american academy of pediatrics-section on neurological ... · guidelines for the acute medical...

Guidelines for the Acute Medical Managementof Severe Traumatic Brain Injury in Infants, Children, and Adolescents-Second Edition

Author Affiliations

Patrick M. Kochanek, MD, FCCM, Professor and Vice Chair, Department of Critical Care Medicine, University of Pittsburgh School of Medicine

Nancy Carney, PhD, Associate Professor, Department of Medical Informatics and Clinical Epidemiology, Oregon Health & Science University

P. David Adelson, MD, FACS, FAAP, Director, Barrow Neurological Institute at Phoenix Children’s Hospital, Chief, Pediatric Neurosur-gery/ Children’s Neurosciences

Stephen Ashwal, MD, Distinguished Professor of Pediatrics and Neurology, Chief of the Division of Child Neurology, Department ofPediatrics, Loma Linda University School of Medicine

Michael J. Bell, MD, Associate Professor of Critical Care Medicine, University of Pittsburgh School of Medicine

Susan Bratton, MD, MPH, FAAP, Professor of Pediatric Critical Care Medicine, University of Utah School of Medicine

Susan Carson, MPH, Senior Research Associate, Department of Medical Informatics and Clinical Epidemiology, Oregon Health & ScienceUniversity

Randall M. Chesnut, MD, FCCM, FACS, Professor of Neurological Surgery, Orthopedics and Sports Medicine, University of WashingtonSchool of Medicine

Jamshid Ghajar, MD, PhD, FACS, Clinical Professor of Neurological Surgery, Weill Cornell Medical College, President of the Brain TraumaFoundation

Brahm Goldstein, MD, FAAP, FCCM, Senior Medical Director, Clinical Research, Ikaria, Inc., Professor of Pediatrics, University ofMedicine and Dentistry of New Jersey, Robert Wood Johnson Medical School

Gerald A. Grant, MD, Associate Professor of Surgery and Pediatrics, Duke University School of Medicine

Niranjan Kissoon, MD, FAAP, FCCM, Professor of Paediatrics and Emergency Medicine, British Columbia’s Children’s Hospital, Universityof British Columbia

Kimberly Peterson, BSc, Research Associate, Department of Medical Informatics and Clinical Epidemiology, Oregon Health & ScienceUniversity

Nathan R. Selden, MD, PhD, FACS, FAAP, Campagna Professor and Vice Chair of Neurological Surgery, Oregon Health & Science University

Robert C. Tasker, MBBS, MD, FRCP, Chair and Director, Neurocritical Care, Children’s Hospital Boston, Professor of Neurology andAnesthesia, Harvard Medical School

Karen A. Tong, MD, Associate Professor of Radiology, Loma Linda University

Monica S. Vavilala, MD, Professor of Anesthesiology and Pediatrics, University of Washington School of Medicine

Mark S. Wainwright, MD, PhD, Director, Pediatric Neurocritical Care, Associate Professor of Pediatrics, Northwestern University FeinbergSchool of Medicine

Craig R. Warden, MD, MPH, MS, Professor of Emergency Medicine and Pediatrics, Chief, Pediatric Emergency Services, Oregon Health& Science University/Doernbecher Children’s Hospital

Project Management

Cynthia Davis-O’Reilly, BSc, Research Associate, Department of Medical Informatics and Clinical Epidemiology, Oregon Health & ScienceUniversity

Amy Huddleston, MPA, Research Associate, Department of Medical Informatics and Clinical Epidemiology, Oregon Health & Science University

Marci Provins, BS, Editorial Assistant, Pediatric Critical Care Medicine, Pittsburgh

Guest Editor

Hector R. Wong, MD

External Peer Reviewers

Drs. Mark S. Dias, Richard G. Ellenbogen, Stuart H. Friess, Jeffrey P. Greenfield, Anne-Marie Guerguerian, Mary E. Hartman, Mark A. Helfaer,John W. Kuluz, Yi-Chen Lai, Leon E. Moores, Jose A. Pineda, Paul M. Shore, Kimberly D. Statler-Bennett, and Michael J. Whalen

S1Pediatr Crit Care Med 2012 Vol. 13, No. 1 (Suppl.)

Endorsements

The following endorse these Guidelines:

American Academy of Pediatrics-Section on Neurological SurgeryAmerican Association of Neurological Surgeons/Congress of Neurological Surgeons

Child Neurology SocietyEuropean Society of Pediatric and Neonatal Intensive Care

Neurocritical Care SocietyPediatric Neurocritical Care Research Group

Society of Critical Care MedicineThe Paediatric Intensive Care Society (UK)

Society for Neuroscience in Anesthesiology and Critical CareWorld Federation of Pediatric Intensive and Critical Care Societies

The following supported these Guidelines:

We gratefully acknowledge funding from the Brain Trauma Foundation andpartial funding from the Charles Maddock Foundation.

Author Disclosures:

Drs. Adelson, Kochanek and Vavilala have received funding from the National Institutes of Health. Dr. Chesnut has consulted for Integraand Innerspace in the past. Drs. Grant and Kochanek have received funding from the Department of Defense. The remaining authors havenot disclosed any potential conflicts of interest.

For information regarding this article, email [email protected] 2012 Brain Trauma Foundation

Disclaimer of Liability

The information contained in this Guidelines document reflects the current state of knowledge at the time of its completion, December5, 2011. In view of the fact that there will be future developments in both scientific information and technology, it is anticipated that theseGuidelines will be periodically reviewed and updated. These Guidelines are published and distributed with the understanding that theBrain Trauma Foundation and the other organizations that have collaborated and supported their development are not engaged inrendering professional medical services. If medical advice or assistance is required, the services of a competent physician should be sought.The recommendations contained in these Guidelines may not be appropriate for use in all circumstances. The decision to adopt anyparticular recommendation contained in these Guidelines must be made by a treating physician with knowledge of all of the facts andcircumstances in each particular case and on the basis of the available resources.

S2 Pediatr Crit Care Med 2012 Vol. 13, No. 1 (Suppl.)

Chapter 1. Introduction

I. RECOMMENDATIONS

This is the second edition of the Guide-lines for the Acute Medical Manage-ment of Severe Traumatic Brain Injuryin Infants, Children, and Adolescents.The first edition was published in 2003,�8 yrs ago (1). Writing the initialguidelines was an exciting but hum-bling experience, because it quickly be-came apparent that, based on the avail-able literature, it would be difficult tomake recommendations above level IIIfor most categories. Despite this chal-lenge, the guidelines committee main-tained its commitment to produce anevidence-based document and did notcomingle consensus when crafting therecommendations. It was clear that oneof the major contributions of the doc-ument would be to identify key gaps inthe literature as targets for future re-search.

For the second edition we were opti-mistic that sufficient new studies aboutpediatric traumatic brain injury (TBI)had been generated since 2003 to supporta document with higher level evidenceand stronger recommendations than thefirst edition. Without question, severalvaluable new reports in pediatric TBIhave been published since 2003, includ-ing randomized controlled trials of hypo-thermia, additional reports investigatingand/or describing optimal cerebral perfu-sion pressure in children, brain tissueoxygen monitoring, nutrition, cerebro-spinal fluid drainage, and the impact ofhypocarbia, among others (2–12).

After rigorous application of the crite-ria for including studies that were pre-specified by the guidelines committee, wefound 27 new publications for the secondedition. However, 25 publications thatwere included in the 2003 documentfailed to meet the more rigorous criteriain this second edition (Appendix A). Keyreasons for excluding publications were1) no clear specification of admissionGlasgow Coma Scale score; 2) inclusionof patients with pathologies other than

severe TBI; 3) inclusion of adult patientswithout analysis of data by age; and 4)failure to include a relevant health out-come such as mortality or function oreven an important surrogate outcomesuch as intracranial pressure. For exam-ple, a recent study by Bar-Joseph et al(13) on the use of ketamine as a sedativein pediatric brain injury could not beincluded as evidence because the admis-sion Glasgow Coma Scale was not speci-fied, and the sample included childrenwith pathologies other than severe TBI.

It is important to distinguish betweeninclusion criteria and quality criteria.Publications were not excluded based ontheir quality. The purposes of the inclu-sion criteria were to 1) clearly define thetarget patient population; 2) identify theindependent variables (treatments) anddependent variables (outcomes); 3) iden-tify the scope of the treatment phases;and 4) use sample sizes and study designscapable of providing a baseline level ofdata (see “Methods” section). All publica-tions meeting these criteria, regardless oftheir quality, were included in the finallibrary and constitute the body of evi-dence. If a publication did not meet thesecriteria, regardless of its quality, it wasexcluded.

After identification as “included,” eachstudy was then assessed for its qualitybased on the quality criteria provided indetail in the “Methods” section. The pur-pose of the quality criteria is to determinethe potential for bias and uncontrolledconfounding based on 1) study design;and 2) flaws in the conduct of the studies.Regardless of quality (class I, II, or III), allincluded studies were used as evidence.However, the level and strength of therecommendations were derived from thequality of the overall body of evidenceused to address each topic.

We rated the quality of randomizedcontrolled trials using predefined criteriadesigned to assess study design factorsthat are widely accepted as important in-dicators of internal validity: use of ade-quate randomization, allocation conceal-ment, and blinding methods; similarity ofcompared groups at baseline; mainte-nance of comparable groups; use of an

intention-to-treat analysis; overall fol-low-up rate of �85%; and no differentialloss to follow-up. We used separate pre-defined criteria to rate the quality of co-hort and case–control studies designed toreflect the most important aspects ofthose study designs: nonbiased patientselection methods, identification andascertainment of events, adequate sam-ple size, follow-up rate of at least 85%,and use of adequate statistical methodsto control for potentially confoundingvariables.

One of the major problems in craftingguidelines in many fields, and in partic-ular in pediatric TBI, is the lack of Ut-stein-stylea data collection for key param-eters in the published studies. Thisresulted in the inability to include other-wise valuable studies as evidence in thisdocument. Lack of Utstein-style data col-lection also created other difficulties. Forexample, data on intracranial pressurewere collected and/or reported by inves-tigators in a number of manners such aspeak value, mean value, or number ofvalues greater than a given threshold.This lack of a systematic approach to datacollection and reporting created impor-tant problems in a number of chapters forour committee to generate cogent rec-ommendations. Until we have an Utstein-style template for pediatric TBI that iswidely accepted and used to conduct re-search, we strongly encourage the TBIcommunity to consider use of the inclu-sion and quality criteria specified in theseguidelines when designing studies.

There are several new additions and/ormodifications to the second edition: 1) Thelevels of recommendation were changedfrom “standard, guideline, and option” to“level I, level II, and level III,” respectively;2) new chapters include Advanced Neu-romonitoring and Neuroimaging with thefocus of these additions on management

Copyright © 2012 Brain Trauma Foundation

DOI: 10.1097/PCC.0b013e31823f437e

aThe Utstein style is a set of guidelines for uniformreporting that has been used by the American HeartAssociation and other organizations for reporting ofcases of cardiac arrest. The name derives from thelocation of a consensus conference held at the UtsteinAbbey in Norway. This standardized approach hasgreatly facilitated research and registry development inthe field of resuscitation medicine.


Table 1. Changes in recommendations from the first edition to the second edition

Chapter First Edition Second Edition

Cerebral PerfusionPressure

Level II—A CPP �40 mm Hg in children with TBIshould be maintained

Level III—A CPP between 40 and 65 mm Hg probablyrepresents an age-related continuum for the optimaltreatment threshold; there may be exceptions to thisrange in some infants and neonates

Level III—Advanced cerebral physiological monitoringmay be useful to define the optimal CPP in individualinstances

Level III—Hypotension should be avoided

Level III—A minimum CPP of 40 mm Hg may beconsidered in children with TBI

Level III—A CPP threshold 40–50 mm Hg may beconsidered; there may be age-specific thresholdswith infants at the lower end and adolescents atthe upper end of this range

Hyperosmolar Therapy Level III—Hypertonic saline is effective for control ofincreased ICP after severe head injury; effective dosesas a continuous infusion of 3% saline range between0.1 and 1.0 mL/kg of body weight per hour,administered on a sliding scale; the minimum doseneeded to maintain ICP �20 mm Hg should be used

Level III—Mannitol is effective for control of increasedICP after severe TBI; effective bolus doses range from0.25 g/kg of body weight to 1 g/kg of body weight

Level III—Euvolemia should be maintained by fluidreplacement; a Foley catheter is recommended inthese patients to avoid bladder rupture

Level III—Serum osmolarity should be maintainedbelow 320 mOsm/L with mannitol use, whereas alevel of 360 mOsm/L appears to be tolerated withhypertonic saline, even when used in combinationwith mannitol

Level II—Hypertonic saline should be consideredfor the treatment of severe pediatric TBIassociated with intracranial hypertension;effective doses for acute use range between 6.5and 10 mL/kg

Level III—Hypertonic saline should be considered forthe treatment of severe pediatric TBI associated withintracranial hypertension; effective doses as acontinuous infusion of 3% saline range between 0.1and 1.0 mL/kg of body weight per hour,administered on a sliding scale; the minimum doseneeded to maintain ICP �20 mm Hg should beused; serum osmolarity should be maintained below360 mOsm/L

Footnote below recommendations: although mannitolis commonly used in the management of raised ICPin pediatric TBI, no studies meeting inclusioncriteria were identified for use as evidence for thistopic

Temperature Control Level III—Extrapolated from the adult data,hyperthermia should be avoided in children withsevere TBI

Level III—Despite the lack of clinical data in children,hypothermia may be considered in the setting ofrefractory intracranial hypertension

Level II—Moderate hypothermia (32–33°C)beginning early after severe TBI for only 24 hrsduration should be avoided

Level II—Moderate hypothermia (32–33°C)beginning within 8 hrs after severe TBI for upto 48 hrs’ duration should be considered toreduce intracranial hypertension

Level II—If hypothermia is induced for anyindication, rewarming at a rate of �0.5°C perhour should be avoided

Level III—Moderate hypothermia (32–33°C)beginning early after severe TBI for 48 hrsduration may be considered

Hyperventilation Level III—Mild or prophylactic hyperventilation (PaCO2

�35 mm Hg) in children should be avoidedLevel III—Mild hyperventilation (PaCO2 30–35 mm Hg)

may be considered for longer periods for intracranialhypertension refractory to sedation and analgesia,neuromuscular blockade, cerebrospinal fluiddrainage, and hyperosmolar therapy

Level III—Aggressive hyperventilation (PaCO2 �30 mmHg) may be considered as a second-tier option in thesetting of refractory hypertension; cerebral bloodflow, jugular venous oxygen saturation, or braintissue oxygen monitoring is suggested to helpidentify cerebral ischemia in this setting

Level III—Aggressive hyperventilation therapy titratedto clinical effect may be necessary for brief periods incases of cerebral herniation or acute neurologicdeterioration

Level III—Avoidance of prophylactic severehyperventilation to a PaCO2 �30 mm Hg may beconsidered in the initial 48 hrs after injury

Level III—If hyperventilation is used in themanagement of refractory intracranialhypertension, advanced neuromonitoring forevaluation of cerebral ischemia may beconsidered


rather than diagnosis or prognosis; 3) chap-ters from the first edition which were elim-inated from the second edition includeTrauma Systems, Prehospital Airway Man-agement,b Resuscitation of Blood Pressureand Oxygenation,c Intracranial PressureMonitoring Technology,d and the CriticalPathway for Treatment of Intracranial Hy-pertensione’; 4) broader representation onthe committee of the relevant specialties inthe field, including pediatric anesthesiol-ogy, child neurology, and neuroradiology;and 5) international representation on the

guidelines committee includes Drs.Kissoon and Tasker.

As indicated, some publications in-cluded in the first edition were elimi-nated, because the methods team foundthey did not meet criteria (Appendix A,publications from the first edition notincluded in the second edition).

Table 1 summarizes changes in therecommendations from the first editionto the second edition of these guidelines.

The field is moving forward and it isclear that with advances in neuromoni-toring and imaging and the publication,subsequent to the first edition of theguidelines, of the results of the first ma-jor multicentered randomized controlledtrials in pediatric TBI, we are on the righttrack. Given the importance of severe TBIto the overall burden of childhood mor-bidity and mortality, we hope that thesenew guidelines aid caregivers and stimu-late the pediatric TBI community to gen-erate additional answers.

The authors thank the 14 externalpeer reviewers who further improved thequality of this document through inde-pendent review, including Drs. MarkDias, Richard Ellenbogen, Stuart Friess,Jeffrey Greenfield, Ann-Marie Guergue-rian, Mary Hartman, Mark Helfaer, JohnKuluz, Yi-Chen Lai, Leon Moores, JosePineda, Paul Shore, Kimberley Statler-Bennett, and Michael Whalen. The au-thors also thank Dr. Hector Wong, whoserved as the guest editor of this docu-ment for Pediatric Critical Care Medi-cine. Finally, we are extremely grateful tothe Brain Trauma Foundation for takingon this project and providing the re-sources necessary to ensure its successand for their commitment to improvingthe care of infants, children, and adoles-cents with severe TBI.

REFERENCES

1. Adelson PD, Bratton SL, Carney NA, et al:Guidelines for the acute medical manage-

bPrehospital treatment of pediatric patients withTBI is addressed in the Guidelines for PrehospitalManagement of Severe Traumatic Brain Injury (14).

cThere were no publications that met the inclusioncriteria for this topic.

dThis topic is addressed in the Guidelines for theManagement of Severe Traumatic Brain Injury, ThirdEdition (15).

eThe critical pathway will be developed and pub-lished as a separate document.

Table 1.—Continued

Chapter First Edition Second Edition

Corticosteroids Level III—The use of steroids is not recommended forimproving outcome or reducing ICP in pediatricpatients with severe TBI; despite two class II studiesfailing to show efficacy, the small sample sizespreclude support for a treatment guideline for thistopic

Level II—The use of corticosteroids is notrecommended to improve outcome or reduceICP for children with severe TBI

Analgesics, Sedatives,and NeuromuscularBlockade

Level III—In the absence of outcome data, the choiceof dosing and sedatives, analgesics, andneuromuscular-blocking agents used in themanagement of infants and children with severe TBIshould be left to the treating physician; however, theeffect of individual sedatives and analgesics on ICP ininfants and children with severe TBI can be variableand unpredictable

Level III—Etomidate may be considered to controlsevere intracranial hypertension; however, therisks resulting from adrenal suppression mustbe considered

Level III—Thiopental may be considered tocontrol intracranial hypertension

Footnotes below recommendations:In the absence of outcome data, the specific

indications, choice and dosing of analgesics,sedatives, and neuromuscular-blocking agentsused in the management of infants and childrenwith TBI should be left to the treating physician

As stated by the Food and Drug Administration,continuous infusion of propofol for eithersedation or the management of refractoryintracranial hypertension in infants and childrenwith severe TBI is not recommended)

Glucose and Nutrition Level III—Replace 130% to 160% of resting metabolismexpenditure after TBI in pediatric patients

Level II—The evidence does not support the use ofan immune-modulating diet for the treatmentof severe TBI to improve outcome

Antiseizure Prophylaxis Level II—Prophylactic use of antiseizure therapy is notrecommended for children with severe TBI forpreventing late posttraumatic seizures

Level III—Prophylactic antiseizure therapy may beconsidered as a treatment option to prevent earlyposttraumatic seizure in young pediatric patients andinfants at high risk for seizures after head injury

Level III—Prophylactic treatment with phenytoinmay be considered to reduce the incidence ofearly posttraumatic seizures in pediatric patientswith severe TBI

CPP, cerebral perfusion pressure; TBI, traumatic brain injury; ICP, intracranial pressure.


ment of severe traumatic brain injury in in-fants, children and adolescents. Pediatr CritCare Med 2003; 4:S1–S75

2. Hutchison JS, Ward RE, Lacroix J, et al:Hypothermia Pediatric Head Injury Trial In-vestigators and the Canadian Critical CareTrials Group. Hypothermia therapy aftertraumatic brain injury in children. N EnglJ Med 2008; 358:2447–2456

3. Adelson PD, Ragheb J, Kanev P, et al: PhaseII clinical trial of moderate hypothermia af-ter severe traumatic brain injury in children.Neurosurgery 2005; 56:740–754

4. Curry R, Hollingworth W, Ellenbogen RG, etal: Incidence of hypo- and hypercarbia insevere traumatic brain injury before and af-ter 2003 pediatric guidelines. Pediatr CritCare Med 2008; 9:141–146

5. Morris KP, Forsyth RJ, Parslow RC, et al:Intracranial pressure complicating severetraumatic brain injury in children: monitor-ing and management. Intensive Care Med2006; 32:1606–1612

6. Carter BG, Butt W, Taylor A: ICP and CPP:Excellent predictors of long term outcome in

severely brain injured children. Childs NervSyst 2008; 24:245–251

7. Wahlstrom MR, Olivecrona M, KoskinenL-OD, et al: Severe traumatic brain injury inpediatric patients: Treatment and outcomeusing an intracranial pressure targeted ther-apy—The Lund concept. Intensive Care Med2005; 31:832–839

8. Catala-Temprano A, Claret Teruel G, CambraLasaosa FJ, et al: Intracranial pressure andcerebral perfusion pressure as risk factors inchildren with traumatic brain injuries.J Neurosurg 2007; 106:463–466

9. Figaji AA, Zwane E, Thompson C, et al: Braintissue oxygen tension monitoring in pediat-ric severe traumatic brain injury. Part 1: Re-lationship with outcome. Childs Nerv Syst2009; 25:1325–1333

10. Shore PM, Thomas NJ, Clark RS, et al: Con-tinuous versus intermittent cerebrospinalfluid drainage after severe traumatic braininjury in children: Effect on biochemicalmarkers. J Neurotrauma 2004; 21:1113–1122

11. Briassoulis G, Filippou O, Kanariou M, et al:

Temporal nutritional and inflammatorychanges in children with severe head injuryfed a regular or an immune-enhancing diet:A randomized, controlled trial. Pediatr CritCare Med 2006; 7:56–62

12. Jagannathan J, Okonkwo DO, Yeoh HK, etal: Long-term outcomes and prognosticfactors in pediatric patients with severetraumatic brain injury and elevated intra-cranial pressure. J Neurosurg Pediatr2008; 2:240 –249

13. Bar-Joseph G, Guilburd Y, Tamir A, et al:Effectiveness of ketamine in decreasing in-tracranial pressure in children with intracra-nial hypertension. J Neurosurg Pediatr 2009;4:40–46

14. Badjita N, Carney N, Crosso TJ, et al: Guide-lines for prehospital management of severetraumatic brain injury. Prehospital Emer-gency Care 2007; 1:S1–S52.

15. Bratton S, Bullock R, Carney N, et al: Guide-lines for the management of severe trau-matic brain injury, 3rd edition. J Neu-rotrauma 2007; 24(Suppl 1):S1–S105


Chapter 2. Methods

I. TOPIC REFINEMENT

The Brain Trauma Foundation (BTF)and BTF Center for Guidelines Manage-ment (Center) convened a virtual meetingof previous guidelines authors and col-leagues new to the project. The panelconsisted of 15 clinicians and three meth-odologists. They specified which previoustopics would be maintained and agreedon new topics to include. Topics were notincluded in the second edition if theywere adequately addressed in otherguidelines documents (e.g., prehospitalmanagement of pediatric patients withtraumatic brain injury is addressed in theGuidelines for Prehospital Managementof Severe Traumatic Brain Injury [1]) orif there was no literature meeting inclu-sion criteria to support any level of rec-ommendation. Specification of new top-ics of interest was determined by panelconsensus. Previous topics that were up-dated are Indications for IntracranialPressure Monitoring, Intracranial Pres-sure Treatment Threshold, Cerebral Per-fusion Pressure, Antiseizure Prophylaxis,Hyperventilation, Cerebrospinal FluidDrainage, Hyperosmolar Therapy, De-compressive Craniectomy, Barbiturates,Analgesics–Sedatives–NeuromuscularBlockades, and Steroids. Topics from thefirst edition not included in this updateare Trauma Systems and PediatricTrauma Centers, Prehospital AirwayManagement, Resuscitation of BloodPressure and Oxygenation, IntracranialPressure Monitoring Technology, and theCritical Pathway. New topics are Ad-vanced Neuromonitoring and Neuroim-aging. The previous topic of TemperatureControl was expanded to Hypothermiaand Temperature Control, and the previ-ous topic of Nutrition was expanded toGlucose and Nutrition.

II. LITERATURE SEARCH ANDRETRIEVAL

Center staff worked with a doctoral-level research librarian to construct elec-tronic search strategies for each topic(Appendix B). For new topics, the litera-ture was searched from 1950 to 2009 andfor previous topics from 1996 to 2009. Asecond search was conducted for 2009–2010 to capture any new relevant litera-ture. Strategies with the highest likeli-hood of capturing most of the targetedliterature were used, which resulted inthe acquisition of a large proportion ofnonrelevant citations.

Two contributing authors (coauthors)were assigned to each topic, and a set ofabstracts was sent to each coauthor.Blinded to each other’s work, they readthe abstracts and eliminated citations us-ing the prespecified inclusion/exclusioncriteria. Center staff compared the coau-thors’ selections and identified and re-solved discrepancies either through con-sensus or through use of a third reviewer.A set of full-text publications was thensent to each coauthor. Again blinded toeach other’s work, they read the publica-tions and selected those that met theinclusion criteria.

Results of the electronic searches weresupplemented by recommendations ofpeers and by reading reference lists ofincluded studies. Relevant publicationswere added to those from the originalsearch, constituting the final library ofstudies that were used as evidence in thisdocument. The yield of literature fromeach phase of the search is presented inAppendix C.

III. STUDY SELECTION

Inclusion Criteria

Inclusion criteria consisted of severetraumatic brain injury (Glasgow ComaScale score �9); human subjects; Englishlanguage publications; pediatric patients(age �18 yrs); randomized controlled tri-als (N �25)a; cohort studies, prospectiveor retrospective (N �25)b; case–controlstudies (N �25); and case series (N �5).

The intervention (independent vari-able) must be specific to the topic.

The outcome must be a relevant healthoutcome (morbidity or mortality) or asurrogate outcome that associates with ahealth outcome.

Minimum sample sizes were identifiedto circumscribe the body of literature andmanage the scope of the project. There isno evidence that the selected cutoffs as-sociate with levels of confidence in thereported results.

Exclusion Criteria

Exclusion criteria consisted of penetrat-ing brain injury; animal studies; cadaverstudies; non-English language publica-tions; and adult patients (age �18 yrs).

Also excluded were studies in whichthe sample contained �15% of adult pa-tients or �15% of patients with patholo-gies other than traumatic brain injurywithout separate analysis (Appendix D).c

Case studies/editorials/comments/letters were excluded.

For each topic, relevant informationfrom the Guidelines for the Managementof Severe Traumatic Brain Injury (2) isreviewed. The panel agreed that datafrom the adult guidelines would not beused to contribute to recommendationsfor this document.

Inclusion of Direct and IndirectEvidence

Figures 1 and 2 illustrate differentlinks in a “causal pathway” that representeither direct or indirect evidence. In Fig-ure 1, arc A represents direct evidence,derived from a comparative study, of theinfluence of an intervention on an impor-tant health outcome (like functional sta-tus). Arc B represents direct evidence ofthe influence of an intervention on a sur-rogate outcome (like partial pressure ofbrain tissue oxygen), and arc C representsa correlation between measures on thesurrogate outcome and the importanthealth outcome. Taken together, arcs Band C represent indirect evidence of theinfluence of the intervention on an im-portant health outcome. Studies wereincluded if they contained direct evi-

aOne randomized controlled trial had a sample of24 patients (Kloti, 1987) and one a sample of 18(Fisher, 1992).

bOne retrospective review had a sample of 24patients (Pfenninger, 1983).

cOne study included 16% of patients with moder-ate traumatic brain injury (Downard, 2000).


DOI: 10.1097/PCC.0b013e31823f43df


dence or if they contained both compo-nents of the indirect evidence illus-trated in Figure 1.

Figure 2 illustrates a second kind ofindirect evidence that we included. Insome studies, an intervention was intro-duced to the entire study sample (withouta comparison group). Change in an im-portant health outcome was measured,and then authors looked for associationsbetween surrogate measures and thehealth outcome. For example, in the

chapter on Cerebral Perfusion Pressure,Downard et al (3) conducted a retrospec-tive analysis of 118 patients who were alltreated for severe traumatic brain injuryand assessed for Glasgow Outcome Scalescore at �3 months, dichotomized as“good” or “poor.” Then, using a logisticregression analysis, they looked for sig-nificant associations between cerebralperfusion pressure, a surrogate measure,and outcome. Lower cerebral perfusionpressure was associated with poorer out-

comes. This association was used as weakclass III evidence for the chapter’s recom-mendations. In Figure 2, arc A� repre-sents an uncontrolled association be-tween an intervention and an importanthealth outcome, and arc B� represents acorrelation between measures on the sur-rogate outcome and the important healthoutcome. Studies were included if theycontained both components of the indi-rect evidence illustrated in Figure 2.

IV. DATA ABSTRACTION ANDSYNTHESIS

Remaining blinded, coauthors readeach publication and abstracted data us-ing an evidence table template (AppendixE). They compared results of their dataabstraction and through consensus final-ized the data tables that constitute theevidence on which the recommendationsare based. As a result of heterogeneity ofstudies within topics, and the lack of lit-erature of adequate quality, data were notcombined quantitatively.

Coauthors drafted manuscripts foreach topic. The entire team gathered for a2-day work session to discuss the litera-ture base and craft the recommendations.Manuscripts were revised. Virtual meet-ings were held with a subset of the coau-thors to complete the editing process.The final draft manuscript was circulatedto the peer review panel and was revisedincorporating selected peer review input.

V. QUALITY ASSESSMENT OFINDIVIDUAL STUDIES ANDCLASSIFICATION OF EVIDENCE

In April of 2004, the BTF established aformal collaboration with the Evidence-Based Practice Center from OregonHealth & Science University. Center staffworked with two Evidence-Based PracticeCenter epidemiologists to develop criteriaand procedures for the quality assess-ment of individual studies and classifica-tion of level of evidence provided by eachincluded study. These criteria are de-signed to assess risk of bias for individualstudies based on study design and con-duct. Criteria for classification of evi-dence are in Table 1 and are derived fromcriteria developed by the U.S. PreventiveServices Task Force (4), the NationalHealth Service Centre for Reviews andDissemination (U.K.) (5), and the Co-chrane Collaboration (6). These criteriawere used to assess the literature.

Figure 1. Direct and indirect evidence.

Figure 2. Indirect evidence.


Three members of the Center staff,two of whom are Evidence-Based PracticeCenter epidemiologists, conducted all ofthe quality assessments. Two assessors,blinded to each other’s work and to pub-lication identification, read the selectedstudies and classified them as class I, II,or III based on the criteria in Table 2.Discrepancies were resolved throughconsensus or through a third person’sreview.

Class I evidence is derived from ran-domized controlled trials. However, somerandomized controlled trials may bepoorly designed, lack sufficient patientnumbers, or suffer from other methodo-logic inadequacies.

Class II evidence is derived from clin-ical studies in which data were collectedprospectively and retrospective analysesthat were based on clearly reliable data.Comparison of two or more groups mustbe clearly distinguished. Types of studiesinclude observational, cohort, preva-lence, and case–control. Class II evidencemay also be derived from flawed random-ized controlled trials.

Class III evidence is derived from pro-spectively collected data that are purelyobservational and retrospectively col-lected data. Types of studies include caseseries, databases, or registries. Class IIIevidence may also be derived from flawedrandomized controlled trials or flawedobservational, cohort, prevalence, or case–control studies.

VI. QUALITY OF BODY OFEVIDENCE AND STRENGTH OFRECOMMENDATIONS

At the beginning of each recommen-dation section in this document, the rec-ommendations are categorized in termsof strength and quality of evidence. Thestrength of the recommendation is de-rived from the overall quality of the bodyof evidence used to assess the topic.

Quality of Body of Evidence

The underlying methods for assessingrisk of bias for individual studies are rep-resented in Table 1. However, ultimately

the individual studies must be consideredin aggregate, whether through meta-analyses or through qualitative assess-ment. Thus, the strength of recommen-dations must be derived from the qualityof the overall body of evidence used toaddress the topic.

Consistent with recommendations forgrading a body of evidence adopted by theAgency for Healthcare Research andQuality (7), we assessed the overall qual-ity of the body of evidence consideringthe domains of 1) risk of bias from indi-vidual studies; 2) consistency of findingsacross studies; 3) directness of evidence;and 4) precision of estimates of effect.The quality of the overall body of evi-dence for each recommendation in thisdocument is classified as high, moderate,or low. Factors that may decrease thequality include potential bias, differingfindings across studies, the use of indi-rect evidence, or lack or precision. Forexample, if two or more class I studiesdemonstrate contradictory findings for aparticular topic, the overall quality mostprobably will be low because there is un-

Table 1. Criteria for assessment of risk of bias and classification of evidence

Class ofEvidence Study Design Quality Criteria

I Good-quality RCT Adequate random assignment methodAllocation concealmentGroups similar at baselineOutcome assessors blindedAdequate sample sizeIntention-to-treat analysisFollow-up rate � 85%No differential loss to follow-upMaintenance of comparable groups

II Moderate or poor-quality RCT Violation of one or more of the criteria for a good quality RCTa

II Good-quality cohort Blind or independent assessment in a prospective study or use of reliableb

data in a retrospective studyComparison of two or more groups must be clearly distinguishedNonbiased selectionFollow-up rate �85%Adequate sample sizeStatistical analysis of potential confoundersc

II Good-quality case–control Accurate ascertainment of casesNonbiased selection of cases/controls with exclusion criteria applied

equally to bothAdequate response rateAppropriate attention to potential confounding variables

III Moderate or poor-quality RCT or cohort Violation of one or more criteria for a good-quality RCT or cohorta

III Moderate or poor-quality case–control Violation of one or more criteria for a good-quality case–controla

III Case series, databases, or registries Prospective collected data that are purely observational and retrospectivelycollected data

RCT, randomized controlled trial.aAssessor needs to make a judgment about whether one or more violations are sufficient to downgrade the class of study based on the topic, the

seriousness of the violation(s), their potential impact on the results, and other aspects of the study. Two or three violations do not necessarily constitutea major flaw. The assessor needs to make a coherent argument why the violation(s) either do, or do not, warrant a downgrade; breliable data are concretedata such as mortality or reoperation; cpublication authors must provide a description of robust baseline characteristics and control for those that areunequally distributed between treatment groups.


certainty about the effect. Similarly, classI or II studies that provide indirect evi-dence may only constitute low-quality ev-idence, overall.

Strength of Recommendations

Consistent with methods generated bythe Grading of Recommendations, As-sessment, Development, and Evaluation(GRADE) Working Group, recommenda-tions in this document are categorized aseither strong or weak. As stated in theAmerican Thoracic Society’s officialstatement (8), in which they endorsed theGRADE methods for their guidelines en-deavors, “The strength of a recommenda-tion reflects the degree of confidence thatthe desirable effects of adherence to arecommendation outweigh the undesir-able effects.”

Strong recommendations are derivedfrom high-quality evidence that provides

precise estimates of the benefits or down-sides of the topic being assessed. Withweak recommendations, 1) there is lackof confidence that the benefits outweighthe downsides; 2) the benefits and down-sides may be equal; and/or 3) there isuncertainty about the degree of benefitsand downsides.

REFERENCES

1. Badjatia N, Carney N, Crocco T, et al: Guide-lines for the prehospital management of se-vere traumatic brain injury, 2nd edition. Pre-hosp Emerg Care 2007; 12:S1–S52

2. Bratton S, Bullock R, Carney N, et al: Guide-lines for the management of severe traumaticbrain injury, 3rd edition. J Neurotrauma2007; 24(Suppl 1):S1–S105

3. Downard C, Hulka F, Mullins RJ, et al: Relation-ship of cerebral perfusion pressure and survivalin pediatric brain-injured patients. J Trauma2000; 49:654–658; discussion 658–659

4. Harris RP, Helfand M, Woolf SH, et al: Cur-

rent methods of the third U.S. PreventiveServices Task Force. Am J Prev Med 2001;20:21–35

5. Anonymous: Undertaking Systematic Reviewsof Research on Effectiveness: CRD’s Guidancefor Those Carrying Out or Commissioning Re-views. CRD Report Number 4. Second Edi-tion). York, UK, NHS Centre for Reviews andDissemination, 2001, p 4

6. Mulrow CD, Oxman AD: How to Conduct aCochrane Systematic Review. Version 3.0.2.Presented at Cochrane Collaboration, San An-tonio, TX, 1997

7. Owens D, Lohr K, Atkins D, et al: Grading thestrength of a body of evidence when compar-ing medical interventions—Agency forHealthcare Research and Quality and the Ef-fective Health Care Program. J Clin Epidemiol2010; 63:513–523

8. Schunemann H, Jaeschke R, Cook D, et al: Anofficial ATS statement: grading the quality ofevidence and strength of recommendations inATS guidelines and recommendations. Am J Re-spir Crit Care Med 2006; 174:605–614


Chapter 3. Indications for intracranial pressure monitoring

I. RECOMMENDATIONS

Strength of Recommendations: Weak.Quality of Evidence: Low, from poor

and moderate-quality class III studies.

A. Level I

There are insufficient data to supporta level I recommendation for this topic.

B. Level II

There are insufficient data to supporta level II recommendation for this topic.

C. Level III

Use of intracranial pressure (ICP)monitoring may be considered in infantsand children with severe traumatic braininjury (TBI).

II. EVIDENCE TABLE (see Table 1)

III. OVERVIEW

Secondary injury to the brain aftersevere TBI occurs, in part, as a result ofreduced perfusion of surviving neural tis-sue, resulting in reduced oxygen and me-tabolite delivery and reduced clearance ofmetabolic waste and toxins. Secondaryinjury also occurs as the result of cerebralherniation syndromes, resulting in focalischemic injury and brain stem compres-sion along with other mechanisms. Intra-cranial hypertension represents a keypathophysiological variable in each of thesesecondary injury mechanisms (1–3).

Since the late 1970s, significant im-provements in both survival and func-tional outcome after severe TBI havebeen achieved using intensive care man-agement protocols that center on themeasurement of ICP and medical andsurgical treatment of intracranial hyper-tension (4). A study by Tilford and col-leagues (5) demonstrated that an inten-sive care unit with higher incidence ofICP monitoring in severely brain-injured

children, plus certain medical interven-tions, had a trend toward lower mortalitythan two other pediatric intensive careunits. Similarly, a study by Tilford andcolleagues (4) demonstrated improvedoutcomes after severe TBI in an era dur-ing which the overall rates of ICP moni-toring in these patients increased. At-tempts to evaluate the independentbenefit of direct ICP measurement to im-prove outcomes, per se, are confoundedby the numerous therapeutic interven-tions that have been simultaneously in-troduced and have not been subjectedindividually to controlled trials. Theseconfounders include protocol-driven pre-hospital care, tracheal intubation and ox-ygenation, aggressive treatment of sys-temic hypotension and hypovolemia,osmolar treatment of cerebral edema,rapid cranial computed tomography (CT)imaging to detect mass lesions, improvedenteral and parenteral nutrition, amongothers.

Several studies demonstrate an associa-tion between intracranial hypertensionand/or systemic hypotension and poor out-come after severe TBI (6–8). It is less clear,however, whether intracranial hyperten-sion or reduced cerebral perfusion second-ary to intracranial hypertension is the pri-mary mechanism of secondary injury.Cerebral perfusion pressure (equals meanarterial pressure minus ICP) is the simplestcorrelate of global cerebral perfusion (9–12). The relative value of ICP monitoring asa means of evaluating and manipulatingcerebral perfusion pressure, vs. avoidanceof cerebral herniation events, is also un-clear (13).

The lack of controlled trials on ICPmonitoring limited the strength of therecommendations contained in the firstedition of the Guidelines for the Manage-ment of Severe TBI in Children (14). Thisdearth of strong evidence is associatedwith mixed adoption of guidelines-directed management in the UnitedStates and abroad (15–17). In a 2007 sur-vey of U.S. neurosurgeons and nonneu-rosurgeons caring for such patients,Dean et al (15) found approximately 60%agreement and conformity with guide-lines recommendations. In the United

Kingdom, only 59% of children present-ing with severe TBI underwent ICP mon-itoring with only half of clinical unitscaring for such children using monitor-ing technology (16, 17). The use of mon-itoring in children �2 yrs of age withsevere TBI may be even less likely. Astudy by Keenan et al (18) observed use ofICP monitoring in only 33% of patientsin this young age group at multiple cen-ters in the state of North Carolina. Thereis also significant variability in the inci-dence of using various interventions for thetreatment of intracranial hypertension atdifferent centers (5).

IV. PROCESS

For this update, MEDLINE was searchedfrom 1996 through 2010 (Appendix B forsearch strategy), and results were supple-mented with literature recommended bypeers or identified from references lists. Of36 potentially relevant studies, seven stud-ies were added to the existing table andused as evidence for this topic.

V. SCIENTIFIC FOUNDATION

Two moderate and 14 poor-qualityclass III studies met the inclusion criteriafor this topic and provide evidence tosupport the recommendation (9, 19–33).

Are Children With Severe TBI atRisk of IntracranialHypertension?

A number of small studies demon-strate a high incidence of intracranial hy-pertension in children with severe TBI(20, 21, 23, 24, 26, 28, 31, 33). Some ofthese studies identify in preliminary fash-ion other clinical factors that, in combi-nation with severe TBI in a child, areindicative of a high incidence of intracra-nial hypertension. In these patients, “dif-fuse cerebral swelling” on CT scan is 75%specific for the presence of intracranialhypertension (26). In a study of 56 se-verely brain-injured patients (39 of whomhad severe TBI), 32% of children had aninitial ICP measurement �20 mm Hg but50% had ICP maximum �20 mm Hg atsome point during their intensive care


DOI: 10.1097/PCC.0b013e31823f440c


Table 1. Evidence table

Reference Study DescriptionData Class, Quality,

and Reasons Results and Conclusion

Studies from previousguidelines

Alberico et al, 1987(19)

Design: single-center, prospective,observational study

N � 100Age: 0–19 yrsGlasgow Coma Scale score: �7Purpose: Assessment of relationship

between ICP and outcomeOutcome: GOS score at 3 months and

1 yr

Class IIIPoor quality: no control for

confounders

Reducible intracranial hypertensionwas significantly associated withbetter outcome than nonreducibleintracranial hypertension

Barzilay et al, 1988(20)

Design: retrospective case series withanalysis of minimum ICP

N � 56Age: mean 6.2 yrsPurpose: assessment of relationship

between ICP and outcome in patientstreated for high ICP withhyperventilation and medicalmanagement

Outcome: survival at hospital discharge


confounders

For children with severe TBI, ICPmaximum was 16.9 � 3.1 insurvivors (N � 32) and 53.7 �10.8 in nonsurvivors (N � 9);p � 0.01

Bruce et al, 1979(21)

Design: single-center, observational studyN � 85, 40 had ICP monitoringAge: 4 months to 18 yrsPurpose: assess relationship between ICP

monitoring and medical managementin a protocol emphasizinghyperventilation therapy to controlintracranial hypertension, but alsoincluding barbiturates, mannitol, and/or surgery

Outcome: dichotomized GOS at 6months


confounders

Intracranial hypertension (ICP �20mm Hg) was more prevalent inchildren without (80%) than with(20%) spontaneous motorfunction

Of the total group (N � 85): 87.5%of children achieved goodrecovery or moderate disability;3.5% persistent vegetative state,9% died.

