cumulative doses administered to children in
for cases of intracranial hypertension resistant to first-tier
both groups were similar.
medical and surgical care.
Seemingly Inconsistent Recommendations. Furthermore,
Diffuse brain swelling and generalized hyperemia after
the published studies targeting the effect of hypothermia on
severe TBI are more common and more lethal (67) in young
long-term outcomes in pediatric severe TBI implemented it
children compared with older children and adults (172–174).
early after injury and did not specifically randomize patients
The rationale for treatment with barbiturate coma is based on
with intracranial hypertension to hypothermia versus another
the logic that uncontrolled ICP leads to ongoing secondary
second-tier therapy. It was used in a prophylactic manner. We
brain injury, and a higher risk of poor cognitive outcome or
did not identify studies comparing the efficacy of second-tier
death. Near maximum reduction in cerebral metabolism and
therapies implemented for refractory raised ICP. It would thus
be premature to dismiss hypothermia in this setting based on CBF occur when burst suppression is induced. Pentobarbital
the available evidence. is the most commonly reported medication used in children
and is dosed to achieve burst suppression, so continuous elec-
Indications From Adult Guidelines troencephalogram monitoring is required to monitor optimal
The clinical investigators do not consider the recommenda- dosing. Although high-dose barbiturates are reserved for a
tions about temperature control from the adult guidelines high-risk group, their use in severe pediatric TBI care is not
applicable to guide treatment decisions in children. rare.

BARBITURATES Quality of the Body of Evidence. Studies included for this
Recommendations topic addressed the influence of barbiturate therapy on out-
Strength of Recommendations: Weak comes and ICP. The evidence consists of one relatively large
observational study (85) and three small treatment series (47,
Levels I and II 167, 175). Although consistency and precision were moderate,
There was insufficient evidence to support a level I or II rec- the overall quality is low because the studies were rated class 3
ommendation for this topic. (Table 25).
Level III Applicability. Two of the included studies were over 25 years
For ICP Control. III.1. High-dose barbiturate therapy is sug- old (47, 175). Three were small and were conducted at single
gested in hemodynamically stable patients with refractory sites (47, 167, 175), whereas the larger study was conducted at
intracranial hypertension despite maximal medical and surgi- multiple sites (85). All were conducted in the United States and
cal management. included a range of ages from infants to teens. Applicability of
Safety Recommendation. When high-dose barbiturate ther- these studies is limited.
apy is used to treat refractory intracranial hypertension, con-
tinuous arterial blood pressure monitoring and cardiovascular Summary of Evidence
support to maintain adequate CPP are required because car- Four class 3 studies, two new (85, 167) and two from the Sec-
diorespiratory instability is common among patients treated ond Edition (47, 175), provided evidence to support the rec-
with barbiturate coma. ommendation (Table 26).

Supplement
TABLE 25. Barbiturates: Quality of the Body of Evidence

Quality
of Evidence
Analysis No. of (High, Directness (High, Moderate,
No. of Studies Recommen Possible Subjects Moderate, (Direct or Moderate, Low, or
High-dose 1 retrospective III.1. No 310 Moderate Direct Moderate Low

barbiturates
for refractory 3 treatment Safety Due to study
intracranial series recommendation designs
hypertension
a
Evidence Synthesis severe TBI management. They tested the relationship between
Influence of Barbiturate Therapy on Outcomes and ICP. Three rates of adherence and in-hospital mortality and discharge
treatment series (47, 167, 175) and one multicenter observa- GOS score. Ninety-four children (82%) developed high ICP
tional study provided data for this topic (85). The treatment (either ICP > 20 mm Hg or clinical signs of intracranial hyper-
series included a total of 74 patients who received barbiturate tension within the first 72 hr of admission). Cerebral edema
therapy for refractory intracranial hypertension. Mortality was based on radiologic imaging was present in 127 (53.8%). The
43%. Two of the three studies assessed 43 patients for function estimated decrease in mortality was 78% (adjusted hazard
at various timepoints, and reported that 19 (44%) had poor ratio, 0.22; 95% CI, 0.18–0.25) when ICU patients received
outcomes (167, 175). They also reported that of the total 63 barbiturates for refractory ICP as recommended.
patients, ICP control was achieved in 24 (38%). When high- These studies provide data to support the recommendation
dose barbiturates were added to additional therapies, rates for for this topic. However, there is insufficient evidence to rec-
control of refractory intracranial hypertension were 28% and ommend use of a particular barbiturate agent or regimen over
52%, respectively. The third study reported a steep decline in another to treat refractory intracranial hypertension.
ICP for patients treated with pentobarbital compared with
those not treated and those treated with mannitol (47). One Indications From Adult Guidelines
study found a nonsignificant lower risk of death in patients The recent Fourth Edition of the adult guidelines does not fur-
with controlled intracranial hypertension (risk reduction, 0.2;
ther inform the pediatric guidelines for this topic (14).
95% CI, 0.03–1.3), and significantly better median discharge
Pediatric Cerebral Performance Category (PCPC) scores for
Decompressive Craniectomy
that group (p < 0.05) (167). They reported significantly bet-
ter PCPC scores at 3–12 months follow-up (p < 0.05), but the Recommendations
analysis excluded patients who died (n = 14), and used dis- Strength of Recommendation: Weak
charge PCPC as the follow-up score for seven patients without
follow-up data. Levels I and II
Patient and disease characteristics, treatment variation, and There was insufficient evidence to support a level I or II recom-
the uncontrolled nature of these studies limited the ability to mendation for this topic.
associate the findings with the intervention. Mellion et al (167)
reported that among patients who had ICP controlled, high- Level III
dose barbiturate therapy was employed significantly later after For ICP Control. III.1. Decompressive craniectomy (DC) is
TBI (76 vs 29 median hours) compared with children whose suggested to treat neurologic deterioration, herniation, or
ICP remained uncontrolled. Kasoff et al (47) reported that intracranial hypertension refractory to MM.
more than 90% of treated patients received ionotropic infu- Changes From Prior Edition. The specification in the rec-
sions. In two studies, despite monitoring and infusion thera- ommendation from the Second Edition, “. . . with duraplasty,
pies, 80% of patients experienced episodes of hypotension or a leaving the bone flap out . . .” has been removed, and for this
fall in CPP below the goal level (47, 167). However, reports of edition, the recommendation is made specifically for ICP con-
both efficacy and toxicity information for all three studies are trol. One class 3 RCT from the First Edition which was removed
based on reports from single centers and relatively few patients. from the Second Edition was returned to this edition (176).
An observational analysis of 236 patients treated at five pedi- Fourteen new class 3 studies—five retrospective comparisons
atric trauma centers by Vavilala et al (85) studied adherence to (177–181) and nine treatment series (182–190)—were added
a set of clinical indicators to the 2003 Pediatric Guidelines for to the evidence base for this topic.

Kochanek et al
TABLE 26. Barbiturates: Summary of Evidence

Class 3 Studies
Study Design
n
Reference Age (yr)
Type of Trauma Center Mean (Range)
High-dose barbiturates for refractory intracranial hypertension: recommendation III.1.

Vavilala et al (85)a Retrospective Class 3 Discharge survival
Five pediatric trauma n = 236 Selection, blinding, In ICU: barbiturates used with high ICP:
centers Age: mean, 8.0; range, NR differential loss to adjusted hazard ratio, 0.22 (95% CI,
Seattle, WA; Pittsburg PA; In-hospital mortality follow-up, and baseline 0.18–0.26)
Chicago, IL; Torrance, differences unclear; Reference: no barbiturates used with high
CA; Columbus, OH did not control for all ICP
relevant confounders (Also measured in OR: not significant;
overall, ICU and OR combined: not
significant)
Mellion et al (167)a Treatment series Class 3 Mortality
Level I pediatric trauma n = 36 Uncontrolled series 22 of 36 survived
center Age: Survival more common in responders than
Salt Lake City, UT Controlled intracranial nonresponders but not significant (relative
hypertension: median, risk of death, 0.2; 95% CI, 0.03–1.3)
10.7; range, NR PCPC
Uncontrolled intracranial 19 with favorable outcome (PCPC less
hypertension: median, 6.4; than 3) at 3 mo or longer after injury;
range, NR three returned to normal function
Mortality; PCPC at discharge ICP
and follow-up (3–12 mo); 10 out of 36 (28%) had controlled
ICP; harms including refractory intracranial hypertension
infections, impaired Of 14 deaths, 13 were without control of
oxygenation, and CPP below ICP
target
No control of refractory intracranial
hypertension was significantly
associated with poor scores on PCPC
(p < 0.05), but with dichotomized PCPC,
function did not differ significantly
between patients with and without
controlled intracranial hypertension.
Pittman et al (175) Treatment series Class 3 Mortality
Cardinal Glennon n = 27 total Uncontrolled series 6 of 27
Memorial Hospital for n = 7 for outcomes GOS
Children Age: mean, 9.0; range, 2 mo Good recovery: 3
Missouri to 15 yr Moderate disability: 2
Mortality; GOS at 1, 6, and 12 Vegetative: 2
mo postinjury; ICP CPP ICP
14/27 (52%) achieved ICP < 20 mm Hg
Of 13 with persistently elevated ICP, six
died (22%)
CPP
Outcome not related to CPP
Kasoff et al (47) Treatment series Class 3 Mortality
Westchester County n = 11 Uncontrolled series 4/11 (36%)
Medical Center Age: mean, 8.8; range, 3 mo Hypotension (mean arterial pressure) <
New York to 17 yr 80 mm Hg occurred in 9/11 (82%)
Mortality
CPP = cerebral perfusion pressure, GOS = Glasgow Outcome Scale, ICP = intracranial pressure, NR = not reported, OR = operating room,
PCPC = Pediatric Cerebral Performance Category.
New study.
a

Supplement
TABLE 27. Decompressive Craniectomy: Quality of the Body of Evidence

Quality of
Meta- Evidence
Analysis Total Consistency Precision (High,
No. of Possible No. of (High, Directness (High, Moderate,
Studies Recommen (Yes or Subjects Moderate, (Direct or Moder- Low, or
Topic Study Design dation Noa) (n) Low) Indirect) ate, Low) Insufficient)
Effect of DC 1 randomized No No 161 (DC: 67; Low Direct Low Insufficient

vs medical controlled recommen medical
management on trial dation management:
ICP, mortality, 94)
and outcomes 1 prospective
4 retrospective
Effect of DC on 16 treatment III.1. No 190 Low Direct Low Low
ICP, mortality, series
and outcomes
Outcomes of DC 1 retrospective No NA 37 NA Direct Low Insufficient
by mechanism recommen
of injury dation
DC = decompressive craniectomy, ICP = intracranial pressure, NA = not applicable.
a
Introduction demonstrated on CT imaging, or may be focused on creating

DC for TBI is a controversial procedure that has become more the maximum possible compartmental expansion to increase
widely considered as a treatment option. Controversy results compliance. Technique varies between surgeons and in
from its invasiveness without clearly defined indications, response to clinical context.
lack of an optimally specified surgical technique, variability Two RCTs treated refractory intracranial hypertension in
in reported outcomes, and significant risk for complications adults with DC (199, 200). No similar trials are available for
(191, 192). DC may be performed solely to treat ongoing or the pediatric population.
refractory intracranial hypertension (“therapeutic DC”), or
concomitantly with the removal of a mass lesion in order to Evaluation of the Evidence
either treat observed brain swelling or prophylaxis against Quality of the Body of Evidence. Studies included for this
anticipated swelling (“incidental DC”). For therapeutic DC, topic addressed questions about the effect of DC on ICP
the timing of the procedure may be predicated on the neuro- control, mortality, and functional outcomes, and outcomes
logic examination, an episode of neurologic deterioration, the of DC in treating accidental versus abusive head trauma.
degree of initial ICP, or the resistance of intracranial hyper-
Subtopics addressed were effect of timing of the procedure
tension to medical treatment. These two general categories of
and complications. One poor-quality RCT (176), one poor-
indication for DC are referred to using a variety of sometimes
quality prospective study (201), and four poor-quality retro-
confusing terms in the literature (e.g., each has been referred
spective studies (177, 178, 180, 181) compared ICP, mortality,
to as either “primary” or “secondary”) (193–195). This section
focuses on therapeutic DC (often but not always referred to as and outcomes for patients who received DC versus those
“primary DC”). who received MM. Sixteen poor-quality treatment series
Published techniques for DC vary. Bone may be removed compared ICP before and after DC, and reported mortal-
unilaterally or bilaterally and may include or exclude sub- ity and outcomes without comparators (182–190, 202–208).
temporal decompression. Hemispheric craniectomies vary in One moderate-quality retrospective study compared out-
recommended size (from relatively small to nearly complete comes after DC between patients who sustained abusive head
subtotal) and circumferential or bifrontal craniectomies are trauma injury with those sustaining other mechanisms of
also used (196). Management of the dura also varies, includ- injury (179). The overall quality of the body of evidence for
ing no manipulation, simple scoring, or wide opening (with this topic is low (Table 27).
or without expansile duraplasty) (197, 198). There is no con- Applicability. Of the 23 included studies conducted in 14
sistent relationship between choice of craniectomy and dural different countries, all but one were single center; 18 had sam-
opening techniques. The design of a procedure for any indi- ples sizes less than 25 patients, and 16 did not have compara-
vidual patient often depends on the underlying pathology as tors. Applicability of the individual studies is questionable.

Kochanek et al
TABLE 28. Decompressive Craniectomy: Summary of Evidence

Class 3 Studies
Study Design
n
Reference Age (yr)
Type of Trauma Center Mean
Geographic (Range)
Effect of DC vs MM on ICP, mortality, and outcomes: no recommendation

Mhanna et al Retrospective Class 3 Mortality
(178)a n = 34 Selection of patients to DC 5/17 vs control 3/17, p = 0.34
Metro Health DC = 17 receive DC was based No significant difference between
Medical Center MM = 17 on physician discretion; groups for mortality or GOS.
Cleveland, OH inadequate control for GOS
Age: confounders; baseline
DC: mean, 10.2; range, NR differences between groups. Median GOS score
MM: mean, 12.4; range, NR DC = 4 (IQR, 3–5) vs control 3
Mortality; GOS 4 yr postinjury (IQR, 3–4), p = 0.09
(IQR, 1–6 yr) Comparing patients who died
or had a disability, there was a
significant difference between
the DC and control groups
(71% [12/17] vs 100% [17/17],
respectively; p = 0.022).
Thomale et al Retrospective Class 3 GOS
(181)a n = 53 No control for confounding; No significant difference in median
Campus Virchow DC = 14 baseline differences GOS at 3, 6, and 12 mo
Medical Center No DC = 39 between groups ICU days
Germany Age: Significantly greater days in ICU for
All patients: median, 8; range, 0–16 DC group (p = 0.026)
No DC: median, 7; range, 0–16 DC group had significantly lower
initial GCS but comparable
DC: median, 12; range, 1–16 outcomes.
GOS at 3, 6, and 12 mo (mean follow-
up 5.2 ± 2.4 yr [range, 1–10.5]); ICU
days
Rubiano et al Retrospective Class 3 Mortality for pediatric patients
(180)a DC = 16; 7 pediatric Controlled for confounders DC = 1/7 (14%)
Simon Bolivar Control = 20; 5 pediatric for total sample with mixed Non-DC = 2/5 (40%)
Hospital Pediatric patients: Age: mean, 5.86; ages GOS
Colombia range, 1–15 For pediatric patients:
Mortality; GOS 6 mo postinjury DC Non-DC
1 2/7 (14%) 2/5 (40%)
2 0 0
3 0 3/5 (60%)
4 2/7 (29%) 0
5 3/7 (57%) 0
For total sample (all ages):
mean GOS at 6 mo postinjury
significantly higher in DC group
than control group (unadjusted
t = 4.26; p = 0.0002)
(Continued)

Supplement
TABLE 28. (Continued). Decompressive Craniectomy: Summary of Evidence

Class 3 Studies
Study Design
Reference n
Geographic Mean (Range)
Josan and Retrospective Class 3 Statistical analysis not performed.