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/persistentvegetative state; 22% died

ICP �20 �40 (N � 17): 88% goodrecovery/moderate disability; 6%severe disability/persistentvegetative state; 6% died

ICP �40 (N � 14): 57% goodrecovery/moderate disability; 7%severe disability/persistentvegetative state; 36% died

Chamberset al, 2001 (9)

Design: single-center, observational studyN � 84Age: 0–16 yrsPurpose: assessment of relationship

between ICP and CPP and outcomeOutcome: GOS at 6 months


confounders; unclear ifpatient selection wasunbiased

ICP maximum predictive of pooroutcome was �35 mm Hg inadults and children

Downardet al, 2000 (22)

Design: retrospective reviewN � 118Age: �15 yrsGlasgow Coma Scale score: mean 6, 84% �8Purpose: assess relationship among ICP, CPP,

and outcome in children with severe TBIin two trauma centers

Outcome: the final available GOS in themedical record

Class IIIPoor quality: as an intervention

study; moderate quality as aprognosis study: logisticregression performed todetermine factors associatedwith GOS, but nocomparison of groups basedon any intervention

In a stepwise logistic regressionanalysis, ICP �20 mm Hg wassignificantly associated with anincreased risk of death





Esparza et al, 1985(23)

Design: single-center, observational studyN � 56Age: 3 months to 14 yrsPurpose: assessment of relationship

between ICP monitoring and surgicaland medical therapy and outcome aftersevere TBI in children

Outcome: GOS dichotomized as good(mild disability) or poor (disability,persistent vegetative state, or death)


confounders; unclear ifoutcome assessment wasunbiased

Outcomes were as follows:93% good, 3% poor for patients

with ICP maximum �20 mm Hg,71% good, 29% poor for patientswith ICP maximum 20–40 mmHg; 0% good, 100% poor forpatients with ICP maximum 40–60 mm Hg and 0% good, 100%poor for patients with ICPmaximum �60 mm Hg (nosignificance test reported)

Kasoff et al, 1988(24)

Design: single-center, retrospective,observational study

N � 25Age: 3 months to 17 yrsPurpose: assess relationship between ICP

and outcome in children treated withmannitol and if refractory, mannitolplus barbiturates

Outcome: mortality


confounders; unclear ifpatients selection wasunbiased

Mortality rate was 20%Children with elevated ICP had a

lower survival rate than childrenwith normal ICP, although nostatistical analysis is presented

Mean highest ICP of those who diedwas 81 mm Hg (range, 55–120);for ICP only group 18.7 (range,10–30), for mannitol group 42.11(range, 10–70), for pentobarbitaland mannitol group 72 (range,30–120)

Four children had normal ICP anddid not require medical therapy;nine required mannitol therapyand eleven mannitol and thenbarbiturate therapy for sustainedintracranial hypertension

Michaudet al, 1992 (25)

Design: single-center, observational studyN � 51Age: 3 months to 14 yrsPurpose: assessment of relationship

between ICP and outcomeOutcome: GOS at discharge

Class IIIModerate quality: no power

calculation; otherwise met allcriteria

94% of children with ICP maximum�20 mm Hg vs. 59% with ICPmaximum �20 mm Hg survived(p � 0.02)

48% of children with ICP elevation�1 hr survived compared to 89%of children with ICP elevated for�1 hr

Outcome was also better in childrenwith ICP elevation for �1 hr

No statistically significantrelationship was found betweenpeak ICP and degree of disability

Shapiro andMarmarou, 1982(26)

Design: retrospective case seriesN � 22Age: 3 months to 15 yrsPurpose: study the use of pressure

volume index assessment usingexternal ventricular drains

Outcome: GOS—time of assessment notindicated

Class IIIPoor quality (diagnostic study):

narrow spectrum of patientsenrolled; small sample size;unclear if reliability of testassessed

86% of children with severe TBI hadICPs exceeding 20 mm Hg

“Diffuse cerebral swelling” oncomputed tomography scan was75% specific for the presence ofintracranial hypertension

Intracranial hypertension could becontrolled in 14 of the 16children whose pressure volumeindex was measured, and in thosepatients, there were no deaths





New studiesAdelson et al, 2005

(27)Design: randomized controlled trial of

hypothermia treatmentN � 75Age: �17 yrsPurpose: ICP monitoring and

randomized, controlled trial ofmoderate hypothermia vs.normothermia plus medicalmanagement of intracranialhypertension

Outcome: GOS at 3 and 6 months


confounders. (class II forhypothermia trial)

High (no. not specified) incidence ofintracranial hypertension

ICP �20 was most sensitive andspecific for poor outcome

Low mean ICP, percent time ICP�20 and mean CPP were allsignificantly associated with goodoutcome

Mean ICP was lower in patients whohad a good outcome versus thosewith a poor outcome (good, 11.9mm Hg; poor, 24.9 mm Hg; p �.036)

The percent time less than 20 mmHg differed between outcomegroups (good, 90.8% � 10.8%;poor, 68.6% � 35.0%; p � .01)

Cruz et al, 2002(28)

Design: single-center retrospective studyN� 45Age: 1–12 yrsPurpose: assessment of the effect of ICP

monitoring and medical therapy onoutcome; also examined relationship tooxygen metabolism through jugularbulb catheter

Outcome: GOS at 6 months


confounders

82% had favorable outcome, 17.8%unfavorable; 4.4% died; 13.3%had severe disability

Higher ICP (p � .02) for days 1–5was significantly associated withdecreased cerebral oxygenextraction and worse clinicaloutcome

Grinkeviciute et al,2008 (29)

Design: single-center prospectiveobservational study

N� 48Age: 2.4 months to 18 yrsPurpose: examination of relationship

between ICP, CPP, and outcome inchildren including 13 treated withdecompressive craniectomy formedically refractory intracranialhypertension

Outcome: GOS at 6 months


confounders

Survival rate was 97.9% (1 death);favorable outcome in 89.6%

There was no difference in ICPmaximum in groups with good(22.2 mm Hg) vs. poor (24.6 mmHg) outcomes

Jagannathan et al,2008 (30)

Design: single-center observational studyN � 96Age: 3–18 yrsPurpose: assessment of relationship

between ICP, treatment and outcomein patients treated with variablecombination of evacuation of masslesions, ventricular drainage, medicalmanagement and decompressivecraniectomy

Outcome: GOS at 2 yrs

Class IIIModerate quality: unclear if

analysis of ICP monitoringcontrolled for confounders

Death was associated with refractoryraised ICP (p � .0001), but notwith ICP maximum, irrespectiveof the surgical or medicalmethods(s) used for successfulreduction of intracranialhypertension

Outcome: quality of life was relatedto medical management ofelevated ICP (p � .04)

Long-term outcomes were notcorrelated with peak ICP

Pfenninger andSanti, 2002 (31)

Design: retrospective single-centerobservational study

N � 51Age: 1 month to 16 yrsPurpose: assess relationship between ICP,

medical or surgical management orjugular venous monitoring andoutcome

Outcome: GOS at 6–12 months


confounders

Moderate to severe intracranialhypertension (mean sustained ICP�20 mm Hg) was associated withpoor outcome (p � .05)

69% of monitored patients hadsustained ICP �20 mm Hg


course (20). Intracranial hypertension(ICP �20 mm Hg) may also be signifi-cantly more prevalent in children withsevere TBI who do not demonstrate spon-taneous motor function (80%) than thosewho do (20%) (21).

These studies suggest that childrenpresenting with severe TBI are at notablerisk of intracranial hypertension. No spe-cific markers have been identified thatreliably determine the presence or ab-sence of intracranial hypertension with-out monitoring in this population.

Are ICP Data Useful inManaging Pediatric Severe TBI?

Fifteen studies involving 857 pediatricpatients demonstrated an association be-tween intracranial hypertension (gener-ally �20 mm Hg) and poor neurologicoutcome or death (9, 19–28, 30–33).

One small study of 48 patients failed todemonstrate a clear association betweenintracranial hypertension and poor out-come (29). Specifically, a study byGrinkeviciute et al reported similar meanICP in children with good and poor out-come. In their study, however, childrenwith higher peak ICP were immediatelyand successfully treated with decompres-sive craniectomy.

These studies suggest that ICP is animportant prognostic variable. It alsoplays a strong role both independentlyand as a component of cerebral perfusion

pressure in directing the management ofpediatric patients with severe TBI.

Does ICP Monitoring andTreatment Improve Outcome?

Two studies of combined treatmentstrategies also suggest that improved clin-ical outcomes are associated with success-ful control of intracranial hypertension (19,30). A prospective observational study of100 children with severe TBI treated withvarying combinations of hyperventilation,diuretics, cerebrospinal fluid drainage, se-dation, pharmacologic paralysis, and barbi-turates reported that children whose intra-cranial hypertension was successfullylowered had better 1-yr outcomes thanchildren whose intracranial hypertensionwas uncontrollable (but worse than thosewithout intracranial hypertension) (19). Aretrospective review of a prospectively ac-quired TBI database showed that reducedsurvival and worsened outcome in childrenwith severe TBI were associated with intra-cranial hypertension refractory to treat-ment rather than peak ICP per se (30). Inthis study, successful control of intracranialhypertension, irrespective of treatment mo-dality (osmolar therapy, cerebrospinal fluiddrainage, decompression, etc.), was deemedto be important.

Although they represent only class IIIevidence for long-term outcome relatedto ICP monitoring and are only correla-tive, these studies support the association

of successful ICP monitor-based manage-ment of intracranial hypertension with im-proved survival and neurologic outcome.

VI. INFORMATION FROMOTHER SOURCES

A. Indications From the AdultGuidelines

The adult guidelines offer the follow-ing recommendation.

Level II: ICP should be monitored in allsalvageable patients with a severe TBI(Glasgow Coma Scale score of 3–8 af-ter resuscitation) and an abnormal CTscan. An abnormal CT scan of the headis one that reveals hematomas, contu-sions, swelling, herniation, or com-pressed basal cisterns.

Level III: ICP monitoring is indicatedin patients with severe TBI with a nor-mal CT scan if two or more of thefollowing features are noted at admis-sion: age �40 yrs, unilateral or bilat-eral motor posturing, or systolic bloodpressure �90 mm Hg.

What Patients Are at High Risk ofICP Elevation? Patients with severe TBI(Glasgow Coma Scale �8) are at highrisk for intracranial hypertension (8,34). The combination of severe TBI andan abnormal head CT scan suggests ahigh likelihood (53% to 63%) of raised




Wahlstromet al, 2005 (32)

Design: single-center observational studyN � 41Age: 3 months to 14.2 yrsPurpose: assess affect of ICP

management using the Lund protocolon outcome

Outcome: GOS assessed between 2.5 and26 months


confounders

Survival rate was 93%; favorableoutcome (GOS 4 and 5) in 80%

ICP in 3 nonsurvivors wassignificantly higher than in 38survivors (mean 43 � 26 mm Hgvs. 13 � 4 mm Hg)

The relationship between ICP andoutcome in survivors was notstatistically analyzed

White et al, 2001(33)

Design: retrospective observational studyN � 136 admitted to pediatric intensive

care unit; 37 with ICP monitoringAge: 0–17 yrsPurpose: assess relationship between ICP

and survivalOutcome: survival


confounders for ICP analysis

14% of survivors and 41% ofnonsurvivors had ICP �20 mmHg in the first 72 hrs

Those with lower mean ICP weremore likely to be survivors (p �.005)

ICP maximum and ICP measured 6,12, and 24 hrs after admissionwere all significantly lower insurvivors

ICP, intracranial pressure; GOS, Glasgow Outcome Scale score; CPP, cerebral perfusion pressure; TBI, traumatic brain injury.


ICP (34). However, even with a normaladmission CT scan, intracranial hyper-tension may be present (35, 36). Datacollected predominantly in adult patientssuggest that detection and treatment of in-tracranial hypertension may protect cere-bral perfusion pressure, avoid cerebral her-niation, and improve neurologic outcome(8, 11, 34, 37–39).

In certain conscious patients with CTfindings suggesting risk of neurologic de-terioration (hematomas, contusions, swell-ing, herniation, or compressed basal cis-terns), however, monitoring may beconsidered based on the opinion of thetreating physician (35, 38). Inability to per-form serial neurologic examinations, be-cause of pharmacologic sedation or anes-thesia, may also influence a clinician’sdecision to monitor ICP in an individualpatient (40, 41).

How Does ICP Data Influence PatientManagement? ICP data allow the man-agement of severe TBI by objective crite-ria. This is particularly important be-cause many, perhaps all, medical andsurgical measures for the treatment ofintracranial hypertension have signifi-cant potential adverse consequences (2,7, 42). Thus, ICP monitoring allows thejudicious use of interventions such as hy-perosmolar therapy, sedatives, neuro-muscular blockade, barbiturates, ventila-tor management, etc., with a defined endpoint that is correlated with clinical out-come. This may avoid potentially harm-ful, overly aggressive treatment.

Does ICP Monitoring Improve Out-come? In adults, intensive managementprotocols for severe TBI, including ICPmonitoring, have been associated withlowered mortality rates as comparedwith historical controls or centers inother countries not using monitoringtechniques (8, 43– 45). A study byEisenberg et al (46) reported that im-proved ICP control was associated withimproved outcome in severely head-injured patients with medically intrac-table intracranial hypertension. Finally,in a small, single-institution study ofpatients triaged according to the at-tending neurosurgery call schedule,mortality was over four times higher innonmonitored than in monitored pa-tients with severe TBI (47).

B. Information Not Included asEvidence

Various class III studies have demon-strated improved outcomes, vs. historical

controls, in the era of ICP monitor-directed intensive therapy of patientswith severe TBI (11, 35, 43, 48, 49). Twospecific ICP monitor-directed therapieseffective in treating acute intracranial hy-pertension have been associated with im-proved survival and clinical outcomes af-ter severe TBI in children. As indicated inthe evidence table, a study by Bruce et al(1) reported that aggressive therapy withhyperventilation and/or barbiturates totreat intracranial hypertension in 85 chil-dren with severe TBI resulted in 87.5%good outcomes and only 9% mortality.Not included as evidence, Peterson et al(50) performed a retrospective study ofsevere TBI in 68 infants and children,which showed that effective treatment ofrefractory intracranial hypertension us-ing continuous infusion of hypertonic(3%) saline resulted in a mortality rate(15%) lower than expected as a result oftrauma severity score (40%). There wereonly three deaths in this study (4%) re-sulting from uncontrolled intracranialhypertension.

VII. SUMMARY

Four lines of evidence support the useof ICP monitoring in children with severeTBI: a frequently reported high incidenceof intracranial hypertension in childrenwith severe TBI, a widely reported asso-ciation of intracranial hypertension andpoor neurologic outcome, the concor-dance of protocol-based intracranial hy-pertension therapy and best-reportedclinical outcomes, and improved out-comes associated with successful ICP-lowering therapies. Evidence reviewed inthe adult guidelines mirrors that for pe-diatric patients, further suggesting thatICP monitoring is of clinical benefit inpatients with severe TBI.

Intracranial hypertension is both dif-ficult to diagnose and is associated withpoor neurologic outcomes and death ininfants and young children. Intracranialhypertension may be present in childrenwith open fontanelles and sutures (18).ICP monitoring is of significant use inthese patient populations.

The presence of intracranial hyperten-sion can also be influenced by the type ofpathology on CT such as diffuse injury orspecific etiologies such as traumatic si-nus thrombosis.

By contrast, ICP monitoring is notroutinely indicated in children with mildor moderate TBI. Treating physiciansmay, however, in some circumstances,

choose to use ICP monitoring in con-scious children who are at relative riskfor neurologic deterioration as a result ofthe presence of traumatic mass lesions orin whom serial neurologic examination isprecluded by sedation, neuromuscularblockade, or anesthesia.

VIII. KEY ISSUES FOR FUTUREINVESTIGATION

• Studies of specific subpopulations ofpediatric patients with TBI in whomICP monitoring is indicated; in partic-ular, in the categories of infants andyoung children with abusive headtrauma and/or infants with open fon-tanelles and sutures.

• Studies of the incidence of intracranialhypertension based on clinical and ra-diologic parameters in children of dif-ferent ages and injury mechanisms.

• Focused multivariate analyses of chil-dren with intracranial hypertension topredict those who respond better tospecific ICP-lowering therapies.

• Careful monitoring of the impact ofadoption of ICP monitoring-directedprotocols by hospitals and health sys-tems should be undertaken to providefurther evaluation of the impact ofthese measures on outcome as well assystem performance variables.

• Studies are also needed to determinewhether the type of ICP monitor (e.g.,ventricular, parenchyma) or approachto monitoring (e.g., continuous or in-termittent with cerebrospinal fluiddrainage) influences outcome.

REFERENCES

1. Bruce DA, Alavi A, Bilaniuk L, et al: Diffusecerebral swelling following head injuries inchildren: The syndrome of ‘malignant brainedema.’ J Neurosurg 1981; 54:170–178

2. Muizelaar JP, Marmarou A, DeSalles AA, etal: Cerebral blood flow and metabolism inseverely head-injured children. Part 1: Rela-tionship with GCS score, outcome, ICP, andPVI. J Neurosurg 1989; 71:63–71

3. Sharples PM, Stuart AG, Matthews DS, et al:Cerebral blood flow and metabolism in chil-dren with severe head injury. Part 1: Relationto age, Glasgow Coma Score, outcome, intra-cranial pressure, and time after injury. J NeurolNeurosurg Psychiatry 1995; 58:145–152

4. Tilford JM, Aitken ME, Anand KJ, et al: Hos-pitalizations for critically ill children withtraumatic brain injuries: A longitudinal anal-ysis. Crit Care Med 2005; 33:2074–2081

5. Tilford JM, Simpson PM, Yeh TS, et al: Vari-ation in therapy and outcome for pediatric


head trauma patients. Crit Care Med 2001;29:1056–1061

6. Schoon P, Benito Mori L, Orlandi G, et al:Incidence of intracranial hypertension re-lated to jugular bulb oxygen saturation dis-turbances in severe traumatic brain injurypatients. Acta Neurochir Suppl 2002; 81:285–287

7. Chesnut RM, Marshall LF, Klauber MR, et al:The role of secondary brain injury in deter-mining outcome from severe head injury.J Trauma 1993; 34:216–222

8. Marmarou A, Anderson R, Ward J, et al: Im-pact of ICP instability and hypotension onoutcome in patients with severe headtrauma. J Neurosurg 1991; 75:S59–S66

9. Chambers IR, Treadwell L, Mendelow AD:Determination of threshold levels of cerebralperfusion pressure and intracranial pressurein severe head injury by using receiver-operating characteristic curves: An observa-tional study in 291 patients. J Neurosurg2001; 94:412–416

10. Chambers IR, Treadwell L, Mendelow AD:The cause and incidence of secondary insultsin severely head-injured adults and children.Br J Neurosurg 2000; 14:424–431

11. Miller JD, Butterworth JF, Gudeman SK, et al:Further experience in the management of se-vere head injury. J Neurosurg 1981; 54:289–299

12. Rosner MJ, Daughton S: Cerebral perfusion pres-sure management in head injury. J Trauma1990; 30:933–940; discussion 940–941

13. Prabhakaran P, Reddy AT, Oakes WJ, et al: Apilot trial comparing cerebral perfusion pres-sure-targeted therapy to intracranial pres-sure-targeted therapy in children with severetraumatic brain injury. J Neurosurg 2004;100:454–459

14. Adelson PD, Bratton SL, Carney NA, et al:Guidelines for the acute medical managementof severe traumatic brain injury in infants, chil-dren, and adolescents. Chapter 5. Indicationsfor intracranial pressure monitoring in pediat-ric patients with severe traumatic brain injury.Pediatr Crit Care Med 2003; 4:S19–S24

15. Dean NP, Boslaugh S, Adelson PD, et al:Physician agreement with evidence-basedrecommendations for the treatment of severetraumatic brain injury in children. J Neuro-surg 2007; 107:387–391

16. Morris KP, Forsyth RJ, Parslow RC, et al:Intracranial pressure complicating severetraumatic brain injury in children: monitor-ing and management. Intensive Care Med2006; 32:1606–1612

17. Segal S, Gallagher AC, Shefler AG, et al:Survey of the use of intracranial pressuremonitoring in children in the United King-dom. Intensive Care Med 2001; 27:236–239

18. Keenan HT, Nocera M, Bratton SL: Frequencyof intracranial pressure monitoring in infantsand young toddlers with traumatic brain in-jury. Pediatr Crit Care Med 2005; 6:537–541

19. Alberico AM, Ward JD, Choi SC, et al: Out-come after severe head injury. Relationshipto mass lesions, diffuse injury, and ICP

course in pediatric and adult patients. J Neu-rosurg 1987; 67:648–656

20. Barzilay Z, Augarten A, Sagy M, et al: Vari-ables affecting outcome from severe braininjury in children. Intensive Care Med 1988;14:417–421

21. Bruce DA, Raphaely RC, Goldberg AI, et al:Pathophysiology, treatment and outcomefollowing severe head injury in children.Childs Brain 1979; 5:174–191

22. Downard C, Hulka F, Mullins RJ, et al: Rela-tionship of cerebral perfusion pressure andsurvival in pediatric brain-injured patients.J Trauma 2000; 49:654 – 658; discussion658–659

23. Esparza J, M-Portillo J, Sarabia M, et al:Outcome in children with severe head inju-ries. Childs Nerv Syst 1985; 1:109–114

24. Kasoff SS, Lansen TA, Holder D, et al: Ag-gressive physiologic monitoring of pediatrichead trauma patients with elevated intracra-nial pressure. Pediatr Neurosci 1988; 14:241–249

25. Michaud LJ, Rivara FP, Grady MS, et al: Pre-dictors of survival and severity of disabilityafter severe brain injury in children. Neuro-surgery 1992; 31:254–264

26. Shapiro K, Marmarou A: Clinical applicationsof the pressure–volume index in treatment ofpediatric head injuries. J Neurosurg 1982;56:819–825

27. Adelson PD, Ragheb J, Kanev P, et al: PhaseII clinical trial of moderate hypothermia af-ter severe traumatic brain injury in children.Neurosurgery 2005; 56:740–754; discussion740–754

28. Cruz J, Nakayama P, Imamura JH, et al:Cerebral extraction of oxygen and intracra-nial hypertension in severe, acute, pediatricbrain trauma: Preliminary novel manage-ment strategies. Neurosurgery 2002; 50:774–779; discussion 779–780

29. Grinkeviciute DE, Kevalas R, Matukevicius A,et al: Significance of intracranial pressureand cerebral perfusion pressure in severe pe-diatric traumatic brain injury. Medicina(Kaunas, Lithuania) 2008; 44:119–125

30. Jagannathan J, Okonkwo DO, Yeoh HK, et al:Long-term outcomes and prognostic factorsin pediatric patients with severe traumaticbrain injury and elevated intracranial pres-sure. J Neurosurg Pediatr 2008; 2:240–249

31. Pfenninger J, Santi A: Severe traumatic braininjury in children—Are the results improv-ing? Swiss Med Wkly 2002; 132:116–120

32. Wahlstrom MR, Olivecrona M, Koskinen LO,et al: Severe traumatic brain injury in pedi-atric patients: Treatment and outcome usingan intracranial pressure targeted therapy—The Lund concept. Intensive Care Med 2005;31:832–839

33. White JR, Farukhi Z, Bull C, et al: Predictorsof outcome in severely head-injured chil-dren. Crit Care Med 2001; 29:534–540

34. Narayan RK, Kishore PR, Becker DP, et al:Intracranial pressure: To monitor or not tomonitor? A review of our experience with

severe head injury. J Neurosurg 1982; 56:650–659

35. Marshall L, Gautille T, Klauber M: The out-come of severe closed head injury. J Neuro-surg 1991; 75:S28–S36

36. O’Sullivan MG, Statham PF, Jones PA, et al:Role of intracranial pressure monitoring in se-verely head-injured patients without signs ofintracranial hypertension on initial computer-ized tomography. J Neurosurg 1994; 80:46–50

37. Eisenberg HM, Gary HE Jr, Aldrich EF, et al:Initial CT findings in 753 patients with se-vere head injury. A report from the NIHTraumatic Coma Data Bank. J Neurosurg1990; 73:688–698

38. Levin HS, Eisenberg HM, Gary HE, et al:Intracranial hypertension in relation tomemory functioning during the first yearafter severe head injury. Neurosurgery 1991;28:196–199; discussion 200

39. Gopinath S, Constant C, Robertson C: Criti-cal thresholds for physiological parametersin patients with severe head injury. Congressof Neurological Surgeons, Vancouver, BC,Canada, 1993

40. Humphreys RP, Hendrick EB, Hoffman HJ: Thehead-injured child who ‘talks and dies.’ A reportof 4 cases. Childs Nerv Syst 1990; 6:139–142

41. Lobato RD, Rivas JJ, Gomez PA, et al: Head-injured patients who talk and deteriorateinto coma. Analysis of 211 cases studied withcomputerized tomography. J Neurosurg1991; 75:256–261

42. Kaufmann AM, Cardoso ER: Aggravation ofvasogenic cerebral edema by multiple-dosemannitol. J Neurosurg 1992; 77:584–589

43. Becker DP, Miller JD, Ward JD, et al: Theoutcome from severe head injury with earlydiagnosis and intensive management. J Neu-rosurg 1977; 47:491–502

44. Colohan AR, Alves WM, Gross CR, et al: Headinjury mortality in two centers with differentemergency medical services and intensivecare. J Neurosurg 1989; 71:202–207

45. Jennett B, Teasdale G, Galbraith S, et al:Severe head injuries in three countries.J Neurol Neurosurg Psychiatry 1977; 40:291–298

46. Eisenberg HM, Frankowski RF, Contant CF,et al: High-dose barbiturate control of ele-vated intracranial pressure in patients withsevere head injury. J Neurosurg 1988; 69:15–23

47. Ghajar J, Hariri R, Patterson R: Improvedoutcome from traumatic coma using onlyventricular CSF drainage for ICP control.Adv Neurosurg 1993; 21:173–177

48. Patel HC, Menon DK, Tebbs S, et al: Special-ist neurocritical care and outcome from headinjury. Intensive Care Med 2002; 28:547–553

49. Ritter AM, Muizelaar JP, Barnes T, et al:Brain stem blood flow, pupillary response,and outcome in patients with severe headinjuries. Neurosurgery 1999; 44:941–948

50. Peterson B, Khanna S, Fisher B, et al: Pro-longed hypernatremia controls elevated in-tracranial pressure in head-injured pediatricpatients. Crit Care Med 2000; 28:1136–1143


Chapter 4. Threshold for treatment of intracranial hypertension

I. RECOMMENDATIONS

Strength of Recommendation: Weak.Quality of Evidence: Low, from poor-

quality class III studies.

A. Level I


B. Level II


C. Level III

Treatment of intracranial pressure(ICP) may be considered at a threshold of20 mm Hg.


III. OVERVIEW

In children with severe traumatic braininjury (TBI), mortality is often the result ofa refractory increase in ICP. Furthermore,the need to prevent raised ICP is recog-nized as central to current neurocriticalcare of children after severe TBI. Manage-ment of severe TBI in the pediatric inten-sive care unit is largely focused on the man-agement of raised ICP and preservation ofcerebral perfusion pressure (CPP). Brief in-creases in ICP that return to normal in �5mins may be insignificant; however, sus-tained increases of �20 mm Hg for �5mins likely warrant treatment (1). Based inlarge part on studies in adults, an ICP treat-ment threshold of 20 mm Hg has been usedin most centers for decades. However, theoptimal ICP target or targets for pediatricTBI remain to be defined. Normal valuesfor mean arterial blood pressure and henceCPP are lower in children, particularly ininfants and young children. It has also beenshown in anesthetized children withoutTBI that the lower CPP limit of autoregu-lation of cerebral blood flow is, surpris-

ingly, similar in young children vs. olderchildren—and does not decrease below ap-proximately 60 mm Hg (2). Thus, youngchildren have less autoregulatory reservethan older children—i.e., the difference inCPP between normal and the lower limit ofautoregulation is smaller in infants andyoung children than it is in older children.This suggests the possible need to set alower ICP therapeutic target for infants andyoung children than older children oradults with TBI. As shown in the “ScientificFoundation” section, most of the evidencespecific to pediatrics supports an ICPthreshold of 20 mm Hg; however, individ-ual reports do support lower ICP thresholds(as low as 15 mm Hg). However, somepediatric studies suggest higher thresholds(35 or even 40 mm Hg). Thus, although anICP threshold of 20 mm Hg is generallyused, and even lower threshold may phys-iologically make sense for infants andyoung children, the optimal threshold forICP-directed therapy and whether or not itshould be adjusted for children of differentages deserves additional investigation. Itshould also be recognized that some of thestudies defining the ICP threshold usedtherapies that are not contemporary suchas aggressive hyperventilation. Finally, inlight of the heterogeneity of the pathologyand pathophysiology in pediatric TBI, ICPmanagement may need to be individualizedin some cases.

IV. PROCESS

For this update, MEDLINE wassearched from 1996 through 2010 (Ap-pendix B for search strategy), and resultswere supplemented with literature rec-ommended by peers or identified fromreference lists. Of 60 potentially relevantstudies, nine were added to the existingtable and used as evidence for this topic.


Eleven poor-quality class III studiesmet the inclusion criteria for this topicand provide evidence to support the rec-ommendations (3–13).

No prospective or retrospective stud-ies were identified that specifically com-pared the effect of ICP-directed therapy

on outcome (either short or long-term)using two (or more) predefined thresh-olds in pediatric TBI. One study examinedthis issue within the context of a random-ized controlled trial (9).

We are thus left with various studies,both prospective and retrospective, that ex-amined the association between outcomeand different ICP thresholds in patientswho, for the most part (5 of 13) (3–5, 8, 11),were managed with a therapeutic goal of�20 mm Hg for ICP-directed therapy. Ofthe 11 studies in the evidence table, onestudy used an ICP treatment threshold of15 mm Hg (10) and another specificallyused CPP rather than ICP treatmentthresholds to guide treatment (13). Onestudy described target “ranges” for the ICPthreshold including 15–25 mm Hg and20–25 mm Hg (12), and one used an age-dependent ICP treatment threshold rang-ing from 15 mm Hg in infants to 20 mmHg in older children (9). Thus, it must berecognized that most of the studies in thisevidence table have an inherent bias—ICP�20 mm Hg was the a priori therapeutictarget for some or all of the patients. Inaddition to this limitation, statistical ap-proaches to adjust for confounding vari-ables in examining the association betweenICP and outcome were variably used. An-other important limitation of these studiesis that there was no consistent approach toassess the relationship with outcome be-tween either the time of assessment of ICPafter TBI or the duration of time ICP wasabove a given threshold value. Generallymean or peak values or ICP values within agiven epoch were used.

Although defining a safe ICP thresholdhas proved elusive, all but one of the 11 stud-ies report that sustained intracranial hyper-tension is associated with mortality or pooroutcome in children after severe TBI.

A study by Pfenninger et al (4) retro-spectively reviewed 24 patients with se-vere TBI. The stated goal of the treatmentwas “to maintain ICP �20 mm Hg andabolish ICP elevations that were �25–30mm Hg that lasted for �3 min.” Thetreatment regimen that was used in-cluded severe hyperventilation (PaCO2

25–30 mm Hg), fluid restriction, mildhypothermia (rectal temperature 35.5–


DOI: 10.1097/PCC.0b013e31823f4424



Reference Study Description Data Class, Quality, and Reasonsa Results and Conclusion


Esparza et al, 1985 (3) Design: single-center retrospective reviewN � 56Age: 3 months to 14 yrsTreatment threshold set at ICP �20

mm HgProtocol: Treatment regimen not

contemporary and included severehyperventilation and dexamethasone


confounders; unclear if outcomeassessment was unbiased

No statistical comparison madebetween groups

13 of 13 (100%) patients with ICP �40 mmHg had poor outcome (severe disability,vegetative, or dead) and all the patientswith poor outcome died

4 of 14 (approximately 28%) patients withICP 20–40 mm Hg had poor outcome

2 of 29 (approximately 7%) patients with ICP0–20 mm Hg had poor outcome

Pfenninger et al, 1983 (4) Design: single-center retrospective reviewN � 24Age: 3 month to 14 yrsGCS: �7Protocol: Treatment threshold was defined

to maintain ICP � 20 mm Hg, abolishICP �25–30 mm Hg (sustained for �3mins) and maintain CPP �50 mm Hg


confounders

ICP �40 mm Hg was associated with highermortality (p � .001)

13 of 16 patients with ICP 20–40 mm Hg hadgood outcome or moderate disability; threeof 3 patients with ICP �20 mm Hg hadgood outcome or moderate disability

Studies from other chaptersof previous guidelines

Alberico et al, 1987 (5) Design: single-center, prospective,observational study

N � 100Age: 0–19 yrsGCS: �7Protocol: treatment threshold set at 20

mm HgTreatment regimen included severe

hyperventilationOutcome: GOS


confounders

70% good outcome GOS in children with ICP�20 mm Hg with treatment vs. 8% goodoutcome in children with ICP refractory totreatment (�20 mm Hg), p � .05

Chambers et al, 2001 (6) Design: single-center retrospective studyN � 84Age: 3 months to 16 yrsProtocol: the ICP threshold for treatment

and the specific treatments used werenot provided

Data recordings were made for a median of41 hrs; ROC curves calculated todetermine threshold value for ICP andCPP; for ICP, the ROC curves werecreated over 1-mm Hg intervals over therange of 0–90 mm Hg

Outcome: dichotomized 6-month GOS


confounders; unclear if patientselection was unbiased

Overall, thresholds of 35 mm Hg for ICP and45 mm Hg for CPP were the bestpredictors of outcome

The ROC-defined cutoffs varied depending onthe Marshall computed tomographyclassification and ranged from 21 mm Hgto 59 mm Hg

Downard et al, 2000 (7) Design: two-center retrospective studyN � 118Age: �15 yrsGCS: �8 in 99 patients (84%)Protocol: no standard treatment protocol;

sedation, mannitol, hyperventilation,vasopressors, ventriculostomy, ordecompressive craniectomy used at thediscretion of the treating physician

Class IIIPoor quality as an intervention

studyModerate quality as a prognosis

study: logistic regressionperformed to determine factorsassociated with GOS; but nocomparison of groups based onany intervention

In a stepwise logistic regression analysis,mean ICP �20 mm Hg in the initial 48hrs was significantly associated with anincreased risk of death

It was not indicated whether or not other ICPthresholds were investigate

Kassof et al, 1988 (8) Design: single-center, retrospective,observational study

N � 25Age: 3 months to 17 yrsProtocol: treatment threshold set at

20 mm HgTreatment regimen not contemporary and

included severe hyperventilation anddexamethasone


confounders; unclear if patientsselection was unbiased

Mean of peak ICP in patients whodied (N � 5) was 81 mm Hg (range, 55–120mm Hg); in contrast, mean of peak ICPwas 18.7 mm Hg (range, 10–30 mm Hg) inpatients who did not require additionaltreatment for ICP and there were nodeaths; no statistical analysis waspresented



Reference Study Description Data Class, Quality, and Reasonsa Results and Conclusion

New studiesAdelson et al, 2005 (9) Design: prospective multicenter randomized

controlled trial of moderate hypothermiavs. normothermia plus medicalmanagement

N � 47Age: �13 yrsProtocol: ICP treatment threshold varied

with age and included 15 mm Hg, 18mm Hg, or 20 mm Hg, for children 0–24 months, 25–96 months, and 97–156months, respectively

Contemporary guidelines-based treatmentregimen including randomized use ofmoderate hypothermia

Outcome: post hoc analysis of relationshipbetween ICP and outcome (3- and6-month GOS)


confounders in ICP analysis

Mean ICP was lower in children withgood (11.9 � 4.7 mm Hg) vs.poor (24.9 � 26.3 mm Hg, p � .05)outcome; the percent time with ICP �20mm Hg differed significantly in thegood (90.8% � 10.8%) vs.poor (68.6% � 35.0%, p � .05) outcomegroups

ICP �20 mm Hg was the most sensitive andspecific for poor outcome

Cruz et al, 2002 (10) Design: single-center prospective studyN � 45; children with ICP �15 mm Hg were

excludedAge: 1–12 yrsProtocol: the ICP threshold for treatment

was � 15 mm HgTreatment protocol to maintain normalized

ICP, CPP, and cerebral oxygenextraction, includedsedation, mannitol, severehyperventilation (PaCO2 �30 mmHg), barbiturates, and decompression for ICP�25 mm Hg

Outcome: 6-month GOS


confounders

ICP peaked on day 4 in both groupsICP was significantly higher (p � .02) on

days 2–5 in children with unfavorable vs.favorable outcomes

Daily mean ICP values ranged between 15and 21 mm Hg on days 2–5 in thefavorable outcome group and between 19and 26 mm Hg on days 2–5 in theunfavorable outcome group

Uncontrolled ICP �40 mm Hg occurred inthe two children who died

82% of the patients had a favorable outcome

Grinkeviciute et al,2008 (11)

Design: single-center prospective studyN � 48Age: 2.4 months to 18 yrsProtocol: treatment threshold defined as

ICP �20 mm HgPatients were treated according to ICP-

targeted protocol of the management ofsevere pediatric traumatic brain injury

27.1% of patients with decompressivecraniectomy

Outcome: 6-month dichotomized GOS


confounders. Insufficient powerto detect outcome

The survival rate was remarkably high at97.9% for children admitted to thepediatric intensive care unit

Differences in peak ICP (22.2 mm Hg vs. 24.6 mmHg, respectively) in groups with favorable vs.unfavorable outcomes were not statisticallysignificant; also no difference was seen betweengroups in minimal CPP

Only 5 patients were described as having pooroutcome

Pfenninger and Santi,2002 (12)

Design: single-center retrospectiveobservational study

N � 51 of whom 26 underwent ICPmonitoring and critical caremanagement

Age: 1 month to 16 yrsProtocol: an ICP target of 20–25 mm Hg

was usedICP-directed therapy included diuretics,

hypertonic saline, hyperventilation,pressors, and barbiturate coma

Outcome: dichotomized 6- to 12-monthGOS


confounders, potential selectionbias in children who receivedICP monitoring

Mean sustained ICP � 20 mm Hg wasassociated with poor outcome (p � .05)

White et al, 2001 (13) Design: single-center retrospectiveobservational study

N � 136; 37 of these patients underwentICP monitoring

Age: 0–17 yrsProtocol: an ICP target was not identified,

but a CPP target of � 50 mm Hg ininfants and � 70 mm Hg for childrenwas used

A contemporary treatment regimen wasused


confounders for ICP analysis,potential selection bias inpatients who received ICPmonitoring

14% of survivors and 41% of nonsurvivorshad ICP �20 mm Hg in the first 72 hrs;no other threshold was specificallyexamined. ICP maximum and ICPmeasured 6, 12, and 24 hrs after admissionwere all significantly lower in survivors

ICP, intracranial pressure; GCS, Glasgow Coma Scale; CPP, cerebral perfusion pressure; GOS, Glasgow Outcome Scale; ROC, receiver operatingcharacteristic.

aNo study provided data for a comparison between specific ICP thresholds for initiation of therapy on outcome.


36.5°C), dexamethasone, and barbiturateinfusion for refractory ICP �20 mm Hg.Sustained ICP �40 mm Hg was associatedwith death (p � .001). Thirteen of 16 pa-tients with sustained ICP 20–40 mm Hghad a good outcome or moderate disability.The three patients with ICP �20 mm Hghad a good outcome or moderate disability.

A study by Esparza et al (3) was aretrospective review of 56 pediatric chil-dren with severe TBI. The study includedtwo victims of abusive head trauma.Treatment threshold was also ICP �20mm Hg. The treatment regimen usedagain was not contemporary in that itincluded severe hyperventilation anddexamethasone. The group of patientswith an ICP �40 mm Hg had a mortalityrate of 100%, those with ICP �20–40mm Hg had a mortality rate of 28%,whereas those with ICP 0–20 mm Hg hadan incidence of poor outcome (severe dis-ability, vegetative, or dead) of only ap-proximately 7%, supporting use of an ICPtreatment threshold �20 mm Hg.

A study by Alberico et al (5) was car-ried out as a prospective, observationalstudy of 100 children (age range, 0–19yrs) with severe TBI using an ICP treat-ment threshold again of 20 mm Hg. Thetreatment regimen once again includedsevere hyperventilation, limiting some-what the ability to generalize the findingsto current treatment. Patients with ICPmaintained �20 mm Hg had a 70% goodoutcome (Glasgow Outcome Scale) incontrast to those with refractory intracra-nial hypertension who had only an ap-proximately 8% good outcome. Thisstudy also supports an ICP treatmentthreshold of 20 mm Hg.

A retrospective, observational study byKasoff et al (8) reported on data from 25children (ages 3 months to 17 yrs). Thepatients included one victim of abusivehead trauma. An ICP treatment thresholdwas again set at 20 mm Hg along with aCPP threshold of 40 mm Hg. The treat-ment regimen again was not contempo-rary and included both severe hyperven-tilation and dexamethasone. Only alimited amount of data on the relation-ship between ICP and outcome was pre-sented, although it was clear from thestudy that severe refractory ICP was as-sociated with poor outcome. The mean ofpeak ICP in patients who died (n � 5) was81 mm Hg (range, 55–120 mm Hg),whereas the mean of peak ICP was 18.7mm Hg (range, 10–30 mm Hg) and therewere no deaths in patients who did notrequire ICP-directed therapies. However,

no statistical analysis was performed.Given the practicalities associated withclinical use of a threshold, for the pur-pose of making guideline recommenda-tions, we have categorized this study assupporting a threshold of 20 mm Hg.

A study by Downard et al (7) was aretrospective observational study that in-cluded two level I trauma centers in thestate of Oregon, the Oregon Health Sci-ences University trauma registry and theLegacy Emanuel Hospital and HealthCenter. A total of 118 children �15 yrs ofage were included. Glasgow Coma Scalescore was �8 in 99 patients (84%). Nostandard treatment protocol was de-scribed; therapies including sedation,mannitol, hyperventilation, vasopressors,ventriculostomy, or decompressive crani-ectomy were used at the discretion of thetreating physician. It was not indicatedthat an ICP of 20 mm Hg was the treat-ment threshold; however, it was the onlyICP threshold that was included in thelogistic regression. A stepwise logistic re-gression analysis showed that a mean ICP�20 mm Hg in the initial 48 hrs wassignificantly associated with an increasedrisk of death with an odds ratio of 2.39. Itwas not indicated whether or not otherICP thresholds were investigated. A CPP�40 mm Hg was also associated withpoor outcome.

A study by Chambers et al (6) retro-spectively reviewed data on 84 childrenwith severe TBI and receiver operatingcharacteristic curves were calculated todetermine threshold values for CPP andICP. ICP treatment thresholds and spe-cific therapies used were not specified.Data recordings were made for a medianof 41.2 hrs. Using receiver operatingcharacteristic curves, an ICP threshold of35 mm Hg was determined to correlatebest with Glasgow Outcome Scale at 6months. The receiver operating charac-teristic-defined cutoffs varied greatly de-pending on the Marshall computed to-mography classification. Specifically, fortypes I, II, and III diffuse injury, the ICPcutoffs were 21 mm Hg, 24 mm Hg, and33 mm Hg, respectively. For evacuatedand unevacuated mass lesion categories,the cutoffs were 40 mm Hg and 59 mmHg, respectively. Thus, this report sup-ports an ICP treatment threshold of 35mm Hg and also suggests that optimalICP thresholds may differ for differentcomputed tomography classifications.However, caution is advised given thatthe sample size in the various subgroupsis limited, the study was retrospective,

and biological plausibility for some of thethresholds is questionable. Nevertheless,this study raises the question as towhether or not ICP thresholds should bedifferent in children with diffuse injuryvs. focal contusion.

A retrospecitve study by White et al(13) reported on 136 children with severeTBI (Glasgow Coma Scale score �8) ofwhom 37 were managed with ICP moni-toring. A CPP-directed protocol with tar-gets of �50 mm Hg in infants and �70mm Hg in children, rather than a specificICP target, was used to direct therapy. Acontemporary approach to treatment wasused with only mild or no hyperventila-tion along with sedation, osmolar ther-apy, barbiturates, and vasopressors. Hy-pothermia was not used. They reportedthat 41% of nonsurvivors vs. 14% of thesurvivors had ICP �20 mm Hg in thefirst 72 hrs. No other ICP threshold wasspecifically examined. The highest re-corded ICP value was 26 � 18 mm Hg vs.59 � 33 mm Hg in survivors vs. nonsur-vivors (p � .03). ICP values measured 6,12, and 24 hrs after admission were allsignificantly lower in survivors (19 � 29,18 � 18, 16 � 24 mm Hg) vs. nonsurvi-vors (43 � 27, 45 � 27, 43 � 34 mm Hg),respectively. It is not clear why only 27%of children with severe TBI received ICPmonitoring in this study. Nevertheless,they revealed an association between ICP�20 mm Hg and mortality in a cohort ofpatients that were not specifically treatedusing an ICP target of 20 mm Hg.

A study by Cruz et al (10) reporteddata from 45 children with severe TBIprospectively studied using a unique pro-tocol that targeted an ICP threshold of�15 mm Hg along with prevention ofincreased cerebral oxygen extraction as asurrogate marker of cerebral ischemia as-sessed with a jugular venous bulb cathe-ter. The treatment protocol included se-dation, fast high-dose mannitol (0.7–1.2g/kg), barbiturates, and surgical decom-pression for refractory ICP �25 mm Hg.Outcome was defined using dichotomized6-month Glasgow Outcome Scale score.ICP peaked on day 4 in both groups andwas significantly (p � .05) higher on days2–5 in children with unfavorable vs. fa-vorable outcomes (6-month GlasgowOutcome Scale). Daily mean ICP valuesranged between 15 and 21 mm Hg ondays 2–5 in the favorable outcome groupand between 19 and 26 on days 2–5 in theunfavorable outcome group. Uncon-trolled ICP �40 mm Hg occurred in thetwo children who died. This article is


unique in using an ICP treatment thresh-old of 15 mm Hg across the age spectrumin pediatric TBI, although they did notnecessarily achieve that target. Neverthe-less, using that threshold for treatment,favorable outcome was seen in tier anal-ysis at a value of �19 mm Hg. Given thepracticalities associated with clinical use ofa numerical threshold, for the purpose ofmaking guideline recommendations, wehave categorized this study as supporting athreshold of 20 mm Hg rather 19 mm Hg.

A study by Pfenninger and Santi (12)retrospectively reviewed data from 51children with severe TBI and compared itwith data from two historical cohorts(1994–1998 vs. 1978–83 and 1988–92).Within the more contemporary cohort,51% of the children underwent ICP mon-itoring. Nonmonitored patients were notsalvageable (n � 5), underwent immedi-ate decompressive craniectomy for earlydeterioration (n � 2), or underwent jug-ular bulb venous saturation monitoringinstead (n � 17). ICP-directed therapyincluded diuretics, hypertonic saline, hy-perventilation, and barbiturate coma tar-geting an ICP of 20–25 mm Hg. Neitherdecompressive craniectomy nor hypo-thermia was used to control ICP. Meansustained ICP �20 mm Hg was associ-ated with poor outcome (p � .05) definedas a dichotomized 6- to 12-month Glas-gow Outcome Scale score. Favorable out-come was observed in all eight childrenwith maximum mean sustained ICP �20mm Hg, eight of 15 with ICP 20–40 mmHg, and one of three children with ICP�40 mm Hg. This study is complex inthat the treatment threshold appeared tobe a range of 20–25 mm Hg and suggestsand ICP threshold of �20 mm Hg asso-ciated with a favorable outcome.

A study by Adelson et al (9) was aprospective multicentered randomizedcontrolled trial of moderate hypothermiavs. normothermia plus medical manage-ment in 47 children �13 yrs of age withsevere TBI. The study was unique in thatthe ICP treatment threshold varied withage using 15 mm Hg, 18 mm Hg, or 20mm Hg for children 0–24 months, 25–96months, and 97–156 months, respec-tively. Although this study was rated levelII for assessment of effect of hypother-mia, it was rated level III as evidencepertaining to ICP threshold. A contempo-rary guidelines-based treatment regimenwas used that also included randomizedtreatment with or without moderate hy-pothermia. Post hoc analysis of the rela-tionship between ICP and outcome (3-

and 6-month Glasgow Outcome Scale)was carried out with the expressed pur-pose (stated in the text) of addressing agap in the pediatric TBI guidelines. Thatanalysis examined the association be-tween outcome and ICP from 0 to 90 mmHg in children treated with this specificregimen. Mean ICP was lower in childrenwith good (11.9 � 4.7 mm Hg) vs. poor(24.9 � 26.3 mm Hg, p � .05) outcome.The percent time with ICP �20 mm Hgdiffered significantly in the good (90.8% �10.8%) vs. poor (68.6% � 35.0%, p �.05) outcome groups. ICP �20 mm Hgwas the most sensitive and specific forpoor outcome. Based on these findings,this study supports an ICP treatmentthreshold of 20 mm Hg. However, itshould be recognized that hypothermiacould represent an important confounderin the report. This study also representsthe only study in the evidence table toguide ICP-directed treatment with age,although the impact of these age-depen-dent thresholds on outcome within thethree age categories was not assessed.

A prospective study by Grinkeviciuteet al (11) of 48 children with severe TBIwho underwent ICP monitoring and ICP-directed therapy at a target of 20 mm Hgusing a contemporary therapeutic regi-men included decompressive craniec-tomy in �27% of cases. The survival ratewas remarkably high at 97.9% for chil-dren admitted to the pediatric intensivecare unit, although the total denomina-tor for all severe TBI victims presentingto the emergency department was notprovided. Surprisingly, differences inpeak ICP (22.2 mm Hg vs. 24.6 mm Hg,respectively) in groups with favorable andunfavorable outcomes (6-month dichoto-mized Glasgow Outcome Scale score)were not statistically significant. Simi-larly, no difference between outcomegroups was seen for minimal CPP. How-ever, despite the fact that many patientshad ICP �25 mm Hg, �90% of the pa-tients had a favorable outcome and thisstudy included only five patients with apoor outcome; thus, the statistical powerto examine the relationship of raised ICPacross outcomes was limited. In addition,the use of peak ICP could also limit datainterpretation.

Only class III studies are available andalthough the studies support several dif-ferent thresholds for ICP treatment,given that eight of the 11 class III studiessupported a threshold of approximately20 mm Hg, that level represents the moststrongly supported value for ICP and thus

is the threshold supported as a level IIIrecommendation. We recognize that ad-ditional studies in pediatric patients withTBI are needed to determine the optimalICP threshold or thresholds for infantsand children and also define whether ornot the threshold is dependent on age,injury mechanism, computed tomogra-phy injury pattern, location of the mon-itor, and/or other factors.



There is level II evidence that treat-ment should be initiated at an ICPthreshold of 20 mm Hg as stated in therecommendations section of the adultguidelines document (14). In addition, inthe summary section of the adult guide-lines, it is stated that current data support20–25 mm Hg as an upper threshold toinitiate treatment. There are no large ran-domized trials in adults that directly com-pare different ICP treatment thresholds.

In a study by Marmarou et al (15), 428patients with severe TBI were prospec-tively analyzed for monitoring parame-ters that determined outcome and theirICP threshold values. Using logistic re-gression, the threshold value of 20 mmHg best correlated with 6-month GlasgowOutcome Scale score. The proportion ofhourly ICP reading �20 mm Hg was asignificant independent determinant ofoutcome. There are small, noncontrolledstudies that suggest a range of 15–25 mmHg. In one of these studies, Saul andDucker (16) changed the ICP thresholdfrom 25 to 15 mm Hg in two sequentiallytreated groups of patients and found adecrease in mortality from 46% to 28%.In the same study by Chambers et al (4)(see evidence table; Table 1) for which weused data from pediatric patients as evi-dence for this document, 207 adult pa-tients were also assessed. They had ICPand CPP monitoring, and receiver oper-ating characteristic curves were used todetermine whether there were significantthresholds for the determination of out-come. The sensitivity for ICP rose forvalues �10 mm Hg, but it was only 61%at 30 mm Hg. In a smaller prospectivestudy by Ratanalert et al (17) of 27 pa-tients grouped into ICP treatmentthresholds of 20 or 25 mm Hg, there wasno difference in outcome between this


narrow range of treatment threshold. Fi-nally, in a report by Schreiber et al (18) of233 patients with ICP monitoring analyzedprospectively, an opening ICP �15 mm Hgwas identified as one of five risk factorsassociated with a higher mortality rate.

Any chosen ICP threshold must beclosely and repeatedly corroborated withthe clinical examination and computedtomography imaging in an individual pa-tient because pupillary abnormalities oc-curred in patients with ICP values as lowas 18 mm Hg (19).

In addition, the critical value of ICPand its interaction with CPP and withother measures (jugular venous oxygensaturation, partial pressure of brain tis-sue oxygen, cerebral blood flow) remainsunknown. The adult guidelines concludethat because the importance of theseother parameters is recognized, the abso-lute value of ICP may be less important(14). The relationship between partialpressure of brain tissue oxygen and ICPin children has only recently begun to beexplored (20).

VII. SUMMARY

There is evidence (eight of 11 class IIIstudies) that sustained elevations in ICP(�20 mm Hg) are associated with pooroutcome in children after severe TBI, andthus the level III recommendation. Whatis not well established is the absolutetarget for ICP-directed therapy that isneeded to maximize outcome since thiswas not specifically addressed prospec-tively in any of the studies reviewed. Al-though one of these studies was carriedout in the setting of a randomized con-trolled trial, no randomized controlledtrial has directly compared the effectof two or more thresholds for ICP-directed therapy on outcome in pediatricTBI. There are also individual poor-quality level III studies that support ei-ther lower (a range of 15–25 mm Hg) orhigher (35 or 40 mm Hg) threshold val-ues than 20 mm Hg, although thresholds�20 mm Hg do, as discussed previously,have theoretical support for infants andyoung children. Finally, based on the factthat normal values of blood pressure andICP are age-dependent, it is anticipatedthat the optimal ICP treatment thresholdmay be age-dependent. However, data onthis point are extremely limited; only a

single study on this topic in children thatmet the inclusion criteria varied the ICPtreatment threshold with age using 15mm Hg, 18 mm Hg, or 20 mm Hg forchildren 0–24 months, 25–96 months,and 97–156 months, respectively (9).


● A direct comparison of two specific ICPtreatment thresholds on outcome in chil-dren, particularly values �20 mm Hg.

● Investigation to determine whetherthreshold values for ICP-directed ther-apy are age-dependent.

● Determination whether or not injurymechanism (e.g., abusive head trauma)or computed tomography patternchanges the optimal ICP treatmentthreshold.

● Examination of physiological and bio-chemical surrogates (e.g., microdialy-sis, partial pressure of brain tissue ox-ygen, pressure volume index) ofoutcome are needed either to comple-ment or supplant ICP-directed therapyin children.

● Assessment as to whether the treat-ment threshold for ICP-directed ther-apy changes with either time after in-jury or duration of intracranialhypertension.

● Investigations that better define therelative value of ICP- vs. CPP-directedtherapy in pediatric TBI.

REFERENCES

1. McLaughlin MR, Marion DW: Cerebral bloodflow and vasoresponsivity within and aroundcerebral contusions. J Neurosurg 1996; 85:871–876

2. Vavilala MS, Lee LA, Lam AM: The lowerlimit of cerebral autoregulation in childrenduring sevoflurane anesthesia. J NeurosurgAnesthesiol 2003; 15:307–312

3. Esparza J, M-Portillo J, Sarabia M, et al.Outcome in children with severe head inju-ries. Childs Nerv Syst 1985; 1:109–114

4. Pfenninger J, Kaiser G, Lutschg J, et al: Treat-ment and outcome of the severely head injuredchild. Intensive Care Med 1983; 9:13–16

5. Alberico AM, Ward JD, Choi SC, et al: Out-come after severe head injury. Relationshipto mass lesions, diffuse injury, and ICPcourse in pediatric and adult patients. J Neu-rosurg 1987; 67:648–656

6. Chambers IR, Treadwell L, Mendelow AD: De-termination of threshold levels of cerebral per-fusion pressure and intracranial pressure in

severe head injury by using receiver-operatingcharacteristic curves: An observational study in291 patients. J Neurosurg 2001; 94:412–416


8. Kasoff SS, Lansen TA, Holder D, et al: Aggres-sive physiologic monitoring of pediatric headtrauma patients with elevated intracranialpressure. Pediatr Neurosci 1988; 14:241–249


10. Cruz J, Nakayama P, Imamura JH, et al:Cerebral extraction of oxygen and intracra-nial hypertension in severe, acute, pediatricbrain trauma: Preliminary novel manage-ment strategies. Neurosurgery 2002; 50:774–779; discussion 779–780

11. Grinkeviciute DE, Kevalas R, Matukevicius A,et al: Significance of intracranial pressureand cerebral perfusion pressure in severe pe-diatric traumatic brain injury. Medicina(Kaunas, Lithuania) 2008; 44:119–125

12. Pfenninger J, Santi A: Severe traumatic braininjury in children—Are the results improv-ing? Swiss Med Wkly 2002; 132:116–120

13. White JR, Farukhi Z, Bull C, et al: Predictorsof outcome in severely head-injured chil-dren. Crit Care Med 2001; 29:534–540

14. Bratton SL, Chestnut RM, Ghajar J, et al:Guidelines for the management of severetraumatic brain injury. VIII. Intracranialpressure thresholds. J Neurotrauma 2007;24(Suppl 1):S55–58

15. Marmarou A, Anderson RL, Ward JD, et al:Impact of ICP instability and hypotension onoutcome in patients with severe headtrauma. J Neurosurg 1991; 75:S59–S66

16. Saul TG, Ducker TB: Effect of intracranialpressure monitoring and aggressive treat-ment on mortality in severe head injury.J Neurosurg 1982; 56:498–503

17. Ratanalert S, Phuenpathom N, Saeheng S, etal: ICP threshold in CPP management ofsevere head injury patients. Surg Neurol2004; 61:429–434; discussion 434–435

18. Schreiber MA, Aoki N, Scott BG, et al: Deter-minants of mortality in patients with severeblunt head injury. Arch Surg 2002; 137:285–290

19. Marshall LF, Barba D, Toole BM, et al: Theoval pupil: Clinical significance and relation-ship to intracranial hypertension. J Neuro-surg 1983; 58:566–568



Chapter 5. Cerebral perfusion pressure thresholds

I. RECOMMENDATIONS

Strength of Recommendations: Weak.Quality of Evidence: Low, from poor-


A. Level I


B. Level II


C. Level III

A minimum cerebral perfusion pres-sure (CPP) of 40 mm Hg may be consid-ered in children with traumatic brain in-jury (TBI).

A CPP threshold 40–50 mm Hg maybe considered. There may be age-specificthresholds with infants at the lower endand adolescents at the upper end of thisrange.


III. OVERVIEW

Global or regional cerebral ischemia is animportant secondary insult to the acutelyinjured brain. CPP, as defined by meanarterial pressure (MAP) minus the meanintracranial pressure (ICP), is the pres-sure gradient driving cerebral blood flow,which, in turn, in the normal state isautoregulated and coupled with cerebralmetabolic rate for oxygen. Autoregula-tion and coupling between cerebral bloodflow and cerebral metabolic rate for oxy-gen may be disrupted in the brain afterTBI, and a decrease in CPP may thereforeinduce cerebral ischemia. With the use ofcontinuous monitoring capabilities in-cluding invasive blood pressure and ICPequipment, the CPP could be manipu-lated by treatment in an attempt to avoidboth regional and global ischemia. The

optimal CPP threshold and therapeuticapproach to achieve it both remain to bedefined.

There are age-related differences inMAP, cerebral blood flow, and cerebralmetabolic rate for oxygen from infancythrough to adulthood. Because pediatricvalues are, in the main, lower thanadult values, we need to know whetherthere are age-specific thresholds or tar-gets for CPP that should be used duringthe critical care management of pediat-ric severe TBI.

Cerebral perfusion pressure is rela-tively easy to measure. The main reasonfor undertaking the invasive monitoringrequired for calculating this number is totitrate treatment using the level of eachof the constituent parameters as a guide(i.e., CPP, ICP, and MAP) (1, 2). There arethree main limitations in comparing CPPdata from various studies for the purposeof identifying whether low CPP is harm-ful or whether there is an age-related“critical threshold” that should be tar-geted in treatment.

First, there may be a problem with themeasurement of CPP, particularly whenit is not standardized. Theoretically, tocalculate actual CPP both MAP and ICPneed to be zero-calibrated to the samelevel. Intraparenchymal fiber-tip sensorsmeasure ICP at the tip of the device inrelation to atmospheric pressure and noadjustment is possible. When using othertypes of devices, it is common practice tocalibrate blood pressure to the rightatrium and ICP to the level of the fora-men of Monro. The calculation of CPPwill underestimate actual CPP by an errorproportional to the distance between thetwo zeroing points multiplied by thesine of the angle of bed elevation. For agiven bed elevation, this error increaseswith increasing size of the patient, andfor a given size of child, this error in-creases with increasing bed elevation.From first principles, across the pediat-ric age range, at bed elevation of 30°,adolescents will have almost double theerror of infants (11 vs. 6 mm Hg). At agiven size, increasing the bed elevationby 30° will double the error when ado-lescents are compared with infants (10

vs. 5 mm Hg). The studies included asevidence for this chapter describe prac-tice in children covering the full pedi-atric age range. Bed elevation is onlydescribed in two studies: at 0 –30° ele-vation (3) and 15–30° elevation (4). TheICP monitoring devices that were usedare not described in two studies (5, 6)and three studies used cerebral intrapa-renchymal monitoring (4, 7, 8). Thereference levels for zero calibration ofICP or blood pressure are not describedin any of the reports.

Second, the real-time numerical valueof CPP not only reflects intracranial tis-sue and fluid dynamics, but also the CPPlevel that is being targeted by those at thebedside. Four studies do not describe anyICP- or CPP-directed strategy in theirmanagement (5, 9–11). One study usedan ICP threshold of 20 mm Hg to directtherapy (7). The other four studies usedan age-related scale in threshold for CPP-directed intervention. In two studies, thelower limit of the scale that was used was40 mm Hg (6, 8), and in the other twostudies, it was 45 mm Hg (3, 4). Theupper threshold in the scale was 50 –70mm Hg. It is evident from these datathat low level of CPP will, therefore,also indicate failure to achieve the CPPtarget as well as a failure to respond totreatment. Table 2 provides a summaryof the targets and treatments used ineach of the studies included in the ev-idence table.

Third, the CPP summary statistic thatis used in the analysis is different in manyof the studies (Table 2). Minimum or low-est CPP during monitoring is used in fourstudies (5, 6, 9, 11). The other five studiesreport mean CPP: as an initial value (4),average in the first 24 hrs and daily for 5days (3), or as an average for the wholeperiod of monitoring (7, 8, 10). Anotherimportant consideration in regard to thesummary statistic is whether or not pre-terminal data in nonsurvivors were in-cluded. Only one report describes exclud-ing preterminal data (10); the rest of thereports do not discuss whether these dataare included or excluded.

Taken together, caution should be ap-plied when interpreting the results from


DOI: 10.1097/PCC.0b013e31823f4450



Reference Study Description Data Class, Quality, and Reasons Results and Conclusion

Studies from previous guidelinesBarzilay et al, 1988 (9) Design: retrospective case series with analysis

of minimum CPPN � 56Age: mean age 6.2 yrs; 41 with severe TBI, 5

with central nervous system infection, and10 miscellaneous conditions

GCS: 54 cases with GCS �8Purpose: patients were treated for increased

or decreased CPPProtocol: CPP management protocol was not

specifiedOutcome: survival at hospital discharge


confounders

Among 41 patients with severe TBI:CPP was 65.5 � 8.5 mm Hg for survivors vs.

6.0 � 3.9 mm Hg for nonsurvivors (p � .01)

Downard et al, 2000 (12) Design: retrospective case series with analysisof mean hourly CPP calculation in thefirst 48 hrs of care

N � 118Age: mean 7.4 � 4.6 yrsGCS: mean GCS 6 � 3 (99 severe cases), 50%

with space-occupying lesionsProtocol: intracranial pressure monitors

established within 24 hrs of admissionOutcome: last recorded GOS in records

at � 3 months, and dichotomized to“good” and “poor” outcomes

Class IIIModerate quality: outcome

assessment methods not clearlydescribed, otherwise met allcriteria

All children with mean CPP �40 mm Hgdied

No significant difference in GOS when meanCPP was divided into deciles from 40 to�70 mm Hg

More patients had a good outcome thanpoor outcome when mean CPP was �50mm Hg, but there was no analysis of thisin the publication

Kaiser and Pfenninger, 1984 (10) Design: retrospective case series with analysisof minimum CPP

N � 24Age: mean 6.3 yrsGCS: all with GCS �8, 21.5% with

intracranial hemorrhageProtocol: CPP management included

intubation, hyperventilation, control ofbody temperature, dexamethasone,barbiturates, and osmotic agents

Outcome: GOS follow-up at mean 2.5yrs (range, 1.5–4.4 yrs) after injury

Class IIIPoor quality: unclear if selection

methods unbiased; no controlfor confounders

All survivors (N � 19) had minimum CPP�50 mm Hg; 3 of the 5 children whodied also had CPP �50 mg Hg

New studiesAdelson et al, 2005 (3) Design: randomized controlled trial of

hypothermia therapy with analysis ofaverage CPP over the first 5 days of care

N � 102Age: �17 yrs (mean age in two-part study

6.89 and 6.95 yrs)CPP management goal was targeted by age

using 45–50 mm Hg, 50–55 mm Hg, and55–60 mm Hg for children aged 0–24months, 25–96 months, and 97–156months (first cohort) or 97–214 (secondcohort), respectively

Protocol: not specifiedOutcome: GOS was dichotomized at 6

months after injury


confounders in CPP analysis(for hypothermia, this is a class II

study)

Mean CPP on day 1 was higher in thehypothermia group (70.75 mm Hg) thanthe normothermia group (64.84 mmHg), p � .037

There were no statistically significantdifferences between groups on days2 to 5, and GOS was not assessed inrelation to differences in CPP on day 1

Average CPP was 69.19 � 11.96 mm Hg forfavorable vs. 56.37 � 20.82 mm Hg forunfavorable (p � .0004) outcome groups;the percent time with CPP �50 mm Hgwas 94.2% � 16.9% for favorable vs.87.3% � 29.5% for unfavorable (p �.0001)

Barlow et al, 1999 (11) Design: retrospective case series with analysisof lowest CPP

N � 17Age: 1–20 months (mean, 5.1 months) with

inflicted TBIProtocol: not specified; increased intracranial

pressure and decreased CPP were treatedin all cases

Outcome: a 6-point outcome scale assessed3–122 months (mean, 33 months)postinjury


confounders; unclear if selectionand outcome assessmentmeasures were unbiased

Lowest CPP correlated with pooroutcome (p � .005)


Table 1. —Continued


Chaiwat et al, 2009 (5) Design: retrospective case seriesanalysis of lowest CPP in the first72 hrs after severe TBI; a Doppler-derived cerebral blood flowautoregulatory index was alsostudied and calculated as percentchange in estimatedcerebrovascular resistance perpercent change in CPP

N � 36 patients (2 inflicted TBI)Age: 9.1 � 5.3 yrs (range, 0.8–16 yrs)Protocol: when ICP was not

monitored, CPP or mean arterialblood pressure was increasedaccording to whichever followingvariable was greater: 1) 20% abovebaseline; or 2) a set value of 80mm Hg for the group �9 yrs and90 mm Hg for the group aged9–16 yrs, respectively

Outcome: GOS dichotomized at 6months after discharge

Class IIIModerate quality: the methods for

outcome were adequate andnonbiased but the adequacy ofthe sample size is unclear

On univariate analysis CPP �40 mm Hgduring the first 72 hrs had no associationwith poor outcome

When logistic regression was performed,using a number of factors, only impairedautoregulatory index remained anindependent predictor of poor outcome

Chambers et al, 2001 (13) Design: retrospective case series withanalysis of CPP

N � 84Age: 3 months to 16 yrs (median, 10

yrs)Outcome: GOS dichotomized at 6

months


confounders; unclear if patientselection was unbiased

Poor outcome in all 8 cases with CPP �40mm Hg; more patients had goodoutcome than poor outcome when meanCPP was �40 mm Hg

Figaji et al, 2009 (4) Design: prospective case series withanalysis of CPP data

N � 52Age: �15 yrs (median, �7 yrs)Protocol: patient management based

on treatment recommendations inprevious edition of the PediatricGuidelines; target values for CPPwere �50 mm Hg in children �2yrs old and �45 mm Hg inchildren �2 yrs old

Outcome: GOS was dichotomizedinto “favorable” and “unfavorable”outcome � 6 months after injury

Class IIIModerate quality: outcome

assessment methods not clearlydescribed, otherwise met allcriteria

Median (interquartile range) for lowest CPPwas significantly lower in unfavorableoutcome patients: 29 (20–45) mm Hg vs.44 (35–51) mm Hg, p � .023

Unfavorable outcome patients also had moreepisodes of CPP �40 mm Hg: 3 (0–10 vs.0 0–1), p � .03

There was no difference in the number ofepisodes of CPP �50 mm Hg

Kapapa et al, 2010 (6) Design: retrospective case series withanalysis of CPP in relation to age-specific lower limit (up to 1month, �40 mm Hg; 2 monthsup to 1 yr, �45 mm Hg; 1 yr upto 7 yrs, �50 mm Hg; �7 yrs, 55–60 mm Hg)

N � 16Age: 0–16 yrsGCS: �9Protocol: treatment algorithm

including CPP management wasused

Outcome: GOS was dichotomized atvaried times after injury

Class IIIPoor quality: small sample size

with inadequate case selectionand outcome measures; uncleardetails of the regression analysisreporting the relationshipbetween CPP and outcome

Patients with CPP value below the age-specific lower limit for just a singleoccurrence had a significantly worseoutcome (p � .013)

Narotam et al, 2006 (7) Design: prospective case series withanalysis of mean CPP

N � 16Age: 1.5–18 yrs (mean, 14 yrs)GCS: 3–12 (mean, 5; 15 cases were

severe)Protocol: patients were managed for

prevention of cerebral ischemiawith ventilation, vasopressors,respiratory treatments, etc.

Outcome: GOS at 3 months afterinjury


confounders for GOS analysis;unclear if selection andoutcome assessment methodsunbiased

All survivors had good outcome; mean CPPwas 81.52 � 16.1 mm Hg for survivorsvs. 50.33 � 31.7 mm Hg fornonsurvivors (p � .033)


the pediatric TBI CPP studies and apply-ing the information to treatment strate-gies for TBI.

IV. PROCESS

For this update, MEDLINE wassearched from 1996 through 2010 (Ap-pendix B for search strategy), and resultswere supplemented with literature rec-ommended by peers or identified fromreference lists. Of 77 potentially relevantstudies, eight were added to the existingtable and used as evidence for this topic.


Three moderate-quality class III stud-ies and eight poor-quality class III studiesabout CPP met the inclusion criteria forthis topic and provide evidence to supportthe recommendations (3–13).

A randomized controlled trial of hypo-thermia (32–33°C) therapy, a class IIstudy for the evidence about hypother-mia, but class III for the evidence aboutCPP, reported average CPP over the first5 days of care as well as for the total 5days of care (3). The study was performedin two parts: part 1, 48 cases of pediatric

TBI with Glasgow Coma Scale score �8,aged 6.89 � 3.46 yrs; and part 2, 27 casesof pediatric TBI with Glasgow Coma Scalescore �8, aged 6.95 � 5.68 yrs. The au-thors used dichotomized Glasgow Out-come Scale outcome (good in 28 cases,14 hypothermia patients and 14 normo-thermia patients; poor in 40 cases, 18hypothermia patients and 22 normother-mia patients) assessed at 6 months afterinjury to examine differences in CPP. Theaverage CPP for all 5 days was higher inthe good outcome group: good outcome69.19 � 11.96 mm Hg vs. poor outcome56.37 � 20.82 mm Hg (p � .0004). In



Stiefel et al, 2006 (8) Design: retrospective case series with analysisof mean daily CPP

N � 6Age: 6–16 yrsProtocol: treatment targeted age-appropriate

CPP (�40 mm Hg)Outcome: GOS was dichotomized at

discharge


confounders, very small sample,unclear if selection methodsunbiased

Mean daily CPP in survivors was75.63 � 11.73 mm Hg

CPP, cerebral perfusion pressure; TBI, traumatic brain injury; GCS, Glasgow Coma Scale; GOS, Glasgow Outcome Scale.

Table 2. Summary of treatments, cerebral perfusion pressure target, and cerebral perfusion pressure statistics used in studies

Reference

Treatments Used

CPP TargetStrategy CPP StatisticHyperventilation

InducedHypothermia Barbs

DecompressiveCraniectomy

Adelson et al, 2005(3)

No Yes Yes Yes Age-related 45mm Hg

Mean CPP

Barlow et al, 1999(11)

TH TH TH TH TH Lowest CPP

Barzilay et al, 1988(9)

Yes Yes Yes No — Lowest CPP

Chaiwat et al, 2009(5)

— — — — — Lowest CPP

Chambers et al,2001 (13)

TH TH TH TH TH Minimum CPP

Downard et al,2000 (12)

Yes No No Yes — Mean CPP 48 hrs

Figaji et al, 2009(4)

Yes Yes Yes Yes Age-related 45mm Hg

Initial and lowestCPP

Kaiser andPfenninger,1984 (10)

Yes Yes Yes No — Mean CPP

Kapapa et al, 2010(6)

Yes Yes Yes Yes Age-related 40mm Hg

Lowest CPP

Narotam et al,2006 (7)

Yes No No No Intracranialpressure-related

Mean CPP

Stiefel et al, 2006(8)

— — — — Age-related 40mm Hg

Mean CPP

CPP, cerebral perfusion pressure.In the studies that describe therapy: “Yes” denotes use of therapy and ”No” denotes where treatment is not used. “TH” denotes where the study is aimed

at defining a threshold about burden from CPP insult and outcome rather than it being an intervention study. Dashes (—) indicate where no informationis given in the report.


addition, the percent time with CPP �50mm Hg was higher in the good outcomegroup: good outcome 94.2% � 16.9% vs.poor outcome 87.3% � 29.5% (p �.0001).

Five studies found higher CPP associ-ated with better outcomes and their find-ings were as follows. A study by Barzilayet al (9) studied 41 consecutive TBI ad-missions to their pediatric intensive careunit with coma for at least 6 hrs beforeadmission. Survivors had higher mini-mum CPP than nonsurvivors: 65.5 � 8.5vs. 6.0 � 3.9 mm Hg, respectively (p �.01). A study by Figaji et al (4) studiedprospectively 52 children with TBI andfound median lowest CPP experiencedduring the course of monitoring washigher in those with better outcome. Byusing dichotomized Glasgow OutcomeScale outcome assessed at least 6 monthsafter injury, those with favorable out-come had lowest CPP median (interquar-tile range) of 44 (35–51) mm Hg vs. 29(20–45) mm Hg in those with an unfa-vorable outcome (p � .023). A study byNarotam et al (7) analyzed data from 16children aged 1.5–18 yrs (mean, 14 yrs),15 of whom had Glasgow Coma Scalescore �8. All ten survivors had an excel-lent recovery at 3 months (Glasgow Out-come Scale score 5). Mean CPP washigher in survivors (81.52 � 16.1 mmHg) than nonsurvivors (50.33 � 31.7 mmHg, p � .033). A study by Stiefel et al (8)studied brain tissue oxygen monitoringin six patients (aged, 6–14 yrs, GlasgowComa Scale score 3–7) and found meandaily CPP in the five survivors was75.63 � 11.73 mm Hg. Last, in a sampleof TBI cases restricted to 17 young chil-dren with inflicted injury (aged 1–20months; mean, 5.1 months), Barlow et al(11) reported that higher lowest CPP dur-ing intensive care was associated withbetter outcomes in a 6-point scale 3–122months (mean, 33 months) after injury(p � .0047).

Four studies reported findings in rela-tion to a threshold in CPP of 40 mm Hg.In the study reported by Figaji et al (4)(see previously), the authors found thatmore episodes of CPP �40 mm Hg wereobserved in those with an unfavorable (3[0–10]) vs. favorable (0 [0–1]) outcome(p � .03). In the other study, a morecomplex relationship between CPP andoutcome involved data from autoregula-tion of cerebral blood flow. A study byChambers et al (13) analyzed 84 childrenaged 3 months to 16 yrs (median, 10 yrs)and examined minimum CPP in relation

to dichotomized Glasgow Outcome Scaleat 6 months. Sixty-three of 76 cases withCPP �40 mm Hg had good outcome andall eight cases with CPP �40 mm Hg hada poor outcome (p � .0001, Fisher’s exacttest). A study by Downard et al (12) ana-lyzed 118 pediatric TBI cases aged up to15 yrs (mean age, 7.4 yrs; 99 cases withGlasgow Coma Scale score 3–8) and re-ported dichotomized Glasgow OutcomeScale score at 3 months or later in rela-tion to CPP thresholds. Seventy-two of 96patients with CPP �40 mm Hg had agood outcome, whereas all 22 cases withCPP �40 mm Hg died. The difference inmortality was statistically significant(p � .0001, Fisher’s exact test). A studyby Chaiwat et al (5) analyzed 36 cases ofTBI for predictors of poor outcome. ICPof �20 mm Hg and CPP �40 mm Hgduring the first 72 hrs were not associ-ated with outcome. However, on logisticregression, an estimate of impaired cere-bral blood flow autoregulation usingDoppler ultrasonography—the autoregu-latory index—was an independent predic-tor of poor outcome (adjusted odds ratio,23.1; 95% confidence interval, 1.9 –279.0). Impaired autoregulatory indexwas an independent risk factor when theauthors entered CPP �40 mm Hg, sys-tolic blood pressure lower than the fifthpercentile for age and gender during thefirst 72 hrs after TBI, low middle cerebralartery velocity, and impaired autoregula-tory index into the model (adjusted oddsratio, 29.8; 95% confidence interval, 1.7–521.4). Because autoregulatory index iscalculated as the percent change in cere-brovascular resistance per percentchange in CPP, and cerebrovascular re-sistance is defined as the ratio of CPP tomiddle cerebral artery velocity, it is im-possible to disentangle the relationshipbetween outcome and CPP. Autoregula-tory index represents a research tool.

Five class III studies contain data con-cerning CPP threshold �40 mm Hg. Tworetrospective case series support the ideathat there may be an age-related CPPthreshold �40 mm Hg. A study by Ka-papa et al (6) analyzed 16 children aged�16 yrs and reported dichotomized Glas-gow Outcome Scale in relation to age-specific lower limits in CPP (i.e., �40mm Hg, infants up to 1 month; �45 mmHg, infants aged 2 months to 1 yr; �50mm Hg, children aged between 1 and 7yrs; 55–60 mm Hg, children aged �7yrs). The authors found that patients withCPP values below the age-specific lowerlimit for just a single occurrence had a

significantly worse outcome (p � .013). Astudy by Kaiser and Pfenninger (10) re-ported findings in 24 consecutive admis-sions to their pediatric intensive care unitof patients with a Glasgow Coma Scalescore �8, average age � 6.3 yrs (tenpatients between 1 and 5 yrs) and showedthat all survivors had CPP �50 mm Hg(p � .005, Fisher’s exact test). The tworemaining studies in this group of fourdid not observe a threshold �40 mm Hg.In the study reported by Figaji et al (4)(see previously), the authors also re-ported outcome in relation to the num-ber of episodes during monitoring thatCPP was �50 mm Hg; there was no dif-ference in the number episodes in thosewith an unfavorable (8 [2–18.5]) vs. fa-vorable (3 [0–8.8]) outcome (p � .137).Of note, two-thirds of the children in thisseries were �8 yrs. As discussed, in thestudy reported by Downard et al (12),100% of children with mean CPP �40mm Hg died as compared with only 25%of children who had a CPP �40 mm Hg.The difference in mortality was statisti-cally significant (p � .0001, Fisher’s exacttest). Last, in the study of young childrenwith inflicted TBI reported by Barlow etal (11) (see previously), only one infant inthe series of 17 had the lowest CPP of�50 mm Hg.

These studies, in aggregate, suggestthat in the pediatric age range, there maybe an age-related threshold between 40and 50 mm Hg with infants at the lowerend and adolescents at the upper end ofthis range. Finally, studies specifically fo-cused on assessment of the optimal upperlimit for CPP management in pediatricTBI were lacking.



In adults, with respect to CPP, it ap-pears that the critical threshold for cere-bral ischemia generally lies in the regionof 50–60 mm Hg and can be furtherdelineated in individual patients by ancil-lary monitoring (14). It is becoming in-creasingly apparent that elevating theCPP through pressors and volume expan-sion is associated with serious systemictoxicity, may be incongruent with fre-quently encountered intracranial condi-tions, and is not clearly associated withany benefit in terms of general outcome.


A study by Clifton et al (15) was a posthoc analysis of the data on CPP withinthe data set from 392 patients in therandomized controlled trial of therapeu-tic hypothermia for severe TBI. Whenthey analyzed individual predictive vari-ables separately, they found CPP of �60mm Hg to be associated with an in-creased proportion of patients with pooroutcome. They found similar associationsfor ICP �25 mm Hg, MAP �70 mm Hg,and fluid balance ��594 mL. Whenthese variables were combined into astepwise logistic regression model, how-ever, low CPP had no effect on outcome,although the other three variables re-mained within the group of most power-ful variables in determining outcome.Based on a purely pragmatic assessmentof these data, the authors noted that aCPP target threshold should be set ap-proximately 10 mm Hg above what isdetermined to be a critical threshold toavoid dips below the critical level (15). Theoverall assessment of the adult CPP guide-lines therefore suggests “a general thresh-old in the realm of 60 mm Hg, with furtherfine-tuning in individual patients based onmonitoring of cerebral oxygenation andmetabolism and assessment of the status ofpressure autoregulation” (14).

The adult guidelines state that thereare insufficient data to support a level Irecommendation for this topic. Under“Options,” it states the following: aggres-sive attempts to maintain CPP �70 mmHg with fluids and pressors should beavoided because of the risk of adult respi-ratory distress syndrome; CPP of �50mm Hg should be avoided; and CPP val-ues to target lie within the range of50–70 mm Hg. Patients with intact pres-sure autoregulation tolerate higher CPPvalues and ancillary monitoring of cere-bral parameters including blood flow, ox-ygenation, or metabolism may facilitateCPP management.

VII. SUMMARY

Survivors of severe pediatric TBI un-dergoing ICP monitoring consistentlyhave higher CPP values vs. nonsurvivors,but no study demonstrates that activemaintenance of CPP above any targetthreshold in pediatric TBI reduces mor-tality or morbidity. In comparing the

findings from pediatric and adult TBIstudies, there does appear to be an age-related difference in CPP threshold.Whether these differences are the resultof differences in measurements, goal inCPP management, or the makeup in agerange of the small numbers in the pedi-atric studies remains unclear. CPPshould be determined in a standard fash-ion with ICP zeroed to the tragus (as anindicator of the foramen of Monro andmidventricular level) and MAP zeroed tothe right atrium with the head of the bedelevated 30°.


● A standard method for measuring CPPlevel and duration and reporting datawould be useful across pediatric TBIstudies that focus on targeting CPP.

● Multimodal neuromonitoring studiesto help determine the relationships be-tween CPP and autoregulation and be-tween CPP and ischemia in individualpatients.

● Controlled, prospective, randomizedstudies in children to determine opti-mal level of CPP based on ischemiamonitoring in various pediatric agegroups and mechanisms of injury.

● Long-term (�1 yr), age-appropriatefunctional outcome studies to assessthe relative importance of ICP- andCPP-targeted therapies as well as anal-yses evaluating outcomes in relation totreatment responders and nonre-sponders.

● Studies to determine whether a CPPtarget threshold set above (e.g., 10 mmHg) what is determined to be a criticalthreshold could avoid dips below thecritical CPP level.

REFERENCES

1. Adelson PD, Bratton SL, Carney NA, et al:Guidelines for the acute medical manage-ment of severe traumatic brain injury in in-fants, children, and adolescents. Chapter 7.Intracranial pressure monitoring technol-ogy. Pediatr Crit Care Med 2003; 4:S28–S30

2. Adelson PD, Bratton SL, Carney NA, et al:Guidelines for the acute medical manage-ment of severe traumatic brain injury in in-fants, children, and adolescents. Chapter 4.Resuscitation of blood pressure and oxygen-ation and prehospital brain-specific therapies

for the severe pediatric traumatic brain in-jury patient. Pediatr Crit Care Med 2003;4:S12–S18



5. Chaiwat O, Sharma D, Udomphorn Y, et al:Cerebral hemodynamic predictors of poor6-month Glasgow Outcome Score in severepediatric traumatic brain injury. J Neu-rotrauma 2009; 26:657–663

6. Kapapa T, Konig K, Pfister U, et al: Headtrauma in children, part 2: Course and dis-charge with outcome. J Child Neurol 2010;25:274–283

7. Narotam PK, Burjonrappa SC, Raynor SC, etal: Cerebral oxygenation in major pediatrictrauma: Its relevance to trauma severity andoutcome. J Pediatr Surg 2006; 41:505–513

8. Stiefel MF, Udoetuk JD, Storm PB, et al:Brain tissue oxygen monitoring in pediatricpatients with severe traumatic brain injury.J Neurosurg 2006; 105:281–286

9. Barzilay Z, Augarten A, Sagy M, et al: Vari-ables affecting outcome from severe braininjury in children. Intensive Care Med 1988;14:417–421

10. Kaiser G, Pfenninger J: Effect of neurointen-sive care upon outcome following severehead injuries in childhood—A preliminaryreport. Neuropediatrics 1984; 15:68–75

11. Barlow KM, Minns RA: The relation betweenintracranial pressure and outcome in non-accidental head injury. Dev Med Child Neu-rol 1999; 41:220–225


13. Chambers IR, Treadwell L, Mendelow AD:Determination of threshold levels of cerebralperfusion pressure and intracranial pressurein severe head injury by using receiver-operating characteristic curves: An observa-tional study in 291 patients. J Neurosurg2001; 94:412–416

14. Bratton SL, Chestnut RM, Ghajar J, et al:Guidelines for the management of severetraumatic brain injury. IX. Cerebral perfu-sion thresholds. J Neurotrauma 2007;24(Suppl 1):S59–S64

15. Clifton GL, Miller ER, Choi SC, et al: Fluidthresholds and outcome from severe braininjury. Crit Care Med 2002; 30:739–745


Chapter 6. Advanced neuromonitoring

I. RECOMMENDATIONS

Strength of Recommendation: Weak.Quality: Low, from one moderate- and

one poor-quality class III study.

A. Level I


B. Level II


C. Level III

If brain oxygenation monitoring isused, maintenance of partial pressure ofbrain tissue oxygen (PbtO2) �10 mm Hgmay be considered.


III. OVERVIEW

Children with severe traumatic brain in-jury (TBI) frequently have abnormal cere-bral hemodynamics, including intracranialhypertension, cerebral hypoxia, delayedand/or altered processing of electrophysiolog-ical signals, and impaired cerebral autoregu-lation. In addition to intracranial pressure(ICP) monitoring, advanced neuromonitor-ing techniques such as microdialysis, elec-trophysiological assessments, and examina-tion of cerebral autoregulation may helpidentify and treat patients with these de-rangements after TBI. The development ofadvanced monitoring systems to provideinformation regarding both cerebrovascu-lar and metabolic function after TBI iscritical to providing optimal neurocriticalcare. If treatment preventing unwantedcerebral pathophysiological processes isshown to improve outcome in childrenwith severe TBI, the use of monitoringsystems, beyond ICP monitoring, willmark an important advance in the care ofpatients with TBI. Advanced neuromoni-

tors may provide useful informationabout derangements in cerebral oxygen-ation, blood flow and metabolism, auto-regulation, and function after severe pe-diatric TBI.

IV. PROCESS

For this new topic, MEDLINE wassearched from 1950 through 2010 (Ap-pendix B for search strategy), and resultswere supplemented with literature rec-ommended by peers or identified fromreference lists. Of 44 potentially relevantstudies, two were included as evidence forthis topic.


Two class III publications met the in-clusion criteria for this topic and provideevidence to support the recommenda-tions (1, 2). The recommendations on theuse of advanced neuromonitoring in thischapter are for patients with no contra-indications for neuromonitoring such ascoagulopathy (brain oxygenation) and forpatients who do not have a diagnosis ofbrain death.

In 2009, a study by Figaji et al (1)reported the relationship between PbtO2

and long-term outcome in 52 childrenwith severe TBI. Patients with compro-mised PbtO2 were treated to a threshold�20 mm Hg. Overall mortality wasnearly 10%. After considering other con-ventional predictors, authors reportedthat PbtO2 �5 mm Hg for �1 hr or �10mm Hg for �2 hrs were associated with asignificantly increased risk of unfavorableoutcome (Glasgow Outcome Scale andPediatric Cerebral Performance Categoryscores) and mortality, independent ofother factors that were also significant(e.g., ICP, cerebral perfusion pressure,Glasgow Come Scale, computed tomog-raphy classification, and systemic hyp-oxia). This study provided no comparisongroup. All patients with compromisedPbtO2 were treated to maintain the tar-geted threshold, and at the same timethey may have received various treat-ments depending on other physiologicalvariables such as ICP, cerebral perfusionpressure, systemic oxygen, and hemoglo-

bin. What can be inferred is that in thissample of patients, those with higherPbtO2 and fewer episodes of PbtO2 �10mm Hg had better outcomes. We cannotsay that this relationship is a direct re-sponse to treatment.

In 2006, a study by Narotam et al (2)described changes in PbtO2 in relation tochanges in cerebral perfusion pressure,FIO2, and PaO2 in 15 children rangingfrom 1.5 to 18 yrs and Glasgow ComaScale score �8. Like with the previousstudy, patients were managed to main-tain a PbtO2 level �20 mm Hg. In addi-tion, the authors aimed to assess a treat-ment protocol (Critical Care Guide) formanipulation of physiological factors thatinfluence oxygen delivery to the brain.Survival was associated with normal ini-tial PbtO2 (�10 mm Hg). There was nodifference in the mean initial PbtO2

among the ten survivors and six deaths at3 months. Final PbtO2 in survivors washigher than that in nonsurvivors (meanPbtO2, 22.7 � 9.05 vs. 7.2 � 7.85 mm Hg;p � .0045). However, only six patientshad elevated ICP, making the relation-ship between ICP and PbtO2 difficult tointerpret. Like with the previous study,we cannot infer from this study that re-sponse to treatment influenced outcome.

In these two studies, a treatmentthreshold for PbtO2 of 20 mm Hg wasused; however, they both reported an as-sociation between unfavorable outcomeand PbtO2 �10 mm Hg. Although thestudy by Figaji et al (1) reported an evenstronger association between PbtO2 �5mm Hg and unfavorable outcome, untilproven otherwise, if this advanced moni-toring modality is used, it would be pru-dent to target the more conservativethreshold of �10 mm Hg.


Several articles on advanced neu-romonitoring in the pediatric TBI litera-ture were identified in the search butexcluded from the evidentiary table be-cause they simply described use of a givenadvanced neuromonitoring device ratherthan targeting a treatment value for thatmonitor (i.e., a threshold parameter on


DOI: 10.1097/PCC.0b013e31823f659a


the advanced monitoring device was notspecifically manipulated). Given that thisguidelines document is focused on treat-ment, for these reports, a treatment rec-ommendation regarding the monitoringdevice could not be given. The devices inthose studies included brain microdialysis(3), cerebral blood flow and autoregulationmonitors (4–7), signal processing of hemo-dynamic and hydrostatic signals (8), andjugular venous oxygen saturation monitor-ing (9).


Evidence from the adult guidelines(10) supported a level III recommenda-tion for use of jugular venous saturationand PbtO2 monitoring, in addition tostandard ICP monitors, in the manage-ment of adults with severe TBI. Evidencesuggests that episodes of jugular venousdesaturation (saturation �50%) are asso-ciated with poor outcome and that thisvalue represents a treatment thresholdwhen using this monitoring technique.Similarly, low values of PbtO2 (�15 mm

Hg) and the extent of their duration (�30mins) are associated with high rates ofmortality and that 15 mm Hg representsa treatment threshold value for PbtO2.However, the accuracy of jugular venoussaturation and PbtO2 monitoring was notevaluated. Although many technologies in-cluding cerebral microdialysis, thermal dif-fusion probes, transcranial Doppler, andnear-infrared spectroscopy were recognizedto hold promise in advancing the care ofadults with severe TBI, there was insuffi-cient evidence to comment on the use ofthese advanced neuromonitors in this pop-ulation.