Sgouros (177)a n = 12 Small sample; baseline Mortality
Birmingham Early DC (< 24 hr) = 6 differences between DC = 0/6
Children’s Non-DC = 6 groups; no control for Non-DC = 2/6 (33%). One of
Hospital confounding whom received late DC at 9 d.
Age:
United Kingdom GOS
Early DC: mean, 13; range, 2–16
Non-DC: mean, 11.5; range, 7–15 DC Non-DC
Mortality; GOS 1 yr postinjury 1 0 2/6 (33%)
2 0 0
3 0 1/6 (17%)
4 2/6 (33%) 0
5 4/6 (67%) 3/6 (50%)
More patients with better outcomes in
DC group than non-DC group
Taylor et al Randomized controlled trial Class 3 GOS and HSUI
(176) n = 27 Allocation concealment, There was no significant difference
Royal Childrens’ DC = 13 blinding, and adequate between groups in GOS or HSUI 6
Hospital Medical = 14 sample size unclear. mo postinjury.
Melbourne, VIC, Baseline differences NR. MM DC
Australia Age: median, 120.9 mo; range, No intent-to-treat analysis
13.6–176.4 mo GOS
One patient GCS = 11 in DC group Fav. 2 7
GOS and HSUI 6 mo postinjury; ICP Unf. 12 6
after randomization/surgery HSUI
Fav. 1 6
Unf. 1 37
ICP 48 hr after randomization or
surgery
Range (sd)
DC 17.4 (3.4) vs MM 21.9 (8.5)
Number of episodes > 20 mm Hg
DC-107 vs MM-223
Number of episodes > 30 mm Hg
DC-9 vs MM-29
There was no significant difference
between groups in mean ICP 48 hr
after randomization or surgery.
(Continued)

Kochanek et al

Class 3 Studies
Study Design
Reference n
Geographic Mean (Range)
Cho et al (201) Prospective Class 3 Mortality

Taichung Veterans n = 23 No control for confounders; Three patients died—all from group
Hospital Group A: low ICP/medical = 6 very small sample; no power B.
Taiwan (Republic Group B: high ICP/medical = 7 calculation; significant COS
of China) differences in baseline ICP Mean COS for group B significantly
Group C: high ICP/DC = 10 and Children’s Coma Scale
Age: mean, 5.91 mo; range, 2–14 mo worse than groups A or C (p =
scores 0.0058)
Group A: mean, 5.33 mo; range,
2–11 mo ICP
Group B: mean, 7.42 mo; range, In group C, DC lowered the mean
3–14 mo ICP measurements from 54.9 to
11.9 mm Hg. Effect of medical
Group C: mean, 5.2 mo; range, treatment on ICP for groups A and
2–14 mo B was NR, so group differences
Mortality; dichotomized scores on COS unknown.
measured between 6 mo and 6 yr Although DC was performed
postinjury (mean, 3.2 yr) based on ICP elevation alone, a
(COS ranges from 1 [best] to 5 mean of 32 mL of subdural blood
[worst]). was removed during the surgery,
COS 1 to 2 = good indicating that this sample includes
COS 3 to 5 = poor both primary and secondary DC
ICP patients.
Outcomes of DC by mechanism of injury: no recommendation

Oluigbo et al Retrospective Class 3 Mortality
(179)a n = 37 Outcome assessors not 5 of 14: abusive head trauma,
Children’s n = 14 abusive head trauma blinded; baseline difference 35.7% (p < 0.05)
Hospital and n = 23 other mechanisms in age for accidental trauma 1 of 23: other mechanisms 4.3%
University of OR = 12.2 (p = 0.02) for abusive
Colorado Age: mean, 6; range, 9 wk to 15 yr
Abusive head trauma: mean, 2.2; head trauma
Denver/Aurora, KOSCHI poor outcome (score of 1,
CO range, NR
Accidental: mean, 8.4; range, NR 2, or 3)
Mortality; KOSCHI (scored same as Nonaccidental: 57.2%
GOS) at mean 23.9 mo (range, Other mechanisms: 30.4%
1–94 mo) Authors report no significant
difference between groups for
poor outcome and an OR for poor
outcome of 3.04 for the abusive
head trauma group.
COS = Children’s Outcome Scale, DC = decompressive craniectomy, Fav. = favorable, GCS = Glasgow Coma Scale, GOS = Glasgow Outcome Scale, HSUI
= Health State Utility Index, ICP = intracranial pressure, IQR = interquartile range, KOSCHI = King’s Outcome Scale for Childhood Head Injury, MM = medical
management, NR = not reported, OR = odds ratio, Unf. = unfavorable.
New study.
a

Supplement
TABLE 29. Decompressive Craniectomy: Summary of Evidence

Treatment Series
Study Design
Reference n
Type of Trauma Age (yr)
Center Mean (Range)
Effect of DC on ICP, mortality, and outcomes: recommendation III.1.

Pechmann et al Treatment series Class 3 Mortality 1/12 (8%)
(188)a n = 12 Small GOS
University Medical Age: mean, 8.5; range, uncontrolled 1 = 1/12 (8%)
Center Freiburg 2–14 series 2 = 2/12 (17%)
Germany Mortality; GOS mean 3.3 3 = 2/12 (17%)
mo (± 2 mo) postinjury; 4 = 5/12 (42%)
range, 3–83 mo; ICP;
complications 5 = 2/12(16%)
ICP
Initial decrease of ICP < 20 mm Hg in all children. Three
of 12 showed a secondary increase in ICP.
Complications
• Formation of hygroma (83%)
• Aseptic bone resorption of the reimplanted bone flap (50%)
• Posttraumatic hydrocephalus (42%)
• Infection: secondary infection or dysfunction of
ventriculoperitoneal shunt (25%) or cranioplasty (33%)
• Epilepsy (33%)
Due to complications, 75% of patients required further surgery
in addition to cranioplasty with up to eight interventions.
Prasad et al (209)a Treatment series Class 3 Mortality
All India Institute of n = 71 Small 18/36 severe (50%)
Medical Sciences Severe = 36 uncontrolled GOS-E
New Delhi, India Age: mean, 1.6; range, 1 mo series 7 to 8 for all who survived except 1 (reported for all
to 3 yr severities)
Mortality; GOS-E; mean Complications (reported for all severities)
19.6 mo (range, 2–42 Ventilator-associated pneumonia in 22 of 71 cases
mo); complications Late onset seizures in 2
Septicemia in 6
Wound infection in 7
Subdural hygroma in 11
Hydrocephalus in 13
Desgranges et al Treatment series Class 3 Mortality
(184)a n = 12 Small 4/12 (33%)
Hospital Femme Mere Age: mean, 8; range, NR uncontrolled ICP
Enfant Mortality; ICP; complications series Before DC (n = 12) After DC (n = 12)
Lyon, France (IBL); Mortality 46 ± 18 mm Hg 10 ± 4 mm Hg
DC induced significant decrease in ICP (p = 0.0005), mean
arterial pressure (p = 0.04) between the immediate pre-
and postoperative periods.
Complications
Median IBL during DC was 49% (17–349%) of EBV.
Children with IBL ≥ 50% of EBV had higher preoperative
ICP (p = 0.03) and INR (p = 0.02) than those with an IBL
< 50% of EBV.
Mortality
IBL ≥ 50% IBL<50%
3/6 50% 1/6 (17%)
(Continued)

Kochanek et al

Treatment Series
Study Design
n
Reference Age (yr)
Khan et al (186)a Treatment series Class 3 Mortality (of entire sample n = 25)
University Hospital n = 25, 21 severe Small 9 patients died (36%)
Karachi, Pakistan Age: mean, 6; range, 1–15 uncontrolled GOS (of entire sample n = 25)
Mortality; dichotomized series 16 had good outcome (GOS 4 [n = 6] and 5 [n = 10]).
GOS at mean 5 mo (± 2 Complications
sd); complications (blood
Significantly greater mortality among patients with >
loss) 300 mL operative blood loss (p = 0.001)
Csókay et al (183)a Treatment series Class 3 Mortality
3 Children’s Hospitals n=8 Small 2/8 (25%)
Hungary and Wales Age: mean, 7.13; range, uncontrolled GOS
1–12 series Discharge 1 yr
Mortality; GOS at 1 yr 1 = 2/8 (25%) 1 = 2/8 (25%)
postinjury; ICP; effect of 2 = 1/8 (12.5%) 2 = 1/8 (13%)
timing
3 = 1/8 (112.5%) 3 = 0/8
4 = 4/8 (50%) 4 = 2/8 (25%)
5 = 0 5 = 3/8 (37%)
ICP
For six patients with pre- and postoperative ICP
measures, the average preoperative ICP was 23.3 mm
Hg compared with 15.3 mm Hg postoperative ICP
(calculated from data provided in Tables 1 and 2).
Timing
At above 20 mm Hg, fast progression of ICP (within
15 min) can occur, limiting the time available to perform
DC with a successful patient outcome.
Perez Suarez et al Treatment series Class 3 Mortality
(190)a n = 14 Small 2/14 (14%)
Nino Jesus Pediatric Age: mean, 5.5; range, 11 uncontrolled GOS
Children’s Hospital mo to 15 yr series 1 = 2/14 (14%)
Madrid, Spain Mortality; GOS at 2 2 = 0
yr postinjury; ICP; 3 = 0
complications
4 = 7/14 (50%)
5 = 5/14 (36%)
ICP
In 13 patients, craniectomy initially decreased ICP to <
25 mm Hg.
Complications:
Hygroma 8/14 (57%)
Infections 3/14 (21%)
Aseptic resorption of bone flap 3/14 (21%)
(Continued)

Supplement

Treatment Series
Study Design
Reference n
Center Geographic Mean (Range)
Adamo et al (182)a Treatment series Class 3 Mortality

Albany Medical n=7 Small All patients survived
Center Hospital Age: mean, 13.86 mo; uncontrolled KOSCHI
New York range, 2–24 mo series Discharge 1 yr
Mortality; KOSCHI 3a = 4/7 (57%) 4a = 3/7 (43%)
(scored same as GOS); 3b = 3/7 (43%) 4b = 1/7 (14%)
complications
3b = 3/7 (43%)
Complications
7/7 (100%) hydrocephalus
2/7 (29%) bone resorption
Jagannathan et al Treatment series Class 3 Mortality
(204) n = 23 Small 7 (30%; one intraoperative; five postoperative; one
University of Virginia Age: mean, 11.9; range, uncontrolled in rehabilitation) Mortality primarily in patients with
Health System 2–19 series multisystem trauma.
Virginia Mortality; GOS 5 yr mean GOS
(range, 11–126 mo); ICP Mean 4.2 (range, 1–5)
n Score
9 5
5 4
1 3
1 2
7 1
ICP
ICP control in 19/23 patients. High ICP associated with
increased mortality.
Figaji et al (185)a Treatment series Class 3 Mortality
Red Cross War n = 12 Small 1/12 (8%)
Memorial Children’s Age: mean, 8.125; range, uncontrolled GOS
Hospital 5–12 series 1 = 1/12 (8%)
Cape Town, South Mortality; GOS at median 2 = 0
Africa 35 mo postinjury 3 = 0
(range, 1–6 yr); ICP;
complications 4 = 6/12 (50%)
5 = 5/12 (42%)
ICP
Control of ICP and clinical improvement was seen in
all but one patient. Mean sustained ICP reduction
postcraniectomy was 53.5% (range, 39–61%) in both
pre- and postoperative ICP monitoring patients.
Complications
1/12 (8%) bone flap sepsis after replacement
1/12 (8%) slight subsidence of the flap at follow-up
2/12 (17%) asymptomatic subdural hygromas
Kan et al (205) Treatment series Class 3 Mortality
Primary Children’s n = 51 Small Five of six patients died (83%).
Medical Center n = 6 for intractable ICP uncontrolled ICP
Utah Age: mean, 6.6; range, NR series Three of the four patients with postoperative ICP
Mortality; ICP monitoring had ICP < 20 mm Hg.
(Continued)

Kochanek et al

Treatment Series
Study Design
Reference n
Center Mean (Range)
Rutigliano et al (207) Treatment series Class 3 Mortality

Level 1 regional n=6 Small All patients survived.
trauma center Age: mean, 14.5; range, uncontrolled FIM
Stoney Brook 12–19 series 3: Independent
University Health Mortality; FIM score at 2: Assistance
Center hospital discharge; ICP 1: Dependent
New York
ICP
Five of the six patients had sustained postoperative
ICP < 20 mm Hg. One had ICP elevations requiring a
second surgery for debridement, with no subsequent ICP
elevations.
Skoglund, 2006 Treatment series Class 3 Mortality
(208) n = 8 Peds (19 total) Small 1/8 (12.5%)
Neonatal ICU Age: Mean 12.6 Range: 7 uncontrolled GOS
Sahlgrenska to 16 series 1: 1 (12.5%)
University Hospital Mortality; GOS at minimum 2: 0
Goteborg, Sweden 1 yr postinjury (range 1 3: 3 (37.5%)
to 6 yr).
4: 1 (12.5%)
(ICP and effect of size of
DC on outcome were 5: 3 (37.5%)
also considered, but in
only 9 patients, and ages
not specified.)
Figaji et al (202) Treatment series Class 3 GOS
Red Cross War n=5 Small All patients had early clinical improvement after surgery
Memorial Children’s Age: mean, 8; median, 6; uncontrolled and were GOS 4 or 5 at long-term follow-up (14–40 mo).
Hospital range, 5–12 series ICP
Cape Town, South GOS at a range of 14–42 In the four patients with postoperative ICP monitoring, two
Africa mo; ICP; complications had no ICP elevations and two had mild, easily controlled
elevations.
Complications
1 subdural hygroma
1 cerebrospinal fluid leak and bone flap sepsis
Messing-Jünger et al Treatment series Class 3 Mortality
(187)a (publication is Small 2/7 (29%) died between 24 and 48 hr and showed signs
Heinrich University comparative but seven uncontrolled of decerebration at admission.
Hospital pediatric patients in the series GOS (in 6 of 7 patients; 1 lost to follow-up)
Dusseldorf, Germany sample were reported as
a treatment series) 1 = 2/7 (29%)
n=7 2 = 0
Age: mean, NR; range, 3 = 1/7 (14%)
1–16 4 = 1/7 (14%)
Mortality, GOS (follow-up 5 = 2/7 (29%)
time not specified)
(Continued)