VII. SUMMARY

Overall, advanced neuromonitors havebeen subjected to very limited clinicalinvestigation in pediatric TBI, particu-larly study of their use specifically toguide therapy. Most of the medical liter-ature on these agents is composed of ob-servational studies on relatively smallnumbers and case series receiving someform of local standard TBI care. The lackof sufficient high-quality pediatric stud-

ies limits the conclusions that can bemade and differences between studycenters in the treatment of TBI andinpatient populations limit the general-izability of findings.


● Examine critical thresholds for eachneuromonitoring modality and de-termine the risk-benefit ratio, cost-effectiveness, comparative effective-ness, and impact of neuromonitorson patient long-term functional out-comes.

● Address issues of single vs. multi-modal neuromonitoring, reliabilityof technology, optimal combinationof monitors, location of neuromoni-tor vs. site of injury (hemispheric,pericontusional), relationship be-tween neuromonitor data and imag-ing data, neuromonitor use for opti-mization of treatment and patient



New studiesFigaji et al, 2009 (1) Design: prospective cohort

N � 52Age: 6.5 � 3.4 yrs (9 months to 14 yrs)Protocol: treatment protocol was used

in patients with compromised PbtO2

to manage to a threshold�20 mm Hg

Purpose: to examine the relationshipbetween factors, including PbtO2,

and outcomeOutcome: mortality; 6 months Glasgow

Outcome Scale score and PediatricCerebral Performance Category

Class IIIModerate quality: unclear if outcome

assessment was unbiased

PbtO2 �5 mm Hg for �1 hr or PbtO2 �10mm Hg for �2 hrs were independentlyassociated with higher risk ofunfavorable outcome defined as severedisability or death (adjusted oddsratio, 27.4; 95% confidence interval,1.9–391), independent of othersignificant factors such as intracranialpressure, computed tomography, lowPaO2, and cerebral perfusion pressure

PbtO2 �5 mm Hg for �1 hr or PbtO2 �10mm Hg for �2 hrs were independentlyassociated with mortality (adjusted oddsratio 26.8; 95% confidence interval,2.7–265)

Narotam et al,2006 (2)

Design: prospective case seriesN � 16Age: 14 yrs (range, 1.5–18 yrs)Glasgow Coma Scale: 3–12; 15

children had Glasgow ComaScale � 8

Protocol: patients with low PbtO2 weremanaged to a threshold�20 mm Hg

Purpose: to direct treatment based oninitial PbtO2 and to examine theeffect of a critical care guide to treatlow oxygen delivery

Outcome: 3-month mortality

Class IIIPoor quality: unclear if sample

selection was unbiased; unclear ifoutcome assessment was unbiased;no control for confounders formortality outcome

None of the patients with normal initialPbtO2 (�10 mm Hg) died

There was no difference in the mean initialPbtO2 among the 10 survivors and 6deaths (measured at 3months) (16.07 � 18.7 vs. 6.76 � 6.69mm Hg, p � .247)

Final PbtO2 in survivors was higher thanthat in nonsurvivors (meanPbtO2, 25.0 � 11.57 vs. 8.53 � 11.0 mmHg; p � .01)

PbtO2, partial pressure of brain tissue oxygen.


prognosis as well as optimal durationof advanced monitoring.

● Evaluate the role of advanced neu-romonitoring on clinical decision-making and patient outcomes.

● Develop additional bedside and non-invasive advanced neuromonitors.

REFERENCES


2. Narotam PK, Burjonrappa SC, Raynor SC, etal: Cerebral oxygenation in major pediatrictrauma: its relevance to trauma severity andoutcome. J Pediatr Surg 2006; 41:505–513

3. Tolias CM, Richards DA, Bowery NG, et al:Extracellular glutamate in the brains of chil-dren with severe head injuries: A pilot mi-crodialysis study. Childs Nerv Syst 2002; 18:368–374

4. Brady KM, Shaffner DH, Lee JK, et al: Con-tinuous monitoring of cerebrovascular pres-sure reactivity after traumatic brain injury inchildren. Pediatrics 2009; 124:e1205–1212

5. Tontisirin N, Armstead W, Waitayawinyu P,et al: Change in cerebral autoregulation as afunction of time in children after severe trau-matic brain injury: A case series. Childs NervSyst 2007; 23:1163–1169

6. Vavilala MS, Tontisirin N, Udomphorn Y, etal: Hemispheric differences in cerebral auto-regulation in children with moderate andsevere traumatic brain injury. NeurocritCare 2008; 9:45–54

7. Vavilala MS, Muangman S, Waitayawinyu P,

et al: Neurointensive care; impaired cerebralautoregulation in infants and young childrenearly after inflicted traumatic brain injury: Apreliminary report. J Neurotrauma 2007; 24:87–96

8. Shapiro K, Marmarou A: Clinical applicationsof the pressure–volume index in treatment ofpediatric head injuries. J Neurosurg 1982;56:819–825

9. Perez A, Minces PG, Schnitzler EJ, et al:Jugular venous oxygen saturation or arte-riovenous difference of lactate content andoutcome in children with severe traumaticbrain injury. Pediatr Crit Care Med 2003;4:33–38

10. Bratton SL, Chestnut RM, Ghajar J, et al:Guidelines for the management of severetraumatic brain injury. X. Brain oxygenmonitoring and thresholds. J Neurotrauma2007; 24(Suppl 1):S65–70


Chapter 7. Neuroimaging

I. RECOMMENDATIONS

Strength of Recommendation: Weak.Quality of Evidence: Low from one

poor-quality class III study.

A. Level I


B. Level II


C. Level III

In the absence of neurologic deterio-ration or increasing intracranial pressure(ICP), obtaining a routine repeat com-puted tomography (CT) scan �24 hrs af-ter the admission and initial follow-upstudy may not be indicated for decisionsabout neurosurgical intervention.


III. OVERVIEW

Early neuroimaging has assumed anincreasingly important role in evaluatingthe extent and severity of traumatic braininjury (TBI) in children (1). CT is impor-tant for the rapid detection of differenttypes of intracranial injury including ex-tra-axial hemorrhage (e.g., subdural orepidural hematomas), acute hydrocepha-lus, fractures, or other intracranial le-sions that may require acute neurosurgi-cal intervention. The early use of CT isalso useful for triage of patients to detectthose who are likely to need neurosur-gery, require management in an inten-sive care unit vs. general hospital settingas well as those who can be safely dis-charged from the emergency departmentand managed at home. Although mag-netic resonance imaging (MRI) sensitivityis understood to be superior to CT forintracranial evaluation, it is not as easily

obtained acutely after injury and has notbeen as widely validated in large studies,particularly regarding influence on man-agement decisions. At the current time,there is little evidence to support the useof MRI in influencing management of pa-tients with severe TBI.

It is understood that acute CT imagingis universally performed in the developedworld for patients with severe TBI. Twostudies (2, 3) show that children withsevere TBI have a high incidence of in-tracranial injury on CT scan (75% and62%, respectively). In these studies, in-tracranial injury included brain contu-sion, extracerebral hematoma, intracere-bral hematoma, diffuse axonal injury,acute brain swelling, penetrating cranio-cerebral injury, pneumocephalus, sub-arachnoid hemorrhage, alterations to cis-terns, midline shift, or fractures. Neitherstudy included treatment-related out-come data related to the findings on CTscan and thus could not be used as spe-cific evidence for this guideline.

Although CT is always obtainedacutely in patients with severe TBI, theuse of two or more CT studies is notagreed on. Repeating a CT scan in chil-dren with severe TBI is usually consid-ered when there is 1) no evidence of neu-rologic improvement; 2) persistent orincreasing ICP; or 3) an inability to assessneurologic status (e.g., sedation, paralyticagents) (4). Studies have reported delayedor progressive lesions in 1% to 50% ofadult/pediatric patients with TBI (5). Be-cause epidural hematoma/subdural he-matoma requiring surgical interventioncan develop hours to days after the acuteinjury, some investigators have suggestedthat a follow-up CT scan be routinelyacquired at 1–3 days postinjury evenwhen clinical deterioration is not evidentunder the assumption that early diagno-sis prompts early intervention leading toa better long-term outcome (4). However,because children with severe TBI aremedically unstable and (if portable CT isnot available) may further deteriorateduring transport to the CT scanner (he-modynamic instability, increased ICP, ox-ygen desaturation), the decision to ordera repeat scan is a treatment decision,

weighing the knowledge gained againstthe risk of additional secondary brain in-jury. Likewise, because of the long-termeffects of CT radiation exposure (lifetimerisk of fatal cancer resulting from onehead CT in a 1-yr-old child is as high asone in 1500), the neurosurgical decisionto order a CT scan also should be consid-ered a treatment decision, weighing theknowledge gained against the risk oflong-term radiation exposure (6). Thisguideline addressed the issue of the valueof routinely acquiring repeat CT scans inchildren with severe TBI.

IV. PROCESS

For this new topic, MEDLINE wassearched from 1950 through 2010 (Ap-pendix B for search strategy), and resultswere supplemented with literature rec-ommended by peers or identified fromreference lists. Of 120 potentially relevantstudies, one was included as evidence forthis topic.


One class III study met the inclusioncriteria for this topic and provides evidenceto support the recommendation (7).

A retrospective study of 40 childrenwith severe TBI (Glasgow Coma Scalescore �8, age 2 months to 17 yrs; US,January 1990 to December 2003) exam-ined whether serial CT scans led to ur-gent neurosurgical operative interven-tion (7). Entry criteria also included ICPmonitoring during hospitalization, nocraniotomy at admission to study, and atleast a second CT scan within the first 48hrs. One hundred fifteen serial CT scanswere ordered (76% routine follow-up,21% increased ICP; 3% neurologicchange). Results of these scans showedno change (53%), improvement (34%),and worsening (13%). Five (4.3%) pa-tients had a surgical intervention basedon the results of the serial CT scan (oneepidural hematoma, craniotomy; onesubdural hemorrhage, burr hole; threefor additional ventriculostomy place-ments). All five scans were ordered basedon a clinical indicator (ICP or neurologicstatus), not as routine follow-up. The au-

Copyright 2012 Brain Trauma Foundation

DOI: 10.1097/PCC.0b013e31823f65e2


thors recommended that a highly selec-tive approach to ordering serial CT scansshould be practiced with the understand-ing that only scans ordered for increasedICP or neurologic change are likely tolead to surgical interventions.


Several additional studies had data re-garding the value of acquiring a secondCT scan in children with severe TBI butnone had information regarding treat-ment-related outcomes and thereforecould not be included as evidence for thisguideline. One retrospective cohort studyof 521 pediatric patients with TBI whomet inclusion criteria from a total of8505 blunt trauma admissions (1994–2003) described the prevalence of wors-ening brain injury on repeat CT, predic-tors of worsening CT findings, and thefrequency of neurosurgical interventionafter the repeat CT (4). Potential predic-tors of worsening CT findings and neuro-surgical intervention were recorded bychart review. Logistic regression and re-cursive partitioning were used to identifypredictors. Patients were grouped intothree categories (moderate/severe, mild,all TBI). In the moderate/severe group(n � 252), 202 (80%, mean GlasgowComa Scale score 3.7) had severe and 50(20%, Glasgow Coma Scale score 10.5)had moderate injury. For children withsevere TBI, the multivariate adjustedodds ratio for worsening or new secondCT findings was 2.4 (95% confidence in-terval, 1.6–3.8). Children with moderate/severe head injuries, especially if they hadintracranial injury, were more likely tohave deteriorating CT findings (107 of248 [43%]) and of these children, 4%(n � 11) required surgery. In contrast,

141 (57%) had stable CT scans and only2% (n � 4) required surgery. In mostsurgical patients, repeat CT was precededby rapid decline in neurologic status orelevated ICP. Four clinical factors wereidentified for stratifying risk of worseningbrain injuries on repeat CT (normal ini-tial CT scan, abnormal initial CT scan,moderate or severe head injury by Glas-gow Coma Scale, and coagulopathy). Thismethod identified 100% of patients whounderwent surgery and 89% of patientswho had worsening brain injuries on re-peat CT.

Another retrospective study of 173consecutive children (ages 8 months to16 yrs; mean 7.1 yrs) with severe (83%)or moderate (17%) TBI (mean GlasgowComa Scale score of 6.8 � 2.1) assessedthe yield of a routine predetermined re-peat CT scan within 24–36 hrs (5). Forty-seven (27%) of the second CT scansshowed new lesions including six withintracranial hypertension, 17 cases ofworsening brain edema, and 18 newlydiagnosed brain contusions. None ofthese findings necessitated surgical inter-vention or any change in therapy. Of the67 patients who underwent a third CTscan, two cases required surgical inter-vention because of new findings on thethird CT. The authors stated that a sec-ond routine prescheduled head CT scanwithin 24–36 hrs after admission in pe-diatric patients with moderate to severehead trauma is unlikely to yield anychange in therapy. Clinically orientedand ICP-directed CT scans may better se-lect and diagnose patients who requirechanges in therapy, including surgery.

Another retrospective study of 351children with severe TBI who had two ormore CT scans within 72 hrs of admissionfound that 41% had delayed and progres-

sive lesions (3). The decision to repeat thescan was based on clinical judgment andalthough the morbidity and mortality ofthese patients were worse, the rate ofsurgical intervention or change in ther-apy after the second CT was not reported;hence, the yield of the imaging is un-known. Injury progression correlatedwith the severity of the initial headtrauma, presence of extracranial injury,and the presence of coagulopathy on ad-mission.

VII. SUMMARY

One study met the criteria for inclu-sion as evidence for this topic given thatwe required that publications about im-aging link the assessment to a treatmentdecision and the decision to an outcome.Our level III recommendation, based onone class III study, questions the use ofrepeat CT scans in the absence of neuro-logic deterioration or increasing ICP.


There is a dearth of information re-garding the use of neuroimaging in di-recting targeted therapies and stratifica-tion. Although MRI is being used morefrequently in the acute evaluation of chil-dren with TBI, particularly for suspectedabusive head trauma, most of the litera-ture is directed at evaluating diagnosticsensitivity or outcome prediction. It isalso known that advanced MRI tech-niques provide unique information aboutbrain function that is not available by CT,but it remains uncertain how this infor-mation can alter management or improvetreatment-related outcomes. Adult TBIliterature also suggests that patterns ofinjury on neuroimaging may be helpful



New studyFigg et al, 2006 (7) Design: case series

N � 40942 screenedAge: mean 9.6 yrs (SD 4.4)Glasgow Coma Scale score: Mean 5.1 (SD 1.5)Purpose: Examined whether serial computed

tomography scans lead to urgentneurosurgical operative intervention

Class IIIPoor quality: no control for potential

confounders

Serial scans after the admission and initialfollow-up study (N � 115 scans)showed: no change (53%), improvement(34%), or worsening (13%)

Five (4.3%) patients had a surgicalintervention based on findings from theserial computed tomography scans;however, all five scans were ordered asa result of clinical indicators(intracranial pressure or neurologicstatus), not as routine follow-up


for improving stratification of injury se-verity and therefore aid in selecting pa-tients for targeted treatment. Importantquestions to address are:

● Do patterns of injury (from findingsprovided by multimodality neuroimag-ing) improve accuracy of injury strati-fication?

● Does MRI provide added value to CT ininfluencing management of childrenwith severe TBI?

● What is the use of neuroimaging (CT,MRI, etc.) in directing targeted thera-pies and improving treatment-relatedoutcomes?

● What is the use of repeat neuroimagingin special settings, such as in patients

who cannot be examined or in the pres-ence of coagulopathy?

REFERENCES

1. Tong K, Oyoyo U, Holshouser B, et al: Chapter7: Evidence-based neuroimaging for trau-matic brain injury in children. In: Evidence-Based Imaging in Pediatrics. Medina L, Apple-gate K, Blackmore C (Eds). New York, NY,Springer, 2010, pp 85–102

2. Levi L, Guilburd JN, Linn S, et al: The asso-ciation between skull fracture, intracranial pa-thology and outcome in pediatric head injury.Br J Neurosurg 1991; 5:617–625

3. Stein SC, Spettell C, Young G, et al: Delayedand progressive brain injury in closed-head trauma: radiological demonstration.Neurosurgery 1993; 32:25–30; discussion30 –31

4. Hollingworth W, Vavilala MS, Jarvik JG, et al:The use of repeated head computed tomogra-phy in pediatric blunt head trauma: factorspredicting new and worsening brain injury.Pediatr Crit Care Med 2007; 8:348–356; CEUquiz 357

5. Tabori U, Kornecki A, Sofer S, et al: Repeatcomputed tomographic scan within 24–48hours of admission in children with moderateand severe head trauma. Crit Care Med 2000;28:840–844

6. Brenner D, Elliston C, Hall E, et al: Estimatedrisks of radiation-induced fatal cancer frompediatric CT. AJR Am J Roentgenol 2001; 176:289–296

7. Figg RE, Stouffer CW, Vander Kolk WE, et al:Clinical efficacy of serial computed tomo-graphic scanning in pediatric severe traumaticbrain injury. Pediatr Surg Int 2006; 22:215–218


Chapter 8. Hyperosmolar therapy

I. RECOMMENDATIONS

Strength of Recommendations: Weak.Quality of Evidence: Moderate, based

on two moderate-quality class II studiesand one poor-quality class III study.

A. Level I


B. Level II

Hypertonic saline should be consid-ered for the treatment of severe pediatrictraumatic brain injury (TBI) associatedwith intracranial hypertension. Effectivedoses for acute use range between 6.5 and10 mL/kg.

C. Level III*

Hypertonic saline should be consid-ered for the treatment of severe pediatricTBI associated with intracranial hyper-tension. Effective doses as a continuousinfusion of 3% saline range between 0.1and 1.0 mL/kg of body weight per houradministered on a sliding scale. The min-imum dose needed to maintain intracra-nial pressure (ICP) �20 mm Hg shouldbe used. Serum osmolarity should bemaintained �360 mOsm/L.

*Although mannitol is commonly used in themanagement of raised ICP in pediatric TBI, nostudies meeting inclusion criteria were identifiedfor use as evidence for this topic.


III. OVERVIEW

Hyperosmolar Therapy forIntracranial Hypertension

Intravenous administration of hyper-osmolar agents was shown to reduce ICPearly in the 20th century (1). A study byWise and Chater (2) introduced mannitolinto clinical use in 1961. Despite wide-spread use of a number of osmolar agents

(mannitol, urea, glycerol) up until thelate 1970s (2), mannitol gradually re-placed other hyperosmolar agents in themanagement of intracranial hyperten-sion. Subsequently, hypertonic saline wasintroduced and now both are used in con-temporary management of intracranialhypertension. In recent studies of hyperos-molar therapy use in pediatric TBI, euvol-emia rather than dehydration has been thegeneral therapeutic target based on fluidbalance and/or central venous pressuremonitoring, and a Foley catheter is rou-tinely used in these patients to quantifyurine output and avoid bladder rupture.

The use of hyperosmolar therapy inthe management of pediatric severe TBIis a topic in which there was investigationshortly before the 2003 pediatric guide-lines, notably studies focused on the useof hypertonic saline for raised ICP (3–5).However, since those guidelines, no newstudy on hyperosmolar therapy met theinclusion criteria for this guideline.

Mannitol

Mannitol is commonly used in themanagement of raised ICP in pediatricand adult TBI (6). In a practice survey inthe United Kingdom in 2001, it was re-ported to be used in 70% of pediatricintensive care units, and recently, even ininfants with severe TBI, mannitol wasreported to be the second most commontherapeutic intervention, surpassed onlyby intubation (7). Despite this fact, manni-tol has not been subjected to controlledclinical trials vs. placebo, other osmolaragents, or other therapies in children. Mostof the investigations on the use of mannitolhave focused on the treatment of adults(8–21). Either children were excluded orthe composition or outcome of the pediat-ric trial was not defined (8–24).

Mannitol can reduce ICP by two distinctmechanisms. Mannitol at 1 g/kg has beenshown to reduce ICP by reducing bloodviscosity. This effect is immediate and re-sults from a viscosity-mediated reflex vaso-constriction (intact autoregulation), whichallows cerebral blood flow to be maintaineddespite a reduced level of cerebral bloodvolume (17, 25–27). Thus, cerebral bloodvolume and ICP both decrease. The effect of

mannitol administration on blood viscosityis rapid but transient (�75 mins) (17).Mannitol administration also reduces ICPby an osmotic effect, which developsmore slowly (over 15–30 mins), as a re-sult of the gradual movement of waterfrom the brain parenchyma into the sys-temic circulation. The effect persists upto 6 hrs and requires an intact blood–brain barrier (28, 29). Mannitol may ac-cumulate in injured brain regions (30),where a reverse osmotic shift may occurwith fluid moving from the intravascularcompartment into the brain paren-chyma, possibly increasing ICP. Thisphenomenon has been suggested to oc-cur when mannitol is used for extendedperiods of time (31). The gap betweenserum and cerebrospinal fluid (CSF) os-molality decreased below baseline insome adult patients treated with man-nitol for �48 – 60 hrs (18). Mannitolpossesses antioxidant effects (32), butthe contribution of this mechanism toits overall efficacy is unclear.

Mannitol is excreted unchanged inurine, and a risk of the development ofacute tubular necrosis and renal failurehas been suggested with mannitol ad-ministration with serum osmolaritylevels �320 mOsm in adults (33–35).However, the literature supporting thisfinding is limited in scope and was gen-erated at a time when dehydration ther-apy was common. A euvolemic hyperos-molar state generally is targeted withcontemporary care.

Hypertonic Saline

In the initial description in 1919 of thereduction in ICP by intravenous admin-istration of hyperosmosal agents, hyper-tonic saline was the agent used (1). Itsuse in the treatment of increased ICP,however, failed to gain clinical accep-tance. Resurgence in interest in thistreatment resulted from the report ofWorthley et al (36), who described twocases in which hypertonic saline (smallvolumes of an extremely hypertonic solu-tion, approximately 29% saline) reducedrefractory ICP elevations. In the last de-cade, many have studied the use of smallvolume hypertonic saline in resuscitation


DOI: 10.1097/PCC.0b013e31823f6621


of hemorrhagic shock and polytraumawith or without TBI in experimentalmodels and in adult humans (37–42).However, the recent National Institutesof Health-funded resuscitation outcomesconsortium trial of hypertonic saline inTBI resuscitation in adults was stoppedfor futility after enrollment of 1073patients (43).

Like mannitol, the penetration of so-dium across the blood–brain barrier islow (39). Sodium thus shares both thefavorable rheologic and osmolar gradienteffects involved in the reduction in ICP byseveral theoretical beneficial effects in-cluding restoration of normal cellularresting membrane potential and cell vol-ume (44, 45), stimulation of arterial na-

triuretic peptide release (46), inhibitionof inflammation (39), and enhancementof cardiac output (47). Possible side ef-fects of hypertonic saline include re-bound in ICP, central pontine myelinol-ysis, renal impairment, subarachnoidhemorrhage, natriuresis, high urinarywater losses, hyperchloremic acidosis,and masking of the development of dia-betes insipidus (39).

Much higher levels of serum osmolar-ity (approximately 360 mOsm) may betolerated in children when induced withhypertonic saline (4, 48) vs. mannitol,although one recent report suggested in-creases in serum creatinine in childrentreated with hypertonic saline when se-rum sodium concentration was allowed

to increase to �160 mmol/L (49). How-ever, the recommendation of an uppersafety threshold of 360 mOsm/L for hy-pertonic saline (in the 2003 pediatric TBIguidelines) (50) was viewed as the itemthat generated the greatest disagreementamong 194 physicians treating pediatricpatients with TBI in a recent survey (51).

In 14 adults with severe TBI, a studyby Lescot et al (11) suggested importantdifferences in the response of contusedvs. noncontused brain tissue to hyper-tonic saline with reductions in the vol-ume of noncontused brain but increasesin the volume of contusions after treat-ment. Studies of regional effects of hyper-osmolar therapy have not been carriedout in pediatric TBI.




Fisher et al, 1992 (3) Design: randomized controlled crossovertrial

N � 18Age: mean 8.3 yrs (range, 0.6–14.5 yrs)Protocol: comparison of 3% saline (sodium

513 mEq/L, 1027 mOsm/L) and 0.9%saline (308 mOsm/L); doses of each agentwere equal and ranged between 6.5 and10 mL/kg in each patient

Purpose: comparison of effect on ICP over 2hrs exposure

Outcome: ICP

Class IIModerate quality: randomization and

allocation concealment methodsnot reported; crossover studylacking reporting on first-periodcomparison of baselinecharacteristics; small sample size

During the 2-hr trial, hypertonic salinewas associated with a lower ICP andreduced need for additionalinterventions (thiopental andhyperventilation) to control ICP

Serum sodium concentration increasedapproximately 7 mEq/L after 3%saline

Peterson et al, 2000 (4) Design: retrospective chart reviewN � 68Age: mean 7.8 � 3.6 yrsProtocol: use of a continuous infusion of 3%

hypertonic saline (513 mEq/L, 1027mOsm/L) titrated to reduce ICP � 20 mmHg; doses of 0.1–1.0 mL�kg�1�hr�1

resulting in mean daily dosages betweenapproximately 11 and 27 mL�kg�1�day�1

were usedPurpose: assess effect of continuous infusion

of hypertonic saline on acute and long-term outcome

Outcome: ICP, 6-month Glasgow OutcomeScale score


confounders

Survival rate was higher than expectedbased on Trauma and InjurySeverity Score (41 predicted, 58actual)

53% had good outcome, 20%moderate, 10% severe, 0.1%vegetative, and 15% died; 3 died ofuncontrolled ICP

No patients developed renal failureCentral pontine myelinolysis,

subarachnoid hemorrhage, orrebound increases in ICP were notobserved

Simma et al, 1998 (5) Design: randomized controlled trialN � 35Age: mean 87 months (� 42; range, 12–173

months)Protocol: comparison of hypertonic saline vs.

lactated Ringer’s solutionPurpose: comparison of 1.7% hypertonic

saline (sodium 268 mmol/L, 598 mOsm/L)vs. lactated Ringer’s solution (sodium 131mmol/L, 277 mOsm/L) as a continuousinfusion for maintenance fluidadministration over a 3-day exposure

Outcome: ICP, cerebral perfusion pressure,need for other interventions, fluidrequirements, intensive care unit stay,survival rate

Class IIModerate quality: not blinded,

insufficient power

There was no difference between groupsin survival rate and length of hospitalstay

Patients treated with hypertonic salinerequired fewer interventions thanthose treated with lactated Ringer’ssolution to maintain ICP control(p � .01)

The hypertonic saline treatment grouphad shorter length of intensive careunit stay (p � .04), shorter durationof mechanical ventilation (p �.10), and fewer complications than thelactated Ringer’s-treated group (p �.09 for two or more complications, notsignificant, without p value reported forone complication)

ICP, intracranial pressure.


A second use of hypertonic saline is totreat hyponatremia resulting from cere-bral salt wasting (CSW) if it develops inpediatric patients after TBI. Hyponatre-mia can result in cell swelling and sei-zures, both of which can compromise theinjured brain (52). Hyponatremia in pe-diatric TBI can result from several mech-anisms including CSW, the syndrome ofinappropriate antidiuretic hormone se-cretion, sodium losses (from renal, CSFdrainage, or other sources), or iatrogeniccauses. It can manifest between 48 hrsand 11 days after injury, and the mecha-nistic underpinnings appear to involveincreases in atrial natriuretic peptide (53,54). Confirmation of the diagnosis is es-sential because management of CSW candiffer greatly from the syndrome of inap-propriate antidiuretic hormone or othercauses of hyponatremia (55). The diagno-sis is made by demonstrating hyponatre-mia and increased urine sodium concen-tration in the face of polyuria andhypovolemia (56). Dramatic examples ofCSW in pediatric TBI show profound hy-ponatremia (serum sodium as low as 98mmol/L) and marked polyuria (�15 mL/kg/hr) requiring large volumes of combi-nations of 0.9% and 3.0% saline to matchurinary losses and address the hyponatre-mia (53, 54). Some have suggested tolimit the rate of correction of serum so-dium concentration to �12 mmol/L perday (50) related to concerns about my-elinolysis. The optimal rate of correctionof hyponatremia in a child with severeTBI is unclear.

IV. PROCESS

For this update, MEDLINE wassearched from 1996 through 2010 (Ap-pendix B for search strategy), and resultswere supplemented with literature rec-ommended by peers or identified fromreference lists. Of 35 potentially relevantstudies, no new studies were added to theexisting table and used as evidence forthis topic.


Two class II studies (3, 5) and oneclass III study (4) met the inclusion cri-teria for this topic and provide evidenceto support the recommendations.

Hypertonic Saline

A study by Fisher et al (3) was carriedout as a double-blind crossover study

comparing 3% saline (513 mEq/L, 1027mOsm/L) and 0.9% saline (154 mEq/L,308 mOsm/L) in 18 children with severeTBI. Bolus doses of each agent were equaland ranged between 6.5 and 10 mL/kg.During the 2-hr trial, hypertonic salineuse was associated with an approximate7-mEq/L increase in serum sodium con-centration, lower ICP, and reduced needfor other interventions. Concomitanttherapies used for patient management inthis study included thiopental, dopamine,mannitol, and hyperventilation. CSFdrainage was not used. As a result ofdesign flaws (see evidence table; Table 1),the evidence from this study is class II.

A study by Simma et al (5) was carriedout as a randomized controlled trial of1.7% hypertonic saline (sodium 268mmol/L, 598 mOsm/L) vs. lactated Ring-er’s solution (sodium 131 mmol/L, 277mOsm/L) administered over the initial 3days in 35 children with severe TBI. Pa-tients treated with hypertonic saline re-quired fewer interventions (includingmannitol use) to control ICP than thosetreated with lactated Ringer’s solution.Patients in the hypertonic saline treat-ment group also had a shorter length ofpediatric intensive care unit stay (p �.04), shorter duration of mechanical ven-tilation (p � .10), and fewer complica-tions than the lactated Ringer’s-treatedgroup (p � .09 for two or more compli-cations, nonsignificant for one complica-tion). As a result of design flaws and in-sufficient power (see evidence table), theevidence from this study is class II.

A study by Peterson et al (4) was aretrospective study on the use of a con-tinuous infusion of 3% saline (sodium513 mEq/L, 1027 mOsm/L) titrated toreduce ICP to �20 mm Hg in 68 infantsand children with TBI. The mean dailydoses of hypertonic saline over a 7-dayperiod ranged between 11 and 27mEq�kg�1�day�1. There was no controlgroup. Three patients died of uncon-trolled ICP, and mortality rate was lowerthan expected based on Trauma and In-jury Severity Score categorization. No pa-tient with a serum sodium concentration�180 mEq/L had a good outcome. Nopatients developed renal failure. Concom-itant therapies included sedation, neuro-muscular blockade, mannitol, hyperven-tilation, and barbiturates. CSF drainagewas used in three children. The meandaily dose of mannitol was 1–2g�kg�1�day�1. Rebound in ICP, centralpontine myelinolysis, and subarachnoidhemorrhage was not seen.

In the three papers cited as evidencefor hypertonic saline, several limitationsshould be recognized. These studies orig-inated from only two centers and therewas limited use of ventriculostomy cathe-ters and CSF drainage; instead, hyperven-tilation and barbiturates were used. Also,the children were enrolled between 16 and26 yrs ago. Finally, the report by Simma etal (5) compared 1.7% hypertonic salinewith lactated Ringer’s solution, which ishypotonic. It should be recognized that thetherapeutic window, safety profile, and op-timal doses or osmolar levels of hypertonicsaline remain to be determined.


A. Indications From AdultGuidelines

Based on an evidence table in theadult guidelines (57) (one class II andseven class III studies), mannitol wasdeemed to be effective for controlling in-creased ICP after severe TBI at dosesranging from 0.25 g/kg to 1 g/kg of bodyweight. Serum osmolarity �320 mOsm/Lwas recommended with mannitol use.Several key studies were cited. In the oneclass II study, Eisenberg et al (58) re-ported that a therapeutic regimen withmannitol was effective for ICP control in78% of patients (n � 73). In addition, astudy by Schwartz et al (21) was carriedout as a randomized comparison of man-nitol vs. barbiturates in 59 adults withsevere TBI. Cerebral perfusion pressurewas better maintained in the mannitol-treated group. Use of mannitol for TBIwas subjected to Cochrane review, and noconclusion could be reached regardingefficacy vs. placebo or any other therapy(59). Two class III level studies of hyper-tonic saline were cited in the adult guide-lines (57). The body of work on hyper-tonic saline in pediatric TBI showingbeneficial effects on ICP was discussed aswas the pediatric guidelines level III rec-ommendation of continuous infusion of3% saline. However, it was stated thatlimited data on hypertonic saline inadults with severe TBI did not allow forconclusions.


Mannitol. When constructing an evi-dence-based document on the use of hy-perosmolar therapy to control ICP in pe-


diatric TBI, one must recognize thatevidence supporting the use of mannitolin adults relies on studies that often in-cluded but did not explicitly define theproportion of children. Mannitol wasused concomitantly to control ICP in theaforementioned studies of hypertonic sa-line in the evidence table. One must thusweigh the value of long-standing clinicalacceptance and safety of a therapy (man-nitol) that has no evidentiary support forits efficacy against a newer therapy (hy-pertonic saline) with less clinical experi-ence but reasonably good performance incontemporary clinical trials (two class IIstudies for ICP and one class III study)(3–5, 48).

There is no study that met the inclu-sion criteria for this guideline, for eitherICP or neurologic outcome, that docu-ments efficacy of mannitol in infants andchildren with severe TBI. In several re-ports (29, 60–62), the specific effect ofmannitol on ICP or outcome was notreported, the sample size was very small,or mannitol was shown to reduce ICPreliably, but the sample represented amixture of adults and children (29).

Hypertonic Saline. One additionalstudy was not included as evidence be-cause it represented a prospective obser-vational study with an inadequate samplesize (n � 10). A study by Khanna et al(48) administered 3% saline (sodium 513mEq/L, 1027 mOsm/L) on a sliding scaleto maintain ICP �20 mm Hg in ten chil-dren with increased ICP resistant to con-ventional therapy. The maximal rate ofincrease in serum sodium was 15mEq�L�1�day�1, and the maximal rate ofdecrease in serum sodium was 10mEq�L�1�day�1. A reduction in ICP spikesand an increase in cerebral perfusionpressure were seen during treatmentwith 3% saline. The mean duration oftreatment was 7.6 days, and the meanhighest serum sodium concentration andosmolarity were 170.7 mEq/L and 364.8mOsm/L, respectively. The maximum se-rum osmolarity in an individual patientwas 431 mOsm/L. Sustained hyperna-tremia and hyperosmolarity were gener-ally well tolerated in the children. Twopatients, both with sepsis and/or multipleorgan failure, developed acute renal fail-ure. Both received continuous veno-venous hemofiltration and recovered re-nal function. One patient died ofuncontrolled intracranial hypertension.Despite its exclusion from the evidencetable, the findings of this report are con-

sistent with our recommendations sup-porting the use of 3% saline.

Hypertonic saline in pediatric patientswith severe TBI is also used in the man-agement of hyponatremia from CSW. Nopublications meeting the inclusion crite-ria for this guideline and addressingtreatment of CSW were identified. Mostreports suggest aggressive replacement ofurine salt and water losses, but only casereports in pediatric TBI or case serieswith various diagnoses (including severeTBI) have been reported (53, 54, 63). Thesodium replacement used ranged in dosebetween 0.1 and 2.4 mmol/kg/hr.

VII. SUMMARY

There is class II evidence supportingthe use of hypertonic saline (3%) for theacute treatment of severe pediatric TBIassociated with intracranial hypertensionand class III evidence to support its use asa continuous infusion during the inten-sive care unit course. There is insufficientevidence to support or refute the use ofmannitol, concentrations of hypertonicsaline �3%, or other hyperosmolaragents for the treatment of severe pediat-ric TBI. One must thus weigh the value oflongstanding clinical acceptance and safetyof mannitol, which has no evidence to sup-port its efficacy that met the inclusion cri-teria for this guideline, against hypertonicsaline, for which there is less clinical expe-rience but reasonably good performance incontemporary clinical trials.


● Documentation of the effect of hyper-osmolar therapy on ICP, cerebral per-fusion pressure, and outcome in stud-ies of infants and children.

● Studies comparing mannitol adminis-tration with hypertonic saline, particu-larly studies evaluating long-term out-come. This should include assessmentof the combination of mannitol andhypertonic saline.

● Study of the use of hyperosmolar ther-apy vs. other therapies such as CSFdrainage or barbiturates, including in-vestigation of both control of ICP andlong-term outcome.

● Studies as to whether or not hyperos-molar therapy can be effective in thesetting of herniation.

● Study of the prevention of intracranialhypertension by continuous infusion ofhypertonic saline vs. treatment in re-

sponse to spikes and its impact onlong-term outcome.

● Additional mechanistic studies in chil-dren with severe TBI examining issuessuch as the serum–CSF osmolar gap,regional effects of hyperosmolar thera-pies on contused vs. noncontused braintissue using computed tomography oradvanced magnetic resonance imaging,and their effects on other surrogatemarkers of brain injury such as bloodflow, metabolism, and biomarkers.

● Studies of the use of hyperosmolar ther-apy across various etiologies (abusive vs.nonabusive) and head computed tomog-raphy injury patterns (contusion vs. dif-fuse injury) in children.

● Optimal dosing and better definitionsof treatment threshold for the develop-ment of nephrotoxicity, rebound intra-cranial hypertension or hyponatremia,central pontine myelinolysis, and othercomplications with mannitol and hy-pertonic saline.

● Studies on the use of hypertonic salinein the management of CSW and othercauses of hyponatremia in pediatric pa-tients with TBI.

REFERENCES

1. Weed L, McKibben P: Pressure changes inthe cerebro-spinal fluid following intrave-nous injection of solutions of various con-centrations. Am J Physiol 1919; 48:512–530

2. Wise BL, Chater N: Use of hypertonic man-nitol solutions to lower cerebrospinal fluidpressure and decrease brain bulk in man.Surg Forum 1961; 12:398–399

3. Fisher B, Thomas D, Peterson B: Hypertonicsaline lowers raised intracranial pressure inchildren after head trauma. J Neurosurg An-esthesiol 1992; 4:4–10

4. Peterson B, Khanna S, Fisher B, et al: Pro-longed hypernatremia controls elevated in-tracranial pressure in head-injured pediatricpatients. Crit Care Med 2000; 28:1136–1143

5. Simma B, Burger R, Falk M, et al: A prospec-tive, randomized, and controlled study offluid management in children with severehead injury: Lactated Ringer’s solution ver-sus hypertonic saline. Crit Care Med 1998;26:1265–1270

6. Segal S, Gallagher AC, Shefler AG, et al:Survey of the use of intracranial pressuremonitoring in children in the United King-dom. Intensive Care Med 2001; 27:236–239

7. Keenan HT, Nocera M, Bratton SL: Fre-quency of intracranial pressure monitoringin infants and young toddlers with traumaticbrain injury. Pediatr Crit Care Med 2005;6:537–541

8. Biestro A, Alberti R, Galli R, et al: Osmoth-erapy for increased intracranial pressure:Comparison between mannitol and glycerol.


Acta Neurochir (Wien) 1997; 139:725–732;discussion 732–733

9. Bingham WF: The limits of cerebral dehydra-tion in the treatment of head injury. SurgNeurol 1986; 25:340–345

10. Fortune JB, Feustel PJ, Graca L, et al: Effectof hyperventilation, mannitol, and ventricu-lostomy drainage on cerebral blood flow afterhead injury. J Trauma 1995; 39:1091–1097;discussion 1097–1099

11. Lescot T, Degos V, Zouaoui A, et al: Opposedeffects of hypertonic saline on contusionsand noncontused brain tissue in patientswith severe traumatic brain injury. Crit CareMed 2006; 34:3029–3033

12. Marshall LF, RW SM, Rauscher LA, et al:Mannitol dose requirements in brain-injuredpatients. J Neurosurg 1978; 48:169–172

13. McGraw CP, Alexander E Jr, Howard G: Ef-fect of dose and dose schedule on the re-sponse of intracranial pressure to mannitol.Surg Neurol 1978; 10:127–130

14. McGraw CP, Howard G: Effect of mannitol onincreased intracranial pressure. Neurosur-gery 1983; 13:269–271

15. Mendelow AD, Teasdale GM, Russell T, et al:Effect of mannitol on cerebral blood flow andcerebral perfusion pressure in human headinjury. J Neurosurg 1985; 63:43–48

16. Miller JD, Leech P: Effects of mannitol andsteroid therapy on intracranial volume–pressure relationships in patients. J Neuro-surg 1975; 42:274–281

17. Muizelaar JP, Lutz HA 3rd, Becker DP: Effectof mannitol on ICP and CBF and correlationwith pressure autoregulation in severelyhead-injured patients. J Neurosurg 1984; 61:700–706

18. Polderman KH, van de Kraats G, Dixon JM,et al: Increases in spinal fluid osmolarityinduced by mannitol. Crit Care Med 2003;31:584–590

19. Procaccio F, Menasce G, Sacchi L, et al: Ef-fects of thiopentone and mannitol on cere-bral perfusion pressure and EEG in headinjured patients with intracranial hyperten-sion. Agressologie 1991; 32:381–385

20. Rosner MJ, Coley I: Cerebral perfusion pres-sure: A hemodynamic mechanism of manni-tol and the postmannitol hemogram. Neuro-surgery 1987; 21:147–156

21. Schwartz ML, Tator CH, Rowed DW, et al:The University of Toronto head injury treat-ment study: A prospective, randomized com-parison of pentobarbital and mannitol. CanJ Neurol Sci 1984; 11:434–440

22. Rosner MJ, Rosner SD, Johnson AH: Cerebralperfusion pressure: Management protocoland clinical results. J Neurosurg 1995; 83:949–962

23. Nara I, Shiogai T, Hara M, et al: Comparativeeffects of hypothermia, barbiturate, and os-motherapy for cerebral oxygen metabolism,intracranial pressure, and cerebral perfusionpressure in patients with severe head injury.Acta Neurochir Suppl 1998; 71:22–26

24. Smith HP, Kelly DL Jr, McWhorter JM, et al:Comparison of mannitol regimens in pa-

tients with severe head injury undergoingintracranial monitoring. J Neurosurg 1986;65:820–824

25. Levin AB, Duff TA, Javid MJ: Treatment ofincreased intracranial pressure: A compari-son of different hyperosmotic agents and theuse of thiopental. Neurosurgery 1979;5:570–575

26. Muizelaar JP, Wei EP, Kontos HA, et al: Man-nitol causes compensatory cerebral vasocon-striction and vasodilation in response toblood viscosity changes. J Neurosurg 1983;59:822–828

27. Muizelaar JP, Wei EP, Kontos HA, et al: Ce-rebral blood flow is regulated by changes inblood pressure and in blood viscosity alike.Stroke 1986; 17:44–48

28. Bouma GJ, Muizelaar JP: Cerebral bloodflow, cerebral blood volume, and cerebrovas-cular reactivity after severe head injury.J Neurotrauma 1992; 9(Suppl 1):S333–348

29. James HE: Methodology for the control ofintracranial pressure with hypertonic man-nitol. Acta Neurochir (Wien) 1980; 51:161–172

30. Kaieda R, Todd MM, Cook LN, et al: Acuteeffects of changing plasma osmolality andcolloid oncotic pressure on the formation ofbrain edema after cryogenic injury. Neuro-surgery 1989; 24:671–678

31. Kaufmann AM, Cardoso ER: Aggravation ofvasogenic cerebral edema by multiple-dosemannitol. J Neurosurg 1992; 77:584–589

32. Kontos HA, Hess ML: Oxygen radicals andvascular damage. Adv Exp Med Biol 1983;161:365–375

33. Becker D, Vries J: The alleviation of increasedintracranial pressure by the chronic admin-istration of osmotic agents. In: IntracranialPressure. MB, HD (Eds). Berlin, Germany,Springer, 1972, pp 309–315

34. The Brain Trauma Foundation: The Ameri-can Association of Neurological Surgeons.The Joint Section on Neurotrauma and Crit-ical Care. Use of mannitol. J Neurotrauma2000; 17:521–525

35. Feig PU, McCurdy DK: The hypertonic state.N Engl J Med 1977; 297:1444–1454

36. Worthley LI, Cooper DJ, Jones N: Treatmentof resistant intracranial hypertension withhypertonic saline. Report of two cases. J Neu-rosurg 1988; 68:478–481

37. Vassar MJ, Fischer RP, O’Brien PE, et al: Amulticenter trial for resuscitation of injuredpatients with 7.5% sodium chloride. The ef-fect of added dextran 70. The MulticenterGroup for the Study of Hypertonic Saline inTrauma Patients. Arch Surg 1993; 128:1003–1011; discussion 1011–1013

38. Prough DS, Whitley JM, Taylor CL, et al:Regional cerebral blood flow following resus-citation from hemorrhagic shock with hyper-tonic saline. Influence of a subdural mass.Anesthesiology 1991; 75:319–327

39. Qureshi AI, Suarez JI: Use of hypertonic sa-line solutions in treatment of cerebral edemaand intracranial hypertension. Crit Care Med2000; 28:3301–3313

40. Shackford SR, Bourguignon PR, Wald SL, etal: Hypertonic saline resuscitation of patientswith head injury: A prospective, randomizedclinical trial. J Trauma 1998; 44:50–58

41. Walsh JC, Zhuang J, Shackford SR: A com-parison of hypertonic to isotonic fluid in theresuscitation of brain injury and hemor-rhagic shock. J Surg Res 1991; 50:284–292

42. Zornow MH, Prough DS: Fluid managementin patients with traumatic brain injury. NewHoriz 1995; 3:488–498

43. Bulger EM, May S, Brasel KJ, et al: Out-of-hospital hypertonic resuscitation followingsevere traumatic brain injury: A randomizedcontrolled trial. JAMA 2010; 304:1455–1464

44. McManus ML, Soriano SG: Rebound swellingof astroglial cells exposed to hypertonic man-nitol. Anesthesiology 1998; 88:1586–1591

45. Nakayama S, Kramer GC, Carlsen RC, et al:Infusion of very hypertonic saline to bledrats: Membrane potentials and fluid shifts.J Surg Res 1985; 38:180–186

46. Arjamaa O, Karlqvist K, Kanervo A, et al:Plasma ANP during hypertonic NaCl infusionin man. Acta Physiol Scand 1992; 144:113–119

47. Moss GS, Gould SA: Plasma expanders. Anupdate. Am J Surg 1988; 155:425–434

48. Khanna S, Davis D, Peterson B, et al: Use ofhypertonic saline in the treatment of severerefractory posttraumatic intracranial hyper-tension in pediatric traumatic brain injury.Crit Care Med 2000; 28:1144–1151

49. Dominguez TE, Priestley MA, Huh JW: Cau-tion should be exercised when maintaining aserum sodium level �160 meq/L. Crit CareMed 2004; 32:1438 –1439; author reply1439–1440

50. Adelson PD, Bratton SL, Carney NA, et al:Guidelines for the acute medical manage-ment of severe traumatic brain injury in in-fants, children, and adolescents. Chapter 8.Cerebral perfusion pressure. Pediatr CritCare Med 2003; 4:S31–S33

51. Dean NP, Boslaugh S, Adelson PD, et al:Physician agreement with evidence-basedrecommendations for the treatment of severetraumatic brain injury in children. J Neuro-surg 2007; 107:387–391

52. Rivkees SA: Differentiating appropriate anti-diuretic hormone secretion, inappropriateantidiuretic hormone secretion and cerebralsalt wasting: The common, uncommon, andmisnamed. Curr Opin Pediatr 2008; 20:448–452

53. Donati-Genet PC, Dubuis JM, Girardin E, etal: Acute symptomatic hyponatremia and ce-rebral salt wasting after head injury: An im-portant clinical entity. J Pediatr Surg 2001;36:1094–1097

54. Jimenez R, Casado-Flores J, Nieto M, et al:Cerebral salt wasting syndrome in childrenwith acute central nervous system injury.Pediatr Neurol 2006; 35:261–263

55. Albanese A, Hindmarsh P, Stanhope R: Man-agement of hyponatraemia in patients withacute cerebral insults. Arch Dis Child 2001;85:246–251


56. Carlotti AP, Bohn D, Rutka JT, et al: Amethod to estimate urinary electrolyte excre-tion in patients at risk for developing cere-bral salt wasting. J Neurosurg 2001; 95:420–424

57. Bratton SL, Chestnut RM, Ghajar J, et al:Guidelines for the management of severetraumatic brain injury. II. Hyperosmolartherapy. J Neurotrauma 2007; 24(Suppl 1):S14–S20

58. Eisenberg HM, Frankowski RF, Contant CF,et al: High-dose barbiturate control of ele-

vated intracranial pressure in patients withsevere head injury. J Neurosurg 1988; 69:15–23

59. Schierhout G, Roberts I: Mannitol for acutetraumatic brain injury. Cochrane DatabaseSyst Rev 2000; 1:CD001049

60. Eder HG, Legat JA, Gruber W: Traumaticbrain stem lesions in children. Childs NervSyst 2000; 16:21–24

61. Kasoff SS, Lansen TA, Holder D, et al: Ag-gressive physiologic monitoring of pediatrichead trauma patients with elevated intracra-

nial pressure. Pediatr Neurosci 1988; 14:241–249

62. Miller JD, Piper IR, Dearden NM: Manage-ment of intracranial hypertension in headinjury: Matching treatment with cause.Acta Neurochir Suppl (Wien) 1993; 57:152–159

63. Berkenbosch JW, Lentz CW, Jimenez DF, etal: Cerebral salt wasting syndrome followingbrain injury in three pediatric patients: Sug-gestions for rapid diagnosis and therapy. Pe-diatr Neurosurg 2002; 36:75–79


Chapter 9. Temperature control

I. RECOMMENDATIONS

Strength of Recommendations: Weak.Quality of Evidence: Moderate, from

class II and III studies with some contra-dictory findings.