Supplement

Treatment Series
Study Design
n
Reference Age (yr)
Ruf et al (206) Treatment series Class 3 Mortality

PICU n=6 Small All patients survived.
Justus-Liebig Age: mean, 7.8; range, 5–11 uncontrolled Neurologic Status
University Medical Mortality; neurologic status series Normal: 3
Centre at 6 mo posttrauma; ICP; Hemiparesis: 1
Giessen, Germany complications Spasticity: 1
In rehabilitation: 1
ICP
ICP decreased to < 12 mm Hg in five patients. Sixth
patient required contralateral subsequent DC, then ICP
was maintained at ≤ 20 mm Hg.
Complications
1 (late aseptic necrosis of bone flap: 1 patient)
Hejazi et al (203) Treatment series Class 3 Mortality
Landeskrankenhaus n=7 Small All patients survived.
Hospital Age: mean, 8.6; range, uncontrolled GCS
Feldkrich, Australia 5.5–14 series All patients had a GCS of 15.
Mortality; GCS within 5 wk ICP
postinjury; ICP Decrease in ICP to < 20 mm Hg in all patients.
Postoperative ICP rose to maximum 29 mm Hg in five
patients.
DC = decompressive craniectomy, EBV = estimated blood volume, FIM = Functional Independence Measure, GCS = Glasgow Coma Scale, GOS = Glasgow
Outcome Scale, GOS-E = Glasgow Outcome Scale Extended, IBL = intraoperative blood loss, ICP = intracranial pressure, INR = international normalized ratio,
KOSCHI = King’s Outcome Scale for Childhood Head Injury, NR = not reported.
New study.
a
Summary of the Evidence (n = 6); B) High ICP/MM (n = 7); and C) High ICP/DC (n
Sixteen class 3 treatment series, nine new (182–190) and seven = 10). The mean Children’s Outcome Scale for group B was
(202–208) from the Second Edition, provided evidence to sup- significantly worse than for groups A and C (p = 0.0058), and
port the recommendation (Tables 28 and 29). the three mortalities occurred in group B—patients with high
ICP who were treated with MM.
Evidence Synthesis These studies varied in criteria for DC, selection criteria
Effect of DC on Mortality and Functional Outcomes. for inclusion in the study, the DC techniques used, and their
Although mortality and functional outcomes are reported in outcome parameters. In addition, none of the investigations
the treatment series in this report, they are not used to sup- defined the study population to an extent adequate to allow
port a recommendation because 1) there are no comparators rigorous inter-study comparisons. The lack of internal com-
for mortality and outcomes in these studies and 2) compara- parison groups or matched controls weakens the analyses that
tive studies are available that address these outcomes. One can be applied and preclude making a recommendation for
class 3 RCT (176) and two class 3 retrospective studies (178, this topic.
181) reported no significant difference between DC and MM Effect of DC on ICP Control. The issue with respect to the
in mortality and/or functional outcomes. In a comparison efficacy of DC in lowering ICP is not the statistical significance
of six patients treated with DC and six with MM, Josan and of the change in ICP from prior to surgery to the postoperative
Sgouros (177) reported better survival and outcomes for the state; it is in lowering severe or medically intractable ICP eleva-
DC group, but a statistical analysis was not performed. In a tion with respect to the treatment threshold such that intracra-
retrospective pre-/post-DC comparison, Rubiano et al (180) nial hypertension is no longer encountered (optimal outcome)
found the mean GOS at 6 months postinjury to be signifi- or easily controlled following surgery.
cantly better in the DC group (t = 4.26; p = 0.0002). Cho et al Two comparative studies reported postsurgical decreases
(201) compared outcomes for three groups: A) Low ICP/MM in ICP (176, 201). Although Taylor et al (176) reported more

Kochanek et al
episodes of ICP greater than 20 and greater than 30 mm Hg in TABLE 30. Complications From
the MM group (> 20 DC-107, MM-223; > 30 DC-9, MM-29), Decompressive Craniectomy
there was no significant difference between groups in mean
ICP 48 hours after randomization or surgery. Cho et al (201) Occurrence
Complication (Total n = 164) Percent
reported mean ICP decreased in the DC group from 54.9 mm
Hg presurgery to 11.9 mm Hg postsurgery. Changes in ICP for Hygroma 32 19.5
the MM group were not reported, so a comparison between
Hydrocephalus 25 15
groups could not be made.
Eleven treatment series compared ICP before and after Ventilator-associated 22 13.4
DC (183–185, 188, 190, 202–207). Two reported a decrease in pneumonia
mean ICP after DC; in Desgranges et al (184), the decrease was Aseptic bone resorption 12 7
statistically significant and Csókay et al (183) did not provide Infection 10 6
a statistical analysis (183, 184). In the other nine studies, of 88
patients with postoperative ICP monitoring, the ICP dropped Septicemia 6 3.7
to acceptable levels in 77 patients (87.5%) (185, 188, 190, 202– Epilepsy/seizures 6 3.7
207). Although these studies did not compare ICP between DC Infection or dysfunction 4 2.4
and MM groups, they provided weak and limited evidence that of cranioplasty
DC may be effective in lowering ICP to below the threshold for
Infection or dysfunction of 3 2
treatment in patients’ refractory to MM, and are the evidence ventriculoperitoneal shunt
base to support the level III.1. recommendation. This limited
conclusion would support choosing to perform DC for ICP Bone flap sepsis 2 1.2
control when intracranial hypertension is resistant to nonsur- Slight subsidence of flap 1 0.6
gical management and observed ICP levels are considered haz- n indicates sample size.
ardous to the patient.
Outcomes of DC by Mechanism of Injury. One moderate-
quality class 3 retrospective study compared outcomes of chil- total sample of 71 patients, 36 of which were severe; all patients
dren with severe TBI, with and without abusive head trauma, in the other samples are severe (189). These data indicate a
all of whom received DC (179). Significantly, more patients high occurrence of complications among patients who receive
with abusive injury died compared with those whose TBIs DC and suggest that potential complications should be taken
were sustained by other mechanisms (p < 0.05; OR, 12.2 [p = into account when making decisions about DC as a treatment.
0.02]). Outcome measured between 1 and 94 months with the As with the subtopic of timing of DC, there are insufficient
King’s Outcome Scale for Childhood Head Injury showed no data in these reports to assess patient characteristics that might
significant difference between groups for poor outcome, and be risk factors for complications. Of note, many of the nonsur-
an OR of poor outcome of 3.04 for the abusive head trauma gical complications listed are also associated with severe TBI,
group. Given this is one relatively small study with internal itself, and relevant data from comparison group is not always
validity concerns, it was considered insufficient to support a available (Table 30).
recommendation.
Effect of Timing of the DC Procedure. One small class 3 Indications From Adult Guidelines
treatment series noted that the ICP in two of eight patients Because the findings of the literature review for pediatric TBI
rose rapidly to greater than 25 mm Hg within 15 minutes of provide levels of evidence and recommendations specific to
onset of increase, resulting in death in both cases (183). The the pediatric population, the clinical investigators do not think
report did not contain sufficient data to assess what charac- that the recommendations about DC from the adult guidelines
teristics of these patients might have been indications for very are required to guide treatment decisions in children (14).
early DC.
Complications. Of nine class 3 treatment series that system- Nutrition
atically assessed complications, two reported blood loss associ-
ated with mortality (182, 184–186, 188–190, 202, 206). In one, Recommendations
of 12 patients, one of six with intraoperative blood loss (IBL)
less than 50% of estimated blood volume (EBV) died, whereas Strength of Recommendations: Weak
three of six with IBL greater than 50% of EBV died (184). The Level I
second study (n = 25; 21 severe) found significantly greater mor- There was insufficient evidence to support a level I recommen-
tality in patients with greater than 300 mL operative blood loss
dation for this topic.
(p = 0.001) (186).
Tables 7 and 8 contain cumulative frequencies for complica- Level II
tions reported in the remaining seven studies (182, 185, 188– To Improve Overall Outcomes. II.1. Use of an immune-modu-
190, 202, 206). One of the studies reported complications for a lating diet is not recommended.

Supplement
Level III remained a concern that studies had not enrolled a subset of
To Improve Overall Outcomes. III.1. Initiation of early enteral sufficiently ill patients to demonstrate a benefit from tighter
nutritional support (within 72 hr from injury) is suggested to glucose control. The most recent report restricted enrollment
decrease mortality and improve outcomes. to hyperglycemia patients treated with either mechanical ven-
Changes From Prior Edition. The level III recommendation tilation or vasoactive medications and the primary outcome
from the Second Edition has been removed. Recommendation was the mean number of 28-day ICU-free days. The study was
III.1. is new to this Third Edition. One new class 3 retrospec- terminated early due to low probability of benefit and signifi-
tive observational study was added to the evidence base for this cantly greater rates of severe hypoglycemia (5.2% vs 2.0%) in
topic (210). the tight control group (216). Currently, there appears to be no
benefit to targeting glucose concentrations lower than a range
Introduction of 150–180 mg/d among pediatric critical care patients (216,
Similar to adults, children with severe TBI require energy to 218).
support recovery (211, 212). Although the exact mechanisms Nevertheless, the severity and duration of posttraumatic
remain unclear, TBI causes an increase in metabolism, which hyperglycemia are consistently associated with worse outcomes
thereby requires increased caloric support during the critical that likely reflect worse injury and greater stress (219, 220). No
phase of injury. In addition, developing children have greater studies of glucose control have focused solely on children with
nutritional needs for normal growth and development. The severe TBI. Currently, there are insufficient data to recommend
decision to administer nutritional support, including the tim- for or against tight glucose control for children with severe TBI
ing, the quantity, the manner, and the composition of such and persistent hyperglycemia.
support, may have effects on short- and long-term outcome. Although enteral nutrition is preferred for critically injured
Hyperglycemia is a consistent stress response to severe patients, some may receive parenteral nutrition due to other
illness or injury including severe TBI (213). Although treat- abdominal injuries or concerns regarding aspiration risk. A
ment of hyperglycemia with insulin to achieve glucose con- recent large randomized clinical trial in general critically ill
trol has been studied in critically ill and injured pediatric pediatric patients demonstrated increased rates of early infec-
patients, initial studies found conflicting results. The first- tion and prolonged length of stay among patients treated with
reported randomized trial found significantly lower infection early initiation of parenteral nutrition (within 24 hr of ICU
rates, length of stay, and mortality, but greater risk of hypo- admission) compared with those with initiation of parenteral
glycemia among pediatric patients treated with tight glucose nutrition after 1 week of critical illness. Both groups received
control compared with those treated with insulin to achieve early initiation of enteral nutrition, which was increased in
higher glucose targets (214). Subsequent studies among gen- accordance to local guidelines. The study found that early
eral pediatric critically ill and injured patients as well as car- enteral feeding was tolerated as well as IV micronutrients
diac surgical patients found no influence on mortality from (trace elements, minerals, and vitamins) starting from day 2
tight control compared with higher glucose targets, but con- and continuing until the enteral nutrition provided reached
sistently reported greater rates of severe hypoglycemia with 80% of the caloric targets (221). Currently, most pediatric
concern for harm (215–217). critical care providers attempt enteral nutrition in preference
Because mortality was lower among pediatric patients to parenteral in trauma patients unless there are severe inju-
with critical illness and injury compared with adults, there ries to the bowel.
TABLE 31. Nutrition: Quality of the Body of Evidence

Meta- Consistency Precision Quality of
Analysis No. of (High, Directness (High, Evidence (High,
No. of Studies Recommen Possible Subjects Moderate, (Direct or Moderate, Moderate, Low, or

Immune- One II.1. NA 40 NA Direct Low Moderate
modulating diet randomized
controlled
trial
New topic: timing One III.1. NA 90 NA Direct High Low
of nutritional retrospective
support
a

Kochanek et al
TABLE 32. Nutrition: Summary of Evidence

Study Design
Reference n
Geographic Location Mean (Range) Outcomes Data Class Results
Class 2 Study
Immune-modulating diet: recommendation II.1.
Briassoulis et al (222) Randomized controlled trial Class 2 IE vs regular formula
PICU University Hospital n = 40 Attrition NR; unclear Mortality
Athens, Greece Age: mean, 120 mo; range, if intention-to-treat Survival: 80% vs 95%; no significant
72–126 mo analysis conducted difference
IE: mean, 127 mo; range, NR Infection
Standard: mean, 112 mo; range, Fewer positive gastric cultures in IE
NR group (p < 0.02), but no significant
Hospital mortality; infection; LOS; difference in infections
metabolic indices LOS
16.7 vs 12.2 d; no significant difference
Length of mechanical ventilation
11 vs 8 d; no significant difference
Metabolic indices
The IE group was more likely to have
positive nitrogen balance at 5 d (69% vs
31%; p < 0.05).
Class 3 study
Timing of nutritional support: recommendation III.1.
Taha et al (210) a Retrospective Class 3 Effect of timing on neurologic status
Level 1 pediatric trauma n = 109; 90 analyzed (19 died Unclear if groups Time to initiation/time to full caloric intake
center before nutritional support) were similar at by discharge status (mean ± sd in days)
California Age: median, 13; range, 8–18; baseline; no control Home: 1.51 ± 1.23/3.38 ± 3.32
under 8 excluded for confounders; Disability: 3.08 ± 2.70/6.99 ± 4.73
Effect of time of initiation on outcome assessors
not blinded. Coma/death: 1.88 ± 0.99/7.12 ± 6.44
neurologic status at discharge Early initiation and achieving full caloric
(normal, disability, death) and intake significantly related to more
ICU LOS favorable discharge status (p < 0.05).
Effect of timing on ICU LOS
Early initiation and achieving full caloric
intake significantly related to shorter ICU
LOS (p < 0.01).
IE = immune enhancing, LOS = length of stay, NR = not reported.
New study.
a
Evaluation of the Evidence were less severely injured. Thus, the applicability of this
Quality of the Body of Evidence. Studies included for this topic study is in question.
addressed questions about the effect on overall outcomes of
an immune-modulating diet and the timing of nutritional Summary of the Evidence
support. One small class 2 RCT provided moderate-quality One class 2 study from the Second Edition (222) and one new
evidence to support recommendation II.1. (222). One class 3 class 3 study (210) provided evidence to support the recom-
observational study provided low quality of evidence to sup- mendation Table 32.
port recommendation III.1. (210) (Table 31).
Applicability. The RCT supporting the level II recom- Evidence Synthesis
mendation was small, and conducted at a single site in Immune-Modulating Diet. No new evidence has been found
Greece, limiting its applicability (222). The single-center since the Second Edition of these guidelines that suggests best
study supporting the level III recommendation did not practice in the quantity, manner, and composition of nutritional
include infants (210). Due to its retrospective design, it is not support for pediatric patients with TBI. Briassoulis et al (222), a
known whether the observed positive results were because class 2 RCT, showed no difference in outcomes for children pro-
children do better if they were fed, or were fed because they vided an immune-enhancing diet versus regular formula.

Supplement
TABLE 33. Corticosteroids: Quality of the Body of Evidence

Components of Overall Quality: Class 3 Study
Quality of
Evidence
No. of Analysis No. of (High, Directness (High, Moderate,
Studies Recommen Possible Subjects Moderate, (Direct or Moderate, Low, or
Topic Study Design dation (Yes or Noa) (n) Low) Indirect) Low) Insufficient)
Use of One III.1. NA 25 NA Direct Low Low

corticosteroids randomized
to improve controlled
outcomes trial with two
publications
a
Timing of Nutritional Support. The retrospective study and CSF production (225, 226), and diminish free radical pro-
by Taha et al (210) suggested that initiation of early (enteral) duction (225). These effects provide a rationale for potential
nutrition may reduce mortality and morbidity. Conducted in steroid benefit in neurologic diseases. This topic summarizes
children 8–18 years old, the study found that when full caloric the clinical evidence for glucocorticoid administration as a
intake was achieved, a shortened length of ICU stay with therapy to improve outcome in pediatric severe TBI.
improved outcomes at the time of discharge could be demon- Treatment of refractory hypotension (i.e., shock) with cor-
strated. However, it is not known whether the children who ticosteroids was not addressed by the included studies. Studies
were fed had better outcomes because they were fed, or if they of adult patients with critical illness–related corticosteroid
were fed because they were less severely injured. Thus, the rec- insufficiency (CIRCI) due to major trauma (defined by change
ommendation it supports is a weak level III. in baseline hydrocortisone with an adrenocorticotropic hor-
mone stimulation test) did not demonstrate a mortality benefit
Indications From Adult Guidelines with steroid therapy. However, steroid treatment was associ-
The clinical investigators do not think that the recommenda- ated with increased risk of ventilator-associated pneumonia
tions about nutrition from the adult guidelines can be used to and significant fewer ventilator-free days, and current inter-
guide treatment decisions in children. national guidelines recommend against treatment for adult
patients with major trauma and CIRCI (227). Similar trials are
Corticosteroids lacking in children. However, children with known primary
Recommendations or secondary adrenal insufficiency and major trauma should
Strength of the Recommendation: Weak receive corticosteroid replacement therapy to avoid acute adre-
nal insufficiency.
Levels I and II
Quality of the Body of Evidence. Two reports of a single small
Level III class 3 RCT were included for this topic (228, 229). They each
To Improve Overall Outcomes addressed the question of the use of corticosteroids to improve
III.1. The use of corticosteroids is not suggested to improve outcomes for children with TBI. Because of the small sample
outcome or reduce ICP. size, low precision and class 3 evidence, the overall quality of
Note: Recommendation III.1. is not intended to circumvent evidence is low (Table 33).
use of replacement corticosteroids for patients needing chronic Applicability. Both reports are over 25 years old, and the dose
steroid replacement therapy, those with adrenal suppression, of steroids is greater than the dose used for acute laryngotracheo-
and those with injury to the hypothalamic-pituitary steroid axis. bronchitis (230) and below that used for reversal of vasodilatory
Changes From Prior Edition. The Level II recommendation septic shock (231). Thus, the relevance of the evidence is limited
from the Second Edition has been downgraded to Level III. The and applicability is a concern. Questions about randomization,
note about use of replacement corticosteroids is new to this edi- blinding, sample size, and statistical analysis render the publica-
tion. No new studies were added to the evidence base for this topic. tions class 3. However, in adults, a large class 1 RCT, the Corti-
costeroid randomisation after significant head injury (CRASH)
Introduction trial, provided strong evidence against the use of steroids for
Corticosteroids can restore altered vascular permeability (223), acute care of severe TBI. This evidence confirms the findings
inhibit tumor-induced angiogenesis (224), decrease edema from the older trial, mitigating the applicability concerns (232).

Kochanek et al
TABLE 34. Corticosteroids: Summary of Evidence

Class 3 Study (Same Sample)
Study Design
n
Reference Age (yr)
Use of corticosteroids to improve outcomes: recommendation III.1.

Fanconi et al (228), Randomized controlled trial Class 3 GOS, ICP, CPP, duration of intubation
Klöti et al (229)
n = 25; 13 steroid, 12 placebo Randomization Steroid treatment resulted in no
ICU (Fanconi et al [228]) and allocation differences vs placebo in 6 mo GOS,
concealment ICP, CPP, duration of ICP monitoring, or
Zúrich, Switzerland n = 24; 12/12 (Klöti et al [229]) methods not duration of intubation
(same patients) described; unclear if
outcome assessors Free cortisol levels
Age:
were blinded; Steroid treatment vs placebo
Steroid group: mean, 7.5; range, unclear if sample significantly suppressed endogenous
1.8–14.6 size was adequate; free cortisol levels from day 1 to day 6.
Placebo group: mean, 7.4; range, did not use intent-
to-treat analysis. Complications
1.4–15.8
Steroid treatment resulted in a trend
GOS at 6 mo postinjury, ICP, CPP, toward increased bacterial pneumonia
duration of monitoring and (7/13 vs 2/12 vs placebo, respectively,
intubation; free cortisol levels; p = 0.097).
and complications
CPP = cerebral perfusion pressure, GOS = Glasgow Outcome Score, ICP = intracranial pressure.
Summary of the Evidence reducing ICP (level I) (14). In adult patients with severe TBI,
Two reports of one class 3 study provide evidence to support high-dose methylprednisolone was associated with increased
the recommendation (228, 229) (Table 34). mortality and is therefore contraindicated (232). As the
CRASH trial’s findings were similar to the small, older pediat-
Evidence Synthesis ric trial, they support the existing recommendation and lessen
Dexamethasone and ICP/CPP. Fanconi et al (228) performed a the concerns about the applicability of the older trials.
randomized, prospective, placebo-controlled clinical trial on 25
pediatric patients with severe TBI using dexamethasone at 1 mg/
ONGOING AND FUTURE RESEARCH
kg/d for 3 days (n = 13) versus placebo (n = 12). Baseline character-
istics did not differ between groups. Dexamethasone treatment did Climbing the Mountain With No Top
not influence ICP (mean of 14 mm Hg in both groups), CPP, num- As stated in the “Methods section” of this document, evidence-
ber of interventions required, duration of intubation, or 6-month based guidelines rarely (if ever) contain enough data to fully
GOS versus placebo. However, steroid treatment versus placebo populate a clinical protocol. This is certainly the case with the
significantly suppressed endogenous free cortisol levels up to day 6. treatment of severe pediatric TBI. The goal is to position the
In addition, steroid treatment resulted in a trend toward increased guidelines in a dynamic process as illustrated in Figure 1.
bacterial pneumonia versus placebo (7/13 vs 2/12, respectively, p = Available evidence is used to generate treatment guidelines.
0.097). Limitations included use of the Richmond screw to assess The guidelines provide recommendations based on the avail-
ICP, fluid restriction, and the use of hyperventilation to a Paco2 of able evidence and identify gaps that become the future research
25–30 mm Hg as part of standard care. agenda. In the interim, those gaps can be filled by creating clinical
Klöti et al (229) reported on 24 of the same 25 patients from protocols using consensus where evidence is lacking. Together the
the study described above. Additional outcomes in this report gaps and protocols provide structure and identify patient samples
included duration of ICP monitoring; steroid treatment produced for the generation of new research. The new research populates
no difference between groups. The small sample size for this trial the evidence base which can then be used to further develop the
limited the ability to make definitive conclusions regarding neuro- guidelines. It is a recursive cycle—a mountain with no top.
logic outcomes or complications. However, suppression of corti- The primary goal of this team has been to generate evidence-
sol production by steroid treatment was clearly documented. based guidelines for the treatment of pediatric TBI. Secondary
goals—as important—have been to include and train new clinical
Indications From Adult Guidelines investigators and methodologists in the technology, and to gener-
There is class 1 evidence in the adult guidelines that the use ate strong research in response to gaps identified in the guidelines
of steroids is not recommended for improving outcome or documents. For the Second Edition of these guidelines, 11 new

Supplement
TABLE 35. Current Status of Pediatric Traumatic Brain Injury Guidelines

Direct/Indirect/ Ongoing
Topic Recommendations Treatment Series Research
ICP monitoring To improve overall outcomes 19 Studies None known
Level III All class 3
1. Use of ICP monitoring is suggested. 3 Direct: all retrospective
16 Indirect
• 1 RCT
• 2 Prospective
• 10 Retrospective
• 3 Treatment series
Advanced To improve overall outcomes 4 Direct studies ADAPT hypotheses
neuromonitoring
Level III All class 3 • Pbro2 monitoring
is associated with
1. If Pbro2 monitoring is used, maintaining a level 1 Prospective favorable outcomes.
> 10 mm Hg is suggested.
3 Treatment series • A threshold Pbro2 value
is associated with
favorable outcomes.
Neuroimaging To improve overall outcomes 3 Direct studies None known
1. Excluding the possibility of elevated ICP on 2 Retrospective
the basis of a normal initial (0–6 hr after injury)
CT examination of the brain is not suggested in 1 Treatment series
comatose pediatric patients.
2. Routinely obtaining a repeat CT scan >24 hr
after the admission and initial follow-up is not
suggested for decisions about neurosurgical
intervention, unless there is either evidence of
neurologic deterioration or increasing ICP.
ICP thresholds To improve overall outcomes 12 Direct studies None known
1. Treatment of ICP targeting a threshold of < 3 Prospective
20 mm Hg is suggested.
9 Retrospective
CPP thresholds To improve overall outcomes 15 Direct studies None known
Level III 1 Class 2, 14 class 3
1. Treatment to maintain a CPP at a minimum of 1 Prospective
40 mm Hg is suggested.
4 Retrospective
2. A CPP target between 40 to 50 mm Hg is
suggested to ensure that the minimum value of 10 Treatment series
40 mm Hg is not breached. There may be age-
specific thresholds with infants at the lower end
and adolescents at or above the upper end of this
range.
(Continued)

Kochanek et al
TABLE 35. (Continued). Current Status of Pediatric Traumatic Brain Injury Guidelines
Hyperosmolar For ICP control 9 Direct studies ADAPT hypotheses

therapy
Level II 3 Class 2, 6 class 3 • Hyperosmolar therapies
improve outcomes.
1. Bolus hypertonic saline (3%) is recommended 2 RCTs
in patients with intracranial hypertension. • Hypertonic saline is
Recommended effective doses for acute use 1 Prospective more effective than
range between 2 and 5 mL/kg over 10–20 min. 5 Retrospective mannitol for ICP control.
For ICP control 1 Treatment series
Level III
1. Continuous infusion hypertonic saline is
suggested in patients with intracranial hypertension.
Suggested effective doses as a continuous
infusion of 3% saline range from between 0.1 and
1.0 mL/kg of body weight per hour, administered
on a sliding scale. The minimum dose needed to
maintain ICP < 20 mm Hg is suggested.
2. Bolus of 23.4% hypertonic saline is suggested
for refractory ICP. The suggested dose is 0.5 mL/
kg with a maximum of 30 mL.
Analgesics, For ICP control 6 Studies None known
sedatives, and
neuromuscular Level III All class 3
blockade 1. With use of multiple ICP-related therapies, as 1 Direct prospective
well as appropriate use of analgesia and sedation
in routine ICU care, avoiding bolus administration 1 Indirect prospective
of midazolam and/or fentanyl during ICP crises is 2 Direct retrospective
suggested due to risks of cerebral hypoperfusion.
2 Direct treatment series
CSF For ICP control 5 Studies ADAPT hypotheses
Level III All class 3 • Continuous CSF drainage
improves outcomes.
1. CSF drainage through an external EVD is 1 Direct retrospective
suggested to manage increased ICP. • Continuous CSF
3 Direct treatment series drainage reduces other
1 Indirect treatment series ICP therapies.
Seizure prophylaxis For seizure prevention (clinical and subclinical) 4 Studies None known
1. Prophylactic treatment is suggested to reduce 1 Direct retrospective
the occurrence of early (within 7 d) PTS.
1 Indirect prospective
1 Indirect retrospective
1 Indirect treatment series
Ventilation To improve overall outcomes 2 Studies ADAPT hypotheses
Level III Both class 3 • Prophylactic
hyperventilation
1. Prophylactic severe hyperventilation to a Paco2 1 Indirect retrospective (Co2 < 30 mm Hg)
< 30 mm Hg in the initial 48 hr after injury is not is associated with
suggested. 1 Direct treatment series
2. If hyperventilation is used in the management • Moderate hyperventilation
of refractory intracranial hypertension, advanced is associated with
neuromonitoring for evaluation of cerebral favorable outcomes.
ischemia is suggested.
(Continued)

Supplement
TABLE 35. (Continued). Current Status of Pediatric Traumatic Brain Injury Guidelines
Temperature To improve overall outcomes 10 Studies None known

control
Level II 1 Class 1, 4 class 2, 3 class
3, 2 meta-analyses
1. Prophylactic moderate (32–33°C) hypothermia
is not recommended over normothermia to 2 Direct meta-analyses
improve overall outcomes.
5 Direct RCTs
For ICP control
1 Indirect RCT
Level III
1 Direct retrospective
1. Moderate (32–33°C) hypothermia is suggested
for ICP control. 1 Indirect retrospective
Barbiturates For ICP control 4 Direct studies None known

1. High-dose barbiturate therapy is suggested in 1 Retrospective
hemodynamically stable patients with refractory
intracranial hypertension despite maximal medical 3 Treatment series
and surgical management.
DC For ICP control 23 Direct studies None known
1. DC is suggested to treat neurologic 1 RCT
deterioration, herniation, or intracranial
hypertension refractory to medical management. 1 Prospective
5 Retrospective
16 Treatment series
Nutrition To improve overall outcomes 2 Direct studies ADAPT hypotheses
Level II 1 Class 2, 1 class 3 • Increased caloric intake
is associated with
1. Use of an immune-modulating diet is not 1 RCT favorable outcomes.
recommended.
1 Retrospective • Early institution of
To improve overall outcomes feeding is associated
Level III with favorable
outcomes.
1. Initiation of early enteral nutritional support
(within 72 hr from injury) is suggested to decrease • Early glucose
mortality and improve outcomes. administration is
associated with
• A threshold value
for hyperglycemia
is associated with
favorable outcomes.
Corticosteroids To improve overall outcomes 1 Direct study None known
Level III Class 3
1. The use of corticosteroids is not suggested to 1 RCT with 2 publications
improve outcome or reduce ICP.
ADAPT = Approaches and Decisions in Acute Pediatric TBI, CPP = cerebral perfusion pressure, CSF = cerebrospinal fluid, DC = decompressive craniectomy,
EVD = external ventricular drain, ICP = intracranial pressure, Pbro2 = brain tissue oxygen, PTS = posttraumatic seizures, RCT = randomized controlled trial.