A. Level I


B. Level II

Moderate hypothermia (32–33°C) be-ginning early after severe traumatic braininjury (TBI) for only 24 hrs’ durationshould be avoided.

Moderate hypothermia (32–33°C) be-ginning within 8 hrs after severe TBI forup to 48 hrs’ duration should be consid-ered to reduce intracranial hypertension.

If hypothermia is induced for any in-dication, rewarming at a rate of�0.5°C/hr should be avoided.

C. Level III*

Moderate hypothermia (32–33°C) be-ginning early after severe TBI for 48 hrs,duration may be considered.

*After completion of these guidelines, thecommittee became aware that the Cool Kids trialof hypothermia in pediatric TBI was stopped be-cause of futility. The implications of this devel-opment on the recommendations in this sectionmay need to be considered by the treating phy-sician when details of the study are published.


III. OVERVIEW

The definitions of hypothermia andhyperthermia are controversial. Posttrau-matic hypothermia is often classified as acore body temperature �35°C, whereas atemperature �38.0 –38.5°C representsfever/pyrexia if it results from an alteredthermoregulatory set point and repre-sents hyperthermia if it is imposed on anormal set point. For simplicity, the term

hyperthermia is used to reflect an ele-vated core body temperature throughoutthis chapter. At present, the data in thebasic science literature on adult animalmodels indicate that hyperthermia con-tributes to greater posttraumatic damageby increasing the acute pathophysiologi-cal response after injury through a mul-titude of mechanisms.

The rationale for use of therapeutichypothermia is a reduction in mecha-nisms of secondary injury resulting fromdecreased cerebral metabolic demands,inflammation, lipid peroxidation, excito-toxicity, cell death, and acute seizures.Clinical studies reviewed on temperatureregulation have focused, by definition forthese guidelines, on global functionaloutcome but also the effect on intracra-nial hypertension. The impact of reduc-tion of intracranial pressure (ICP) aftersevere TBI in children on outcome re-mains to be determined. As discussed inprevious chapters, the lowering of se-verely elevated ICP with respect to thetreatment threshold may be a desirableoutcome.

Lastly, based on experimental studiesin animal models and clinical studies inadults, in which hyperthermia was corre-lated with poor outcome, it has been rec-ommended that hyperthermia after TBIin children should be prevented. How-ever, no study of the impact of hyperther-mia on outcome after TBI met the inclu-sion criteria for this guideline. There alsomay be a role for therapeutic hypother-mia in reducing intracranial hyperten-sion in severe pediatric TBI.

IV. PROCESS

For this update, MEDLINE wassearched from 1996 through 2010 (Ap-pendix B for search strategy), and resultswere supplemented with literature rec-ommended by peers or identified fromreference lists. Of 17 potentially relevantstudies, two were added to the existingtable for this topic.


Two moderate-quality class II studiesand one poor-quality class III study met

the inclusion criteria for this topic andprovide evidence to support the recom-mendations (1–3).

Level II Recommendations

Outcome. This review provides a levelII recommendation for the avoidance ofmoderate hypothermia (32–33°C) initi-ated early after severe TBI and applied foronly 24 hrs’ duration followed by rapidrewarming at a rate of �0.5°C/hr. Thiswas based on the Hutchison et al (3)study that reported a phase III multicen-tered randomized trial (225 children withsevere TBI; Glasgow Coma Scale Score3–8) of moderate hypothermia (32–33°C)for 24 hrs followed by rewarming at a rateof 0.5–1.0°C every hour. In this study,hypothermia was used in a prophylacticmanner as a neuroprotective strategywhether or not raised ICP was present.The findings from this study trended to-ward worse outcomes at 6 months afterinjury in children treated with hypother-mia vs. normothermia using the Pediat-ric Cerebral Performance Category score(30% vs. 22%; p � .08) and increasedmortality (21% vs. 14%; p � .06). In thisstudy, the investigators screened the pa-tients within 8 hrs, and the mean time toinitiation of cooling was 6.3 hrs with arange of 1.6–19.7 hrs. As well, the proto-col included a rapid rewarming rate asdescribed previously so that the patientswere normothermic by a mean of 19 hrsor within 48 hrs of injury. They foundthat hypothermia reduced intracranialhypertension with ICP significantly lowerin the hypothermia vs. normothermiagroup during the cooling period, but thiswas followed by a significantly higher ICPin the hypothermia vs. normothermicgroups during rewarming. A potentiallyconfounding factor in this study was thatmarked hyperventilation (PaCO2 �30 mmHg) was used as part of standard manage-ment in �40% of the patients in thestudy and hypertonic (3%) saline use wassignificantly reduced in the hypothermiavs. normothermia group.

Intracranial Hypertension. In con-trast, a level II recommendation wasmade supporting the use of moderate hy-pothermia (32–33°C) in severe pediatric


DOI: 10.1097/PCC.0b013e31823f664b


TBI in the setting of refractory intracra-nial hypertension for 48 hrs’ durationfollowed by slow rewarming at a rate of0.5–1.0°C per 12–24 hrs if the injury oc-curred within 8 hrs. The recommenda-tion was based on two class II studieswith benefit of hypothermia on ICP (2, 3).

As mentioned, Hutchison et al (3)showed that hypothermia reduced intra-cranial hypertension with ICP signifi-cantly lower in the hypothermia vs. nor-mothermia groups during the coolingperiod, although this rebounded tohigher ICP during rewarming. However,hypothermia in that study was used onlyfor up to 24 hrs. A study by Adelson et al(2) was a phase II multicentered random-ized trial in 75 children with severe TBI(Glasgow Coma Scale score 3–8) of mod-erate hypothermia (32–33°C) for 48 hrsfollowed by rewarming at a rate of 0.5–1.0°C every 3–4 hrs. Additional details ofthis study as it relates to outcome areprovided subsequently. Although therewas no overall effect of hypothermia vs.normothermia on ICP, ICP was signifi-cantly decreased in the initial 24 hrs after

TBI in the hypothermia vs. normother-mia groups.

Level III Recommendation

Outcome. Supporting the level III rec-ommendation for early administration oftherapeutic hypothermia for 48 hrs’ dura-tion with slow rewarming, Adelson et al (2)carried out a phase II multicentered ran-domized trial in 75 children with severeTBI (Glasgow Coma Scale score 3–8) ofmoderate hypothermia (32–33°C) for 48hrs followed by rewarming at a rate of 0.5–1.0°C every 3–4 hrs. They reported thathypothermia was safe, associated with a de-creased mortality rate (8% vs. 16%), anddid not increase complications. Once again,in this study, hypothermia was used in aprophylactic manner as a neuroprotectivestrategy whether or not raised ICP waspresent. Similar to the report of Hutchisonet al (3), they also noted rebound intracra-nial hypertension in the previously cooledpatients during rewarming.

In support of this recommendation isHendrick (1), a case series of 19 children

with severe TBI who presented with de-cerebrate posturing that, although beforeactual classification of severity of injury,would translate to a present-day GlasgowComa Scale score of 4. These children weretreated with moderate hypothermia (32–33°C). There were ten long-term survivorswith only one severely impaired. It was con-cluded that systemic cooling after injurywas effective as a “useful adjunct” thatcould improve outcome in children after TBI.



In the third edition of the adult guide-lines (4), there was a chapter entitled“Prophylactic Hypothermia” based on asufficient number of studies assessing theefficacy of therapeutic hypothermia aftersevere TBI in adults. Much of the previ-ous edition of the pediatric guidelineswas based on a scarcity of pediatric spe-


Reference Description of Study Data Class, Quality, and Reasons Results and Conclusion

Studies from previous guidelinesHendrick et al, 1959 (1) Design: uncontrolled case series Class III 8 deaths

N � 18Protocol: patients who presented

with decerebrate posturing werecooled to 32– 33°C; adjunctivetherapies included promethazineand chlorpromazine

Poor quality: no control forconfounders

Among 10 survivors, 4 no disability, 1minimal hearing loss, 2 minimalhemiparesis and aphasia, 1hemiparesis, 1 diplopia and mildpersonality changes, 1 grossintellectual impairment

New studiesAdelson et al, 2005 (2) Design: randomized controlled trial

N � 75Protocol: cooled to 32–33°C within

8 hrs of injury for 48 hrs ascompared with normothermia

Outcome: mortality, 3- and 6-month Glasgow Outcome Scale

Class IIModerate quality: unclear reporting

of randomization methods,allocation concealment methods,and attrition

No difference between groups inmortality or 3- and 6-monthGlasgow Outcome Scale

ICP: overall, there was no statisticaldifference in mean ICP between thegroups during the 5-day period(p � .37) except within the first 24hrs, when the ICP was reduced inthe hypothermia group (p � .024)

No difference between groups incomplication rates

Hutchison et al, 2008 (3) Design: randomized controlled trialN � 225Protocol: Randomized to cooling to

32–33°C within 8 hrs of injuryfor 24 hrs vs. normothermia;patients rewarmed at 0.5°C perhour

Class IIModerate quality: some differences

between groups on baselineprognostic factors

No difference between groups onfunctional outcomes at 6 months

Trend toward increased mortality andmorbidity in the hypothermia group

ICP was lower during cooling in thehypothermia group at 16 hrs and 24hrs (p � .02 and p � .01,respectively)

Significant increase in hypotension andpressor requirements in thehypothermia group



cific information with the use of hypo-thermia to treat patients with severe TBIoriginally reported �50 yrs ago (1). How-ever, its use did not become establishedbecause early studies lacked modern sci-entific methods and adequate outcomemeasures to definitively prove or refuteefficacy, and there were concerns overside effects. Renewed interest in moder-ate hypothermia after severe TBI did notoccur until the past 15–20 yrs, when pre-liminary data from single-center clinicaltrials were published in adults. Studies byShiozaki et al (5), Marion et al (6), Cliftonet al (7), and Marion et al (8) all demon-strated that moderate hypothermia re-duced ICP and tended to improve overalloutcomes. In the first randomized con-trolled trial that followed these single-center studies, Clifton et al (9) reportedlack of effectiveness in adults in a multi-centered clinical trial of moderate hy-pothermia after severe TBI. Despite fail-ure to replicate the earlier single-centerfindings in the larger multicenteredtrial, there was a suggestion of improvedoutcome in those patients who presentedas hypothermic and were then kept cooland in the younger age groups within thestudy (�40 yrs of age). Children (�16yrs) were not included in the Cliftonstudy or in subsequent studies. In therecent adult guidelines (4), it was notedthat although hypothermia is often in-duced prophylactically on admission andused for ICP elevation in the intensivecare unit in many trauma centers, thescientific literature has failed to consis-tently support its positive influence onmortality and morbidity. Four meta-analyses of hypothermia in adult patientswith TBI have been published (10–13).All analyses concluded that the evidencewas insufficient to support routine use ofhypothermia and recommended furtherstudy to determine factors that might ex-plain variation in results. As a result, theauthors of the adult guidelines undertookanother meta-analysis of the six trialsthat were assessed to be of moderatequality (7–9, 14–16). The overall risk re-duction for mortality from this large dataset was not significantly different be-tween hypothermia and normothermiatreatment groups, but hypothermia wasassociated with a 46% increased chanceof good neurologic outcome (relativerisk, 1.46; 95% confidence interval, 1.12–1.92). This led to a level III recommen-dation based on this pooled data. Addi-tionally, preliminary findings suggestedthat a greater decrease in mortality risk is

observed when target temperatures aremaintained for �48 hrs (4). The resultsof these clinical trials of hypothermia inadult patients with TBI and even the re-sultant meta-analyses cannot be extrapo-lated directly to the management of se-vere TBI in children because childrenwere not included in the samples ana-lyzed.


A study by Li et al (17) reported on theuse of local hypothermia (head cooling)rather than systemic cooling to an intra-cranial temperature of 34.5 � 0.5°Cwithin 8 hrs of injury vs. normothermia(37.5–38.5°C) for a period of 72 hrs. Al-though the study showed a positive effectof cooling on intracranial hypertensionand biomarkers, neuron-specific enolase,S-100B, and CK-BB at 8, 24, and 48 hrsafter injury, suggesting neuronal protec-tion, neurologic outcomes could only bedetermined for eight patients (three ofwhom had died), because almost two-thirds were lost to clinical follow-up. Astudy by Biswas et al (18) reported on 21children with severe TBI (Glasgow ComaScale score 3– 8), ten of whom werecooled to moderate hypothermia (32–34°C) within 6 hrs of injury for 48 hrsfollowed by rewarming to normothermiawithin 12 hrs. Although the study showeda positive effect of cooling on intracranialhypertension, there was no significantdifference between hypothermic and nor-mothermic groups in other outcomemeasures including Glasgow OutcomeScale, Pediatric Cerebral PerformanceCategory, or Pediatric Overall Perfor-mance Category. Despite the small sam-ple size overall, of the 21 children treated,11 of 11 and six of six in the normother-mia group had a good outcome at 3 and12 months postinjury, respectively,whereas six of ten and five of eight in thehypothermia group had a good outcomeat 3 and 12 months postinjury. Both ofthese studies supported the level II rec-ommendation despite being poor-qualitystudies for the end points defined.

A study by Aibiki et al (14) evaluatedventilated adults and children with severeTBI (Glasgow Coma Scale score 3– 8)treated with moderate hypothermia (32–33°C) for 3–4 days as compared with nor-mothermia for prostanoid production. Al-though not the focus on the article, dataas to age and outcome (Glasgow OutcomeScale at 6 months) could be abstracted.

There were 11 children treated and two offour in the normothermia vs. six of sevenin the hypothermia groups had a goodoutcome. Similar to the study by Grusz-kiewicz et al (19), confounding this studywas the cotreatment with dexametha-sone. The study by Grinkeviciute andKevalas (20) reported on a prospectivecohort to determine the safety of mildhypothermia after TBI. There were eightpatients included in the study, with amean age of 10.7 yrs, who had severe TBI(Glasgow Coma Scale score 4–8), andwho were treated with mild hypothermia(33–34°C) with a rapid induction of 2–3hrs and maintained for 48 hrs with pas-sive rewarming at 1°C per 4 hrs. Usingthe Glasgow Outcome Scale, all patientshad a good outcome. Average GlasgowOutcome Scale score was 4.13 at 6months after injury.

Finally, a study by Gruszhiewicz et al(19) reported on a prospective, random-ized study of 20 children �16 yrs of agewho had severe TBI presenting with aclinical examination of decerebrate rigid-ity (Glasgow Coma Scale score � 4). Thechildren were randomized to hypother-mia vs. hypothermia combined withdexamethasone (2 mg twice a day). Therewas no normothermic group. Sixteen ofthese 20 patients were hyperthermic atpresentation and sustained various mech-anisms of injury. Outcome was deter-mined by duration of coma and time until“recovery,” although the length of fol-low-up was �7 months in all instances.Although no statistical analysis was per-formed, the authors described a similarduration of coma and neurologic recov-ery for the two groups, although thedepth and duration of the hypothermiadiffered. Some patients were cooled to30 –32°C, whereas others to 35–36°C.There was also variability of applicationfrom 18 hrs to 17 days. There were 19survivors. Adjunctive therapies includedpromethazine, chlorpromazine, manni-tol, and lumbar puncture to reduce ICP.

Although no study on the effect ofhyperthermia after TBI in children metthe inclusion criteria for these guide-lines, a study by Heindl and Laub (21)reported that posttraumatic hyperther-mia (defined as a temperature �38.2°C,lasting for at least 1 wk) was associatedwith a poor outcome vs. normothermiain an extremely severe cohort of 82 pa-tients who remained in a persistent veg-etative state at least 30 days postinjury. Inthis purely observational study, patientswith hyperthermia had a poorer outcome


vs. those that did not (81% vs. 19%; p �.01). The time window that hyperthermiamay contribute to secondary injury aftersevere TBI and the best approach to pre-venting or treating it were not addressed,although this study suggests that it maybe important to prevent or treat hyper-thermia after pediatric TBI.

VII. SUMMARY

Considerable uncertainty exists re-garding the specifics of the use of tar-geted temperature management in pedi-atric TBI. A number of studies, includingtwo new studies with class II evidence,show that mild or moderate hypother-mia, vs. normothermia, can attenuate in-tracranial hypertension. However, the ef-ficacy of this therapy vs. others as eithera first-line agent or to treat refractoryintracranial hypertension remains un-clear. Similarly, conflicting results havebeen obtained regarding the effect of hy-pothermia on mortality and/or neuro-logic outcomes. It appears that details ofthe protocols used both to induce andmaintain hypothermia and rewarm maybe extremely important with short (24-hr)periods of cooling and rapid rewarming ex-hibiting the most complications. Finally,no study of the effect of hyperthermia onoutcome after TBI in children met the in-clusion criteria to allow a recommendationon this aspect of management.


● The effect of temperature control, in-cluding prevention of hyperthermia,on outcome after pediatric TBI needsto be further studied.

● Issues such as the duration of vulner-ability to hyperthermia and the optimalway to prevent or treat it should beaddressed.

● The role of therapeutic hypothermia,both as a neuroprotective measure andfor refractory intracranial hyperten-sion, deserves investigation in pediatricTBI. Direct comparisons to other ther-apies should be conducted.

● Evaluations of therapeutic hypother-mia should be age-stratified. Additionaldocumentation of the effect of hypo-thermia and temperature regulation instudies restricted to infants and chil-dren are needed.

● Studies of the effect of hypothermia onspecific TBI pathologies such as contu-sion, diffuse injury, and abusive headtrauma are needed.

● Studies addressing both the therapeu-tic window and optimal duration areneeded.

● Studies of the optimal timing and rateof rewarming are also needed.

● Studies are needed to better under-stand the effect of temperature regula-tion on key physiological and pharma-cologic parameters (e.g., ICP, cerebralperfusion pressure, cardiac output, im-mune status, drug metabolism anddrug dosing, etc.) and how these effectsmight influence long-term outcome.

REFERENCES

1. Hendrick EB: The use of hypothermia insevere head injuries in childhood. Arch Surg1959; 79:362–364


3. Hutchison JS, Ward RE, Lacroix J, et al:Hypothermia therapy after traumatic braininjury in children. N Engl J Med 2008; 358:2447–2456

4. Bratton SL, Chestnut RM, Ghajar J, et al:Guidelines for the management of severetraumatic brain injury. III. Prophylactic hy-pothermia. J Neurotrauma 2007; 24(Suppl1):S21–S25

5. Shiozaki T, Sugimoto H, Taneda M, et al:Effect of mild hypothermia on uncontrol-lable intracranial hypertension after severehead injury. J Neurosurg 1993; 79:363–368

6. Marion DW, Obrist WD, Carlier PM, et al: Theuse of moderate therapeutic hypothermiafor patients with severe head injuries: Apreliminary report. J Neurosurg 1993; 79:354 –362

7. Clifton GL, Allen S, Barrodale P, et al: Aphase II study of moderate hypothermia insevere brain injury. J Neurotrauma 1993;10:263–271; discussion 273

8. Marion DW, Penrod LE, Kelsey SF, et al:Treatment of traumatic brain injury withmoderate hypothermia. N Engl J Med 1997;336:540–546

9. Clifton GL, Miller ER, Choi SC, et al: Lack ofeffect of induction of hypothermia after acutebrain injury. N Engl J Med 2001; 344:556–563

10. Alderson P, Gadkary C, Signorini DF: Ther-apeutic hypothermia for head injury. Co-chrane Database Syst Rev 2004; 4:CD001048

11. Harris OA, Colford JM Jr, Good MC, et al: Therole of hypothermia in the management ofsevere brain injury: A meta-analysis. ArchNeurol 2002; 59:1077–1083

12. Henderson WR, Dhingra VK, Chittock DR, etal: Hypothermia in the management of trau-matic brain injury. A systematic review andmeta-analysis. Intensive Care Med 2003; 29:1637–1644

13. McIntyre LA, Fergusson DA, Hebert PC, et al:Prolonged therapeutic hypothermia aftertraumatic brain injury in adults: A system-atic review. JAMA 2003; 289:2992–2999

14. Aibiki M, Maekawa S, Yokono S: Moderatehypothermia improves imbalances of throm-boxane A2 and prostaglandin I2 productionafter traumatic brain injury in humans. CritCare Med 2000; 28:3902–3906

15. Jiang J, Yu M, Zhu C: Effect of long-termmild hypothermia therapy in patients withsevere traumatic brain injury: 1-year fol-low-up review of 87 cases. J Neurosurg 2000;93:546–549

16. Qiu WS, Liu WG, Shen H, et al: Therapeuticeffect of mild hypothermia on severe traumatichead injury. Chin J Traumatol 2005; 8:27–32

17. Li H, Lu G, Shi W, et al: Protective effect ofmoderate hypothermia on severe traumaticbrain injury in children. J Neurotrauma2009; 26:1905–1909

18. Biswas AK, Bruce DA, Sklar FH, et al: Treat-ment of acute traumatic brain injury in chil-dren with moderate hypothermia improvesintracranial hypertension. Crit Care Med2002; 30:2742–2751

19. Gruszkiewicz J, Doron Y, Peyser E: Recoveryfrom severe craniocerebral injury with brainstem lesions in childhood. Surg Neurol 1973;1:197–201

20. Grinkeviciute D, Kevalas R: Induced mildhypothermia in children after brain injury.Rev Neurosci 2009; 20:261–266

21. Heindl UT, Laub MC: Outcome of persistentvegetative state following hypoxic or trau-matic brain injury in children and adoles-cents. Neuropediatrics 1996; 27:94–100


Chapter 10. Cerebrospinal fluid drainage

I. RECOMMENDATIONS

Strength of Recommendations: Weak.Quality of Evidence: Low from poor-

and moderate-quality class III studieswith some contradictory findings.

A. Level I


B. Level II


C. Level III

Cerebrospinal fluid (CSF) drainagethrough an external ventricular drainmay be considered in the management ofincreased intracranial pressure (ICP) inchildren with severe traumatic brain in-jury (TBI).

The addition of a lumbar drain may beconsidered in the case of refractory intra-cranial hypertension with a functioningexternal ventricular drain, open basal cis-terns, and no evidence of a mass lesion orshift on imaging studies.


III. OVERVIEW

With the use of the external ventricu-lar drain as a common means of measur-ing ICP of patients with TBI, the potentialadded therapeutic benefits of CSF drain-age is of interest. Before the use of theexternal ventricular drain in TBI, theprincipal use of CSF drainage was in pa-tients with hydrocephalus, but the abilityof this procedure to potentially affect ICPled to its increased use as a therapeuticdevice for TBI. The role of CSF drainageis to reduce intracranial fluid volume andthereby lower ICP. Both intermittent andcontinuous drainage approaches havebeen reported in the pediatric literature

(1). Therapy may be associated with anincreased risk of complications fromhemorrhage and malpositioning.

IV. PROCESS

For this update, MEDLINE wassearched from 1996 through 2010 (Ap-pendix B for search strategy), and resultswere supplemented with literature rec-ommended by peers or identified fromreference lists. Of six potentially relevantstudies, one was added to the existingtable and used as evidence for this topic.


Four class III studies met the inclu-sion criteria and are used as evidence forthis topic (2–5). Ventricular drainagealone was used in two studies, and lum-bar drainage in combination with an ex-ternal ventricular drain was used in theother two.

A study by Shapiro and Marmarou (4)retrospectively studied 22 children withsevere TBI defined as a Glasgow ComaScale score of �8, all of whom weretreated with ventricular drainage. Param-eters measured included ICP, pressure–volume index, and mortality. DrainingCSF increased pressure–volume indexand decreased intracranial hypertension.Two neurologic deaths occurred in pa-tients with refractory intracranial hyper-tension; however, the ICP of the otherthree patients who died, and the foursurvivors with severe disability, is not re-ported. Consequently, the absolute influ-ence of CSF drainage in this sample can-not be determined.

A study by Jagannathan et al (5) ret-rospectively studied 96 children with se-vere TBI comparing management of ICPalone vs. ICP along with surgery usingeither external ventricular drainage oroperative treatment (evacuation of hema-toma or decompressive craniectomy).ICP control was achieved in 82 patients(85%). Methods used to achieve ICP con-trol included maximal medical therapy(sedation, hyperosmolar therapy, andneuromuscular blockade) in 34 patients(35%), external ventricular drain in 23patients (24%), and surgery in 39 pa-

tients (41%). Refractory ICP resulted in100% mortality. Authors concluded thatcontrolling elevated ICP is an importantfactor in patient survival after severe pe-diatric TBI. The modality used for ICPcontrol appears to be less important. Nolong-term follow-up to determine neuro-cognitive sequelae was performed.

Drainage of CSF is not limited to theventricular route. The other level III rec-ommendation is that although CSFdrainage can be accomplished through anexternal ventricular drain catheter aloneor in combination with a lumbar drain,the addition of lumbar drainage shouldonly be considered in the case of refrac-tory intracranial hypertension with afunctioning external ventricular drain,open basal cisterns, and no evidence of amajor mass lesion or shift on imagingstudies. A study by Baldwin and Rekate(2) reported a series of five children withsevere TBI, in whom lumbar drains wereplaced after failure to control ICP withboth ventricular drainage and barbituratecoma. Three children had quick and last-ing resolution of raised ICP, two of themwith good outcome and one with moder-ate remaining disability. In the other twocases, there was no effect on ICP and bothchildren died.

In a later paper from the same insti-tution, Levy et al (3) reported the effecton outcome of controlled lumbar drain-age with simultaneous external ventricu-lar drainage in 16 pediatric patients withsevere TBI. In two patients, ICP was un-affected and both died. The remaining 14survived, eight having a good outcome,three with moderate disability, and threehaving severe disability. Although therewas no direct outcome study or analysison the use of barbiturates in this series,the authors proposed that barbituratecoma and its associated morbidity couldbe avoided by the use of lumbar drainage,based on their findings in this series thatnot all patients were given barbiturates(five of 16 patients received no barbitu-rates and six of 16 received only intermit-tent dosing). The use of lumbar drainage,however, was contraindicated in the set-ting of a focal mass lesion or shift and theauthors recommended the use of lumbar


DOI: 10.1097/PCC.0b013e31823f665f


drainage only in conjunction with a func-tioning external ventricular drain in thesetting of open basal cisterns based onimaging.



The adult guidelines do not addressCSF drainage as a treatment for TBI.


In one study that was not included asevidence because it did not report func-tional outcomes, Anderson et al (6) ret-rospectively studied 80 children with se-

vere TBI, all of whom were treated withan ICP monitor or an external ventriculardrain (EVD) or both. The authors ob-served a fourfold increase in the risk ofcomplications for EVD as compared witha fiberoptic monitor (p � .004). Theseincluded: greater hemorrhagic complica-tions with the EVD in 12 of 62 (17.6%);fiberoptic in four of 62 (6.5%) (p � .025);malposition of the EVD requiring re-placement in six of 68 (8.8%); and infec-tion in one of 62 (1.5%). They concludedthat the use of an EVD may be associatedwith increased risk of complications fromhemorrhage and malposition.

Following earlier reports of an effecton ICP by drainage of CSF, Ghajar et al(7) performed a prospective study, with-out randomization, of the effect of CSFdrainage in adults with TBI. Treatment

was selected by the admitting neurosur-geon and, after evacuation of mass le-sions, patients either received ventriculo-stomies with drainage if ICP exceeded 15mm Hg along with medical management(group 1) or medical management only(group 2). The medical management con-sisted of mild hyperventilation to PCO2 35mm Hg, head-of-bed elevation, normo-volemia, and mannitol (although only onadmission). Patients in group 2 had noICP monitor of any kind. The outcomemeasures were mortality and degree ofdisability. Mortality was 12% in group 1vs. 53% in group 2. Of the patients ingroup 1, 59% were living independentlyat follow-up vs. 20% of group 2.

A study by Fortune et al (8) studiedthe effect of hyperventilation, mannitol,and CSF drainage on cerebral blood flow





Baldwin and Rekate,1991 (2)

Design: case seriesN � 5Age: 8–14 yrsProtocol: external ventricular drain, then lumbar

drain; lumbar drain should only be consideredin the case of refractory intracranialhypertension with a functioning externalventricular drain, open basal cisterns, and nomass lesion or shift on imaging studies

Class IIIPoor quality: no control

for confounders, smallsample size

3 of 5 survived; (1 moderate disability, 2good recovery) all had decrease in ICPafter lumbar drainage

Levy et al, 1995 (3) Design: case seriesN � 16Age: 1–15 yrsProtocol: external ventricular drain, then lumbar

drain; lumbar drain should only be consideredin the case of refractory intracranialhypertension with a functioning externalventricular drain, open basal cisterns, and nomass lesion or shift on imaging studies

Class IIIPoor quality: no control

for confounders, smallsample size

ICP lowered in 14 of 16; 2 of 16 died, bothof whom had uncontrolled ICP

Of 14 survivors, 8 had good recovery; 3moderate disability, 3 severe disability

Shapiro and Marmarou,1982 (4)

Design: case seriesN � 22Age: 2–15 yrsProtocol: external ventricular drainage, ICP/ PVI

measured

Class IIIPoor quality: small sample

size with narrowspectrum of patients

5 of 22 died 4 of 17 survivors were severelydisabled; 13 of 17 had a good outcome orwere moderately disabled

16 of 22 patients had PVI measured beforeand after therapy. Drainage increased PVIand decreased ICP in 14 of 16. 2 of the 5deaths were due to uncontrolledintracranial hypertension

New studyJagannathan et al,

2008 (5)Design: case seriesN � 96Age: 3–18 yrs, mean 15.1 yrsProtocol: compared management of ICP alone

(N � 34) vs. ICP along with surgery using anexternal ventricular drain (N � 23) oroperative treatment (N � 39; 14 mass lesionevacuation, 25 decompressive craniectomy)

Class IIIModerate quality: control

for confounders unclearfor ICP

ICP control achieved in 82 of 96 (85%) overall20 of 23 (87%) achieved ICP control with

external ventricular drain; of 3 not achievingICP control, 2 died, 1 had craniectomy

Refractory ICP was associated with 100%mortality; the method used to control ICPhad no correlation with mortality

ICP, intracranial pressure; PVI, pressure–volume index.


in TBI. Twenty-two patients were studiedwith a mean age of 24 yrs (range, 14–48yrs). Children were not reported sepa-rately. Although patient outcome was notreported, this study established that CSFdrainage, hyperventilation, and intermit-tent mannitol were all effective in reduc-ing ICP. They also found that mannitoluse increased cerebral blood flow, CSFdrainage had a negligible impact on ce-rebral blood flow, and hyperventilationdecreased cerebral blood flow.

VII. SUMMARY

Four class III studies provide the evi-dence base for this topic resulting in alevel III recommendation for the thera-peutic use of CSF drainage for the man-agement of intracranial hypertension.Two of these studies supported the use ofventricular CSF drainage. Although mostcommonly achieved with an EVD, a ran-domized controlled trial comparing theefficacy of treatment of intracranial hy-pertension in pediatric TBI with or with-out CSF drainage has not been carriedout. In the setting of refractory intracra-nial hypertension, a lumbar drain may beconsidered but only in conjunction with afunctional ventricular drain in patientswith open cisterns on imaging and with-out major mass lesions or shift. This wasalso supported only as a level III recom-mendation. A randomized controlled trial

comparing the different available ap-proaches to the treatment of refractoryintracranial hypertension has also notbeen carried out. Overall, it is possiblethat control of refractory ICP may be themost important aspect of treatment inchildren with severe TBI and may notdepend on a single modality of treatment,i.e., in this case, CSF drainage.


● Prospective studies as to the risks andbenefits of placement of an ICP moni-tor alone vs. placement of an EVD cath-eter.

● Prospective studies on the outcomebenefits of CSF drainage vs. other ther-apies.

● Role of surrogate markers of outcomeusing CSF drainage.

● Studies to compare CSF drainage withother therapeutic modalities used inTBI management such as osmolartherapy, barbiturates, or surgery.

● Studies about the technical aspects ofdrain use such as continuous vs. inter-mittent drainage, age-specific use, anduse related to mechanism of injury.

● Comparison of lumbar drainage withother second-tier therapies such as de-compressive craniotomy/craniectomy.

● Study of the potential role of subgalealdrainage in infants.

REFERENCES

1. Shore PM, Thomas NJ, Clark RS, et al: Con-tinuous versus intermittent cerebrospinalfluid drainage after severe traumatic brain in-jury in children: Effect on biochemical mark-ers. J Neurotrauma 2004; 21:1113–1122

2. Baldwin HZ, Rekate HL: Preliminary experi-ence with controlled external lumbar drainagein diffuse pediatric head injury. Pediatr Neu-rosurg 1991; 17:115–120

3. Levy DI, Rekate HL, Cherny WB, et al: Con-trolled lumbar drainage in pediatric head in-jury. J Neurosurg 1995; 83:453–460

4. Shapiro K, Marmarou A: Clinical applicationsof the pressure–volume index in treatment ofpediatric head injuries. J Neurosurg 1982; 56:819–825

5. Jagannathan J, Okonkwo DO, Yeoh HK, et al:Long-term outcomes and prognostic factorsin pediatric patients with severe traumaticbrain injury and elevated intracranial pres-sure. J Neurosurg Pediatr 2008; 2:240–249

6. Anderson RC, Kan P, Klimo P, et al: Compli-cations of intracranial pressure monitoring inchildren with head trauma. J Neurosurg 2004;101:53–58

7. Ghajar JBG, Hariri RJ, Patterson RH: Im-proved outcome from traumatic coma usingonly ventricular cerebrospinal fluid drainagefor intracranial pressure control. Adv Neuro-surg 1993; 21:173–177

8. Fortune JB, Feustel PJ, Graca L, et al: Effect ofhyperventilation, mannitol, and ventriculos-tomy drainage on cerebral blood flow afterhead injury. J Trauma 1995; 39:1091–1097;discussion 1097–1099


Chapter 11. Barbiturates

I. RECOMMENDATIONS

Strength of Recommendation: Weak.Quality of Evidence: Low from poor-


A. Level I


B. Level II


C. Level III

High-dose barbiturate therapy may beconsidered in hemodynamically stablepatients with refractory intracranial hy-pertension despite maximal medical andsurgical management.

When high-dose barbiturate therapy isused to treat refractory intracranial hy-pertension, continuous arterial bloodpressure monitoring and cardiovascularsupport to maintain adequate cerebralperfusion pressure are required.


III. OVERVIEW

Children with severe traumatic braininjury (TBI) may develop intracranial hy-pertension resistant to medical and sur-gical management. Reported rates of re-fractory intracranial hypertension vary(21% to 42%) (1–6). A recent study of132 children from South America foundthat 43% experienced refractory intracra-nial hypertension that was treated witheither high-dose barbiturates or decom-pressive craniectomy (7). Children havemore diffuse swelling and higher rates ofgeneralized hyperemia after severe TBIand compared with adults (8, 9) andyoung children have greater risk of in-tractable intracranial hypertension com-pared with older children (7).

Barbiturates lower intracranial pres-sure (ICP) when first-tier medical andsurgical management have not resultedin adequate control. However, cardiore-spiratory side effects are very commonand potentially toxic, including decreasedcardiac output, hypotension, and in-creased intrapulmonary shunt resultingin lower cerebral perfusion pressure andhypoxia. Thus, high-dose barbituratetherapy has been reserved for extremecases of intracranial hypertension resis-tant to first-tier medical and surgicalcare.

The use of high-dose barbiturates isbased on the logic that uncontrolled in-tracranial hypertension leads to ongoingsecondary brain injury and a high risk ofdeath or poor cognitive outcomes. Thus,control of ICP may improve patient sur-vival and outcome. A recent randomizedcontrolled study of 225 traumatic brain-injured children that used a tiered ther-apy protocol for management of ICP andcerebral perfusion pressure treated 16%of patients with barbiturates as a latetherapy (10). So, although high-dose bar-biturates are reserved for a high-riskgroup, use in North American pediatricsevere TBI care is common.

High-dose barbiturates lower ICPthrough two distinct mechanisms: sup-pression of metabolism and alteration ofvascular tone (11–13). Barbiturate ther-apy improves coupling of regional bloodflow to metabolic demands resulting inhigher brain oxygenation (14) with lowercerebral blood flow and decreased ICPfrom decreased cerebral blood volume.Other brain protective mechanisms in-clude inhibition of oxygen radical medi-ated lipid peroxidation as well as inhibi-tion of excitotoxicity (15).

Few studies have evaluated high-barbiturate pharmacokinetics and phar-macodynamics in head-injured children(16–19). Clearance appears to vary widelyand may be increased with duration ofbarbiturate administration (17). Barbitu-rate levels are poorly correlated with elec-trical activity (18, 19). Monitoring elec-trographic patterns to achieve burstsuppression is thought to be more reflec-tive of therapeutic effect than drug levels.

Near maximum reduction in cerebral me-tabolism and cerebral blood flow occurswhen burst suppression is induced.

High-dose barbiturates suppress me-tabolism and, although both use of pen-tobarbital and thiopental have been re-ported, there is insufficient informationabout comparative efficacy to recom-mend one over another, except in rela-tion to their particular pharmacologicproperties.

IV. PROCESS

For this update, MEDLINE wassearched from 1996 through 2010 (Ap-pendix B for search strategy), and resultswere supplemented with literature rec-ommended by peers or identified fromreference lists. Of 47 potentially relevantstudies, none were added to the existingtable and used as evidence for this ques-tion.


Two class III studies met the inclusioncriteria for this topic and provide evi-dence to support the recommendations(1, 20).

A study by Kasoff et al (1) reported acase series of 25 children with severe TBI.ICP was monitored in all patients andsurgical lesions treated. Standard care forelevated ICP (in 1988) included targetedhyperventilation to partial pressure of ar-terial carbon dioxide 25–30 mm Hg, ad-ministration of dexamethasone, andmannitol for osmolar therapy. If ICP re-mained �20 mm Hg, patients receivedpentobarbital as an initial bolus of 4–7mg/kg followed by a continuous infusionof 1–4 mg/kg/hr to a goal of clinicalcoma. All patients treated with high-dosebarbiturates (n � 11) were monitoredwith a pulmonary artery catheter. Goalswere to maintain ICP �20 mm Hg, cere-bral perfusion pressure �40 mm Hg, andhemodynamic stability. Ninety-one per-cent (ten of 11) required dopamine tomaintain blood pressure goals comparedwith 11% of children who did not receivebarbiturate therapy. Eighty-two percent(nine of 11) developed hypotension(mean arterial pressure �80 mm Hg).


DOI: 10.1097/PCC.0b013e31823f6696


The authors noted that children treatedwith high-dose barbiturates had dimin-ished cardiac output, lower systemic vas-cular resistance, decreased left ventricu-lar stroke volume, and increasedintrapulmonary shunt. Thirty-seven per-cent of children treated with high-dosebarbiturates died. The effects of barbitu-rate therapy on ICP and cerebral perfu-sion pressure were not reported.

A study by Pittman et al (20) reported acase series of 27 children with severe TBItreated with addition of pentobarbital if ICPremained �30 mm Hg after treatment withhyperventilation to a goal arterial carbondioxide 25–30 mm Hg, serum osmolality�300 mOsm, cerebrospinal fluid drain-age, and evacuation of surgical mass le-sions. A pentobarbital dose of 5 mg/kgfollowed by an infusion of 1–2 mg/kg/hrwith a goal barbiturate level of 30–40mg% was used. Fourteen children (52%)responded to pentobarbital and ICP wascontrolled (�20 mm Hg). Mortality inthis subgroup was not reported. Thirteenchildren had persistent intracranial hy-pertension despite addition of high-dosebarbiturates. Six died within 48 hrs ofstarting barbiturates. Seven children hadprolonged (�2 days) duration of elevatedICP with pressure �35 mm Hg for “ex-tended” periods of time. Glasgow Out-come Scale score was assessed at 6months and 1 yr after injury in the sevenchildren who survived. Three improvedto make a good recovery, two were left

with severe disability, and two were vegeta-tive. Among children with elevated ICP de-spite the addition of high-dose barbiturates,poor outcome (severe disability–death) wasreported in ten of 13 (77%). Glasgow Out-come Scale score was not reported for the14 children with controlled ICP. In thisseries of children with intractable intracra-nial hypertension, addition of high-dosebarbiturates controlled ICP in just over halfthe patients; however, the authors did notreport rates of cardiovascular complica-tions, mortality, or survival with neurologicmorbidity, preventing conclusions regard-ing barbiturate-related control of ICP andits effect on outcome. Among children withuncontrolled ICP despite addition of high-dose barbiturates, good survival was possi-ble (33%); however, this estimate is basedon a small number (n � 13).



There are no published studies of pro-phylactic barbiturate use in children withsevere TBI. The studies in adults are sum-marized in the Guidelines for the Man-agement of [Adult] Severe TraumaticBrain Injury (21). There are two random-ized clinical trials that examined earlyprophylactic administration of barbitu-

rates. Neither reported clinical benefit(22, 23). A study by Schwartz et al (22)did not define the lower age limit in theirstudy and although the study by Ward etal (23) included adolescents aged �12yrs, they did not separately report theeffects of early barbiturate therapy amongchildren. The study by Ward et al (23)reported that 54% of barbiturate-treatedpatients developed hypotension com-pared with 7% of control patients. Hypo-tension is a well-described risk factor formortality and neurologic morbidity inhead-injured pediatric patients (24).