Kochanek et al
TABLE 36. Roadmap to Future Research: Pediatric Traumatic Brain Injurya

Overall Direct vs
Level of Outcomes vs. Class of Indirect Number of
Topic Recommendation Intermediate Studies Evidence Studies
ICP monitoring
Advanced neuromonitoring
Neuroimaging
ICP thresholds
Cerebral perfusion pressure thresholds
Hyperosmolar therapy
Analgesics, sedatives, neuromuscular block
Cerebrospinal fluid drainage
Seizure prophylaxis
Ventilation therapies
Temperature control
Barbiturates
Decompressive craniectomy
Nutrition
Corticosteroids
ICP = intracranial pressure.
a
Lighter cells indicate stronger body of evidence.
Full explanation of shading is provided in FUTURE DIRECTIONS section.

investigators joined the team, and multiple new studies have been develops agendas and priorities for future research. For example,
generated by our team as well as by other members of the TBI the clinical research community may benefit from reconsidera-
clinical research community. As mentioned in the “Introduction tion of the question about the use of monitors to measure ICP
section,” one of the most important studies of pediatric TBI, and a reframing of that question in the context of goal-directed
designed and executed by a guidelines clinical investigator, is therapy. Indeed, for clinical environments that are fully resourced,
concluding—the ADAPT. We look forward to strong class 2 pub- goal-directed therapy might be an appropriate context in which
lications from this project that will address 12 a priori hypoth- all topics and questions are reconsidered, and new ones generated.
eses across five guidelines topics (Table 35), as well as post hoc Furthermore, for cells populated with studies from ADAPT, the
analyses of other topics and questions. ADAPT is an important direction of future research may be defined by the findings from
example of the utilization of a strong research design (compara- those studies; however, the questions may remain unanswered).
tive effectiveness) to address questions that have eluded examina-
tion via RCTs. It is also an important demonstration that part of FUTURE DIRECTIONS
the value of a guideline is found in both the evidence-based rec- Table 36 illustrates target priorities for future pediatric TBI
ommendations and in highlighting what cannot be said due to research. The rows are the current topics. The columns are
lack of evidence; those gaps provide opportunities for innovation characteristics of a body of evidence used to quantify quality.
and direction for research capable of fully populating a clinical The cells that intersect row and column are coded in shades of
protocol with evidence-based recommendations. gray. The range of shades from dark to light indicates:
Current Status ●● Level III to II to I

Table 35 outlines the current status of research for each of the ●● Intermediate outcomes to relevant patient outcomes
topics in this edition in terms of target outcomes, level of recom- ●● Class 3 to 2 to 1
mendations, class and quantity of studies, and current research ●● Indirect evidence to direct evidence
●● Smaller number of studies to larger number of studies
in progress with a design capable of generating class 2 evidence.
●● Thus, the lighter the cell, the stronger the body of evidence
Blank cells in the “Future Research” column are obvious target
for the topic and characteristic; the darker the cell, the
priorities, with a caveat. A blank cell may indicate that there is no
greater the need for change in research approach.
longer a need for further research of that topic, or that the ques-
tions within the topic may require reconsideration and refram- The dominance of lighter cells (indicating higher quality)
ing; this is what the TBI community needs to determine as it in the columns for “Overall Outcomes vs Intermediate” and

Supplement
“Direct vs Indirect Evidence” is due in part to the criteria we severe traumatic brain injury–results from an automated data collec-
tion system time-synched to drug administration. Pediatr Crit Care
used to include studies into the evidence base. We only included Med 2016; 17:236–245
studies with intermediate outcomes when there were insuf- 3. Welch TP, Wallendorf MJ, Kharasch ED, et al: Fentanyl and midazolam
ficient data for overall outcomes; similarly, we only included are ineffective in reducing episodic intracranial hypertension in severe
indirect evidence when direct evidence was lacking or scarce. pediatric traumatic brain injury. Crit Care Med 2016; 44:809–818
Table 36 illustrates that for 11 of 15 topics, only class 3 stud- 4. Bell MJ, Adelson PD, Hutchison JS, et al; Multiple Medical Therapies
for Pediatric Traumatic Brain Injury Workgroup: Differences in medical
ies were included, and 12 topics had only level III recommen- therapy goals for children with severe traumatic brain injury-an interna-
dations. Clearly, the TBI clinical research community needs to tional study. Pediatr Crit Care Med 2013; 14:811–818
continue to find innovative methods to generate high-quality 5. Bell MJ, Adelson PD, Wisniewski SR; Investigators of the ADAPT
class 2 and 1 studies. What Table 36 does not illustrate is the Study: Challenges and opportunities for pediatric severe TBI-review
of the evidence and exploring a way forward. Childs Nerv Syst 2017;
following: 33:1663–1667
1. The integration of individual treatments in the context of 6. Kurz JE, Poloyac SM, Abend NS, et al; Investigators for the Approaches
and Decisions in Acute Pediatric TBI Trial: Variation in anticonvulsant
goal-directed therapy. Management of patients with TBI selection and electroencephalographic monitoring following severe
is not a function of the application of individual treat- traumatic brain injury in children-understanding resource availability in
ments. No treatment or management approach exists sites participating in a comparative effectiveness study. Pediatr Crit
Care Med 2016; 17:649–657
independent of other treatments and approaches, or 7. Jha RM, Kochanek PM: Adding insight to injury: A new era in neu-
independent of the ecology of the treatment setting. The rotrauma. Lancet Neurol 2017; 16:578–580
design of meaningful and effective future research must 8. Kochanek PM, Bell MJ: Tackling the challenges of clinical trials for
be consistent with this clinical reality. Thus, although severe traumatic brain injury in children: Screening, phenotyping, and
adapting. Crit Care Med 2015; 43:1544–1546
Table 36 points to areas for improvement, it creates an
9. DeKosky ST, Asken BM: Injury cascades in TBI-related neurodegen-
illusion of independence of treatments by illustrating eration. Brain Inj 2017; 31:1177–1182
treatment characteristics with independent cells. 10. Johnson VE, Stewart JE, Begbie FD, et al: Inflammation and white
2. New topics for investigation, or the integration of more matter degeneration persist for years after a single traumatic brain
than one topic as a treatment “approach.” Which topics injury. Brain 2013; 136:28–42
11. Kochanek PM, Jackson TC, Ferguson NM, et al: Emerging therapies
are complete? What new topics should be identified and
in traumatic brain injury. Semin Neurol 2015; 35:83–100
included? How can the clinical reality of multiple treat- 12. Kochanek PM, Tasker RC, Bell MJ, et al: Management of Pediatric
ments be accurately represented in research that inte- Severe Traumatic Brain Injury: 2019 Consensus and Guidelines-
grates topics in ecologically valid designs? Based Algorithm for First and Second Tier Therapies. Pediatr Crit
Care Med 2019; 20:269–279
3. Consistency in data collection across studies. Future research
13. Badjatia N, Carney N, Crocco TJ, et al: Guidelines for prehospital
should include consistency in data collection across management of traumatic brain injury 2nd edition. Prehosp Emerg
research projects, such as utilization of the Common Data Care 2008; 12(Suppl 1):S1–S52
Elements of the National Institutes of Health (233–236). 14. Carney N, Totten AM, O’Reilly C, et al: Guidelines for the manage-
ment of severe traumatic brain injury, fourth edition. Neurosurgery
The clinical investigators and methods team recommend that 2017; 80:6–15
the pediatric TBI research community systematically addresses 15. Guyatt GH, Oxman AD, Kunz R, et al; GRADE Working Group:
these questions through creating a prioritized research agenda GRADE guidelines: 8. Rating the quality of evidence–indirectness. J
Clin Epidemiol 2011; 64:1303–1310
and advocating for additional high-quality research that can 16. Morton SC, Murad MH, O’Connor E, et al: Quantitative Synthesis—
populate the evidence base for future guidelines designed to An Update. Methods Guide for Comparative Effectiveness Reviews.
improve outcomes for children who sustain TBI. (Prepared by the Scientific Resource Center under Contract No. 290-
2012-0004-C). AHRQ Publication No. 18-EHC007-EF. Rockville,
MD: Agency for Healthcare Research and Quality; 2018
ACKNOWLEDGMENTS 17. Alkhoury F, Kyriakides TC: Intracranial pressure monitoring in children
with severe traumatic brain injury: National Trauma Data Bank-based
We thank the following people at the Pacific Northwest Evi- review of outcomes. JAMA Surg 2014; 149:544–548
dence-based Practice Center at Oregon Health & Science 18. Bennett TD, Riva-Cambrin J, Keenan HT, et al: Variation in intracranial
University for their invaluable assistance in producing this pressure monitoring and outcomes in pediatric traumatic brain injury.
document: Roger Chou, MD, Elaine Graham, MLS, Andrew Arch Pediatr Adolesc Med 2012; 166:641–647
Hamilton, Hyon Hildebrant, BA, Shaun Ramirez, MPH, and 19. Bennett TD, DeWitt PE, Greene TH, et al: Functional outcome after
intracranial pressure monitoring for children with severe traumatic
Leah Williams, BS. We also thank Jamshid Ghajar, MD, PhD, brain injury. JAMA Pediatr 2017; 171:965–971
from the Brain Trauma Foundation and Stanford University. 20. Bruce DA, Alavi A, Bilaniuk L, et al: Diffuse cerebral swelling following
head injuries in children: The syndrome of “malignant brain edema.” J
Neurosurg 1981; 54:170–178
REFERENCES 21. Muizelaar JP, Marmarou A, DeSalles AA, et al: Cerebral blood flow and
1. Kochanek PM, Carney N, Adelson PD, et al: Guidelines for the acute metabolism in severely head-injured children. Part 1: Relationship with
medical management of severe traumatic brain injury in infants, children, GCS score, outcome, ICP, and PVI. J Neurosurg 1989; 71:63–71
and adolescents—second edition. [Erratum appears in Pediatr Crit Care 22. Sharples PM, Stuart AG, Matthews DS, et al: Cerebral blood flow and
Med. 2012; 13:252]. Pediatr Crit Care Med 2012; 13(Suppl 1):S1–S82 metabolism in children with severe head injury. Part 1: Relation to age,
2. Shein SL, Ferguson NM, Kochanek PM, et al: Effectiveness of phar- Glasgow Coma Score, outcome, intracranial pressure, and time after
macological therapies for intracranial hypertension in children with injury. J Neurol Neurosurg Psychiatry 1995; 58:145–152

Kochanek et al
23. Tilford JM, Aitken ME, Anand KJ, et al: Hospitalizations for critically ill 43. Cruz J, Nakayama P, Imamura JH, et al: Cerebral extraction of oxygen
children with traumatic brain injuries: A longitudinal analysis. Crit Care and intracranial hypertension in severe, acute, pediatric brain trauma:
Med 2005; 33:2074–2081 Preliminary novel management strategies. Neurosurgery 2002;
24. Tilford JM, Simpson PM, Yeh TS, et al: Variation in therapy and 50:774–779
outcome for pediatric head trauma patients. Crit Care Med 2001; 44. Downard C, Hulka F, Mullins RJ, et al: Relationship of cerebral perfu-
29:1056–1061 sion pressure and survival in pediatric brain-injured patients. J Trauma
25. Chesnut RM, Marshall LF, Klauber MR, et al: The role of secondary 2000; 49:654–658
brain injury in determining outcome from severe head injury. J Trauma 45. Grinkeviciūte DE, Kevalas R, Matukevicius A, et al: Significance of
1993; 34:216–222 intracranial pressure and cerebral perfusion pressure in severe pedi-
26. Marmarou A, Anderson R, D Ward J, et al: Impact of ICP instability atric traumatic brain injury. Medicina (Kaunas) 2008; 44:119–125
and hypotension on outcome in patients with severe head trauma. J 46. Jagannathan J, Okonkwo DO, Yeoh HK, et al: Long-term outcomes
Neurosurg 1991; 75:S59–S66 and prognostic factors in pediatric patients with severe traumatic
27. Schoon P, Benito Mori L, Orlandi G, et al: Incidence of intracranial brain injury and elevated intracranial pressure. J Neurosurg Pediatr
hypertension related to jugular bulb oxygen saturation disturbances 2008; 2:240–249
in severe traumatic brain injury patients. Acta Neurochir Suppl 2002; 47. Kasoff SS, Lansen TA, Holder D, et al: Aggressive physiologic moni-
81:285–287 toring of pediatric head trauma patients with elevated intracranial
28. Steiner LA, Andrews PJ: Monitoring the injured brain: ICP and CBF. pressure. Pediatr Neurosci 1988; 14:241–249
Br J Anaesth 2006; 97:26–38 48. Michaud LJ, Rivara FP, Grady MS, et al: Predictors of survival and
29. Chambers IR, Treadwell L, Mendelow AD: The cause and incidence severity of disability after severe brain injury in children. Neurosurgery
of secondary insults in severely head-injured adults and children. Br J 1992; 31:254–264
Neurosurg 2000; 14:424–431
49. Pfenninger J, Santi A: Severe traumatic brain injury in children–are the
30. Chambers IR, Treadwell L, Mendelow AD: Determination of thresh- results improving? Swiss Med Wkly 2002; 132:116–120
old levels of cerebral perfusion pressure and intracranial pressure
50. Wahlström MR, Olivecrona M, Koskinen LO, et al: Severe traumatic
in severe head injury by using receiver-operating characteristic
brain injury in pediatric patients: Treatment and outcome using an
curves: An observational study in 291 patients. J Neurosurg 2001;
intracranial pressure targeted therapy–the Lund concept. Intensive
94:412–416
Care Med 2005; 31:832–839
31. Miller JD, Butterworth JF, Gudeman SK, et al: Further experience in the
management of severe head injury. J Neurosurg 1981; 54:289–299 51. White JR, Farukhi Z, Bull C, et al: Predictors of outcome in severely
head-injured children. Crit Care Med 2001; 29:534–540
32. Rosner MJ, Daughton S: Cerebral perfusion pressure management in
head injury. J Trauma 1990; 30:933–940 52. Barzilay Z, Augarten A, Sagy M, et al: Variables affecting outcome
from severe brain injury in children. Intensive Care Med 1988;
33. Prabhakaran P, Reddy AT, Oakes WJ, et al: A pilot trial comparing
14:417–421
cerebral perfusion pressure-targeted therapy to intracranial pres-
sure-targeted therapy in children with severe traumatic brain injury. J 53. Esparza J, M-Portillo J, Sarabia M, et al: Outcome in children with
Neurosurg 2004; 100:454–459 severe head injuries. Childs Nerv Syst 1985; 1:109–114
34. Adelson PD, Bratton SL, Carney NA, et al; American Association for 54. Shapiro K, Marmarou A: Clinical applications of the pressure-volume
Surgery of Trauma; Child Neurology Society; International Society for index in treatment of pediatric head injuries. J Neurosurg 1982;
Pediatric Neurosurgery; International Trauma Anesthesia and Critical 56:819–825
Care Society; Society of Critical Care Medicine; World Federation 55. Figaji AA, Zwane E, Graham Fieggen A, et al: The effect of increased
of Pediatric Intensive and Critical Care Societies: Guidelines for the inspired fraction of oxygen on brain tissue oxygen tension in children
acute medical management of severe traumatic brain injury in infants, with severe traumatic brain injury. Neurocrit Care 2010; 12:430–437
children, and adolescents. Chapter 5. Indications for intracranial pres- 56. Stippler M, Ortiz V, Adelson PD, et al: Brain tissue oxygen monitoring
sure monitoring in pediatric patients with severe traumatic brain injury. after severe traumatic brain injury in children: Relationship to outcome
Pediatr Crit Care Med 2003; 4:S19–S24 and association with other clinical parameters. J Neurosurg Pediatr
35. Dean NP, Boslaugh S, Adelson PD, et al: Physician agreement with 2012; 10:383–391
evidence-based recommendations for the treatment of severe trau- 57. Friess SH, Kilbaugh TJ, Huh JW: Advanced neuromonitoring and
matic brain injury in children. J Neurosurg 2007; 107:387–391
imaging in pediatric traumatic brain injury. Crit Care Res Pract 2012;
36. Morris KP, Forsyth RJ, Parslow RC, et al; UK Paediatric Traumatic 2012:361310
Brain Injury Study Group; Paediatric Intensive Care Society Study
58. Adelson PD, Srinivas R, Chang Y, et al: Cerebrovascular response
Group: Intracranial pressure complicating severe traumatic brain
in children following severe traumatic brain injury. Childs Nerv Syst
injury in children: Monitoring and management. Intensive Care Med
2011; 27:1465–1476
2006; 32:1606–1612
37. Segal S, Gallagher AC, Shefler AG, et al: Survey of the use of intracra- 59. Ragan DK, McKinstry R, Benzinger T, et al: Depression of whole-brain
nial pressure monitoring in children in the United Kingdom. Intensive oxygen extraction fraction is associated with poor outcome in pediat-
Care Med 2001; 27:236–239 ric traumatic brain injury. Pediatr Res 2012; 71:199–204
38. Chesnut RM, Temkin N, Carney N, et al; Global Neurotrauma 60. Narotam PK, Burjonrappa SC, Raynor SC, et al: Cerebral oxygen-
Research Group: A trial of intracranial-pressure monitoring in trau- ation in major pediatric trauma: Its relevance to trauma severity and
matic brain injury. N Engl J Med 2012; 367:2471–2481 outcome. J Pediatr Surg 2006; 41:505–513
39. Keenan HT, Nocera M, Bratton SL: Frequency of intracranial pressure 61. Figaji AA, Zwane E, Thompson C, et al: Brain tissue oxygen ten-
monitoring in infants and young toddlers with traumatic brain injury. sion monitoring in pediatric severe traumatic brain injury. Part 1:
Pediatr Crit Care Med 2005; 6:537–541 Relationship with outcome. Childs Nerv Syst 2009; 25:1325–1333
40. Adelson PD, Ragheb J, Kanev P, et al: Phase II clinical trial of mod- 62. Bata SC, Yung M: Role of routine repeat head imaging in paediatric
erate hypothermia after severe traumatic brain injury in children. traumatic brain injury. ANZ J Surg 2014; 84:438–441
Neurosurgery 2005; 56:740–754 63. Bailey BM, Liesemer K, Statler KD, et al: Monitoring and prediction
41. Alberico AM, Ward JD, Choi SC, et al: Outcome after severe head of intracranial hypertension in pediatric traumatic brain injury: Clinical
injury. Relationship to mass lesions, diffuse injury, and ICP course in factors and initial head computed tomography. J Trauma Acute Care
pediatric and adult patients. J Neurosurg 1987; 67:648–656 Surg 2012; 72:263–270
42. Bruce DA, Raphaely RC, Goldberg AI, et al: Pathophysiology, treat- 64. Levi L, Guilburd JN, Linn S, et al: The association between skull frac-
ment and outcome following severe head injury in children. Childs ture, intracranial pathology and outcome in pediatric head injury. Br J
Brain 1979; 5:174–191 Neurosurg 1991; 5:617–625