A study by Eisenberg et al (25) re-ported a multicentered randomized clin-ical trial of high-dose barbiturates in se-verely head-injured patients withintractable intracranial hypertension. Pa-tient age ranged from 15 to 50 yrs butresults for the pediatric patients were notseparately reported. ICP was the primaryoutcome and patients in the controlgroup could be crossed over to barbitu-rate therapy at prespecified ICP failurepoints in the study. Among 68 study pa-tients, 32 were randomized to high-dosebarbiturate therapy and 32 of the 36 con-trol patients ultimately crossed over tobarbiturate therapy. Therapy before initi-ating barbiturates included hyperventila-tion, neuromuscular blockade, sedation,osmolar therapy with mannitol, steroids,and cerebrospinal fluid drainage whenpossible. The odds of ICP control weretwofold greater in the barbiturate group



Studies from previous guidelinesKasoff et al, 1988 (1) Design: case series Class III 4 of 11 (36%) died

N � 11Age: 3 months to 17 yrsProtocol: patients were treated with

high-dose pentobarbital forintracranial hypertension despitefirst-line therapy

Outcome: hemodynamicmonitoring, use of inotropicagents, and hospital mortalitywere assessed

Poor quality: observational studywith no control forconfounding

Hypotension (mean arterial pressure �80mm Hg) occurred in 9/11 (82%)

Systemic vascular resistance wasdepressed, left ventricular stroke workdecreased, cardiac index depressed,pulmonary shunt fraction increasedwith pentobarbital therapy

Pittman et al, 1989 (20) Design: case seriesN � 27Age: 2 months to 15 yrsProtocol: patients were treated with

high-dose pentobarbital forintracranial hypertension despitefirst-line therapy

Outcome: cerebral perfusionpressure, intracranial pressure,and 1-yr outcome were reported

Class IIIPoor quality: observational study

with no control forconfounding

14 of 27 (52%) achieved intracranialpressure �20 mm Hg

Of 13 with persistently elevatedintracranial pressure, 6 died (22%), 2were vegetative, 2 had moderaterecovery, and 3 good recovery


and survival 1 month after injury was92% for responders compared with 17%for nonresponders. At 6 months, 36% ofresponders were vegetative comparedwith 90% of nonresponders. The cross-over design of this study precludes firmconclusions about the efficacy of high-dose barbiturates to control intractableICP and improve outcome.

A number of barbiturate dosing regi-mens have been reported. The study byEisenberg et al (25) used the followingregimen for pentobarbital: a loading dose10 mg/kg over 30 mins, then 5 mg/kgevery hour for three doses, and a main-tenance dose of 1 mg/kg/hr.


The Cochrane Review (26) has apooled analysis from three trials of bar-biturates and calculated the pooled riskestimated for barbiturate therapy onmortality was 1.09 (95% confidence in-terval, 0.81–1.47). They found that one infour barbiturate-treated patients devel-oped hypotension and concluded thatthere is no evidence that barbituratetherapy in patients with acute severehead injury improves outcome.

A study by Nordby et al (27) usedthiopental in a study that included chil-dren and adults with loading doses of20 –30 mg/kg and a maintenance of 3–5mg/kg/hr. Doses of thiopental were re-duced if blood pressure fell or ICP was�25 mm Hg.

The duration and optimal method todiscontinue high-dose barbiturates havenot been studied. Clinicians typically waitfor at least a 24-hr period of ICP controlwithout sustained elevations with stimu-lation before beginning to taper the bar-biturate infusion (28).

Refractory IntracranialHypertension

Use of high-dose barbiturates to treatelevated ICP in children with severe TBIhas been reported since the 1970s. Mar-shall et al (29) were the first to reportthat both control of ICP and outcomeswere improved with use of barbiturates;however, patient age was not specified inthe report, which was a case series of 25patients with ICP �40 mm Hg treatedwith high-dose pentobarbital. When ICPwas controlled, mortality was signifi-cantly reduced compared with patientswith persistently elevated ICP despite

addition of barbiturate therapy (21% vs.83%).

VII. SUMMARY

Studies regarding high-dose barbitu-rate administration to treat severe TBI inpediatric patients are limited to two caseseries (class III evidence), which limitsfirm conclusions. The evidence suggeststhat barbiturates effectively lower ICPamong a subset of children with intrac-table intracranial hypertension; however,a beneficial effect on survival or improvedneurologic outcome has not been estab-lished. Administration of high-dose bar-biturates is commonly associated withhypotension and the need for blood pres-sure support in both children and adults.Studies have not evaluated whether therisk of cardiovascular side effects differ bypatient age. Administration of high-dosebarbiturates to infants and children re-quires appropriate monitoring to avoidand rapidly treat hemodynamic instabil-ity and should be supervised by experi-enced critical care providers.

There is no evidence to support use ofprophylactic barbiturates to prevent in-tracranial hypertension or for neuropro-tective effects in children.


● High-dose barbiturates are used totreat intractable intracranial hyperten-sion in children. Studies are needed tobetter quantify their effect on ICP,long-term outcome, the risk of concur-rent hemodynamic instability, andtheir association with morbidity andmortality. In addition, direct compari-son to other therapies for refractoryintracranial hypertension is needed.

● Age-dependent toxicity of high-dosebarbiturates has not been evaluated.

● The effectiveness of high-dose barbitu-rate therapy for children with differentanatomical lesions, including diffuseswelling, has not been evaluated foreither control of ICP or outcome.

● High-dose barbiturate therapy to con-trol intractable intracranial hyperten-sion among infants with abusive headinjury has not been described. Theseinfants have poor cognitive outcomes(30).

REFERENCES

1. Kasoff SS, Lansen TA, Holder D, et al: Ag-gressive physiologic monitoring of pediatric

head trauma patients with elevated intracra-nial pressure. Pediatr Neurosci 1988; 14:241–249

2. Berger MS, Pitts LH, Lovely M, et al: Out-come from severe head injury in childrenand adolescents. J Neurosurg 1985; 62:194–199

3. Bruce DA, Schut L, Bruno LA, et al: Outcomefollowing severe head injuries in children.J Neurosurg 1978; 48:679–688

4. Cordobes F, Lobato RD, Rivas JJ, et al: Post-traumatic diffuse brain swelling: Isolated orassociated with cerebral axonal injury. Clin-ical course and intracranial pressure in 18children. Childs Nerv Syst 1987; 3:235–238

5. Eder HG, Legat JA, Gruber W: Traumaticbrain stem lesions in children. Childs NervSyst 2000; 16:21–24

6. Esparza J, M-Portillo J, Sarabia M, et al:Outcome in children with severe head inju-ries. Childs Nerv Syst 1985; 1:109–114

7. Guerra SD, Carvalho LF, Affonseca CA, et al:Factors associated with intracranial hyper-tension in children and teenagers who suf-fered severe head injuries. J Pediatr (Rio J)2010; 86:73–79

8. Aldrich EF, Eisenberg HM, Saydjari C, et al:Diffuse brain swelling in severely head-injured children. A report from the NIHTraumatic Coma Data Bank. J Neurosurg1992; 76:450–454

9. Zwienenberg M, Muizelaar JP: Severe pediat-ric head injury: The role of hyperemia revis-ited. J Neurotrauma 1999; 16:937–943


11. Demopoulos HB, Flamm ES, PietronigroDD, et al: The free radical pathology and themicrocirculation in the major central ner-vous system disorders. Acta Physiol ScandSuppl 1980; 492:91–119

12. Kassell NF, Hitchon PW, Gerk MK, et al:Alterations in cerebral blood flow, oxygenmetabolism, and electrical activity producedby high dose sodium thiopental. Neurosur-gery 1980; 7:598–603

13. Piatt JH Jr, Schiff SJ: High dose barbituratetherapy in neurosurgery and intensive care.Neurosurgery 1984; 15:427–444

14. Chen HI, Malhotra NR, Oddo M, et al: Bar-biturate infusion for intractable intracranialhypertension and its effect on brain oxygen-ation. Neurosurgery 2008; 63:880–886; dis-cussion 886–887

15. Goodman JC, Valadka AB, Gopinath SP, et al:Lactate and excitatory amino acids measuredby microdialysis are decreased by pentobar-bital coma in head-injured patients. J Neu-rotrauma 1996; 13:549–556

16. Russo H, Bressolle F, Duboin MP: Pharma-cokinetics of high-dose thiopental in pediat-ric patients with increased intracranial pres-sure. Ther Drug Monit 1997; 19:63–70

17. Russo H, Bres J, Duboin MP, et al: Variabilityof thiopental clearance in routine critical


care patients. Eur J Clin Pharmacol 1995;48:479–487

18. Schickendantz J, Funk W, Ittner KP, et al:Elimination of methohexitone after long-term, high-dose infusion in patients withcritically elevated intracranial pressure. CritCare Med 1999; 27:1570–1576

19. Turcant A, Delhumeau A, Premel-Cabic A, etal: Thiopental pharmacokinetics under con-ditions of long-term infusion. Anesthesiol-ogy 1985; 63:50–54

20. Pittman T, Bucholz R, Williams D: Efficacy ofbarbiturates in the treatment of resistant in-tracranial hypertension in severely head-injured children. Pediatr Neurosci 1989; 15:13–17

21. Bratton SL, Chestnut RM, Ghajar J, et al:Guidelines for the management of severetraumatic brain injury. XI. Anesthetics, anal-

gesics, and sedatives. J Neurotrauma 2007;24(Suppl 1):S71–S76

22. Schwartz ML, Tator CH, Rowed DW, et al:The University of Toronto head injury treat-ment study: A prospective, randomized com-parison of pentobarbital and mannitol. CanJ Neurol Sci 1984; 11:434–440

23. Ward JD, Becker DP, Miller JD, et al: Failureof prophylactic barbiturate coma in the treat-ment of severe head injury. J Neurosurg1985; 62:383–388

24. Zebrack M, Dandoy C, Hansen K, et al: Earlyresuscitation of children with moderate-to-severe traumatic brain injury. Pediatrics2009; 124:56–64

25. Eisenberg HM, Frankowski RF, Contant CF, etal: High-dose barbiturate control of elevatedintracranial pressure in patients with severehead injury. J Neurosurg 1988; 69:15–23

26. Roberts I: Barbiturates for acute traumaticbrain injury. Cochrane Database Syst Rev2000; 2:CD000033

27. Nordby HK, Nesbakken R: The effect of highdose barbiturate decompression after severehead injury. A controlled clinical trial. ActaNeurochir 1984; 72:157–166

28. Raphaely RC, Swedlow DB, Downes JJ, et al:Management of severe pediatric head trauma.Pediatr Clin North Am 1980; 27:715–727

29. Marshall LF, Smith RW, Shapiro HM: Theoutcome with aggressive treatment in severehead injuries. Part II: Acute and chronic bar-biturate administration in the managementof head injury. J Neurosurg 1979; 50:26–30

30. Keenan HT, Runyan DK, Nocera M: Child out-comes and family characteristics 1 year aftersevere inflicted or noninflicted traumatic braininjury. Pediatrics 2006; 117:317–324


Chapter 12. Decompressive craniectomy for the treatment ofintracranial hypertension

I. RECOMMENDATIONS

Strength of Recommendations: Weak.Quality of Evidence: Low, from poor


A. Level I


B. Level II


C. Level III

Decompressive craniectomy (DC) withduraplasty, leaving the bone flap out, maybe considered for pediatric patients withtraumatic brain injury (TBI) who areshowing early signs of neurologic deteri-oration or herniation or are developingintracranial hypertension refractory tomedical management during the earlystages of their treatment.


III. OVERVIEW

DC in the setting of TBI is a contro-versial procedure that has recently be-come widely considered as a treatmentoption. It may be performed concomi-tantly with the removal of a mass lesionto either treat observed brain swelling oract as prophylaxis of anticipated swelling(secondary DC). Alternatively, it may beperformed as a standalone procedure forthe purpose of treating cerebral hernia-tion or established intracranial hyperten-sion, wherein the timing of the decom-pression may be predicated on theclinical examination, course of neuro-logic deterioration, initial degree of intra-cranial pressure (ICP) elevation, or theresistance of that elevation to various

thresholds of medical treatment (primaryDC). These two conditions of employ-ment are actually quite different and it isthe second (DC as a primary treatmentfor cerebral swelling) that is the focus onthis section.

The nature of the procedure varieswidely. It may consist of uni- or bilateralsubtemporal decompressions, hemi-spheric craniectomies of varying sizes(from relatively small to quite expansive),circumferential craniectomy, or bifrontalcraniectomy. The choice of proceduremay depend on the underlying pathology,as demonstrated on computed tomogra-phy imaging, or may simply be focusedon developing the maximum possiblecompliance compartment. The manage-ment of the underlying dura also mayvary, ranging from leaving it intactthrough simple scoring to opening itwidely (with or without expansive dura-plasty). Furthermore, the treatment ofthe dura may vary independently with thechoice of bony decompressive procedure.

With respect to the use of DC for ICPcontrol in adults, two randomized con-trolled trials were underway, the DECRATrial (1) (international multicenter ran-domized controlled trial (on Early De-compressive Craniectomy in TraumaticBrain Injury), which recently reportedtheir findings (2) of reduced ICP but sig-nificantly worsened outcomes, and theRescueICP Trial (3) (randomized evalua-tion of surgery with craniectomy for un-controllable elevation ICP). No similarstudies are ongoing for the pediatric pop-ulation.

IV. PROCESS

For this update, MEDLINE wassearched from 1996 through 2010 (Ap-pendix B for search strategy), and resultswere supplemented with literature rec-ommended by peers or identified fromreference lists. Of 20 potentially relevantstudies, seven new studies were includedas evidence for this topic.


Eight class III studies met the inclu-sion criteria for this topic and provideevidence to support the recommenda-tions (4–11). These studies vary in criti-cal areas such as their selection criteriafor DC, the DC techniques used, and theiroutcome parameters. In addition, none ofthem defined the study population to anextent adequate to allow rigorous inter-study comparisons. The lack of internalcomparison groups or matched controlsweakens the analyses that can be applied.

Is Decompressive CraniectomyEffective in Lowering ICP?

The issue with respect to the efficacyof DC in lowering ICP is not the statisti-cal significance of the change in ICP frombefore surgery to the postoperative statebut rather it is in lowering severely ormedically intractable ICP elevation withrespect to the treatment threshold suchthat intracranial hypertension is no lon-ger encountered (optimal outcome) or iseasily controlled after surgery.

A study by Hejazi et al (6) was per-formed investigating early unilateral orbilateral DC with duraplasty for GlasgowComa Scale score of 3–5 in seven pediat-ric patients with TBI within 70 mins fromtrauma resulting from “massive” bilateralor unilateral swelling, compressed supra-tentorial ventricular spaces, and perimes-encephalic cisterns. The DC was fronto-temporal and did not include the parietaland occipital regions. A low craniectomywas performed in all patients to decom-press the brainstem. The initial ICP ex-ceeded 45 mm Hg in all patients. In six ofthe seven, ICP remained �20 mm Hgafter surgery. Persistent intracranial hy-pertension (although not to the level ofpreoperative) in the one patient was con-trolled with medical therapy. This sug-gests that DC might be effective in con-trolling ICP.

A study by Ruf et al (9) was also per-formed on unilateral or bilateral DC withduraplasty when the ICP exceeded 20 mmHg for �30 mins in six pediatric patients


DOI: 10.1097/PCC.0b013e31823f672e




Study from previousguidelines

Cho et al, 1995 (4) Design: case seriesN � 13Age: 2–14 monthsProtocol: medical treatment in first 4 and DC

in 9ICP and scores on COS measured between 6

months and 6 yrs postinjury (mean, 3.2 yrs)DC: bifrontal DC for diffuse swelling, or large

unilateral frontotemporoparietal DCs forunilateral hemispheric swelling

Class IIIPoor quality: no control for confounders,

very small sample and no powercalculation

In the surgical group, DC lowered themean ICP measurements from 54.9mm Hg to 11.9 mm Hg; effect ofmedical treatment on ICP was notreported

For the medically treated group, scoreson the COS, measured at a mean of3.2 years (range, 6 months to 6yrs), were 2 dead (COS 5) and 2vegetative (COS 4); for the surgicalgroup, 2 patients had an “excellent”recovery (COS 1), 2 had a moderaterecovery (COS 2), 4 had severedisability (COS 3), and 1 wasvegetative; notably, although DC wasperformed based on ICP elevationalone, a mean of 32 mL of subduralblood was removed during thesurgery

New studiesFigaji et al, 2003 (5) Design: case series

N � 5Age: 5–12 yrsProtocol: DC for clinical deterioration in patients

presenting with or deteriorating rapidly toGCS � 8 in intensive care unit; ICP notmonitored before surgery

Outcome: GOSDC: unilateral craniotomy with duraplasty either

leaving the bone out or loosely suturing it inplace (floating flap)


very small sample, and no powercalculation

All patients had early clinicalimprovement after surgery and wereGOS 4 or 5 at long term follow-up (14–40 months)

In the 4 patients with postoperativeICP monitoring, 2 had no ICPelevations and 2 had mild, easilycontrolled elevations

Hejazi et al, 2002 (6) Design: retrospective case seriesN � 7Age: 5–14 yrsGCS: 3–5 on admission and bilateral swelling

with compression of the perimesencephaliccisterns on CT; initial ICP �45 mm Hg in allpatients

Protocol: patients with traumatic brain injurytreated with early DC

Outcome: survival, ICPDC: unilateral craniectomy, frontal temporal only

with duraplasty leaving the bone out orbilateral craniectomy with stellate duralopening


very small sample and no powercalculation

All patients survived despite severebaseline intracranial hypertension;decompression decreased ICP from�45 mm Hg to � 20 mm Hgimmediately and it remainedcontrolled in 6 of 7 patients; onepatient later developed intracranialhypertension but not to the levelpresent before decompression

All patients achieved a “completerecovery” on follow-up of �8months although this is not defined

Jagannathan et al, 2007 (7) Design: retrospective case seriesN � 23Age: mean 1.9 yrs (2 patients: 19 yrs old and 21

of 23 patients �19 yrs old)GCS: mean 4.6 (3–9)Protocol: patients with traumatic brain injury

treated with DC done for either 1) ICP �20mm Hg refractory to maximal medicaltherapy; or 2) mass lesion

Outcome: long-term functional outcome andindependence levels were evaluated using theGOS and a Likert patient quality-of-life ratingscale

DC: large, wide with duraplasty; unilateral forhemispheric swelling or bifrontal for diffuseswelling; in bifrontal, the sagittal suture wassuture ligated and falx sectioned



Survival rate of 70%; mortality wasseen primarily in patients withmultisystem trauma

ICP control in 19 of 23 patients; highICP associated with increasedmortality; mean follow-up usingGOS over 5 years was 4.2 (range, 1–5); majority had “good”outcomes (17 of 23) at 2 yrs 13 of17 survivors returned to school


with severe TBI. In five of the six, ICPremained �20 mm Hg after surgery. Per-sistent intracranial hypertension in thesixth patient prompted a return to sur-gery for a contralateral DC, which re-sulted in sustained ICP control. Thissuggests that DC might be effective incontrolling ICP. Unfortunately, furtherinformation on how the choice of oper-ation was made in these patients islacking.

A study by Kan et al (8) was performedto investigate a large unilateral DC withduraplasty in pediatric patients with TBI,either in conjunction with the removal ofa mass lesion (45 patients) or primarilyfor brain swelling (six patients, five forrefractory ICP �25 mm Hg, and one forherniation). The six patients relevant tothis topic were very severely injured with

low admission Glasgow Coma Scalescores, evidence of herniation, or severesecondary insults common among them.For these six patients, three of the fourwho received postoperative ICP monitor-ing had sustained ICP values �20 mmHg. The fourth had intracranial hyper-tension requiring further treatment. Fiveof the six patients died.

A study by Rutigliano et al (10) wasperformed which was a retrospective caseseries of six patients with TBI of age �20yrs who underwent DC for elevated ICP(without a specified definition), whichwas refractory to guidelines-based treat-ment. Five of these patients were �18 yrsof age and could be analyzed separately.They performed wide bifrontal/biparietalcraniectomies with duraplasty. Four ofthe five had no postoperative ICP eleva-

tions. The fifth patient required a returnto surgery for intracranial hypertensionwhereupon debridement of the contusedbrain resulted in resolution.

A study by Jagannathan et al (7) wasperformed as a retrospective case series of23 patients with TBI of age �20 yrs whounderwent DC for initial mass lesion re-quiring evacuation or elevated ICP (�20mm Hg), which was refractory to guide-lines-based treatment. Twenty-one ofthese patients were �18 yrs of age andcould be analyzed separately. They per-formed wide bifrontal/biparietal craniec-tomies with duraplasty and sectioning ofthe falx or unilateral DC if there was amass lesion or unilateral swelling. Ten ofthe 23 patients underwent early DC, 11had later DC, and two even later as aresult of medical instability. Mean ICP



Kan et al, 2006 (8) Design: case seriesN � 6Age: 0.3–14 yrsGCS: mean 4.6Protocol: DC performed in the absence of mass

lesion; all 6 with very severe injuries; DC donein 5 for refractory ICP �25 mm Hg and 1 forherniation

Outcome: mortality and ICPDC: large unilateral craniectomy with duraplasty

Class IIIPoor quality: no control for confounders

5 of 6 patients died3 of the 4 patients with postoperative ICP

monitoring had ICP �20 mm Hg

Ruf et al, 2003 (9) Design: retrospective case seriesN � 6GCS: 3–7Age: 5–11 yrsProtocol: DC for refractory ICP �20 mm Hg for

�30 minsOutcome: 6-month survival and neurological

assessmentDC: unilateral or bilateral craniectomy

(depending on CT) with duraplasty



3 patients were without disability; 2had mild to moderate deficits at6-month follow-up

Postoperative ICP �20 mm Hg in 5 of6 patients; sixth patient requiredcontralateral subsequent DC, thenICP was maintained at � 20 mm Hg

Rutigliano et al, 2006 (10) Design: retrospective case seriesN � 6Age: �20 yrs with 5 �18 yrs (range, 12–15 yrs)

and having distinct dataProtocol: DC done for refractory “elevated ICP”Outcome: Functional Independence Measure

score and ICPDC � bifrontal craniectomies with duraplasty



All 5 had Functional IndependenceMeasurement scores of independentor minimal assistance at discharge

5 of the 6 patients had nopostoperative ICP elevations; 1 hadICP elevations requiring a secondsurgery for débridement, with nosubsequent ICP elevations

Skoglund et al, 2006 (11) Design: retrospective case seriesN � 19Age: 8 �18 yrs (range, 7–16 yrs) and having

distinct dataGCS: mean 7 (3–15), with deterioration, evidence

of herniation, or refractory ICPProtocol: DC done for either 1) ICP �20 mm Hg

refractory to Lund therapy; or 2) acuteneurologic deterioration immediately aftertrauma with CT showing diffuse edema

Outcome: GOS at 1 yrDC: large with duraplasty; unilateral for

hemispheric swelling or bifrontal for diffuseswelling

Class IIIModerate quality: unclear if outcome

assessment methods were unbiased

At � 1 yr follow-up, 3 patients withGOS � 5, 1 GOS � 4, 3 GOS � 3,and 1 dead; 5 of these patients withneurologic deterioration or pupillarychanges at the time of surgery

DC, decompressive craniectomy; ICP, intracranial pressure; COS, Children’s Outcome Scale; GCS, Glasgow Coma Scale; GOS, Glasgow Outcome Scale;CT, computed tomography.


reduced from 30 mm Hg preoperativelyto 18 mm Hg postoperatively. Nineteen of23 patients had control of postoperativeICP elevations with maximal medicalmanagement. Two patients continued tohave refractory ICP.

A study by Cho et al (4) was a caseseries of 23 children �2 yrs of age pre-senting with nonaccidental trauma. Chil-dren were included based on their ICPregardless of their presenting level ofconsciousness. A subgroup of 13 patientswith a Children’s Coma Score equivalentto severe on the Glasgow Coma Scale,and ICP values �30 mm Hg, were treatedmedically (n � 4) or with DC (n � 9)based on either family wishes or beingadmitted before DC became a routinepart of treatment for this disease. On thenine surgical patients, bifrontal DC wasperformed for diffuse swelling or largeunilateral frontotemporoparietal DCs forunilateral hemispheric swelling. They in-cluded a section of the anterior sagittalsinus and an expansive duraplasty. Thedecompression was performed within 24hrs of injury in the majority. In the sur-gical group, DC lowered the mean ICPmeasurements from 54.9 mm Hg to 11.9mm Hg.

In summary, it appears that DC maybe effective in lowering ICP to below thethreshold for treatment in patients re-fractory to medical management. Thislimited conclusion would add some sup-port to choosing to perform DC for ICPcontrol when intracranial hypertension isresistant to nonsurgical management andthe ICP levels maintained are consideredhazardous to the patient.

Does DecompressiveCraniectomy Improve ClinicalOutcomes?

This section focuses on whether DCperformed for severe or intractable intra-cranial hypertension or clinical hernia-tion is associated with a beneficial influ-ence on outcome.

All of the studies in this section areretrospective case series. All used retro-spectively collected data, except for theRutigliano et al (10) study that used aprospectively collected database, whichwas not designed specific to the questionof DC. None of them have internal ormatched external controls and there wereno randomized controlled trials. Com-mon to all of these studies is the absenceof sufficient data on the injury character-istics of the study group to predict their

outcomes independent of the surgical de-compression using predictive modeling.

A study by Hejazi et al (6) reportedthat all of the patients with early DC hada “complete recovery” although this isnot defined. There was no mortality andcomplication rate was low with only sub-dural effusions in four of seven.

A study by Figaji et al (5) reported“early [postoperative] clinical improve-ment” in their decompressed patients. Allfive cases had Glasgow Outcome Scalescores of 4–5 at 14- to 40-month follow-up. The patients had not had preoperativeICP monitoring and had DC performedfor clinical deterioration. The authors feltthat the outcomes were better than ex-pected given that each of the patientshad an initial Glasgow Coma Scalescore �8, each had a documented sec-ondary deterioration, which was be-lieved to be the result of raised ICP,pupillary abnormalities were seen infour, and all demonstrated obliterationof the perimesencephalic cisterns (dif-fuse injury III and IV).

A study by Ruf et al (9) studied sixpediatric patients with TBI undergoingDC for refractory ICP �20 mm Hg. Oneof the six was a posterior fossa DC to treatswelling from a cerebellar contusion. At 6months, all patients had survived, threebeing described as “normal” and the oth-ers having mild-to-moderate residual def-icits.

A study by Rutigliano et al (10) de-scribed six pediatric patients with TBIwho underwent DC. Five of these patientswere �18 yrs of age. A large bilateralfrontoparietal DC with duraplasty wasperformed for “elevated ICP” refractory totier 1 and tier 2 medical management.They reported early signs of clinical im-provement and discharge Functional In-dependence Measurement scores of inde-pendent or minimal assistance for all fivepatients.

A study by Jagannathan et al (7) de-scribed 21 pediatric patients with TBI af-ter undergoing DC either incidentally af-ter evacuation of a mass lesion or fordiffuse swelling refractory ICP to medicalmanagement. Eighteen of 23 were donefor refractory ICP to maximal medicalmanagement, three of whom had pupil-lary changes and did not survive DC.They reported an overall 22% mortalityrate despite ICP �20 mm Hg in two ofthe five patients who died. Mean fol-low-up was 62 months (range, 11–126months) and the mean Glasgow OutcomeScale score was 4.2 (range, 1–5). The

mean score on the quality-of-life ques-tionnaires was 4 (maximum, 5) in theability to perform activities of daily living,general cognition, interpersonal behav-ior, and emotional behavior (range,1–4.75).

In the Cho et al (4) case series, chil-dren �2 yrs of age with severe TBI fromnonaccidental trauma and ICP values�30 mm Hg were treated with medically(n � 4) or with decompressive craniot-omy (n � 9). For the medically treatedgroup, scores on the Children’s OutcomeScale (COS), measured at a mean of 3.2yrs (range, 6 months to 6 yrs), revealedtwo dead (COS 5) and two vegetative(COS 4). For the surgical group, two pa-tients had an “excellent” recovery (COS1), two had a moderate recovery (COS 2),four had severe disability (COS 3), andone was vegetative. Notably, although DCwas performed based on ICP elevationalone, a mean of 32 mL of subdural bloodwas removed during the surgery.

Two studies reported less favorableoutcomes (8, 11). A study by Skoglund etal (11) studied 19 patients with TBI, ofwhom eight were �18 yrs, treated withDC for either refractory ICP �20 mm Hgor acute neurologic deterioration imme-diately after trauma with computed to-mography scan showing diffuse edema.All patients were medically managed us-ing the Lund approach. Five of the eightpediatric patients had neurologic deteri-oration or pupillary changes at the timeof surgery. Outcome at �1 yr after sur-gery was three patients with GlasgowOutcome Scale score of 5, one with Glas-gow Outcome Scale score of 4, three withGlasgow Outcome Scale score of 3, andone death.

A study by Kan et al (8) described 51pediatric patients with TBI undergoingDC, although the craniectomy was inci-dental to surgery to evacuate a mass le-sion in 45. Five cases were done for re-fractory ICP �25 mm Hg and the sixthfor clinical herniation. These patientswere very severely injured. Three wereGlasgow Coma Scale score 3 on admis-sion, three were bilaterally fixed and di-lated, and two others had a unilateralfixed and dilated pupil. The sixth patientpresented with profound hypotension.They reported an 83% mortality rate de-spite ICP �20 mm Hg in three of the fourpatients monitored after surgery. Five ofthe six patients died.

Given the paucity of descriptive statis-tics contained within these studies, it isimpossible to accurately compare the pa-


tients studied between these various pa-pers. Adding in the differences in triggercriteria for DC, variations in DC technique,and the wide variations in outcome mea-surements, no more than simple, qualita-tive summaries may be made. Given theseverity of injury of these children and thephysiological abnormalities required to be-come candidates for DC, cautious interpre-tation of these outcomes suggests that DCmay be effective in improving outcome inpatients with medically intractable intra-cranial hypertension.



The Guidelines for the Surgical Man-agement of TBI (12), published in 2006,found no class I or II evidence on which tobase level I or II recommendations. Thelevel III-equivalent recommendations withrespect to DC were based on class III liter-ature, the most prominent of which werethe reports of Polin et al (13) and Taylor etal (14) (briefly reviewed subsequently).

The recommendations from the adultguidelines regarding DC were:

● Bifrontal DC within 48 hrs of injury isa treatment option for patients withdiffuse, medically refractory posttrau-matic cerebral edema and resultant in-tracranial hypertension.

● Decompressive procedures, includingsubtemporal decompression, temporallobectomy, and hemispheric DC, aretreatment options for patients with re-fractory intracranial hypertension anddiffuse parenchymal injury with clini-cal and radiographic evidence for im-pending transtentorial herniation.

Of note, the recently completed DECRAstudy by Cooper et al (2) for adults withdiffuse severe TBI showed that ICP could beeffectively reduced with early bifrontotem-poroparietal DC but, interestingly, out-comes were worse in the surgery groupthan the clinical management group alone.


Indications From the 2009 CochraneReview on Decompressive Craniectomy.In the 2009 update of the Cochrane Re-view on DC (15), the author found onlyone publication of sufficient rigor to in-clude, that of Taylor et al (14), which

studied a pediatric TBI group. It was con-cluded that “despite the wide confidenceinterval for death and the small samplesize of this one identified study, the treat-ment may be justified in patients belowthe age of 18 yrs when maximal medicaltreatment has failed to control ICP.” Withrespect to the current evidence report,however, this paper must be excluded as aresult of its inclusion of patients withadmissions scores above the cutoff (�8).

VII. SUMMARY

Eight small class III case series suggestthat large decompressive surgeries with du-raplasty may be effective in reversing earlysigns of neurologic deterioration or herni-ation, and in treating intracranial hyper-tension refractory to medical management,and that these effects may be correlatedwith improving outcomes in the criticallyill pediatric patients who develop such in-dications. Limited evidence suggests thatduraplasties, when done, should be large,and consideration should be given to re-moving the bone rather than “floating” it insitu. There is insufficient evidence to allowdefining the patient characteristics that either1) optimize the beneficial effects of these pro-cedures or 2) render them ineffective.


● A primary focus on future researchshould be performing a randomizedcontrolled trial on DC as a method ofcontrolling increased ICP in pediatricpatients with TBI.

● Given the infrequency with which pedi-atric patients with TBI are admitted toany individual center, it would be veryuseful to develop a prospective pediatricTBI database to facilitate class II investi-gations into many of the variables relevantto DC (such as timing, size and placement,and technique), which are unlikely to everbe subject to class I study.

● It would be very useful if the investiga-tors involved in the two adult DC trials,the DECRA trial (1) and the Rescue ICPtrial (3), both of which enrolled patientsoverlapping with the pediatric age group,would parse out this group for separatesubgroup analysis of efficacy and techni-cal details. It would be valuable to designor determine standardized and practicaltechniques to quantify the physiologicalchanges induced by DC, both as a clini-cally useful measure of efficacy and as aresearch parameter.

REFERENCES

1. Rosenfeld JV, Cooper DJ, Kossmann T, et al:Decompressive craniectomy. J Neurosurg2007; 106:195–196; author reply 197

2. Cooper DJ, Rosenfeld JV, Murray L, et al:Decompressive craniectomy in diffuse trau-matic brain injury. N Engl J Med 2011; 364:1493–1502

3. Hutchinson PJ, Corteen E, Czosnyka M, et al:Decompressive craniectomy in traumaticbrain injury: The randomized multicenterRESCUEicp study (www.RESCUEicp.com).Acta Neurochir Suppl 2006; 96:17–20

4. Cho DY, Wang YC, Chi CS. Decompressivecraniotomy for acute shaken/impact baby syn-drome. Pediatr Neurosurg 1995; 23:192–198

5. Figaji AA, Fieggen AG, Peter JC: Early de-compressive craniotomy in children with se-vere traumatic brain injury. Childs Nerv Syst2003; 19:666–673

6. Hejazi N, Witzmann A, Fae P: Unilateral de-compressive craniectomy for children withsevere brain injury. Report of seven cases andreview of the relevant literature. Eur J Pedi-atr 2002; 161:99–104

7. Jagannathan J, Okonkwo DO, Dumont AS, etal: Outcome following decompressive craniec-tomy in children with severe traumatic braininjury: A 10-year single-center experience withlong-term follow up. J Neurosurg 2007; 106:268–275

8. Kan P, Amini A, Hansen K, et al: Outcomesafter decompressive craniectomy for severetraumatic brain injury in children. J Neuro-surg 2006; 105:337–342

9. Ruf B, Heckmann M, Schroth I, et al: Earlydecompressive craniectomy and duraplastyfor refractory intracranial hypertension inchildren: results of a pilot study. Crit Care2003; 7:R133–R138

10. Rutigliano D, Egnor MR, Priebe CJ, et al:Decompressive craniectomy in pediatric pa-tients with traumatic brain injury with in-tractable elevated intracranial pressure. J Pe-diatr Surg 2006; 41:83–87; discussion 83–87

11. Skoglund TS, Eriksson-Ritzen C, Jensen C,et al: Aspects on decompressive craniectomyin patients with traumatic head injuries.J Neurotrauma 2006; 23:1502–1509

12. Bullock MR, Chesnut R, Ghajar J, et al: Surgicalmanagement of acute epidural hematomas. Neu-rosurgery 2006; 58:S7–15; discussion SI–SIV

13. Polin RS, Shaffrey ME, Bogaev CA, et al:Decompressive bifrontal craniectomy in thetreatment of severe refractory posttraumaticcerebral edema. Neurosurgery 1997; 41:84–92; discussion 92–94

14. Taylor A, Butt W, Rosenfeld J, et al: A random-ized trial of very early decompressive craniec-tomy in children with traumatic brain injuryand sustained intracranial hypertension. ChildsNerv Syst 2001; 17:154–162

15. Sahuquillo J, Arikan F: Decompressive craniec-tomy for the treatment of refractory high intra-cranial pressure in traumatic brain injury. Co-chrane Database Syst Rev 2009; 1: CD003983


Chapter 13. Hyperventilation

I. RECOMMENDATIONS

Strength of Recommendations: Weak.Quality of Evidence: Low, from one

poor-quality study and one moderate-quality class III study.

A. Level I


B. Level II


C. Level III

Avoidance of prophylactic severe hy-perventilation to a PaCO2 �30 mm Hgmay be considered in the initial 48 hrsafter injury.

If hyperventilation is used in the man-agement of refractory intracranial hyper-tension, advanced neuromonitoring forevaluation of cerebral ischemia may beconsidered.


III. OVERVIEW

Hyperventilation has been used in themanagement of severe pediatric trau-matic brain injury (TBI) for the rapidreduction of ICP since the 1970s. Thisapproach was based on the assumptionthat hyperemia was common after pedi-atric TBI. Hyperventilation therapy wasthought to benefit the injured brain pri-marily through an increase in perfusionto ischemic brain regions and a decreasein ICP. More recent pediatric studies haveshown that hyperemia is uncommon andalso have raised concerns about the safetyof hyperventilation therapy (1–6).

Research on the effect of hyperventi-lation in children has focused on assess-ment of cerebral physiological variables.The effect of hyperventilation therapy on

outcome in infants and children with se-vere TBI has not been directly comparedwith other therapies such as hyperosmo-lar agents, barbiturates, hypothermia, orearly decompressive craniectomy.

Hyperventilation reduces ICP by pro-ducing hypocapnia-induced cerebral va-soconstriction and a reduction in cere-bral blood flow (CBF) and cerebral bloodvolume, resulting in a decrease in ICP.Recent clinical studies in mixed adult andpediatric populations have demonstratedthat hyperventilation may decrease cere-bral oxygenation and may induce brainischemia (5, 7–9). In addition, after TBI,the CBF response to changes in PaCO2 canbe unpredictable. A study by Stringer etal (10) studied regional CBF using xenoncomputed tomography and vascular reac-tivity before and after hyperventilation in12 patients including three children withsevere TBI. Hyperventilation-inducedCBF reductions affected both injured andapparently intact areas of the brain. Theischemic threshold was defined as a CBFof 23 mL/100 g/min in gray matter andthis occurred in four of 12 patients afterhyperventilation. Changes in ICP, cere-bral perfusion pressure, and mean arte-rial pressure were variable in these pa-tients after hyperventilation. The level ofhyperventilation used in this study wasprofound with end-tidal CO2 values of20–26 mm Hg before and 8–19 mm Hgafter hyperventilation. In addition to re-ducing CBF, prophylactic hypocarbia af-ter TBI has been shown experimentally toreduce the buffering capacity of cerebro-spinal fluid (CSF), an effect that may beas or more important than its effect onCBF (5).

Despite a prior recommendation inthe 2003 guidelines against prophylactichyperventilation, it remains a commonlyused therapy in children (11–13). For ex-ample, �40% of the children in the re-cent Canadian multicentered trial of hy-pothermia in severe pediatric TBI hadPaCO2 �30 mm Hg (12). Similarly, in thestudy by Curry et al (11), 50% of patientswith severe TBI had severe hypocarbia(PaCO2 �30 mm Hg) by arterial blood gasin the first 48 hrs of admission. This find-ing parallels another recent report that

mild hyperventilation was the most com-monly used therapy, having been applied in�90% of patients in the data bank of �500children with severe TBI from the UnitedKingdom and Ireland (14).

IV. PROCESS

For this update, MEDLINE was searchedfrom 1996 through 2010 (Appendix B forsearch strategy), and results were supple-mented with literature recommended bypeers or identified from reference lists. Of15 potentially relevant studies, one wasadded to the existing table and used asevidence for this topic.


Two class III studies met the inclusioncriteria for this topic and provide evi-dence to support the recommendations(11, 15). Neither represented a compari-son of hyperventilation vs. normal venti-lation or to any other therapy targetingcontrol of ICP. Similarly, there were noreports in children specifically addressingthe effects of varying levels of hyperventila-tion on ICP or outcome or studies of thetransient application of hyperventilation inthe setting of impending herniation or ICPcrisis. Lastly, neither study had a standard-ized protocol to assess PaCO2, measuring itonly intermittently.

One report described the effects of hy-perventilation on CBF, brain physiology,and Glasgow Outcome Scale at 6 months(15). A study by Skippen et al (15) wascarried out as a prospective nonrandom-ized, selected case series of 23 children (3months to 16 yrs of age) with isolatedsevere TBI. CBF was measured by xenoncomputed tomography during PaCO2 ad-justments to �35, 25–35, and �25 mmHg. The ischemic threshold was definedas CBF �18 mL/100 g/min. However, theischemic threshold in children is not de-fined and may vary with age. CO2 reactiv-ity of CBF was also assessed. Managementincluded CSF drainage and hyperosmolartherapy but not steroids or barbiturates.As PaCO2 was reduced with hyperventila-tion, CBF decreased in almost all patientsdespite decreased ICP and increased ce-rebral perfusion pressure. A relationship


DOI: 10.1097/PCC.0b013e31823f6765


between the level of hypocarbia and fre-quency of cerebral ischemia was ob-served. The frequency of regional isch-emia was 28.9% during normocapnia andincreased to 59.4% and 73.1% for PaCO2

25–35 mm Hg and �25 mm Hg, respec-tively. However, no statistical analysiswas done. Fifty-two percent had a good ormoderate outcome, 43.5% were severelydisabled or vegetative, and 4.3% died.Again, no analysis was conducted.

A second report examined the associ-ation between hypocarbia and outcome athospital discharge in a large pediatric se-ries of severe TBI victims who were allmechanically ventilated (11). A study byCurry et al (11) was carried out as aretrospective cohort study of 464 patients�15 yrs of age with an admission Glas-gow Coma Scale score �9 and a headAbbreviated Injury Score �3 with a PaCO2

recorded in the first 48 hrs of admissionfor the years 2000–2005. The authors ex-amined the incidence of severe hypocar-bia (PaCO2 �30 mm Hg) and its relation-ship with neurologic outcome before

(375 patients) and after (89 patients) thepublication of the 2003 pediatric TBIguidelines (16). They found a nonsignifi-cant change in the incidence of severe hy-pocarbia from 60% of patients before to52% after (p � .19). Patients with one doc-umented episode of severe hypocarbia, con-trolling for emergency department Glas-gow Coma Scale score, lowest emergencydepartment systolic blood pressure, InjurySeverity Score, PaCO2 sampling frequency,and year of admission, had an adjusted oddsratio (95% confidence interval) for mortal-ity of 1.44 (0.56–3.73) for one episode ofsevere hypocarbia, 4.18 (1.58–11.03) fortwo episodes, and 3.93 (1.61–9.62) for threeor more episodes compared with patientswith mild or no hypocarbia. These findings,although retrospective, show a strong asso-ciation of severe hypocarbia with poor out-comes. However, there might be other con-tributors to hypocarbia such as markedreduction in metabolic rates or acidosisfrom systemic shock. Thus, the exact con-tribution of induced hyperventilation to

poor outcome cannot be clearly definedfrom this study.



The most recent adult guidelines (17)had one level II recommendation: “pro-phylactic hyperventilation (PaCO2 of 25mm Hg or less) is not recommended.”The authors also had several level III rec-ommendations: 1) “hyperventilation isrecommended as a temporizing measurefor the reduction of elevated ICP”; 2) “hy-perventilation should be avoided duringthe first 24 hrs after injury when CBF isoften critically reduced”; and 3) “if hyper-ventilation is used, jugular venous oxygen-ation saturation or brain tissue oxygen ten-sion measurements are recommended tomonitor oxygen delivery.”


Reference Study DescriptionData Class, Quality, and

Reasons Results and Conclusion


Skippen et al,1997 (15)

Design: case seriesN � 23Age: mean 11 yrs (range, 3 months

to 16 yrs)Protocol: CBF measured by xenon-

enhanced computed tomographyduring partial pressure of arterialcarbon dioxide adjustments to�35, 25–35, and �25 mm Hg

Outcome: ischemic threshold definedas �18 mL/100 g/min;

Glasgow Outcome Scale score at 6months


confounders

Areas of CBF below ischemic threshold28.9%, 59.4%, and 73.1%, respectively(not compared statistically)

Mean vasoreactivity 2.7% change in CBFper mm Hg change in arterial carbondioxide (range, �2.3% to 7.1%)

52.2% had good or moderate outcome;43.5% were severe or vegetative; 4.3%died (no analysis)

New studyCurry et al,

2008 (11)Design: retrospective cohort study

(trauma registry) before and after2003 pediatric Brain TraumaFoundation guidelines

N � 464Age: mean 8.0 yrs (range, 0–14 yrs)Protocol: all children had controlled

mechanical ventilationOutcome: incidence of severe

hypocarbia (arterial carbon dioxide�30 mm Hg) during the initial 48hrs and risk of inpatient mortality

Analysis: chi square and logisticregression

Class IIIModerate quality: unclear if

outcome assessmentmethods unbiased;otherwise met all criteria

Severe hypocarbia 60% patients beforeand 52% after (p � .19)

Severe hypocarbia on initial measurementwas more common in infants (�2 yrs)vs. older children (30.8% vs.19.3%, respectively, p � .02)

Incidence of severe hypocarbia in the first48 hrs was similar between agegroups (58.9% for infants vs. 58.0% forolder children; p � .91)

Mortality adjusted odds ratio (95%confidence interval) of 1.44 (0.56–3.73)for 1 episode, 4.18 (1.58–11.03) for 2episodes of severe hypocarbia, and3.93 (1.61–9.62) for �3 episodes

CBF, cerebral blood flow.