Supplement
65. Stein SC, Spettell C, Young G, et al: Delayed and progressive 85. Vavilala MS, Kernic MA, Wang J, et al; Pediatric Guideline Adherence
brain injury in closed-head trauma: Radiological demonstration. and Outcomes Study: Acute care clinical indicators associated with
Neurosurgery 1993; 32:25–30 discharge outcomes in children with severe traumatic brain injury. Crit
66. Maas AI, Hukkelhoven CW, Marshall LF, et al: Prediction of outcome Care Med 2014; 42:2258–2266
in traumatic brain injury with computed tomographic characteristics: 86. van Beek AH, Claassen JA, Rikkert MG, et al: Cerebral autoregulation:
A comparison between the computed tomographic classification and An overview of current concepts and methodology with special focus
combinations of computed tomographic predictors. Neurosurgery on the elderly. J Cereb Blood Flow Metab 2008; 28:1071–1085
2005; 57:1173–1182 87. Gonda DD, Meltzer HS, Crawford JR, et al: Complications asso-
67. Liesemer K, Riva-Cambrin J, Bennett KS, et al: Use of Rotterdam CT ciated with prolonged hypertonic saline therapy in children with
scores for mortality risk stratification in children with traumatic brain elevated intracranial pressure. Pediatr Crit Care Med 2013;
injury. Pediatr Crit Care Med 2014; 15:554–562 14:610–620
68. Bratton SL, Chestnut RM, Ghajar J, et al: Guidelines for the man- 88. Roumeliotis N, Dong C, Pettersen G, et al: Hyperosmolar therapy in
agement of severe traumatic brain injury 3rd edition. J Neurotrauma pediatric traumatic brain injury: A retrospective study. Childs Nerv
2007; 24(Suppl 1):S1–S106 Syst 2016; 32:2363–2368
69. Bullock R, Chestnut R, Clifton L, et al: Guidelines for the management 89. Webster DL, Fei L, Falcone RA, et al: Higher-volume hypertonic
of severe traumatic brain injury 2nd edition. J Neurotrauma 2000; saline and increased thrombotic risk in pediatric traumatic brain
17:451–553 injury. J Crit Care 2015; 30:1267–1271
70. Adelson PD, Bratton SL, Carney NA, et al; American Association for 90. Piper BJ, Harrigan PW: Hypertonic saline in paediatric traumatic
Surgery of Trauma; Child Neurology Society; International Society for brain injury: A review of nine years’ experience with 23.4% hyper-
Pediatric Neurosurgery; International Trauma Anesthesia and Critical tonic saline as standard hyperosmolar therapy. Anaesth Intensive
Care Society; Society of Critical Care Medicine; World Federation Care 2015; 43:204–210
of Pediatric Intensive and Critical Care Societies: Guidelines for the 91. Weed LH, McKibben PS: Pressure changes in the cerebro-spinal
acute medical management of severe traumatic brain injury in infants, fluid following intravenous injection of solutions of various concentra-
children, and adolescents. Chapter 1: Introduction. Pediatr Crit Care tions. Amer J Physiol 1919; 48:512–530
Med 2003; 4:S2–S4 92. Wise BL, Chater N: Use of hypertonic mannitol solutions to lower
71. Hollingworth W, Vavilala MS, Jarvik JG, et al: The use of repeated cerebrospinal fluid pressure and decrease brain bulk in man. Surg
head computed tomography in pediatric blunt head trauma: Factors Forum 1961; 12:398–399
predicting new and worsening brain injury. Pediatr Crit Care Med 93. James HE, Langfitt TW, Kumar VS, et al: Treatment of intracranial
2007; 8:348–356; CEU quiz 357 hypertension. Analysis of 105 consecutive, continuous recordings of
72. Figg RE, Stouffer CW, Vander Kolk WE, et al: Clinical efficacy of intracranial pressure. Acta Neurochir (Wien) 1977; 36:189–200
serial computed tomographic scanning in pediatric severe traumatic 94. Pitlick WH, Pirikitakuhlr P, Painter MJ, et al: Effect of glycerol and
brain injury. Pediatr Surg Int 2006; 22:215–218 hyperosmolality on intracranial pressure. Clin Pharmacol Ther 1982;
73. Mehta A, Kochanek PM, Tyler-Kabara E, et al: Relationship of intracranial 31:466–471
pressure and cerebral perfusion pressure with outcome in young children 95. Fisher B, Thomas D, Peterson B: Hypertonic saline lowers raised
after severe traumatic brain injury. Dev Neurosci 2010; 32:413–419 intracranial pressure in children after head trauma. J Neurosurg
74. Miller Ferguson N, Shein SL, Kochanek PM, et al: Intracranial hyper- Anesthesiol 1992; 4:4–10
tension and cerebral hypoperfusion in children with severe traumatic 96. Bennett TD, Statler KD, Korgenski EK, et al: Osmolar therapy in
brain injury: Thresholds and burden in accidental and abusive insults. pediatric traumatic brain injury. Crit Care Med 2012; 40:208–215
Pediatr Crit Care Med 2016; 17:444–450 97. Muizelaar JP, Lutz HA III, Becker DP: Effect of mannitol on ICP and
75. Stein DM, Hu PF, Brenner M, et al: Brief episodes of intracranial CBF and correlation with pressure autoregulation in severely head-
hypertension and cerebral hypoperfusion are associated with poor injured patients. J Neurosurg 1984; 61:700–706
functional outcome after severe traumatic brain injury. J Trauma 2011; 98. Muizelaar JP, Wei EP, Kontos HA, et al: Mannitol causes compensa-
71:364–373 tory cerebral vasoconstriction and vasodilation in response to blood
76. McLaughlin MR, Marion DW: Cerebral blood flow and vasorespon- viscosity changes. J Neurosurg 1983; 59:822–828
sivity within and around cerebral contusions. J Neurosurg 1996; 99. Bouma GJ, Muizelaar JP: Cerebral blood flow, cerebral blood vol-
85:871–876 ume, and cerebrovascular reactivity after severe head injury. J
77. Vavilala MS, Lee LA, Lam AM: The lower limit of cerebral auto- Neurotrauma 1992; 9(Suppl 1):S333–S348
regulation in children during sevoflurane anesthesia. J Neurosurg 100. James HE: Methodology for the control of intracranial pressure with
Anesthesiol 2003; 15:307–312 hypertonic mannitol. Acta Neurochir (Wien) 1980; 51:161–172
78. Pfenninger J, Kaiser G, Lütschg J, et al: Treatment and outcome of the 101. Kaieda R, Todd MM, Cook LN, et al: Acute effects of changing
severely head injured child. Intensive Care Med 1983; 9:13–16 plasma osmolality and colloid oncotic pressure on the formation of
79. Allen BB, Chiu YL, Gerber LM, et al: Age-specific cerebral perfusion brain edema after cryogenic injury. Neurosurgery 1989; 24:671–678
pressure thresholds and survival in children and adolescents with 102. Kaufmann AM, Cardoso ER: Aggravation of vasogenic cerebral

severe traumatic brain injury. Pediatr Crit Care Med 2014; 15:62–70 edema by multiple-dose mannitol. J Neurosurg 1992; 77:584–589
80. Barlow KM, Minns RA: The relation between intracranial pressure and 103. Polderman KH, van de Kraats G, Dixon JM, et al: Increases in spi-
outcome in non-accidental head injury. Dev Med Child Neurol 1999; nal fluid osmolarity induced by mannitol. Crit Care Med 2003;
41:220–225 31:584–590
81. Chaiwat O, Sharma D, Udomphorn Y, et al: Cerebral hemodynamic 104. The Brain Trauma Foundation. The American Association of

predictors of poor 6-month Glasgow Outcome Score in severe pedi- Neurological Surgeons. The Joint Section on Neurotrauma and
atric traumatic brain injury. J Neurotrauma 2009; 26:657–663 Critical Care. Use of mannitol. J Neurotrauma 2000; 17:521–525
82. Kaiser G, Pfenninger J: Effect of neurointensive care upon outcome 105. Becker DP, Vries JK: The alleviation of increased intracranial pres-
following severe head injuries in childhood–a preliminary report. sure by the chronic administration of osmotic agents. In: Intracranial
Neuropediatrics 1984; 15:68–75 Pressure. Brock M, Dietz H (Eds). Berlin, Heidelberg, Springer,
83. Kapapa T, König K, Pfister U, et al: Head trauma in children, part 1972, pp 309–315
2: Course and discharge with outcome. J Child Neurol 2010; 106. Feig PU, McCurdy DK: The hypertonic state. N Engl J Med 1977;
25:274–283 297:1444–1454
84. Stiefel MF, Udoetuk JD, Storm PB, et al: Brain tissue oxygen monitor- 107. Worthley LI, Cooper DJ, Jones N: Treatment of resistant intracra-
ing in pediatric patients with severe traumatic brain injury. J Neurosurg nial hypertension with hypertonic saline. Report of two cases. J
2006; 105:281–286 Neurosurg 1988; 68:478–481

Kochanek et al
108. Qureshi AI, Suarez JI: Use of hypertonic saline solutions in treatment in children: Effect on biochemical markers. J Neurotrauma 2004;
of cerebral edema and intracranial hypertension. Crit Care Med 21:1113–1122
2000; 28:3301–3313 130. Levy DI, Rekate HL, Cherny WB, et al: Controlled lumbar drainage in
109.
McManus ML, Soriano SG: Rebound swelling of astroglial pediatric head injury. J Neurosurg 1995; 83:453–460
cells exposed to hypertonic mannitol. Anesthesiology 1998; 131. Pearl PL, McCarter R, McGavin CL, et al: Results of phase II leveti-
88:1586–1591 racetam trial following acute head injury in children at risk for post-
110. Nakayama S, Kramer GC, Carlsen RC, et al: Infusion of very hyper- traumatic epilepsy. Epilepsia 2013; 54:e135–e137
tonic saline to bled rats: Membrane potentials and fluid shifts. J Surg 132. Liesemer K, Bratton SL, Zebrack CM, et al: Early post-traumatic sei-
Res 1985; 38:180–186 zures in moderate to severe pediatric traumatic brain injury: Rates,
111. Arjamaa O, Karlqvist K, Kanervo A, et al: Plasma ANP during hyper- risk factors, and clinical features. J Neurotrauma 2011; 28:755–762
tonic NaCl infusion in man. Acta Physiol Scand 1992; 144:113–119 133. Chung MG, O’Brien NF: Prevalence of early posttraumatic seizures
112. Moss GS, Gould SA: Plasma expanders. An update. Am J Surg in children with moderate to severe traumatic brain injury despite
1988; 155:425–434 levetiracetam prophylaxis. Pediatr Crit Care Med 2016; 17:150–156
113. Peterson B, Khanna S, Fisher B, et al: Prolonged hypernatremia con- 134. Yablon SA: Posttraumatic seizures. Arch Phys Med Rehabil 1993;
trols elevated intracranial pressure in head-injured pediatric patients. 74:983–1001
Crit Care Med 2000; 28:1136–1143 135. Holmes GL: Effects of seizures on brain development: Lessons from
114. Lescot T, Degos V, Zouaoui A, et al: Opposed effects of hypertonic the laboratory. Pediatr Neurol 2005; 33:1–11
saline on contusions and noncontused brain tissue in patients with 136. Herman ST, Abend NS, Bleck TP, et al; Critical Care Continuous
severe traumatic brain injury. Crit Care Med 2006; 34:3029–3033 EEG Task Force of the American Clinical Neurophysiology Society:
115. Simma B, Burger R, Falk M, Sacher P, Fanconi S: A prospective, Consensus statement on continuous EEG in critically ill adults and
randomized, and controlled study of fluid management in children children, part I: Indications. J Clin Neurophysiol 2015; 32:87–95
with severe head injury: Lactated Ringer’s solution versus hypertonic 137. Lewis RJ, Yee L, Inkelis SH, et al: Clinical predictors of post-trau-
saline. Ped Crit Care 1998; 26:1265–1270 matic seizures in children with head trauma. Ann Emerg Med 1993;
116. Cottenceau V, Masson F, Mahamid E, et al: Comparison of effects 22:1114–1118
of equiosmolar doses of mannitol and hypertonic saline on cerebral 138. Klein P, Herr D, Pearl PL, et al: Results of phase 2 safety and feasibil-
blood flow and metabolism in traumatic brain injury. J Neurotrauma ity study of treatment with levetiracetam for prevention of posttrau-
2011; 28:2003–2012 matic epilepsy. Arch Neurol 2012; 69:1290–1295
117. Mangat HS, Chiu YL, Gerber LM, et al: Hypertonic saline reduces 139. Skippen P, Seear M, Poskitt K, et al: Effect of hyperventilation on
cumulative and daily intracranial pressure burdens after severe trau- regional cerebral blood flow in head-injured children. Crit Care Med
matic brain injury. J Neurosurg 2015; 122:202–210 1997; 25:1402–1409
118. de Bray JM, Granry JC, Monrigal JP, et al: Effects of thiopental on 140. Adelson PD, Clyde B, Kochanek PM, et al: Cerebrovascular response
middle cerebral artery blood velocities: A transcranial Doppler study in infants and young children following severe traumatic brain injury:
in children. Childs Nerv Syst 1993; 9:220–223 A preliminary report. Pediatr Neurosurg 1997; 26:200–207
119. Bar-Joseph G, Guilburd Y, Tamir A, et al: Effectiveness of ketamine 141. Curry R, Hollingworth W, Ellenbogen RG, et al: Incidence of hypo-
in decreasing intracranial pressure in children with intracranial hyper- and hypercarbia in severe traumatic brain injury before and after
tension. J Neurosurg Pediatr 2009; 4:40–46 2003 pediatric guidelines. Pediatr Crit Care Med 2008; 9:141–146
120. Chin KH, Bell MJ, Wisniewski SR, et al; Pediatric Traumatic Brain 142. Muizelaar JP, Marmarou A, Ward JD, et al: Adverse effects of pro-
Injury Consortium: Hypothermia Investigators: Effect of adminis- longed hyperventilation in patients with severe head injury: A ran-
tration of neuromuscular blocking agents in children with severe domized clinical trial. J Neurosurg 1991; 75:731–739
traumatic brain injury on acute complication rates and outcomes: A
143. Muizelaar JP, van der Poel HG, Li ZC, et al: Pial arteriolar vessel
secondary analysis from a randomized, controlled trial of therapeutic
diameter and CO2 reactivity during prolonged hyperventilation in the
hypothermia. Pediatr Crit Care Med 2015; 16:352–358
rabbit. J Neurosurg 1988; 69:923–927
121. Minardi C, Sahillioğlu E, Astuto M, et al: Sedation and analgesia in
144. Hutchison JS, Ward RE, Lacroix J, et al; Hypothermia Pediatric Head
pediatric intensive care. Curr Drug Targets 2012; 13:936–943
Injury Trial Investigators and the Canadian Critical Care Trials Group:
122. Martinon C, Duracher C, Blanot S, et al: Emergency tracheal intubation Hypothermia therapy after traumatic brain injury in children. N Engl J
of severely head-injured children: Changing daily practice after imple- Med 2008; 358:2447–2456
mentation of national guidelines. Pediatr Crit Care Med 2011; 12:65–70
145. Koenig MA, Bryan M, Lewin JL III, et al: Reversal of transtentorial
123. Kerr ME, Weber BB, Sereika SM, et al: Effect of endotracheal suc- herniation with hypertonic saline. Neurology 2008; 70:1023–1029
tioning on cerebral oxygenation in traumatic brain-injured patients.
146. Qureshi AI, Geocadin RG, Suarez JI, et al: Long-term outcome after
Crit Care Med 1999; 27:2776–2781
medical reversal of transtentorial herniation in patients with supraten-
124. Tume LN, Baines PB, Lisboa PJ: The effect of nursing interventions torial mass lesions. Crit Care Med 2000; 28:1556–1564
on the intracranial pressure in paediatric traumatic brain injury. Nurs 147. Adelson PD, Bratton SL, Carney NA, et al; American Association for
Crit Care 2011; 16:77–84 Surgery of Trauma; Child Neurology Society; International Society for
125. Felmet K, Nguyen T, Clark RS, et al: The FDA warning against
Pediatric Neurosurgery; International Trauma Anesthesia and Critical
prolonged sedation with propofol in children remains warranted. Care Society; Society of Critical Care Medicine; World Federation
Pediatrics 2003; 112:1002–1003 of Pediatric Intensive and Critical Care Societies: Guidelines for the
126. Bramwell KJ, Haizlip J, Pribble C, et al: The effect of etomidate acute medical management of severe traumatic brain injury in infants,
on intracranial pressure and systemic blood pressure in pediat- children, and adolescents. Chapter 14. The role of temperature con-
ric patients with severe traumatic brain injury. Pediatr Emerg Care trol following severe pediatric traumatic brain injury. Pediatr Crit
2006; 22:90–93 Care Med 2003; 4:S53–S55
127. Andrade AF, Paiva WS, Amorim RL, et al: Continuous ventricular 148. Hendrick EB: The use of hypothermia in severe head injuries in child-
cerebrospinal fluid drainage with intracranial pressure monitor- hood. Arch Surg 1959; 79:362–364
ing for management of posttraumatic diffuse brain swelling. Arq 149. Crompton EM, Lubomirova I, Cotlarciuc I, et al: Meta-analysis of ther-
Neuropsiquiatr 2011; 69:79–84 apeutic hypothermia for traumatic brain injury in adult and pediatric
128. Baldwin HZ, Rekate HL: Preliminary experience with controlled
patients. Crit Care Med 2017; 45:575–583
external lumbar drainage in diffuse pediatric head injury. Pediatr 150. Tasker RC, Vonberg FW, Ulano ED, et al: Updating evidence

Neurosurg 1991; 17:115–120 for using hypothermia in pediatric severe traumatic brain injury:
129. Shore PM, Thomas NJ, Clark RS, et al: Continuous versus intermit- Conventional and Bayesian meta-analytic perspectives. Pediatr Crit
tent cerebrospinal fluid drainage after severe traumatic brain injury Care Med 2017; 18:355–362

Supplement
151. Tasker RC, Akhondi-Asl A: Updating evidence for using therapeutic 172. Guerra SD, Carvalho LF, Affonseca CA, et al: Factors associated
hypothermia in pediatric severe traumatic brain injury. Crit Care Med with intracranial hypertension in children and teenagers who suffered
2017; 45:e1091 severe head injuries. J Pediatr (Rio J) 2010; 86:73–79
152. Crompton E, Sharma P: The authors reply. Crit Care Med 2017; 173. Zwienenberg M, Muizelaar JP: Severe pediatric head injury: The role
45:e1091–e1092 of hyperemia revisited. J Neurotrauma 1999; 16:937–943
153. Adelson PD, Wisniewski SR, Beca J, et al; Paediatric Traumatic Brain 174. Pittman T, Bucholz R, Williams D: Efficacy of barbiturates in the treat-
Injury Consortium: Comparison of hypothermia and normothermia ment of resistant intracranial hypertension in severely head-injured
after severe traumatic brain injury in children (Cool Kids): A phase 3, children. Pediatr Neurosci 1989; 15:13–17
randomised controlled trial. Lancet Neurol 2013; 12:546–553
175. Taylor A, Butt W, Rosenfeld J, et al: A randomized trial of very early
154. Beca J, McSharry B, Erickson S, et al; Pediatric Study Group of decompressive craniectomy in children with traumatic brain injury
the Australia and New Zealand Intensive Care Society Clinical Trials and sustained intracranial hypertension. Childs Nerv Syst 2001;
Group: Hypothermia for traumatic brain injury in children-a phase II 17:154–162
randomized controlled trial. Crit Care Med 2015; 43:1458–1466
176. Josan VA, Sgouros S: Early decompressive craniectomy may

155. Empey PE, Velez de Mendizabal N, Bell MJ, et al; Pediatric TBI
be effective in the treatment of refractory intracranial hyperten-
Consortium: Hypothermia Investigators: Therapeutic hypothermia
sion after traumatic brain injury. Childs Nerv Syst 2006; 22:
decreases phenytoin elimination in children with traumatic brain
1268–1274
injury. Crit Care Med 2013; 41:2379–2387
177. Mhanna MJ, Mallah WE, Verrees M, et al: Outcome of children with
156. Hutchison JS, Frndova H, Lo TY, et al; Hypothermia Pediatric Head
Injury Trial Investigators; Canadian Critical Care Trials Group: Impact severe traumatic brain injury who are treated with decompressive
of hypotension and low cerebral perfusion pressure on outcomes in craniectomy. J Neurosurg Pediatr 2015; 16:1–7
children treated with hypothermia therapy following severe traumatic 178. Oluigbo CO, Wilkinson CC, Stence NV, et al: Comparison of out-
brain injury: A post hoc analysis of the Hypothermia Pediatric Head comes following decompressive craniectomy in children with acci-
Injury Trial. Dev Neurosci 2010; 32:406–412 dental and nonaccidental blunt cranial trauma. J Neurosurg Pediatr
157. Bourdages M, Bigras JL, Farrell CA, et al; Canadian Critical Care 2012; 9:125–132
Trials Group: Cardiac arrhythmias associated with severe traumatic 179. Rubiano AM, Villarreal W, Hakim EJ, et al: Early decompressive cra-
brain injury and hypothermia therapy. Pediatr Crit Care Med 2010; niectomy for neurotrauma: An institutional experience. Ulus Travma
11:408–414 Acil Cerrahi Derg 2009; 15:28–38
158. Su E, Bell MJ, Kochanek PM, et al: Increased CSF concentrations 180. Thomale UW, Graetz D, Vajkoczy P, et al: Severe traumatic brain
of myelin basic protein after TBI in infants and children: Absence of injury in children–a single center experience regarding therapy and
significant effect of therapeutic hypothermia. Neurocrit Care 2012; long-term outcome. Childs Nerv Syst 2010; 26:1563–1573
17:401–407 181. Adamo MA, Drazin D, Waldman JB: Decompressive craniectomy
159. Natale JE, Joseph JG, Helfaer MA, et al: Early hyperthermia after and postoperative complication management in infants and tod-
traumatic brain injury in children: Risk factors, influence on length dlers with severe traumatic brain injuries. J Neurosurg Pediatr 2009;
of stay, and effect on short-term neurologic status. Crit Care Med 3:334–339
2000; 28:2608–2615
182. Csókay A, Emelifeonwu JA, Fügedi L, et al: The importance of very
160. Sakurai A, Atkins CM, Alonso OF, et al: Mild hyperthermia worsens early decompressive craniectomy as a prevention to avoid the sud-
the neuropathological damage associated with mild traumatic brain den increase of intracranial pressure in children with severe trau-
injury in rats. J Neurotrauma 2012; 29:313–321 matic brain swelling (retrospective case series). Childs Nerv Syst
161. Dietrich WD, Bramlett HM: The evidence for hypothermia as a neuro- 2012; 28:441–444
protectant in traumatic brain injury. Neurotherapeutics 2010; 7:43–50
183. Desgranges FP, Javouhey E, Mottolese C, et al: Intraoperative blood
162. Polderman KH: Application of therapeutic hypothermia in the ICU: loss during decompressive craniectomy for intractable intracranial
Opportunities and pitfalls of a promising treatment modality. Part 1: hypertension after severe traumatic brain injury in children. Childs
Indications and evidence. Intensive Care Med 2004; 30:556–575 Nerv Syst 2014; 30:1393–1398
163. Salonia R, Empey PE, Poloyac SM, et al: Endothelin-1 is increased 184. Figaji AA, Fieggen AG, Argent A, et al: Surgical treatment for “brain
in cerebrospinal fluid and associated with unfavorable outcomes in compartment syndrome” in children with severe head injury. S Afr
children after severe traumatic brain injury. J Neurotrauma 2010;
Med J 2006; 96:969–975
27:1819–1825
185. Khan SA, Shallwani H, Shamim MS, et al: Predictors of poor out-
164. Biswas AK, Bruce DA, Sklar FH, et al: Treatment of acute traumatic
come of decompressive craniectomy in pediatric patients with
brain injury in children with moderate hypothermia improves intracra-
nial hypertension. Crit Care Med 2002; 30:2742–2751 severe traumatic brain injury: A retrospective single center study
from Pakistan. Childs Nerv Syst 2014; 30:277–281
165. Li H, Lu G, Shi W, et al: Protective effect of moderate hypothermia
on severe traumatic brain injury in children. J Neurotrauma 2009; 186. Messing-Jünger AM, Marzog J, Wöbker G, et al: Decompressive
26:1905–1909 craniectomy in severe brain injury. Zentralbl Neurochir 2003;
64:171–177
166. Mellion SA, Bennett KS, Ellsworth GL, et al: High-dose barbiturates
for refractory intracranial hypertension in children with severe trau- 187. Pechmann A, Anastasopoulos C, Korinthenberg R, et al:
matic brain injury. Pediatr Crit Care Med 2013; 14:239–247 Decompressive craniectomy after severe traumatic brain injury in chil-
167. Demopoulos HB, Flamm ES, Pietronigro DD, et al: The free radical dren: Complications and outcome. Neuropediatrics 2015; 46:5–12
pathology and the microcirculation in the major central nervous sys- 188. Prasad GL, Gupta DK, Mahapatra AK, Sharma BS: Surgical results
tem disorders. Acta Physiol Scand Suppl 1980; 492:91–119 of decompressive craniectomy in very young children: A level one
168. Kassell NF, Hitchon PW, Gerk MK, et al: Alterations in cerebral blood trauma centre experience from India. Brain Inj 2015; 29:1–8
flow, oxygen metabolism, and electrical activity produced by high 189. Perez Suarez E, Serrano Gonzalez A, Perez Diaz C, et al:

dose sodium thiopental. Neurosurgery 1980; 7:598–603 Decompressive craniectomy in 14 children with severe head injury:
169. Piatt JH Jr, Schiff SJ: High dose barbiturate therapy in neurosurgery Clinical results with long-term follow-up and review of the literature.
and intensive care. Neurosurgery 1984; 15:427–444 J Trauma 2011; 71:133–140
170. Chen HI, Malhotra NR, Oddo M, et al: Barbiturate infusion for intrac- 190. Honeybul S, Ho KM, Gillett GR: Long-term outcome following

table intracranial hypertension and its effect on brain oxygenation. decompressive craniectomy: An inconvenient truth? Curr Opin Crit
Neurosurgery 2008; 63:880–886 Care 2018; 24:97–104
171. Aldrich EF, Eisenberg HM, Saydjari C, et al: Diffuse brain swelling 191. Yue JK, Rick JW, Deng H, et al: Efficacy of decompressive craniec-
in severely head-injured children. A report from the NIH Traumatic tomy in the management of intracranial pressure in severe traumatic
Coma Data Bank. J Neurosurg 1992; 76:450–454 brain injury. J Neurosurg Sci 2017 Nov 7. [Epub ahead of print]