VII. SUMMARY

Despite a lack of published evidencesupporting the use of hyperventilation inthe management of pediatric patientswith severe TBI, it continues to be usedcommonly worldwide. No randomizedcontrolled trial has been carried out tostudy the impact of hyperventilation onany aspect of the management of severeTBI in children such as in the setting ofrefractory intracranial hypertension orherniation. The limited evidence, how-ever, supports that prophylactic severehyperventilation to a PaCO2 �30 mm Hgshould be avoided in the initial 48 hrsafter injury. Arguing against the use ofprophylactic hyperventilation, publishedevidence discussed in this report indi-cates that the use of hyperventilation isassociated with CBF reductions and thatprolonged and or significant hypocarbiais associated with poor outcome in pe-diatric patients with severe TBI. As aresult, advanced neuromonitoring forevaluation of cerebral ischemia may beconsidered if hyperventilation is to beused in the management of refractoryintracranial hypertension.


• In the setting of refractory intracranialhypertension or brain herniation, studiesare needed to determine the efficacy ofhyperventilation in comparison to othersecond-tier therapies.

• Studies are needed to determine the op-timal monitoring technique in patientstreated with hyperventilation, includingassessments of markers of cerebral isch-emia, such as CBF, brain tissue oxygen

tension, jugular venous oxygenation sat-uration, transcranial Doppler, near-infrared spectroscopy, serum biomarkersof brain injury, or other advanced neu-romonitoring.

• The effects of hyperventilation on long-term outcome should be addressed.

REFERENCES

1. Adelson PD, Clyde B, Kochanek PM, et al:Cerebrovascular response in infants andyoung children following severe traumaticbrain injury: A preliminary report. PediatrNeurosurg 1997; 26:200–207

2. Sharples PM, Stuart AG, Matthews DS, et al:Cerebral blood flow and metabolism in chil-dren with severe head injury. Part 1: Relationto age, Glasgow Coma Score, outcome, intra-cranial pressure, and time after injury.J Neurol Neurosurg Psychiatry 1995; 58:145–152

3. Sharples PM, Matthews DS, Eyre JA: Cerebralblood flow and metabolism in children withsevere head injuries. Part 2: Cerebrovascularresistance and its determinants. J NeurolNeurosurg Psychiatry 1995; 58:153–159

4. Muizelaar JP, Marmarou A, DeSalles AA, etal: Cerebral blood flow and metabolism inseverely head-injured children. Part 1: Rela-tionship with GCS score, outcome, ICP, andPVI. J Neurosurg 1989; 71:63–71

5. Muizelaar JP, Marmarou A, Ward JD, et al:Adverse effects of prolonged hyperventilationin patients with severe head injury: A ran-domized clinical trial. J Neurosurg 1991; 75:731–739

6. Zwienenberg M, Muizelaar JP: Severe pediat-ric head injury: The role of hyperemia revis-ited. J Neurotrauma 1999; 16:937–943

7. Kiening KL, Hartl R, Unterberg AW, et al:Brain tissue pO2-monitoring in comatose pa-tients: Implications for therapy. Neurol Res1997; 19:233–240

8. Schneider GH, von Helden A, Lanksch WR,et al: Continuous monitoring of jugular bulboxygen saturation in comatose patients—

Therapeutic implications. Acta Neurochir(Wien) 1995; 134:71–75

9. von Helden A, Schneider GH, Unterberg A, etal: Monitoring of jugular venous oxygen sat-uration in comatose patients with subarach-noid haemorrhage and intracerebral haema-tomas. Acta Neurochir Suppl 1993; 59:102–106

10. Stringer WA, Hasso AN, Thompson JR, et al:Hyperventilation-induced cerebral ischemiain patients with acute brain lesions: Demon-stration by Xenon-enhanced CT. AJNR Am JNeuroradiol 1993; 14:475–484

11. Curry R, Hollingworth W, Ellenbogen RG, etal: Incidence of hypo- and hypercarbia insevere traumatic brain injury before and af-ter 2003 pediatric guidelines. Pediatr CritCare Med 2008; 9:141–146


13. Kornecki A, Constantini S, Adani LB, et al:Survey of critical care management of severetraumatic head injury in Israel. Childs NervSyst 2002; 18:375–379

14. Morris KP, Forsyth RJ, Parslow RC, et al:Intracranial pressure complicating severetraumatic brain injury in children: Monitor-ing and management. Intensive Care Med2006; 32:1606–1612

15. Skippen P, Seear M, Poskitt K, et al: Effect ofhyperventilation on regional cerebral bloodflow in head-injured children. Crit Care Med1997; 25:1402–1409

16. Adelson PD, Bratton SL, Carney NA, et al:Guidelines for the acute medical manage-ment of severe traumatic brain injury in in-fants, children, and adolescents. Chapter 14.The role of temperature control followingsevere pediatric traumatic brain injury. Pe-diatr Crit Care Med 2003; 4:S53–S55

17. Bratton SL, Chestnut RM, Ghajar J, et al:Guidelines for the management of severe trau-matic brain injury. XIV. Hyperventilation.J Neurotrauma 2007; 24(Suppl 1):S87–S90


Chapter 14. Corticosteroids

I. RECOMMENDATIONS

Strength of the Recommendation: Weak.Quality of the Evidence: Low, from

two reports of one small, moderate-quality class II study.

A. Level I


B. Level II

The use of corticosteroids is not rec-ommended to improve outcome or re-duce intracranial pressure (ICP) for chil-dren with severe traumatic brain injury(TBI).

C. Level III

There are insufficient data to supporta level III recommendation for this topic.


III. OVERVIEW

Corticosteroids are widely used intreatment of a variety of pediatric ill-nesses, including neurologic conditionssuch as brain tumors and meningitis.Steroids are thought to restore alteredvascular permeability (1), inhibit tumorinduced angiogenesis (2), and decreaseedema and cerebrospinal fluid production(3, 4) as well as diminish free radicalproduction (3). These mechanisms ofaction provide a rationale for potentialbenefit in neurologic diseases. Admin-istration of steroids to patients withsymptomatic brain tumors is standardcare and preoperative administration isbeneficial for patients undergoing resec-tion. However, the efficacy of steroids toattenuate morbidity among pediatric pa-tients with acute bacterial meningitis re-mains controversial (5). A number of cor-ticosteroids are available; however, onlydexamethasone has been reported in

studies of pediatric TBI. This chapter ad-dresses the use of corticosteroids as aneuroprotective agent to treat cerebraledema and improve Glasgow OutcomeScale in pediatric TBI. The question ofthe use of corticosteroids in the treat-ment of refractory hypotension was notaddressed by the studies. Finally, the roleof steroid therapy, both efficacy and tox-icity, remains less well known in childrencompared with adults.

IV. PROCESS

For this update, MEDLINE wassearched 1996 through 2010 (Appendix Bfor search strategy), and results were sup-plemented with literature recommendedby peers or identified from reference lists.Of 20 potentially relevant studies, nonewere added as evidence for this topic.


Two reports of one moderate-qualityclass II trial met the inclusion criteria forthis topic and provide evidence to supportthe recommendation (6, 7).

A study by Fanconi et al (6) was arandomized, prospective, placebo-con-trolled clinical trial on 25 pediatric pa-tients with severe TBI using dexametha-sone at 1 mg/kg/day for 3 days (n � 13)vs. placebo (n � 12). Baseline character-istics did not differ between groups.Dexamethasone treatment did not influ-ence ICP (mean of 14 mm Hg in bothgroups), cerebral perfusion pressure,number of interventions required, dura-tion of intubation, or 6-month GlasgowOutcome Scale vs. placebo. However, ste-roid treatment vs. placebo significantlysuppressed endogenous-free cortisol lev-els up to day 6. In addition, steroid treat-ment resulted in a trend toward increasedbacterial pneumonia vs. placebo (seven of13 vs. two of 12, respectively, p � .097).Although this study appeared to be care-fully performed, limitations included useof the Richmond screw to assess ICP,fluid restriction, and the use of hyperven-tilation to a PaCO2 of 25–30 mm Hg aspart of standard care.

The study by Kloti et al (7) reported on24 of the same 25 patients from the study

described previously. Additional out-comes in this report included duration ofICP monitoring; steroid treatment pro-duced no difference between groups. Thesmall sample size for this trial limits theability to make definitive conclusions re-garding neurologic outcomes or compli-cations. However, suppression of cortisolproduction by steroid treatment wasclearly documented.



The most recent Guidelines for theManagement of [Adult] Severe Trau-matic Brain Injury (8) summarize stud-ies of corticosteroid administration inadults and found that it did not improvefunctional outcome or mortality or lowerICP. They provide a strong class I recom-mendation against administration of ste-roids to improve outcome or lower ICPand caution that use is associated withincreased risk of mortality and thus con-traindicated. However, the studies in theadult guidelines do not specifically reporton steroids use for pediatric patients aftersevere TBI.


Children with severe TBI have beenobserved to have higher rates of general-ized hyperemia and more diffuse swellingafter injury compared with adults, whichcould, theoretically, serve as a basis forthe possible need for different approachesto the management of brain edema (9,10). One report of steroid therapy in pa-tients with severe TBI indicated betteroutcomes for children vs. adults (11);however, this difference may be the resultof age and cannot be directly attributed tosteroid-associated benefit. Several reportsincluded in the 2003 pediatric TBI guide-lines were excluded from this documentbecause they failed to meet the more rig-orous inclusion criteria. A study by Coo-per et al (12) looked at a combined group


DOI: 10.1097/PCC.0b013e31823f67d0


of children and adults with severe andmoderate TBI. Only ten patients were�10 yrs of age. In this subgroup, two offour (50%) in the placebo group com-pared with five of six (83%) in a combinedlow- and high-dose steroid group had agood outcome, which did not reach sta-tistical significance. A study by Gobiet(13) compared two cohorts, one from1972–1974 without steroid treatment anda second from 1975–1976 with steroidtreatment, and suggested a reduction inmortality. However, important differ-ences between groups in ICP monitoringand intensive care unit care were alsodescribed, making it impossible to deter-mine the effect of steroids on outcome. Astudy by James et al (14) reported a caseseries of nine children with severe TBIand compared two doses of dexametha-sone (1 mg/kg or 0.25 mg/kg) vs. no ste-roid in groups with sample sizes of onlytwo in some groups, limiting any abilityto assess for a treatment effect. A study byKretschmer (15) reported a case series of107 children in 1983 with TBI. Fifty-sixreceived steroids and 51 received dexa-methasone in addition to standard ther-apy. Reasons for exclusion of this studywere the inclusion of 29 cases of pene-trating injury, selection bias—24 of 29cases of penetrating injury were in the nosteroid group, and inclusion of patientswith mild or moderate TBI. Overall mor-tality did not differ with treatment (24%vs. 23%). The authors reported a trendtoward reduced mortality with steroiduse in the subgroup of children with in-tracranial hematoma: from 36.8% to11.8% in the placebo vs. steroid groups,

respectively. Although no significant ben-eficial effect of steroids was reported, theexclusion violations in the overall study,small sample size, and major limitationsin the study design preclude the ability tomake meaningful conclusions with re-gard to corticosteroid therapy in pediatricTBI.

VII. SUMMARY

The recommendation regarding ste-roid administration to treat severe TBI inpediatrics is based on two reports of oneclass II trial, which indicates that steroidtreatment is not associated with im-proved functional outcome, decreasedmortality, or reduced ICP. Significantsuppression of endogenous cortisol levelswas documented with dexamethasonetreatment and trends toward increasedincidence of pneumonia were observed.

Given the lack of evidence for benefitin children and the potential for harmfrom infectious complications and knownsuppression of the pituitary adrenal axis,the routine use of steroids to treat chil-dren with severe TBI to lower ICP orimprove functional outcomes or mortal-ity is not recommended.


● Further studies are needed to deter-mine risk factors for pituitary dysfunc-tion and appropriate screening in theacute and chronic phases after severeTBI in children because alterations inthe endogenous steroid response could

have important implications on manage-ment, complications, and outcome (16).

● Future research should consider test-ing the efficacy of the use of cortico-steroids for treatment of severe TBI inpediatric patients as distinct fromadults. However, if a corticosteroidtrial is considered, preliminary data areneeded for careful assessment of poten-tial toxicities, including infectiouscomplications, hyperglycemia, and det-rimental effects on nutritional status.

REFERENCES

1. Maxwell RE, Long DM, French LA: The ef-fects of glucosteroids on experimental cold-induced brain edema. Gross morphologicalalterations and vascular permeabilitychanges. J Neurosurg 1971; 34:477–487

2. Kasselman LJ, Kintner J, Sideris A, et al:Dexamethasone treatment and ICAM-1 defi-ciency impair VEGF-induced angiogenesis inadult brain. J Vasc Res 2007; 44:283–291

3. Pappius HM, McCann WP: Effects of steroidson cerebral edema in cats. Arch Neurol 1969;20:207–216

4. Weiss MH, Nulsen FE: The effect of gluco-corticoids on CSF flow in dogs. J Neurosurg1970; 32:452–458

5. Chavez-Bueno S, McCracken GH Jr: Bacte-rial meningitis in children. Pediatr ClinNorth Am 2005; 52:795–810, vii

6. Fanconi S, Kloti J, Meuli M, et al: Dexameth-asone therapy and endogenous cortisol pro-duction in severe pediatric head injury. In-tensive Care Med 1988; 14:163–166

7. Kloti J, Fanconi S, Zachmann M, et al: Dexa-methasone therapy and cortisol excretion insevere pediatric head injury. Childs NervSyst 1987; 3:103–105

8. Bratton SL, Chestnut RM, Ghajar J, et al:



Studies from previous guidelinesFanconi et al, 1988 (6); Kloti

et al, 1987 (7)Design: randomized prospective,

placebo controlled trialN � 25; 13 steroid, 12 placebo

(Fanconi)N � 24; 12/12 (Kloti)Age: range 1.4–15.8 yrsGlasgow Coma Scale score: �7Protocol: dexamethasone at 1 mg/kg/

day vs. placeboOutcome: 6-month Glasgow Outcome

Scale, ICP, duration of ICPmonitoring, duration of intubation,cerebral perfusion pressure, freecortisol levels, and complications

Class IIModerate quality: randomization and

allocation concealment methodsnot described; unclear if outcomeassessors were blinded

Steroid treatment resulted in nodifferences vs. placebo in ICP,cerebral perfusion pressure,6-month Glasgow OutcomeScale, duration of ICPmonitoring, or duration ofintubation

Steroid treatment vs. placebosignificantly suppressedendogenous free cortisollevels from day 1 to day 6

Steroid treatment resulted in atrend to ward increasedbacterial pneumonia (7 of 13vs. 2 of 12 vs. placebo,respectively, p � .097)



Guidelines for the management of severetraumatic brain injury. XV. Steroids. J Neu-rotrauma 2007; 24(Suppl 1):S91–S95

9. Aldrich EF, Eisenberg HM, Saydjari C, et al:Diffuse brain swelling in severely head-injured children. A report from the NIHTraumatic Coma Data Bank. J Neurosurg1992; 76:450–454

10. Zwienenberg M, Muizelaar JP: Severepediatric head injury: The role of hyper-emia revisited. J Neurotrauma 1999; 16:937–943

11. Hoppe E, Christensen L, Christensen KN:

The clinical outcome of patients with severehead injuries, treated with highdose dexa-methasone, hyperventilation and barbitu-rates. Neurochirurgia (Stuttg) 1981; 24:17–20

12. Cooper PR, Moody S, Clark WK, et al: Dexa-methasone and severe head injury. A pro-spective double-blind study. J Neurosurg1979; 51:307–316

13. Gobiet W: Advances in management of severehead injuries in childhood. Acta Neurochir(Wien) 1977; 39:201–210

14. James HE, Madauss WC, Tibbs PA, et al: The

effect of high dose dexamethasone in chil-dren with severe closed head injury. A pre-liminary report. Acta Neurochir (Wien) 1979;45:225–236

15. Kretschmer H: Prognosis of severe head in-juries in childhood and adolescence. Neuro-pediatrics 1983; 14:176–181

16. Acerini CL, Tasker RC, Bellone S, et al:Hypopituitarism in childhood and adoles-cence following traumatic brain injury:The case for prospective endocrine investi-gation. Eur J Endocrinol 2006; 155:663– 669


Chapter 15. Analgesics, sedatives, and neuromuscular blockade

I. RECOMMENDATIONS

Strength of Recommendations: Weak.Quality of Evidence: Low, from poor-


A. Level I


B. Level II


C. Level III*

Etomidate may be considered to con-trol severe intracranial hypertension;however, the risks resulting from adrenalsuppression must be considered.

Thiopental may be considered to con-trol intracranial hypertension.

*In the absence of outcome data, the specificindications, choice and dosing of analgesics, sed-atives, and neuromuscular-blocking agents usedin the management of infants and children withsevere traumatic brain injury (TBI) should be leftto the treating physician.

*As stated by the Food and Drug Administra-tion, continuous infusion of propofol for eithersedation or the management of refractory intra-cranial hypertension in infants and children withsevere TBI is not recommended.


III. OVERVIEW

Analgesics, sedatives, and neuromus-cular-blocking agents are commonlyused in the management severe pediatricTBI. Use of these agents can be dividedinto two major categories: 1) for emer-gency intubation; and 2) for managementincluding control of elevated intracranialpressure (ICP) in the intensive care unit(ICU). This chapter evaluates theseagents during ICU treatment.

Analgesics and sedatives are believedto favorably treat a number of importantpathophysiological derangements in se-

vere TBI. They can facilitate necessarygeneral aspects of patient care such as theability to maintain the airway, vascularcatheters, and other monitors. They canalso facilitate patient transport for diag-nostic procedures and mechanical venti-latory support. Other proposed benefits ofsedatives after severe TBI include anti-convulsant and antiemetic actions, theprevention of shivering, and attenuatingthe long-term psychological trauma ofpain and stress. Analgesics and sedativesalso are believed to be useful by mitigat-ing aspects of secondary damage. Painand stress markedly increase cerebralmetabolic demands and can pathologi-cally increase cerebral blood volume andraise ICP. Studies in experimental modelsshowed that a two- to threefold increasein cerebral metabolic rate for oxygen ac-companies painful stimuli (1, 2). Noxiousstimuli such as suctioning can also in-crease ICP (3–6). Painful and noxiousstimuli and stress can also contribute toincreases in sympathetic tone with hyper-tension and bleeding from operative sites(7). However, analgesic or sedative-induced reductions in arterial blood pres-sure can lead to cerebral ischemia as wellas vasodilation and can exacerbate in-creases in cerebral blood volume and ICP.In the absence of advanced neuromoni-toring, care must be taken to avoid thiscomplication.

The ideal sedative for patients withsevere TBI has been described as one thatis rapid in onset and offset, easily titratedto effect, has well-defined metabolism(preferably independent of end-organfunction), neither accumulates nor hasactive metabolites, exhibits anticonvul-sant actions, has no adverse cardiovascu-lar or immune actions, and lacks drug–drug interactions while preserving theneurologic examination (8).

Neuromuscular-blocking agents havebeen suggested to reduce ICP by a varietyof mechanisms including a reduction inairway and intrathoracic pressure withfacilitation of cerebral venous outflow andby prevention of shivering, posturing, orbreathing against the ventilator (9). Reduc-tion in metabolic demands by eliminationof skeletal muscle contraction has also

been suggested to represent a benefit. Risksof neuromuscular blockade include the po-tential devastating effect of hypoxemia sec-ondary to inadvertent extubation, risks ofmasking seizures, increased incidence ofnosocomial pneumonia (shown in adultswith severe TBI) (9), cardiovascular sideeffects, immobilization stress (if neuro-muscular blockade is used without ade-quate sedation/analgesia), and increasedICU length of stay (9, 10). Myopathy ismost commonly seen with the combineduse of nondepolarizing agents and corti-costeroids. Incidence of this complicationvaries between 1% and over 30% of cases(5, 11, 12). Monitoring of the depth ofneuromuscular blockade can shorten du-ration of its use in the ICU (13).

IV. PROCESS

For this update, MEDLINE was searchedfrom 1996 through 2010 (Appendix B forsearch strategy), and results were supple-mented with literature recommended bypeers or identified from reference lists. Of46 potentially relevant studies, two wereincluded as evidence for this topic.


The recommendations on the use ofanalgesics, sedatives, and neuromuscular-blocking agents in this chapter are for pa-tients with a secure airway who are receiv-ing mechanical ventilatory support yieldingthe desired arterial blood gas values andwho have stable systemic hemodynamicsand intravascular volume status.

Two class III studies of the use ofanalgesics or sedatives met inclusion cri-teria for this topic and provide evidenceto support the recommendations: onestudy about etomidate and one aboutthiopental. These studies only addressedICP as the outcome (14, 15). No studyaddressed the most commonly used an-algesics and sedatives (narcotics andbenzodiazepines).

Etomidate

A study by Bramwell et al (14) carriedout a prospective unblinded class IIIstudy of the effect of a single dose of


DOI: 10.1097/PCC.0b013e31823f67e3


etomidate (0.3 mg/kg, intravenously) onICP �20 mm Hg in eight children withsevere TBI. Etomidate reduced ICP vs.baseline in each 5-min interval duringthe 30-min study period. The patients inthis study had severe intracranial hyper-tension and etomidate reduced ICP from32.8 � 6.6 mm Hg to 21.2 � 5.2 mm Hg.An increase in cerebral perfusion pres-sure was also seen that was significant forthe initial 25 mins after etomidate ad-ministration. Every patient in the studyexhibited a reduction in ICP with treat-ment. No data were presented on cortisollevels in these patients. However, in thediscussion section of the manuscript, theauthors indicated that at 6 hrs after eto-midate administration, adrenocortico-tropic hormone stimulation tests wereperformed on each patient; four of theeight showed adrenal suppression. It isunclear if this degree of adrenal suppres-sion is different from that normally ob-served in pediatric TBI (16). No patientshowed clinical signs of adrenal insuffi-ciency such as electrolyte disturbances orblood pressure lability, and no patientreceived steroid therapy.

The availability of other sedatives andanalgesics that do not suppress adrenalfunction, small sample size and single-

dose administration in the study dis-cussed previously, and limited safety pro-file in pediatric TBI limit the ability toendorse the general use of etomidate as asedative other than as an option for sin-gle-dose administration in the setting ofraised ICP.

Barbiturates

Barbiturates can be given as a sedativeat doses lower than those required toinduce or maintain barbiturate coma. Noreport specifically addressed their use inthat capacity in pediatric TBI. One reportdid, however, address the effects of bar-biturate administration outside of thesetting of refractory raised ICP. A studyby de Bray et al (15) was a prospectivestudy of the effect of a single dose ofthiopental (5 mg/kg, intravenously) onmiddle cerebral artery flow velocity in tenchildren with severe TBI and comparedthe findings with those seen with thio-pental administration in ten children un-der general anesthesia for orthopedicprocedures. In this small study, effects onICP were assessed in only six of the tenchildren with severe TBI. In those six,thiopental reduced ICP by 48%. Flow ve-locity was reduced by approximately 15%

to 21% in the pediatric patients with TBI.Baseline ICP was 16.5 mm Hg. Cerebralperfusion pressure was not significantlychanged. At the class III level, this studysupports the ability of thiopental, admin-istered as a single dose, to reduce ICP,even when only moderately increased.The effects on flow velocity are also con-sistent with the reduction in cerebralblood volume that would be expected tomediate the reduction in ICP produced bythiopental. No study was identified, how-ever, that specifically addressed barbitu-rate use as a sedative on any other out-come parameter.



In the most recent adult guidelines, achapter on “Anesthetics, Analgesics, andSedatives” identified a class II study torecommend continuous infusion ofpropofol as the agent of choice.

Only case reports or mixed adult andpediatric case series have been publishedsupporting propofol use in pediatric TBI




New studiesBramwell et al,

2006 (14)Design: prospective case seriesN � 8Age: �15 yrsProtocol: single IV dose of etomidate

(0.3 mg/kg)Purpose: determine if etomidate reduces

ICP in the setting of intracranialhypertension (ICP �20 mm Hg)

Outcome: ICP


confounders; very smallsample size

Etomidate administration resulted in a decrease inICP vs. baseline (p � .05) without change inmean arterial pressure, thereby increasingcerebral perfusion pressure at each 5-mininterval; at 6 hrs after etomidate administration,adrenocorticotropic hormone stimulation testsshowed adrenal suppression in 4 of the 8patients; however, no patient required treatmentwith steroids

de Bray et al,1993 (15)

Design: prospective case seriesN � 10 TBI and 10 orthopedic controlsAge: 4–14 yrsProtocol: IV administration of thiopental

and Doppler assessment of middlecerebral artery flow velocity

Purpose: assess the effect of thiopental(5 mg/kg, IV) on ICP and middlecerebral artery flow velocity

Outcome: middle cerebral artery flowvelocity blood velocity, measured atthe time of greatest decrease of meanarterial pressure after thiopentaladministration, compared withbaseline


confounders; unclear ifselection was unbiased;unclear if missing data

Thiopental reduced mean ICP, measured in 6 of the10 patients with TBI, by 48% (p � .01), with nosignificant correlation with middle cerebralartery flow velocity; thiopental also reducedmiddle cerebral artery flow velocity (systolicvelocities �15% � 6.9%, p � .01) and diastolicvelocities (�21% � 6.5%, p � .01) in cases, notcontrols; reduction in middle cerebral artery flowvelocity occurred in 90% cases compared with10% controls; mean ICP, measured in 6 of the 10patients with TBI, was reduced by 48% (p � .01)with no significant correlation with middlecerebral artery flow velocity

IV, intravenous; ICP, intracranial pressure; TBI, traumatic brain injury.


(17, 18). However, a number of reports(in cases not restricted to TBI) suggestthat continuous infusion of propofol isassociated with an unexplained increasein mortality risk in critically ill children.A syndrome of lethal metabolic acidosis(“propofol syndrome”) can occur (19–24).In light of these risks, and with alterna-tive therapies available, continuous infu-sion of propofol for either sedation ormanagement of refractory intracranialhypertension in severe pediatric TBI isnot recommended. The Center for DrugEvaluation and Research Web site of theFood and Drug Administration (25)states, “Propofol is not indicated for pe-diatric ICU sedation as safety has notbeen established.” Based on the Food andDrug Administration recommendationsagainst the continuous infusion of propo-fol for sedation in pediatric critical caremedicine, the recommendation from theadult guidelines cannot be translated topediatric TBI management and repre-sents an important discontinuity betweenpediatric and adult TBI management.

Neuromuscular-blocking agents werenot addressed in the “Anesthetics, Anal-gesics, and Sedatives” chapter of the mostrecent adult guidelines. In the 2000 adultguidelines (26), the initial managementsection cited a study that examined 514entries in the Traumatic Coma Data Bankand reported no beneficial effects of neu-romuscular blockade and an increased in-cidence of nosocomial pneumonia andprolonged ICU stay associated with pro-phylactic neuromuscular blockade (9). Itwas suggested that use of neuromuscu-lar-blocking agents be reserved for spe-cific indications (intracranial hyperten-sion, transport).


Ketamine exhibits neuroprotective ef-fects in experimental models of TBI; how-ever, concerns over its vasodilatory ef-fects and their impact on ICP have longlimited its consideration as a sedative inTBI. Recently, a study by Bar-Joseph et al(27) was carried out, which was a pro-spective study in 30 children with raisedICP, 24 with nonpenetrating TBI. A singledose of ketamine (1–1.5 mg/kg, intrave-nously) was evaluated for its ability toeither 1) prevent further increases in ICPduring a stressful procedure (i.e., suc-tioning); or 2) treat refractory intracra-nial hypertension. Ketamine reduced ICPin both settings. These patients had se-

vere intracranial hypertension with anoverall mean ICP of 25.8 mm Hg. Thestudy did not meet inclusion criteria forthese guidelines for two reasons. First, itfell just below the cutoff of 85% of TBIcases, and second, Glasgow Coma Scalescore was not provided–although it islikely that the children had severe TBIgiven the ICP data.

Regarding the use of etomidate in crit-ical care, including severe TBI and mul-tiple trauma victims (28–31), there aregeneral concerns over adrenal suppres-sion. As stated earlier, the availability ofother sedatives and analgesics that do notsuppress adrenal function, along with thesmall sample size and single-dose admin-istration in the single study in the evi-dence table (Table 1) and limited safetyprofile in pediatric TBI, limit the abilityto endorse the general use of etomidateas a sedative other than as an option forsingle-dose administration in the settingof raised ICP.

VII. SUMMARY

Two studies were identified that metinclusion criteria, rendering reservedclass III recommendations that 1) etomi-date may be considered to decrease intra-cranial hypertension, although the risksresulting from adrenal suppression mustbe considered; and 2) thiopental, given asa single dose, may be considered to con-trol intracranial hypertension.

Despite the common use of analgesicsand sedatives in TBI management, therehave been few studies of these drugs fo-cused on pediatric patients with severeTBI, and studies meeting inclusion crite-ria for the most commonly used agentswere lacking. Similarly, no studies wereidentified meeting inclusion criteria thataddressed the use of neuromuscular block-ade in pediatric patients with severe TBI.Until experimental comparisons amongthese agents are carried out, the choice anddosing of analgesics, sedatives, and neuro-muscular-blocking agents used should beleft to the treating physician. Based on rec-ommendations of the Food and Drug Ad-ministration, continuous infusion of propo-fol is not recommended in the treatment ofpediatric TBI.


• Studies are needed comparing the var-ious sedatives and analgesics in pediat-ric patients with severe TBI, examining

sedative and analgesic efficacy, effectson ICP, other surrogate markers, andfunctional outcome.

• Studies are needed to assess the toxici-ties, including hypotension, adrenal sup-pression, effects on long-term cognitiveoutcomes, and other adverse effects.

• Studies are needed on dosing, duration,and interaction effects with other con-current therapies.

• Optimal sedation after severe TBI maydiffer between infants and older childrenand requires investigation. Specifically,given concerns over the effects of variousanesthetics and sedatives on neuronaldeath in the developing brain (32, 33),studies of various analgesic and sedativeregimens in infants with TBI are needed,including infants who are victims of abu-sive head trauma.

• The specific role of neuromuscular-blocking agents in infants and childrenwith severe TBI needs to be defined.

REFERENCES

1. Nilsson B, Rehncrona S, Siesjo BK: Couplingof cerebral metabolism and blood flow inepileptic seizures, hypoxia and hypoglycae-mia. Ciba Found Symp 1978; 56:199–218

2. Rehncrona S, Siesjo BK: Metabolic and phys-iologic changes in acute brain failure. In:Brain Failure and Resuscitation. Grenvik A,Safar P (Eds). New York, NY, Churchill Liv-ingstone, 1981, pp 11–33

3. Fortune JB, Feustel PJ, Weigle CG, et al:Continuous measurement of jugular venousoxygen saturation in response to transientelevations of blood pressure in head-injuredpatients. J Neurosurg 1994; 80:461–468

4. Kerr ME, Weber BB, Sereika SM, et al: Effectof endotracheal suctioning on cerebral oxy-genation in traumatic brain-injured patients.Crit Care Med 1999; 27:2776–2781

5. Raju TN, Vidyasagar D, Torres C, et al: Intra-cranial pressure during intubation and anes-thesia in infants. J Pediatr 1980; 96:860–862

6. White PF, Schlobohm RM, Pitts LH, et al: Arandomized study of drugs for preventingincreases in intracranial pressure during en-dotracheal suctioning. Anesthesiology 1982;57:242–244

7. Todres ID: Post anesthesia recovery in thepediatric intensive care unit. In: PediatricCritical Care. BP, F, JJ, Z (Eds). St. Louis,MO, Mosby, 1998, pp 1391–1398

8. Prielipp RC, Coursin DB: Sedative and neu-romuscular blocking drug use in critically illpatients with head injuries. New Horiz 1995;3:456–468

9. Hsiang JK, Chesnut RM, Crisp CB, et al:Early, routine paralysis for intracranial pres-sure control in severe head injury: is it nec-essary? Crit Care Med 1994; 22:1471–1476

10. Durbin CG Jr: Neuromuscular blocking


agents and sedative drugs. Clinical uses andtoxic effects in the critical care unit. CritCare Clin 1991; 7:489–506

11. Douglass JA, Tuxen DV, Horne M, et al: My-opathy in severe asthma. Am Rev Respir Dis1992; 146:517–519

12. Rudis MI, Guslits BJ, Peterson EL, et al:Economic impact of prolonged motor weak-ness complicating neuromuscular blockadein the intensive care unit. Crit Care Med1996; 24:1749–1756

13. Rudis MI, Sikora CA, Angus E, et al: A pro-spective, randomized, controlled evaluationof peripheral nerve stimulation versus stan-dard clinical dosing of neuromuscular block-ing agents in critically ill patients. Crit CareMed 1997; 25:575–583

14. Bramwell KJ, Haizlip J, Pribble C, et al: Theeffect of etomidate on intracranial pressureand systemic blood pressure in pediatric pa-tients with severe traumatic brain injury.Pediatr Emerg Care 2006; 22:90–93

15. de Bray JM, Granry JC, Monrigal JP, et al:Effects of thiopental on middle cerebral arteryblood velocities: A transcranial Doppler studyin children. Childs Nerv Syst 1993; 9:220–223

16. Acerini CL, Tasker RC: Endocrine sequelae oftraumatic brain injury in childhood. HormRes 2007; 68(Suppl 5):14–17

17. Farling PA, Johnston JR, Coppel DL: Propo-fol infusion for sedation of patients with headinjury in intensive care. A preliminary re-port. Anaesthesia 1989; 44:222–226

18. Spitzfaden AC, Jimenez DF, Tobias JD:Propofol for sedation and control of intracra-

nial pressure in children. Pediatr Neurosurg1999; 31:194–200

19. Bray RJ: Propofol infusion syndrome in chil-dren. Paediatr Anaesth 1998; 8:491–499

20. Canivet JL, Gustad K, Leclercq P, et al: Mas-sive ketonuria during sedation with propofolin a 12 year old girl with severe head trauma.Acta Anaesth Belg 1994; 45:19–22

21. Cray SH, Robinson BH, Cox PN: Lactic aci-demia and bradyarrhythmia in a child se-dated with propofol. Crit Care Med 1998;26:2087–2092

22. Fong JJ, Sylvia L, Ruthazer R, et al: Predic-tors of mortality in patients with suspectedpropofol infusion syndrome. Crit Care Med2008; 36:2281–2287

23. Hanna JP, Ramundo ML: Rhabdomyolysisand hypoxia associated with prolongedpropofol infusion in children. Neurology1998; 50:301–303

24. Parke TJ, Stevens JE, Rice AS, et al: Meta-bolic acidosis and fatal myocardial failureafter propofol infusion in children: Five casereports. BMJ 1992; 305:613–616

25. The Food and Drug Administration: Centerfor Drug Evaluation and Research. Availableat: http://www.fda.gov/Safety/MedWatch/SafetyInformation/SafetyAlertsforHumanMedicalProducts/ucm172351.htm; http://www.fda.gov/downloads/Safety/MedWatch/SafetyInformation/SafetyAlertsforHumanMedicalProducts/UCM173766.pdf. Accessed February17, 2010

26. The Brain Trauma Foundation; The Ameri-can Association of Neurological Surgeons:

The Joint Section on Neurotrauma and Crit-ical Care. Initial management. J Neu-rotrauma 2000; 17:463–469

27. Bar-Joseph G, Guilburd Y, Tamir A, et al:Effectiveness of ketamine in decreasing in-tracranial pressure in children with intracra-nial hypertension. J Neurosurg Pediatr 2009;4:40–46

28. Cohan P, Wang C, McArthur DL, et al: Acutesecondary adrenal insufficiency after trau-matic brain injury: A prospective study. CritCare Med 2005; 33:2358–2366

29. Cotton BA, Guillamondegui OD, Fleming SB,et al: Increased risk of adrenal insufficiencyfollowing etomidate exposure in critically in-jured patients. Arch Surg 2008; 143:62–67;discussion 67

30. Hildreth AN, Mejia VA, Maxwell RA, et al:Adrenal suppression following a single doseof etomidate for rapid sequence induction: Aprospective randomized study. J Trauma2008; 65:573–579

31. Warner KJ, Cuschieri J, Jurkovich GJ, et al:Single-dose etomidate for rapid sequence in-tubation may impact outcome after severeinjury. J Trauma 2009; 67:45–50

32. Bittigau P, Sifringer M, Pohl D, et al: Apo-ptotic neurodegeneration following traumais markedly enhanced in the immature brain.Ann Neurol 1999; 45:724–735

33. Ikonomidou C, Bittigau P, Koch C, et al:Neurotransmitters and apoptosis in the de-veloping brain. Biochem Pharmacol 2001;62:401–405


http://www.fda.gov/Safety/MedWatch/SafetyInformation/SafetyAlertsforHumanMedicalProducts/ucm172351.htm



http://www.fda.gov/downloads/Safety/MedWatch/SafetyInformation/SafetyAlertsforHumanMedicalProducts/UCM173766.pdf




Chapter 16. Glucose and nutrition

I. RECOMMENDATIONS

Strength of the Recommendation:Weak.

Quality of Evidence: Moderate, fromone moderate-quality class II study.

A. Level 1


B. Level II

The evidence does not support the useof an immune-modulating diet for thetreatment of severe traumatic brain in-jury (TBI) to improve outcome.

C. Level III

In the absence of outcome data, thespecific approach to glycemic control inthe management of infants and childrenwith severe TBI should be left to thetreating physician.


III. OVERVIEW

Providing nutritional support to chil-dren after TBI is a decision with wide-ranging implications. Similar to adults,traumatically injured children requireenergy for wound healing, repair, altera-tions in normal organ function, andother pathologic processes initiated bythe injury. However, children havegreater nutritional needs for normalgrowth and development. The decision toadminister nutritional support, includingthe timing, the quantity, the manner, andthe composition of such support, mayhave profound effects on short- and long-term outcome, and results from studiesin adults may not be applicable to infantsand children.

IV. PROCESS

For this update, MEDLINE was searchedfrom 1996 through 2010 (Appendix B forsearch strategy), and results were supple-mented with literature recommended bypeers or identified from reference lists. Ofthe 104 potentially relevant studies, onewas added to the existing table and used asevidence for this topic.


One class II randomized controlledtrial met the inclusion criteria for thistopic and provides evidence to supportthe recommendation (1). A study by Bri-assoulis et al (1) prospectively studied theeffect of an immune-enhancing formulaon various outcomes after TBI in a cohortof 40 children in a single center. Subjectswith severe TBI (Glasgow Coma Scale[GCS] score �8) without renal or gastro-intestinal disease were eligible. Enteralnutrition was initiated within 12 hrs ofTBI. Children were randomized within ablock design to either a specialized for-mula (Stresson, including supplementalglutamine, arginine, antioxidants, andomega-3 fatty acids) or a more standardformula (Tentrini) through a nasogastrictube. For each 100 mL, the specializedformulation contained greater amountsof protein (7.5 g vs. 3.3 g), fat (13.2 g vs.11.1 g), glutamine (1.3 g vs. 0 g), arginine(0.89 g vs. 0 g), docosahexaenoic acid(0.028 g vs. 0 g), eicosapentaenoic acid(0.072 g vs. 0 g), selenium (14.1 mg vs.4.9 mg), copper (338 �g vs. 0 �g), vita-min E (12.5 mg vs. 1.3 mg), carotenoids(0.38 mg vs. 0.15 mg), and carnitine (7.5mg vs. 3 mg). Furthermore, the experi-mental formula demonstrated an in-creased osmolarity (420 mOsm/L vs. 245mOsm/L) compared with the standardpreparation. Administration of feedingswas targeted based on predicted energyexpenditure (PEE) that included compen-satory increases for various injury fac-tors. The amount of nutritional supportfrom each formula was escalated over thefirst 5 days after TBI based on PEE (0.5%,100%, 125%, 150%, and 150%, respec-tively). In both groups, feeding intoler-ance was treated with gastric-emptying

agents and diarrhea was treated withtemporary discontinuation of feedings.Failure of a regimen was defined as in-ability to follow the prescription outlinedhere. Nitrogen balance, serum nutri-tional indices, and cytokines were deter-mined in each group as the primary out-come parameters. The mean age ofenrolled children was 120 months with amajority being male (71.4%). There werefive deaths (12.5%). There were no sig-nificant differences in outcomes betweenthe two feeding groups for survival (en-hanced vs. standard: 80% vs. 95%),length of stay (16.7 vs. 12.2 days), orlength of mechanical ventilation (11 vs. 8days). Nitrogen balance was achieved in agreater percentage of children receivingthe enhanced formula by day 5 (69.2% vs.30.8%), but zinc, copper, retinol-bindingprotein, and transthyretin were not dif-ferent between the groups throughoutthe study period. The only cytokine mea-sured that was independently associatedwith the enhanced diet was interleukin-8.Levels were lower in the immune en-hanced vs. standard treatment group.



Based on three class II and 11 class IIIstudies (2), a recommendation to obtainfull caloric replacement by 7 days postin-jury was made in the adult guidelines.Overall, studies included within thisguideline addressed the manner of feed-ing, the quantity of calories adminis-tered/expended, hyperglycemia, and min-eral supplementation.

In comparing the manner in whichnutrition is administered, a class II studyby Rapp et al (3) randomized 38 subjectsto total parenteral nutrition (TPN) or en-teral nutrition (EN) and found that theTPN group had decreased mortality (zerovs. eight subjects, p � .001). They alsofound that the TPN group achievedhigher caloric intake and reached full nu-tritional replacement by 7 days postinjury(compared with 14 days postinjury for EN


DOI: 10.1097/PCC.0b013e31823f67fc


group). Class III studies provided comple-mentary information regarding this is-sue. Specifically, a study by Hadley et al(4) demonstrated that subjects random-ized to TPN had higher mean daily nitro-gen intakes and losses, resulting ulti-mately in no difference in nitrogenbalance between the intervention groups.A study by Young et al (5) randomized 51subjects with severe to moderate TBI(GCS 4–10) to TPN or EN and found theTPN group had increased protein intake,improved nitrogen balance, and im-proved outcome at 3 months postinjury.Finally, a study by Borzotta et al (6)found that TPN and EN were equally ef-fective when prescribed based on mea-sured energy expenditure (MEE). Theyfound that MEE remained between 135%and 146% of predicted for the first 4 wkspostinjury and neither feeding regimenwas associated with differences in infec-tion rates or hospital costs.

In comparing various EN strategies, aclass II study by Taylor et al (7) demon-strated that an accelerated EN regimen tomeet goals within the first week postin-jury in mechanically ventilated TBI vic-tims was associated with improved out-come at 3 months (yet no difference at 6months). Furthermore, fewer infectionswere also observed in the accelerated ENgroup. Other class III studies demon-strated that 1) percutaneously placedfeeding tubes could safely administer cal-ories after TBI (8, 9); 2) continuously fedsubjects demonstrated less feeding intol-erance and reached caloric goals morequickly (10); and 3) nasojejunal feedings

permitted increased delivery of calories(11). A study by Clifton et al (12) recom-mended that a nomogram be used to es-timate energy requirements to guide ca-loric intake, whereas other recent studiessuggest that published formulas poorlypredict the energy requirements of adults(13) or children (14).

Regarding supplementation of feed-ings, a class II study showed a nonsignif-icant trend (p � .09) toward decreasedmortality in subjects randomized to re-ceive 12 mg elemental zinc in parenteralnutrition for 15 days followed by 22 mgoral zinc for an additional 15 days (15).Improvements in nutritional markers (al-bumin, prealbumin, and retinol-bindingprotein) were observed in this treatmentgroup compared with the standard sub-jects. Finally, two class III studies dem-onstrated that hyperglycemia early afterTBI was associated with poor outcome(16, 17), although this effect may reflect astress response after injury rather than anutritional effect.

In summary, the adult guidelines (2)suggest that starved patients with TBIlose sufficient nitrogen to reduce weightby 15% per week and support administra-tion of 100% to 140% replacement ofresting energy expenditure with 15% to20% nitrogen calories, which may reducenitrogen loss. The data support full feed-ing at least by the end of the first week. Ithas not been established that any methodof feeding is better than another or thatearly feeding before 7 days improves out-come. Based on the level of nitrogen-wasting documented in patients with

TBI and the nitrogen-sparing effect offeeding, it is a level II recommendationthat full nutritional replacement be in-stituted by day 7 postinjury for adultpatients.


A number of studies have been re-ported on this topic that failed to meetinclusion criteria because they did notcompare specific nutritional regimens.There have been several studies address-ing the effect of TBI on metabolism witha focus on the amount of consumed cal-ories in the immediate post-TBI time pe-riod. This information is thought to be animportant precursor to studies thatwould target the amount of calories re-quired after TBI and the possible effect ofdifferent nutritional support strategies onoverall outcome. Evidence suggests thatunderfed critically ill, nontrauma pediat-ric patients have increased mortality, in-fections, and poor wound healing (14).However, overfeeding is associated withincreased carbon dioxide production andrespiratory complications. Caloric needscan be measured using indirect calorim-etry (MEE) or estimated by various math-ematical formulae (PEE). Because manyfactors after TBI can affect caloric expen-diture (including sedation, neuromuscu-lar blockade, hemodynamic support, sei-zures, temperature, other injuries, andothers), MEE currently represents themost accurate method for determiningenergy requirements (18).