Kochanek et al
192. Jehan F, Azim A, Rhee P, et al: Decompressive craniectomy versus 215. Agus MS, Wypij D, Hirshberg EL, et al; HALF-PINT Study

craniotomy only for intracranial hemorrhage evacuation: A propensity Investigators and the PALISI Network: Tight glycemic control in criti-
matched study. J Trauma Acute Care Surg 2017; 83:1148–1153 cally ill children. N Engl J Med 2017; 376:729–741
193. Kolias AG, Adams H, Timofeev I, et al: Decompressive craniectomy 216. Macrae D, Grieve R, Allen E, et al; CHiP Investigators: A randomized
following traumatic brain injury: Developing the evidence base. Br J trial of hyperglycemic control in pediatric intensive care. N Engl J
Neurosurg 2016; 30:246–250 Med 2014; 370:107–118
194. Tapper J, Skrifvars MB, Kivisaari R, et al: Primary decompressive craniec- 217. Jacobi J, Bircher N, Krinsley J, et al: Guidelines for the use of an
tomy is associated with worse neurological outcome in patients with trau- insulin infusion for the management of hyperglycemia in critically ill
matic brain injury requiring acute surgery. Surg Neurol Int 2017; 8:141 patients. Crit Care Med 2012; 40:3251–3276
195. Kurzbuch AR: Does size matter? Decompressive surgery under
218. Chiaretti A, Piastra M, Pulitanò S, et al: Prognostic factors and out-
review. Neurosurg Rev 2015; 38:629–640 come of children with severe head injury: An 8-year experience.
196. Barthélemy EJ, Melis M, Gordon E, et al: Decompressive craniec- Childs Nerv Syst 2002; 18:129–136
tomy for severe traumatic brain injury: A systematic review. World 219. Elkon B, Cambrin JR, Hirshberg E, et al: Hyperglycemia: An inde-
Neurosurg 2016; 88:411–420
pendent risk factor for poor outcome in children with traumatic brain
197. Nagai M, Ishikawa M: Exploration of the most effective dural incision injury. Pediatr Crit Care Med 2014; 15:623–631
design in a decompressive craniectomy. World Neurosurg 2017;
220. Fivez T, Kerklaan D, Mesotten D, et al: Early versus late parenteral
100:224–229
nutrition in critically ill children. N Engl J Med 2016; 374:1111–1122
198. Cooper DJ, Rosenfeld JV, Murray L, et al; DECRA Trial Investigators;
Australian and New Zealand Intensive Care Society Clinical Trials 221. Briassoulis G, Filippou O, Kanariou M, et al: Temporal nutritional
Group: Decompressive craniectomy in diffuse traumatic brain injury. and inflammatory changes in children with severe head injury fed a
N Engl J Med 2011; 364:1493–1502 regular or an immune-enhancing diet: A randomized, controlled trial.
Pediatr Crit Care Med 2006; 7:56–62
199. Hutchinson PJ, Kolias AG, Timofeev IS, et al; RESCUEicp Trial

Collaborators: Trial of decompressive craniectomy for traumatic 222. Maxwell RE, Long DM, French LA: The effects of glucosteroids on exper-
intracranial hypertension. N Engl J Med 2016; 375:1119–1130 imental cold-induced brain edema. Gross morphological alterations and
200. Cho DY, Wang YC, Chi CS: Decompressive craniotomy for
vascular permeability changes. J Neurosurg 1971; 34:477–487
acute shaken/impact baby syndrome. Pediatr Neurosurg 1995; 223. Kasselman LJ, Kintner J, Sideris A, et al: Dexamethasone treatment
23:192–198 and ICAM-1 deficiency impair VEGF-induced angiogenesis in adult
201. Figaji AA, Fieggen AG, Peter JC: Early decompressive craniotomy in brain. J Vasc Res 2007; 44:283–291
children with severe traumatic brain injury. Childs Nerv Syst 2003; 224. Pappius HM, McCann WP: Effects of steroids on cerebral edema in
19:666–673 cats. Arch Neurol 1969; 20:207–216
202. Hejazi N, Witzmann A, Fae P: Unilateral decompressive craniec- 225. Weiss MH, Nulsen FE: The effect of glucocorticoids on CSF flow in
tomy for children with severe brain injury. Report of seven cases and dogs. J Neurosurg 1970; 32:452–458
review of the relevant literature. Eur J Pediatr 2002; 161:99–104 226. Annane D, Pastores SM, Rochwerg B, et al: Guidelines for the diag-
203. Jagannathan J, Okonkwo DO, Dumont AS, et al: Outcome following nosis and management of critical illness-related corticosteroid insuf-
decompressive craniectomy in children with severe traumatic brain ficiency (CIRCI) in critically ill patients (part I): Society of Critical
injury: A 10-year single-center experience with long-term follow up. J Care Medicine (SCCM) and European Society of Intensive Care
Neurosurg 2007; 106:268–275 Medicine (ESICM) 2017. Intensive Care Med 2017; 43:1751–1763
204. Kan P, Amini A, Hansen K, et al: Outcomes after decompressive cra- 227. Fanconi S, Klöti J, Meuli M, et al: Dexamethasone therapy and endog-
niectomy for severe traumatic brain injury in children. J Neurosurg enous cortisol production in severe pediatric head injury. Intensive
2006; 105:337–342 Care Med 1988; 14:163–166
205. Ruf B, Heckmann M, Schroth I, et al: Early decompressive craniec- 228. Klöti J, Fanconi S, Zachmann M, et al: Dexamethasone therapy and
tomy and duraplasty for refractory intracranial hypertension in chil- cortisol excretion in severe pediatric head injury. Childs Nerv Syst
dren: Results of a pilot study. Crit Care 2003; 7:R133–R138
1987; 3:103–105
206. Rutigliano D, Egnor MR, Priebe CJ, et al: Decompressive craniec-
229. Garbutt JM, Conlon B, Sterkel R, et al: The comparative effectiveness of
tomy in pediatric patients with traumatic brain injury with intractable
prednisolone and dexamethasone for children with croup: A community-
elevated intracranial pressure. J Pediatr Surg 2006; 41:83–87
based randomized trial. Clin Pediatr (Phila) 2013; 52:1014–1021
207. Skoglund TS, Eriksson-Ritzén C, Jensen C, et al: Aspects on decom-
pressive craniectomy in patients with traumatic head injuries. J 230. Gibbison B, López-López JA, Higgins JP, et al: Corticosteroids in
Neurotrauma 2006; 23:1502–1509 septic shock: A systematic review and network meta-analysis. Crit
Care 2017; 21:78
208. Prasad GL, Gupta DK, Mahapatra AK, et al: Surgical results of grow-
ing skull fractures in children: A single centre study of 43 cases. 231. Edwards P, Arango M, Balica L, et al; CRASH trial collaborators:
Childs Nerv Syst 2015; 31:269–277 Final results of MRC CRASH, a randomised placebo-controlled trial
209. Taha AA, Badr L, Westlake C, et al: Effect of early nutritional support of intravenous corticosteroid in adults with head injury-outcomes at
on intensive care unit length of stay and neurological status at dis- 6 months. Lancet 2005; 365:1957–1959
charge in children with severe traumatic brain injury. J Neurosci Nurs 232. Adelson PD, Pineda J, Bell MJ, et al; Pediatric TBI Demographics
2011; 43:291–297 and Clinical Assessment Working Group: Common data elements
210. Malakouti A, Sookplung P, Siriussawakul A, et al: Nutrition support for pediatric traumatic brain injury: Recommendations from the work-
and deficiencies in children with severe traumatic brain injury. Pediatr ing group on demographics and clinical assessment. J Neurotrauma
Crit Care Med 2012; 13:e18–e24 2012; 29:639–653
211. Mtaweh H, Smith R, Kochanek PM, et al: Energy expenditure in chil- 233. Berger RP, Beers SR, Papa L, et al; Pediatric TBI CDE Biospecimens
dren after severe traumatic brain injury. Pediatr Crit Care Med 2014; and Biomarkers Workgroup: Common data elements for pediatric
15:242–249 traumatic brain injury: Recommendations from the biospecimens and
212. Srinivasan V: Stress hyperglycemia in pediatric critical illness: The inten- biomarkers workgroup. J Neurotrauma 2012; 29:672–677
sive care unit adds to the stress! J Diabetes Sci Technol 2012; 6:37–47 234. Duhaime AC, Holshouser B, Hunter JV, et al: Common data elements
213. Vlasselaers D, Milants I, Desmet L, et al: Intensive insulin therapy for neuroimaging of traumatic brain injury: Pediatric considerations. J
for patients in paediatric intensive care: A prospective, randomised Neurotrauma 2012; 29:629–633
controlled study. Lancet 2009; 373:547–556 235. McCauley SR, Wilde EA, Anderson VA, et al; Pediatric TBI Outcomes
214. Agus MS, Steil GM, Wypij D, et al; SPECS Study Investigators: Workgroup: Recommendations for the use of common outcome
Tight glycemic control versus standard care after pediatric cardiac measures in pediatric traumatic brain injury research. J Neurotrauma
surgery. N Engl J Med 2012; 367:1208–1219 2012; 29:678–705


Manejo de Trauma Pediatrico

Uploaded by

Copyright:

Available Formats

Manejo de Trauma Pediatrico

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Manejo de Trauma Pediatrico

Uploaded by

Copyright:

Available Formats

Guidelines for the Management of Pediatric Severe

Traumatic Brain Injury, Third Edition: Update of the

Pediatric Critical Care Medicine www.pccmjournal.org S1

S2 www.pccmjournal.org March 2019 • Volume 20 (Suppl) • Number 3

Pediatric Critical Care Medicine www.pccmjournal.org S3

S4 www.pccmjournal.org March 2019 • Volume 20 (Suppl) • Number 3

Pediatric Critical Care Medicine www.pccmjournal.org S5

S6 www.pccmjournal.org March 2019 • Volume 20 (Suppl) • Number 3

Pediatric Critical Care Medicine www.pccmjournal.org S7

TABLE 1. Intracranial Pressure Monitoring: Quality of the Body of Evidence

Use of ICP 2 retrospective III.1. No 6,191 Low Direct Moderate Low

Probable overlap of patients across Bennett et al (18, 19).

n indicates sample size.

S8 www.pccmjournal.org March 2019 • Volume 20 (Suppl) • Number 3

TABLE 2. Intracranial Pressure Monitoring: Summary of Evidence

Use of information from ICP monitors to inform treatment: recommendation III.1.

Pediatric Critical Care Medicine www.pccmjournal.org S9

TABLE 2. (Continued). Intracranial Pressure Monitoring: Summary of Evidence

S10 www.pccmjournal.org March 2019 • Volume 20 (Suppl) • Number 3

TABLE 2. (Continued). Intracranial Pressure Monitoring: Summary of Evidence

Cruz et al (43) Retrospective Class 3 Mortality

Pediatric Critical Care Medicine www.pccmjournal.org S11

TABLE 2. (Continued). Intracranial Pressure Monitoring: Summary of Evidence

White et al Retrospective Class 3 Survival

TABLE 2. (Continued). Intracranial Pressure Monitoring: Summary of Evidence

Kasoff et al Retrospective Class 3 Mortality

Pediatric Critical Care Medicine www.pccmjournal.org S13

TABLE 2. (Continued). Intracranial Pressure Monitoring: Summary of Evidence

Shapiro and Treatment series Class 3 Mortality

n indicates sample size.

S14 www.pccmjournal.org March 2019 • Volume 20 (Suppl) • Number 3

Pediatric Critical Care Medicine www.pccmjournal.org S15

TABLE 3. Advanced Neuromonitoring: Quality of the Body of Evidence

n indicates sample size.

TABLE 4. Advanced Neuromonitoring: Summary of Evidence

Reference Study Design

Pbro2 monitoring thresholds: recommendation III.1.

TABLE 4. (Continued). Advanced Neuromonitoring: Summary of Evidence

Reference Study Design

Figaji et al (61) Prospective Class 3 Pbro2 < 5 mm Hg for >1 hr or

specificity for favorable outcomes. However, the sensitivity was Neuroimaging

Pediatric Critical Care Medicine www.pccmjournal.org S17

TABLE 5. Neuroimaging: Quality of the Body of Evidence

Use of initial CT 1 treatment series III.1. NA 9 NA Direct NA Low

n indicates sample size.

S18 www.pccmjournal.org March 2019 • Volume 20 (Suppl) • Number 3

TABLE 6. Neuroimaging: Summary of Evidence

Reference Study Design n

Use of initial CT to rule out intracranial hypertension: recommendation III.1.

n indicates sample size.

Pediatric Critical Care Medicine www.pccmjournal.org S19

severe TBI treated over a decade. They evaluated whether Introduction

S20 www.pccmjournal.org March 2019 • Volume 20 (Suppl) • Number 3

TABLE 7. Threshold for Treatment of Intracranial Hypertension: Quality of the Body of

Target intracranial 3 prospective III.1. No 649b Moderate Direct Low Low

n indicates sample size.

TABLE 8. Threshold for Treatment of Intracranial Hypertension: Summary of Evidence

Target ICP threshold to improve outcomes: recommendation III.1.

Cruz et al (43) Retrospective Class 3 Mortality

White et al Retrospective Class 3 Survival

Kasoff et al Retrospective Class 3 Mortality

Shapiro and Treatment series Class 3 Mortality

Figaji et al (61) Prospective Class 3 Pbro2 < 5 mm Hg for >1 hr or

Narotam et al Treatment series Class 3 Mortality

Barzilay et al Treatment series Class 3 Survival (TBI group)

Continuous infusion hypertonic saline to control ICP: recommendation III.1.

Bolus hypertonic saline vs mannitol to control refractory ICP: no recommendation