New studyBriassoulis et al,

2006 (1)Design: randomized controlled trialN � 40GCS: mean 6.2 (SEM 0.5)Age: mean 127 month � 7.9 for immune-

modulating group; 112 months � 14.5for standard group

Protocol: children randomized to immune-enhancing diet containingsupplementation with glutamine,arginine, and antioxidants vs. a regularformula from 12 hrs after admission

Purpose: determine if an immune-enhanceddiet would alter mortality

Outcomes: hospital mortality, length of stay,nutritional indices, and cytokineconcentrations

Class IIModerate quality: attrition not

reported; unclear ifintention-to-treat analysisconducted; otherwise metall criteria

Immune-enhancing vs. regular formulaSurvival: 80% vs. 95%Length of stay: 16.7 vs. 12.2 daysLength of mechanical ventilation: 11 vs. 8

daysP values not reported; no significant

differences between groupsFewer positive gastric cultures in

immune-enhancing group (p � .02),but infections did not differ

The group fed an immune-modulatingdiet; was more likely to have positivenitrogen balance at 5 days (69% vs.31%, p � .05)

GCS, Glasgow Coma Scale; SEM, standard error of mean.


Two studies have reported MEE aftersevere TBI in children. Phillips et al (19)studied the effect of TBI on energy expen-diture (measured by indirect calorimetry),nitrogen excretion, and serum markers ofnutritional adequacy in children with GCS3–8. This observational study followed12 children (aged 2–17 yrs) for the first2 wks after TBI. There was one case ofpenetrating TBI, whereas all others hadclosed TBI. Multiple other injuries aredescribed, yet the “major injury” was tothe brain. Six children developed intra-cranial hypertension that was treatedwith hyperventilation, neuromuscularblockade (n � 4), cerebrospinal fluiddiversion, mannitol, or barbiturates(n � 4). Phenytoin was administeredonly when seizures were observed. Allchildren received antibiotics and anti-pyretics (aspirin/acetaminophen). Nu-trition was administered enterally start-ing 3–12 days after injury (n � 5) orparenterally starting 2–6 days after injury(n � 7). MEE was performed on ninechildren, 1–14 days after TBI. The meanMEE was 130% of PEE derived from theHarris/Benedict formula, and the lowestMEE/PEE was 94%.

Diarrhea (n � 5) and gastric residuals(n � 2) were noted as limitations to en-teral feeding regimen. Mean nitrogen ex-cretion was 307 mg/kg/day for adoles-cents and 160 mg/kg/day for youngerchildren, and nitrogen balance remainednegative throughout the 2-wk period.Mean serum albumin decreased duringthe 2-wk study period (2.9 g/dL to 2.4g/dL), whereas mean serum protein in-creased (5.4–6.0 g/dL) with both beingbelow laboratory normals for the firstweek after TBI. Other nutritional mark-ers (retinol binding protein and prealbu-min) were slightly increased in week 2compared with week 1. Weight loss wasprominent, ranging 2–26 pounds amongall subjects. This represents 9% loss ofbody weight for adolescents and 4% lossof body weight for children. The possibleeffects of neuromuscular blockade, seda-tion, temperature, and seizures were notaddressed.

In another study, Moore et al (20)measured MEE within the first 48 hrsafter TBI in 20 subjects with severe TBI,including seven children. Entry into thisstudy was limited to patients with an In-jury Severity Score for head injurygreater than all other organ systems. Allsubjects underwent pulmonary arterycatheterization for cardiac output moni-toring, 17 received intracranial pressure

monitoring, and two received corticoste-roids. Within the pediatric group (age,3–16 yrs), oxygen consumption was180% of predicted and energy expendi-ture was 173% of predicted. None of thevalues were �100% of predicted. The aver-age respiratory quotient was 0.68, indicat-ing consumption of lipids as a predominantfuel. The mean rectal temperature at thetime of the metabolic testing was 38.2°C.Nutritional support started within 48 hrsafter TBI, but information regarding theadministration of enteral or parenteral nu-trition, neuromuscular blockade, and bar-biturates was not reported.

Two additional manuscripts, compris-ing some of the same patient population,were reported regarding MEE measure-ments (21, 22). Eighteen children aftersevere TBI (GCS �8) were studied and allreceived standard therapies including se-dation and paralysis during the study pe-riod. Nasogastric feedings were begun onday 2 and MEE was determined seriallyfor the first several days after TBI usingthe Douglas bag method. A total of 107MEE measurements were obtained, with1) 82% within the normal referenceranges for resting children (85% to 115%PEE); 2) 4% at �115% PEE; and 3) 14%at �85% PEE. Logistic regression dem-onstrated a significant association be-tween MEE and rectal temperature withan increase of 1°C corresponding to anincrease in MEE by 7.4%. Furthermore,MEE was significantly associated withplasma epinephrine, triiodothyronine,and glucagon concentrations.

Although the precise mechanism un-derlying the association between hyper-glycemia and outcome is still unclear, thepossibility exists that it may be related inpart to nutrient delivery. Two studies re-garding hyperglycemia and TBI includedadmission glucose concentrations amongchildren with TBI. A study by Michaud etal (23) retrospectively studied 54 children(age, �16 yrs) with severe TBI (GCS �8)treated in a single center. Children whodied in the emergency department, thosewith gunshot wounds to the head, andthose who had fatal outcomes from mul-tiple or extracranial injuries (n � 8) wereexcluded. Children who received dex-trose-containing solutions at another in-stitution before serum glucose testingwere separately analyzed. Discharge Glas-gow Outcome Scale (GOS) scores wererecorded. In the 16 children who died orremained in a vegetative state, mean ad-mission glucose concentration was 288mg/dL compared with 194 mg/dL for

those with more favorable outcome (p �.01). Increases in blood glucose were alsoassociated with hypotension, acidosis, ab-normal pupillary responses, lower GCS,and cerebral edema on initial computedtomography scan.

A study by Cochran et al (24) was of170 children with both moderate and se-vere TBI (Abbreviated Injury Score of thehead at admission �3) who had admis-sion serum glucose concentration mea-sured. GOS scores were obtained at hos-pital discharge and mortality rate was9.4%. Children who died had greatermean serum admission glucose concen-tration (267 mg/dL) compared with thosewith severe (GOS � 3; 249 mg/dL), mod-erate (GOS � 4; 168 mg/dL), or milddisability (GOS � 1; 128 mg/dL).

A study by Chiaretti et al (25) retro-spectively analyzed 122 children after se-vere TBI (GCS �8) for various factorsthat might be associated with an adverseneurologic outcome. Inclusive in thesefactors was hyperglycemia, defined asblood glucose concentration �150 mg/dL. Glucose measurements were obtainedat hospital admission and at least twicedaily during the admission. Other factorsconsidered in the analysis were hypoxia(PaO2 �60 mm Hg or SaO2 �90% for atleast 15 mins or apnea/cyanosis noted onexamination), hypotension (arterial pres-sure less than the fifth percent for age forat least 15 mins), radiologic findings (ce-rebral hemorrhages, cerebral edema, andother findings as interpreted by an inde-pendent radiologist), hematologic, coag-ulation, metabolic and seizures. Out-comes were assessed by GOS scores at 6months after TBI and dichotomized intofavorable (GOS 4 –5) and unfavorable(GOS 1–3). Of the children enrolled, themean age was 122 months, 74 had iso-lated head injury, whereas 48 had multi-ple trauma. There were 47 children witha poor outcome (38.5%) at 6 months. Allchildren had admission glucose concen-trations obtained and initial GCS andblood glucose were highly correlated(p � .001). Mean admission serum glu-cose varied by outcomes that includedsome overlap between the groups (GOS4–5, 221 mg/dL � 70; GOS 3–4, 261mg/dL � 102; GOS 1–2, 290 mg/dL �88). Hyperglycemia after TBI was associ-ated with poor outcome based on bivari-ate analysis, which remained significantin multivariate analysis adjusting forGCS, type of trauma (isolated vs. multi-trauma), hypoxia, hypotension, dissemi-nated intravascular coagulation, and


early posttraumatic seizures. Administra-tion of glucose, insulin, and other nutri-tional support was not reported.

VII. SUMMARY

Although multiple studies examinedthe timing, quantity, manner, and com-position of nutritional support for pa-tients with TBI, only one met the inclu-sion criteria for this topic. That class IIrandomized controlled trial showed nodifference in outcomes for children pro-vided an immune-enhancing diet vs. reg-ular formula. There is insufficient evi-dence to recommend the use of glycemiccontrol after severe pediatric TBI to im-prove outcome despite evidence indicat-ing that posttraumatic hyperglycemia isassociated with poor outcome.


• Prospective trials of nutritional sup-port, either enteral or parenteral, todetermine the superiority of variouspotential strategies.

• Measurement of caloric expenditureand other nutritional indices in largerstudies to gain a broader understand-ing of the role of nutrition after TBI sothat novel strategies, including thepossible targeting of caloric expendi-ture, can be tested.

• Prospective trials of glycemic controlwith protocolized administration ofnutrition and insulin.

• More complete reporting of nutritionalstrategies used in large, randomizedcontrolled trials of other therapies inTBI that would increase our under-standing of the effect of nutrition onimportant outcomes.

• Fundamental observational studies ofthe effect of TBI on nutritional mark-ers (including standard indices andmetabolomics) for the design of ratio-nal clinical trials.

REFERENCES

1. Briassoulis G, Filippou O, Kanariou M, et al:Temporal nutritional and inflammatorychanges in children with severe head injuryfed a regular or an immune-enhancing diet:A randomized, controlled trial. Pediatr CritCare Med 2006; 7:56–62

2. Bratton SL, Chestnut RM, Ghajar J, et al:Guidelines for the management of severetraumatic brain injury. XII. Nutrition. J Neu-rotrauma 2007; 24(Suppl 1):S77–S82

3. Rapp RP, Young B, Twyman D, et al: Thefavorable effect of early parenteral feeding onsurvival in head-injured patients. J Neuro-surg 1983; 58:906–912

4. Hadley MN, Grahm TW, Harrington T, et al:Nutritional support and neurotrauma: A crit-ical review of early nutrition in forty-fiveacute head injury patients. Neurosurgery1986; 19:367–373

5. Young B, Ott L, Haack D, et al: Effect of totalparenteral nutrition upon intracranial pres-sure in severe head injury. J Neurosurg 1987;67:76–80

6. Borzotta AP, Pennings J, Papasadero B, et al:Enteral versus parenteral nutrition after se-vere closed head injury. J Trauma 1994; 37:459–468

7. Taylor SJ, Fettes SB, Jewkes C, et al: Pro-spective, randomized, controlled trial to de-termine the effect of early enhanced enteralnutrition on clinical outcome in mechani-cally ventilated patients suffering head in-jury. Crit Care Med 1999; 27:2525–2531

8. Kirby DF, Clifton GL, Turner H, et al: Earlyenteral nutrition after brain injury by percu-taneous endoscopic gastrojejunostomy.JPEN J Parenter Enteral Nutr 1991; 15:298–302

9. Klodell CT, Carroll M, Carrillo EH, et al:Routine intragastric feeding following trau-matic brain injury is safe and well tolerated.Am J Surg 2000; 179:168–171

10. Rhoney DH, Parker D Jr, Formea CM, et al:Tolerability of bolus versus continuous gas-tric feeding in brain-injured patients. NeurolRes 2002; 24:613–620

11. Grahm TW, Zadrozny DB, Harrington T: Thebenefits of early jejunal hyperalimentation inthe head-injured patient. Neurosurgery1989; 25:729–735

12. Clifton GL, Robertson CS, Choi SC: Assess-ment of nutritional requirements of head-

injured patients. J Neurosurg 1986; 64:895–901

13. Walker RN, Heuberger RA: Predictive equa-tions for energy needs for the critically ill.Respir Care 2009; 54:509–521

14. Skillman HE, Wischmeyer PE: Nutritiontherapy in critically ill infants and children.JPEN J Parenter Enteral Nutr 2008; 32:520–534

15. Young B, Ott L, Kasarskis E, et al: Zincsupplementation is associated with improvedneurologic recovery rate and visceral proteinlevels of patients with severe closed headinjury. J Neurotrauma 1996; 13:25–34

16. Lam AM, Winn HR, Cullen BF, et al: Hyper-glycemia and neurological outcome in pa-tients with head injury. J Neurosurg 1991;75:545–551

17. Young B, Ott L, Dempsey R, et al: Relation-ship between admission hyperglycemia andneurologic outcome of severely brain-injuredpatients. Ann Surg 1989; 210:466–472; dis-cussion 472–473

18. Levine JA: Measurement of energy expendi-ture. Public Health Nutr 2005; 8:1123–1132

19. Phillips R, Ott L, Young B, et al: Nutritionalsupport and measured energy expenditure ofthe child and adolescent with head injury.J Neurosurg 1987; 67:846–851

20. Moore R, Najarian MP, Konvolinka CW: Mea-sured energy expenditure in severe headtrauma. J Trauma 1989; 29:1633–1636

21. Matthews DS, Aynsley-Green A, MatthewsJN, et al: The effect of severe head injury onwhole body energy expenditure and its pos-sible hormonal mediators in children. Pedi-atr Res 1995; 37:409–417

22. Matthews DS, Bullock RE, Matthews JN, etal: Temperature response to severe head in-jury and the effect on body energy expendi-ture and cerebral oxygen consumption. ArchDis Child 1995; 72:507–515

23. Michaud LJ, Rivara FP, Longstreth WT Jr, etal: Elevated initial blood glucose levels andpoor outcome following severe brain injuriesin children. J Trauma 1991; 31:1356–1362

24. Cochran A, Scaife ER, Hansen KW, et al:Hyperglycemia and outcomes from pediatrictraumatic brain injury. J Trauma 2003; 55:1035–1038

25. Chiaretti A, Piastra M, Pulitano S, et al: Prog-nostic factors and outcome of children withsevere head injury: An 8-year experience.Childs Nerv Syst 2002; 18:129–136


Chapter 17. Antiseizure prophylaxis

I. RECOMMENDATIONS

Strength of Recommendation: Weak.Quality of Evidence: Low, from one

poor-quality class III study.

A. Level I


B. Level II


C. Level III

Prophylactic treatment with pheny-toin may be considered to reduce theincidence of early posttraumatic seizures(PTS) in pediatric patients with severetraumatic brain injury (TBI).


III. OVERVIEW

Posttraumatic seizures are definedas occurring early, within 7 days of in-jury, or late, beyond 8 days of recovery(1). Risk factors associated with the oc-currence of PTS include location of thelesion, cerebral contusions, retainedbone and metal fragments, depressedskull fracture, focal neurologic deficits,loss of consciousness, Glasgow ComaScale (GCS) score �10, severity of in-jury, length of posttraumatic amnesia,subdural or epidural hematoma, pene-trating injury, chronic alcoholism, andage. Infants and children have lowerseizure thresholds (2), adding to thechallenge of recognition of subtleclinical seizures (3) in critically illchildren.

IV. PROCESS

For this update, MEDLINE wassearched from 1996 through 2010 (Ap-pendix B for search strategy), and re-

sults were supplemented with literaturerecommended by peers or identifiedfrom reference lists. Of 15 potentiallyrelevant new studies, no new studieswere used as evidence for this topic.


One class III study met the inclusioncriteria for this topic and provides evi-dence to support the recommendation.Data from a single center retrospectivecohort study of children ages 3 monthsto 15 yrs identified by InternationalClassification of Diseases, 9th Revisioncode were reported by Lewis et al (4).This study reported a significant reduc-tion in early PTS rate in the severe TBIcases treated with prophylactic pheny-toin compared with patients with severeTBI who were not treated prophylacti-cally (15% vs. 53%, p � .04, one-tailedFisher’s exact test). Limitations of thisstudy include the small size of the se-vere TBI group, the decision to treatbased on individual physician prefer-ence, and the absence of data onlong-term outcome, phenytoin levels,or complications of anticonvulsanttherapy.



Based on data from five studies, theadult guidelines for the prevention of PTSprovide a level II recommendation for theuse of anticonvulsants to decrease theincidence of early PTS (3). Among thesestudies, three compared phenytoin withplacebo, one compared phenobarbitalwith placebo, and one compared pheny-toin with valproate. The use of eitherphenytoin or valproic acid as prophylaxisto reduce the incidence of late PTS is notrecommended. Similar recommenda-tions have been published elsewhere (5).There are no data to show that early PTSare associated with worse outcomes.

A prospective study by Temkin et al(6) was performed as a double-blind,

placebo-controlled study to determinethe effect of treatment with phenytoinon early and late PTS in 404 patients.Importantly, dosages were adjusted tomaintain therapeutic levels. In thetreated group, the incidence of earlyPTS was 3.6%, a significant reduction(p � .001) compared with placebo(14.2%) (risk ratio, 0.27; 95% confi-dence interval [CI], 0.12– 0.62). Treat-ment with phenytoin had no effect oneither late PTS or survival comparedwith placebo.

A randomized, double-blind trial toevaluate the effect of valproic acid on theincidence of PTS compared phenytoinwith valproic acid (7). One hundred thir-ty-two patients were randomized to 1-wktreatment with phenytoin, 120 to 1month of valproic acid, and 126 to 6months of valproic acid. The rates of earlyPTS did not differ between treatmentgroups (1.5% for the phenytoin groupand 4.5% for both arms of the valproicacid group) and there were also no differ-ences in the rate of late PTS. There was atrend toward higher mortality rate in pa-tients treated with valproic acid com-pared with phenytoin (13.4% vs. 7.2%,p � .07; risk ratio, 2.0; 95% CI, 0.9–4.1).


To address the question aboutwhether prophylactic treatment re-duces seizures, various questions/issuesneed to be considered. For example: 1)What is the incidence of PTS? 2) Whatis the right anticonvulsant medication?3) What is the appropriate dose? 4)What is the risk– benefit of the drug inthe context of other morbidities afterTBI? 5) Can and should the drug ther-apy be targeted to a high-risk group?

The following studies provide infor-mation about these questions but donot constitute evidence. It is importantto keep in mind that the various studieshave different case definitions whendiscussing PTS: 0 –24 hrs, 0 – 48 hrs,0 –7 days, or 0 –2 yrs.

Frequency of posttraumatic seizuresin pediatric TBI. A number of studiesthat report the inclusion of pediatric


DOI: 10.1097/PCC.0b013e31823f681d


cases have examined the frequency ofearly and late PTS after severe TBI.

In a four-center study, subjects �6yrs admitted between 1993 and 1998with computed tomography evidence ofTBI or a GCS less �10 24 hrs postinjurywith negative computed tomographywere studied to determine the naturalhistory of later PTS in moderate andsevere TBI (8). The subjects were fol-lowed for 2 yrs or until the first seizure�8 days after TBI, death, or treatmentwith an anticonvulsant. Among the 647subjects, 43% were �30 yrs. Sixty-six(10%) of subjects had a late PTS, al-though 26% of the total were lost tofollow-up. The probability of developinglate PTS at 2 yrs after TBI was 13.8%(66 of 480). The majority (79%) of theseseizures were generalized. The lengthof initial anticonvulsant prophylaxiscorrelated with a greater frequency oflate PTS. The relative risk of seizures at2 yrs after treatment with phenytoin ondays 1–7 was 1.56 compared with 4.27in the subjects treated up to 30 daysafter injury. It is possible this differencereflects differences in the severity ofinjury between these groups.

The incidence of late PTS was exam-ined in two populations in Italy, a ret-rospective study of 55 cases and a pro-spective study of 82 subject all withsevere TBI (9). In the retrospectivegroup (age range, 14 – 62 yrs), ten pa-tients (18%) had PTS of whom half hadbeen treated with an anticonvulsant

(phenobarbital) and half had not. In theprospective part of the study, 84% ofthe subjects were treated with prophy-lactic anticonvulsants during 2-yr fol-low-up and 39% experienced PTS.There were no PTS in the subjects whowere not treated with an anticonvul-sant. This counterintuitive finding mayagain reflect the clinical assessment ofthe need for treatment in the more se-verely impaired subjects.

A retrospective study from two hos-pitals in Turkey examined the risk fac-tors for PTS in children �16 yrs (10).There were 149 cases of PTS (8.4%) inthe 1785 patients in this series. Youngage (�3 yrs), severity of injury, cerebraledema, depressed skull fracture, andhemorrhage were more common in thecases with PTS. A retrospective reviewof traumatic intracranial hemorrhageconfirmed by computed tomographyscan at three centers in Israel identified52 cases (mean age, 50 yrs; range, 8 – 85yrs) with recurrent seizures (11). Onlyfive cases were �19 yrs, all of whomwere reported as mentally handicapped.The patients with seizures or epilepsywere identified only by InternationalClassification of Diseases, 9th Revisioncode and the majority of cases (44) weremale. This study did not define riskfactors for seizures after traumatic in-tracranial hemorrhage, but rather pro-vided a description of the characteris-tics of patients with traumatic

intracranial hemorrhage leading to re-current seizures.

A study of 102 children aged 1.3–15.2 yrs with severe TBI, of which 85%required mechanical ventilation, all ofwhom received inpatient rehabilitationtherapy between 1991 and 1998, exam-ined the prevalence of posttraumaticepilepsy (12). Follow-up in this studyranged from 19 months to 7 yrs, duringwhich nine subjects (9%) developedposttraumatic epilepsy. The intervalfrom insult to first seizure onset rangedfrom 0.7 to 5.2 yrs (median, 2.9 yrs).The presence of early (within the firstweek post-TBI) seizures (p � .002) andGCS score (p � .043) were the onlyfactors at the time of injury related tothe development of posttraumatic epi-lepsy. A series of 318 children ages 1month to 17 yrs treated between 1965and 1991–with an average follow-up of8 yrs, 9 months–reported early seizuresin 19.8% and an incidence of late sei-zures of 29.6% after open head injurycompared with 20.2% after closed headinjury (13).

Effects of treatment with anticonvul-sants. In a randomized, double-blind,placebo-controlled study of the efficacyof phenytoin in preventing late PTS in41 patients, Young et al (14) found nodifference in rate of PTS in the treatedgroup (12%) compared with controlsubjects (6.2%). All seizures occurredwithin the first year after injury. Com-pliance was poor, and by 6 months,





Lewis et al, 1993 (4) Design: retrospective cohort studyN � 194; 31 with severe traumatic

brain injuryAge: ranged from 3 months to 15

yrs; median, 6 yrsGCS: 3–8 (31 �16%�);9–15 (163 �84%�)Protocol: phenytoin within 24 hrs

of hospital admission or noprophylactic anticonvulsantmedication

Purpose: to determine factorsassociated with early PTS

Outcome: occurrence of anyseizure during hospitalization

Class IIIPoor quality: for comparison of

groups based onanticonvulsant medicationuse; moderate for prognosticfactor analysis: control forconfounders only in analysisof predictors of seizure, notfor comparison of groupsbased on seizure prophylaxis

For children with GCS 3–8, treatment withprophylactic phenytoin was associated with areduced rate of seizures (2 of 13 �15%�) comparedwith patients not treated with prophylacticmedication (9 of 17 �53%�) (p � .04 one-tailedFisher’s; p � .057 two-tailed)

Rate of seizures in total group of 194 was 9.3%In 14 of these 18 cases (78%), seizures occurred

within 24 hrs of injuryGCS of 3–8 (p � .01) and abnormal computed

tomography (p � .02) associated with increased riskof early PTS

Logistic regression performed to account forcontribution of abnormal computed tomography,loss of consciousness, and GCS score to risk forPTS showed only association with GCS of 3–8 (p� .001)

GCS, Glasgow Coma Scale; PTS, posttraumatic seizures.


serum levels of phenytoin were avail-able on only 15 (60%) of the treatmentgroup. Among this group, six subjects(40%) had a measured serum drug levelof �10 �g/mL. Notably, no patientswith a serum level �10 �g/mL had aseizure. The study is limited by thesmall size, poor compliance, unclearcriteria for randomization, and lack ofclarity over association between GCSand outcome. Sixteen (39%) of the sub-jects had a GCS of �7. Because theanalysis combined severe and moderatepatients, it did not meet criteria forinclusion as evidence for this topic.

In a prospective cohort study of chil-dren admitted to the pediatric intensivecare units at three centers, Tilford et al(15) identified 138 cases of severe TBIamong the 477 children admitted witha diagnosis of head trauma. There was asignificant variation in anticonvulsantuse (range, 10% to 35%) among thethree centers with an overall incidenceof early PTS of 9.4%. The type of anti-convulsant used was not specified. Theindications for such use, either prophy-laxis or in response to a clinical orelectrographic seizure, were also notspecified. In a stepwise logistic regres-sion model (accounting for GCS, theparticipating site, other therapies), theuse of an anticonvulsant medicationwas associated with a significant reduc-tion in risk of mortality (p � .014; oddsratio, 0.17; 95% CI, 0.04 – 0.70), but theanalysis was not limited to patients inthe severe TBI group.

A study by Young et al (16) reportedno reduction in the rate of PTS within48 hrs of injury in a randomized, dou-ble-blind, placebo-controlled trial ofphenytoin in children with moderate tosevere blunt head injury. Children �16yrs with a GCS of �9 (�4 yrs) or �10(�4 yrs) were enrolled by deferred con-sent within 40 mins of presentation tothe emergency department and drug orplacebo administered within 60 mins ofpresentation. Phenytoin dose was 18mg/kg followed by 2 mg/kg every 8 hrsfor the 48 hrs of the study. Subjectswere stratified by age and GCS. Onehundred three subjects were random-ized with 33% lost at 48-hr follow-upand 36% lost at 30-day follow-up. In thephenytoin-treated group, three patients(7%) had a seizure during the 48-hrobservation period compared with three(5%) in the placebo group. Six patients(one phenytoin, five placebo) had anelectroencephalogram performed. None

showed nonconvulsive seizures. Over30 days recovery, there was no differ-ence in mortality in the treatmentgroup (20% [six of 30]) compared withplacebo (39% [14 of 36]). The majorlimitations of this study are the lowseizure rate and the small sample sizeresulting from early loss of subjects anddecrease in enrollment after ceasing towaive consent.

A study of 318 cases of severe TBIfrom a single center in Germany withmean follow-up of 8 yrs, 9 months iden-tified 68 cases (21%) of late seizureswith a mean latency of 2 yrs, 5 months(17). Approximately half of these caseswere resistant to anticonvulsant ther-apy, although the details of therapy arenot given. The children with PTS had aworse outcome in this series with 60%having disabilities compared with 17%in the other patients.

In a single-center, prospective studyduring the war in Bosnia, 310 patientsbetween 0 and 18 yrs with severe TBIwere treated with either intravenousphenytoin or phenobarbital, dependingon the availability of each drug (18).The primary outcome was the fre-quency of seizure in the first 24 hrsafter admission. This was low, occur-ring only in two cases (0.64%). Al-though the criteria for classification ofcases as severe are not specified, skullfracture was present in 85% of cases.The frequency of seizures is low andthere is no detail on the process formonitoring for seizures, suggestingthis may be an underestimate.

Pharmacokinetic considerations. TBIresults in an increase in hepatic metab-olism and decrease in protein bindingof drugs including anticonvulsants(19), resulting in an increase in plasmaclearance. The free fraction of pheny-toin is elevated (20). The altered phar-macokinetics of phenytoin and otherdrugs may result in levels considered tobe subtherapeutic. As part of a clinicaltrial evaluating the use of valproic acidfor prophylaxis of posttraumatic sei-zures, the time-dependent effects of TBIon the pharmacokinetics of total andunbound valproic acid were evaluated(21). In the trial, 158 adult TBI cases(mean age, 36 yrs; mean GCS, 10;range, 3–15) were treated with a load-ing dose of valproic acid (20 mg/kg)followed by a maintenance dose. TBIresulted in an average 75% increase indrug clearance by 2 and 3 wks of recov-ery, which was associated with in-

creased TBI severity, lower albuminconcentration, tube feeding, and thepresence of ethanol on admission. Ingeneral, there are limited data (22) onthe effect of early age, genetic factors,and other drug interactions affectingpharmacokinetics of anticonvulsantsafter TBI and the contribution of thesefactors to neurologic outcomes.

Mechanisms of epileptogenesis rele-vant to pediatric TBI. Studies of themechanisms of posttraumatic epilepsytraditionally were limited by the lack ofanimal models; however, recent studieshave begun to focus on PTS in develop-ing animals after experimental TBI (23–25). A number of mechanisms of post-traumatic epilepsy have beeninvestigated; many focused on patho-physiological changes in the hippocam-pus including axonal sprouting, im-paired K� buffering by glia, saturationof synaptic long-term potentiation ofSchaffer collaterals, hilar neuron loss,and activation of hippocampal TrkB-ERK1/2-CREB/ELK-1 pathways (23,26). Recent studies have suggested arole for albumin-induced changes inthe electrophysiological properties ofastrocytes mediated by the transform-ing growth factor-� receptor and lead-ing to accumulation of extracellular po-tassium (27, 28).

VII. SUMMARY

The incidence of early PTS in pedi-atric patients with TBI is approximately10% given the limitations of the avail-able data. Based on a single class IIIstudy (4), prophylactic anticonvulsanttherapy with phenytoin may be consid-ered to reduce the incidence of earlyposttraumatic seizures in pediatric pa-tients with severe TBI. Concomitantmonitoring of drug levels is appropriategiven the potential alterations in drugmetabolism described in the context ofTBI. Stronger class II evidence is avail-able supporting the use of prophylacticanticonvulsant treatment to reduce therisk of early PTS in adults. There are nocompelling data in the pediatric TBIliterature to show that such treat-ment reduces the long-term risk of PTSor improves long-term neurologicoutcome.


● Investigation of the frequency of earlyPTS in the setting of contemporary


management and their associationwith acute pathophysiology and long-term neurologic sequelae.

● Investigation of the efficacy, safety, anddrug levels required for the preventionof early posttraumatic seizures.

● Investigation of the efficacy andsafety of new anticonvulsants for thetreatment of early and late posttrau-matic seizures.

● Identification of neuroimaging, elec-troencephalography, or serum bio-markers, which serve to predict pa-tients at increased risk for lateposttraumatic seizures.

● Elucidation of the mechanisms of epi-leptogenesis after TBI and identifica-tion of new therapeutic targets basedon understanding these mechanisms.

● Improvement in the classification ofearly and late seizures, including the useof electroencephalography, to detect andclassify posttraumatic seizures.

● Evaluation of the effect of TBI onchanges in dosage requirements for an-ticonvulsant drugs and the contribu-tion of age and genetically determineddifferences in hepatic and renal drugmetabolism to the efficacy of anticon-vulsants in the treatment of posttrau-matic seizures.

REFERENCES

1. Yablon SA: Posttraumatic seizures. ArchPhys Med Rehabil 1993; 74:983–1001

2. Holmes GL: Effects of seizures on brain de-velopment: Lessons from the laboratory. Pe-diatr Neurol 2005; 33:1–11

3. Bratton SL, Chestnut RM, Ghajar J, et al:Guidelines for the management of severetraumatic brain injury. XIII. Antiseizureprophylaxis. J Neurotrauma 2007;24(Suppl 1):S83–S86

4. Lewis RJ, Yee L, Inkelis SH, et al: Clinicalpredictors of post-traumatic seizures in chil-dren with head trauma. Ann Emerg Med1993; 22:1114–1118

5. Chang BS, Lowenstein DH, Quality Stan-dards Subcommittee of the American Acad-emy of Neurology: Practice parameter: An-tiepileptic drug prophylaxis in severetraumatic brain injury: report of the Qual-ity Standards Subcommittee of the Ameri-can Academy of Neurology. Neurology2003; 60:10 –16

6. Temkin NR, Dikmen SS, Wilensky AJ, et al: Arandomized, double-blind study of phenytoinfor the prevention of post-traumatic sei-zures. N Engl J Med 1990; 323:497–502

7. Temkin NR, Dikmen SS, Anderson GD, et al:Valproate therapy for prevention of posttrau-matic seizures: A randomized trial. J Neuro-surg 1999; 91:593–600

8. Englander J, Bushnik T, Duong TT, et al:Analyzing risk factors for late posttrau-matic seizures: A prospective, multicenterinvestigation. Arch Phys Med Rehabil2003; 84:365–373

9. Formisano R, Barba C, Buzzi MG, et al: Theimpact of prophylactic treatment on post-traumatic epilepsy after severe traumaticbrain injury. Brain Inj 2007; 21:499–504

10. Ates O, Ondul S, Onal C, et al: Post-traumaticearly epilepsy in pediatric age group withemphasis on influential factors. Childs NervSyst 2006; 22:279–284

11. Medvdovsky M, Ifergane G, Wirguin I, et al:Traumatic intracranial hemorrhage in pa-tients with seizures: descriptive characteris-tics. Epilepsy Behav 2006; 8:429–433

12. Appleton RE, Demellweek C: Post-traumaticepilepsy in children requiring inpatient reha-bilitation following head injury. J NeurolNeurosurg Psychiatry 2002; 72:669–672

13. Kieslich M, Jacobi G: Incidence and risk fac-tors of post-traumatic epilepsy in childhood.Lancet 1995; 345:187

14. Young B, Rapp RP, Norton JA, et al: Failureof prophylactically administered phenytointo prevent post-traumatic seizures in chil-dren. Childs Brain 1983; 10:185–192

15. Tilford JM, Simpson PM, Yeh TS, et al: Vari-ation in therapy and outcome for pediatrichead trauma patients. Crit Care Med 2001;29:1056–1061

16. Young KD, Okada PJ, Sokolove PE, et al: Arandomized, double-blinded, placebo-con-trolled trial of phenytoin for the preventionof early posttraumatic seizures in children

with moderate to severe blunt head injury.Ann Emerg Med 2004; 43:435–446

17. Kieslich M, Marquardt G, Galow G, et al:Neurological and mental outcome after se-vere head injury in childhood: A long-termfollow-up of 318 children. Disabil Rehabil2001; 23:665–669

18. Gavranovic M, Konjhodzic F, Zubcevic S, etal: Posttraumatic seizures—Prevention ornot. Bosn J Basic Med Sci 2005; 5:58–60

19. Boucher BA, Hanes SD: Pharmacokineticalterations after severe head injury. Clini-cal relevance. Clin Pharmacokinet 1998;35:209 –221

20. Griebel ML, Kearns GL, Fiser DH, et al: Phe-nytoin protein binding in pediatric patientswith acute traumatic injury. Crit Care Med1990; 18:385–391

21. Anderson GD, Temkin NR, Awan AB, et al:Effect of time, injury, age and ethanol oninterpatient variability in valproic acidpharmacokinetics after traumatic brain in-jury.[ erratum appears in Clin Pharmaco-kinet 2007; 46:447. Note: Winn, Richard Hcorrected to Winn, H Richard]. Clin Phar-macokinet 2007; 46:307–318

22. Frend V, Chetty M: Dosing and therapeuticmonitoring of phenytoin in young adultsafter neurotrauma: Are current practicesrelevant? Clin Neuropharmacol 2007; 30:362–369

23. Kharatishvili I, Pitkanen A: Posttraumaticepilepsy. Curr Opin Neurol 2010; 23:183–188

24. Liesemer K, Bratton SL, Zebrack CM, et al:Early post-traumatic seizures in moderate tosevere pediatric traumatic brain injury: rates,risk factors, and clinical features. J Neu-rotrauma 2011; 28:755–762

25. Pitkanen A, McIntosh TK: Animal models ofpost-traumatic epilepsy. J Neurotrauma2006; 23:241–261

26. Garga N, Lowenstein DH : Posttraumatic ep-ilepsy: A major problem in desperate need ofmajor advances. Epilepsy Curr 2006; 6:1–5

27. Ivens S, Kaufer D, Flores LP, et al: TGF-betareceptor-mediated albumin uptake into as-trocytes is involved in neocortical epilepto-genesis. Brain 2007; 130:535–547

28. van Vliet EA, da Costa Araujo S, Redeker S, etal: Blood–brain barrier leakage may lead toprogression of temporal lobe epilepsy. Brain2007; 130:521–534


APPENDIX A

Publications from the First Edition Not Included in the Second Edition

Topic Reference Reason(s) for Exclusion

Indications for ICPmonitoring

Cho, 1995 Data not relevant to this topicTaylor, 2001 GCS range exceeds 8 with no separate analysis of severeSharples part I, 1995 No direct correlation between ICP and outcomeEder, 2000 Retrospective, N � 21Peterson, 2000 Treatment study about effect of hypertonic saline on ICP

ICP thresholds Cho, 1995 Data not relevant to this topicShapiro and Marmarou, 1982 Data not relevant to this topicSharples part I, 1995 No direct correlation between ICP and outcome

Cerebral perfusion pressurethresholds

Elias-Jones, 1992 GCS range exceeds 8 with no separate analysis of severeSharples part III, 1995 No association between ICP/CPP and outcome

Hyperosmolar therapy James, 1980 Mean age 42 yrs, with no separate analysis of pediatric patientsMiller, 1993 4 of 16 patients are children and relevant data are not provided by ageKhanna, 2000 Prospective cohort, N � 10

Temperature control Gruszkiewicz, 1973 Randomized controlled trial, N � 20Decompressive craniectomy Polin, 1997 Average age 18.7 12.6 yrs with no separate analysis of pediatric patients

Taylor, 2001 GCS range �8 with no separate analysis of severeHyperventilation Stringer, 1993 Case series, N � 3Corticosteroids Gobiet, 1977, Advances in . . . GCS not reported

Gobiet, 1977, Monitoring of . . . Sample includes adults with no separate analysis of pediatric patientsHoppe, 1981 No comparison groupKretschmer, 1983 27% penetrating brain injury without separate analysisJames, 1979 Retrospective, N � 9Cooper, 1979 Prospective cohort, N � 10

Analgesics, sedatives, andneuromuscular blockade

Vernon and Witte, 2000 Includes patients with pathologies other than traumatic brain injury withoutseparate analysis

Glucose and nutrition Phillips, 1987 No analysis of association between any nutritional parameter and any clinical outcomeMoore, 1989 Age range is 3–67 yrs with no separate analysis of pediatric patients

Antiseizure prophylaxis Tilford, 2001 Does not analyze the effect of anticonvulsants within the severe groupYoung, 1983 GCS range �8 with no separate analysis of severe

ICP, intracranial pressure; GCS, Glasgow Coma Scale; CPP, cerebral perfusion pressure.


APPENDIX B

Literature Search Strategies

Indications for Intracerebral PressureMonitoring

Database: Ovid Medline �1996 to 2010�Search Strategy

Line Search

1 exp Craniocerebral Trauma/2 head injur$.mp. [mp � title, original

title, abstract, name of substance word,subject heading word]

3 brain Injur$.mp. [mp � title, originaltitle, abstract, name of substance word,subject heading word]

4 1 or 2 or 35 intracranial pressure.mp. or Exp In-

tracranial Pressure/6 intracranial hypertension.mp. or exp

Intracranial Hypertension/7 5 or 68 4 and 79 Limit 8 to “all Child (0 to 18 Yrs)”

10 (2001$ or 2002$ or 2003$ or 2004$ or2005$ or 2006$ or 2007$ or 2008$ or2009$).ed.

11 10 and 9

Intracerebral Pressure ThresholdsDatabase: Ovid Medline �1996 to 2010�Search Strategy

Line Search



3 brain injur$.mp. [mp � title, originaltitle, abstract, name of substance word,subject heading word]

4 1 or 2 or 35 intracranial pressure.mp. or exp In-

tracranial Pressure/6 intracranial hypertension.mp. or exp

Intracranial Hypertension/7 5 or 68 4 and 79 limit 8 to “all child (0 to 18 yrs)”


11 10 and 9

Cerebral Perfusion Pressure ThresholdsDatabase: Ovid Medline �1996 to 2010�Search Strategy

Line Search




4 1 or 2 or 35 cerebral perfusion pressure.mp.6 cerebrovascular circulation/and blood

pressure/7 5 or 68 4 and 79 Limit 8 to “all child (0 to 18 Yrs)”


11 10 and 9

Advanced NeuromonitoringDatabase: Ovid Medline �1950 to 2010�Search Strategy

Line Search

1 Exp Craniocerebral Trauma/2 ((head or brain$ or cereb$ or cerebell$)

adj3 (wound$ or traum$ or injur$ ordamag$)).mp. [mp � title, original ti-tle, abstract, name of substance word,subject heading word]

3 1 or 24 exp Monitoring, Physiologic/5 exp Intensive Care Units/or Exp In-

tensive Care/6 4 and 3 and 57 exp Oxygen/bl, an [Blood, Analysis]8 licox.mp.9 pbto2.mp.

10 ((oxygen$ or o2 or hypoxi$) adj3(concentrat$ or level$ or monitor$ orpressur$)).mp.

11 exp Oximetry/12 8 or 11 or 7 or 10 or 913 ((transcrani$ adj3 (doppler or ultra-

sono$)) or tcd).mp. [mp � title, orig-inal title, abstract, name of substanceword, subject heading word]

14 ((near infrared adj3 spectrosc$) ornirs).mp. [mp � title, original title,abstract, name of substance word,subject heading word]

15 exp Phosphopyruvate Hydratase/16 exp Nervous System/17 exp Nervous System Diseases/18 17 or 1619 18 and 1520 neuron specific enolase$.mp.

21 nse.mp.22 21 or 20 or 1923 exp S100 Proteins/24 (s100b or S100 �).mp. [mp � title,

original title, abstract, name of sub-stance word, subject heading word]

25 24 or 2326 exp Myelin Basic Proteins/27 (Myelin basic protein$ or mbp).mp.

[mp � title, original title, abstract, nameof substance word, subject heading word]

28 26 or 2729 glutamat$.mp.30 xenon.mp. or exp Xenon/31 ((brain$ or cereb$ or cerebell$) adj5

((interstitial$ or extracellul$) adj3(fluid$ or space$))).mp. [mp � title,original title, abstract, name of sub-stance word, subject heading word]

32 exp Extracellular Space/or exp Extra-cellular Fluid/

33 exp Brain/34 32 and 3335 34 or 3136 microdialysis.mp. [mp � title, origi-

nal title, abstract, name of substanceword, subject heading word]

37 exp Biological Markers/38 (biomarker$ or biological mark-

er$).mp. [mp � title, original title,abstract, name of substance word,subject heading word]

39 37 or 3840 3 and 541 40 and 1242 13 and 4043 14 and 4044 22 and 4045 25 and 4046 28 and 4047 29 and 4048 30 and 4049 35 and 4050 36 and 4051 39 and 4052 50 or 51 or 41 or 48 or 47 or 42 or 49

or 46 or 45 or 43 or 4453 52 or 6 (333)54 limit 53 to “all child (0 to 18 yrs)”55 limit 54 to English language

NeuroimagingDatabase: Ovid Medline �1950 to 2010�Search Strategy

Line Search

1 exp Craniocerebral Trauma/2 ((head or brain$ or cereb$ or cer-

ebell$) adj3 (wound$ or traum$ or


injur$ or damag$)).mp. [mp � title,original title, abstract, name ofsubstance word, subject headingword]

3 1 or 24 exp Tomography, X-Ray Computed/5 exp Magnetic Resonance Imaging/6 ((t2 or t1 or diffusion or susceptibil-

ity) adj weight$ adj3 imag$).mp.[mp � title, original title, abstract,name of substance word, subjectheading word]

7 exp Magnetic Resonance Spectrosco-py/

8 (magnetic$ adj resonan$ adj2 spec-troscop$).mp. [mp � title, originaltitle, abstract, name of substanceword, subject heading word]

9 8 or 710 apparent diffusion coefficient$.mp.11 exp Tomography, Emission-Comput-

ed/12 (positron$ adj emission$ adj2 spec-

troscop$).mp. [mp � title, originaltitle, abstract, name of substanceword, subject heading word]

13 (positron$ adj emission$ adj3 tomo-gra$).mp. [mp � title, original title,abstract, name of substance word,subject heading word]

14 pet scan$.mp.15 11 or 13 or 12 or 1416 6 or 4 or 10 or 9 or 15 or 517 3 and 1618 Limit 17 to (English Language and

Humans)19 Limit 18 to “all child (0 to 18 yrs)”20 exp Intensive Care Units/or exp Inten-

sive Care/21 ((intensiv$ or critical$) adj2 (care or

cared or caring or treat$ or thera-p$)).mp. [mp � title, original title,abstract, name of substance word,subject heading word]

22 (icu or ccu).mp. [mp � title, originaltitle, abstract, name of substanceword, subject heading word]

23 22 or 21 or 2024 23 and 1925 exp Emergency Treatment/26 exp Emergency Service, Hospital/27 25 or 2628 27 and 1929 28 or 24

Hyperosmolar TherapyDatabase: Ovid Medline �1996 to 2010�Search Strategy

Line Search




4 1 or 2 or 35 hyperosmolar therapy.mp.6 hyperosmolar treatment.mp.7 fluid therapy.mp. or exp Fluid

Therapy/8 Saline Solution, Hypertonic/9 Osmolar Concentration/

10 5 or 6 or 7 or 8 or 911 4 and 1012 limit 11 to (English language and hu-

mans)13 limit 12 to “all child (0 to 18 yrs)”14 (2001$ or 2002$ or 2003$ or 2004$ or

2005$ or 2006$ or 2007$ or 2008$ or2009$).ed.

15 13 and 14

Temperature ControlDatabase: Ovid Medline �1996 to 2010�Search Strategy

Line Search

1 exp Craniocerebral Trauma/2 head injur$.mp. [mp � title, origi-

nal title, abstract, name of sub-stance word, subject heading word]

3 brain injur$.mp. [mp � title, origi-nal title, abstract, name of sub-stance word, subject heading word]

4 1 or 2 or 35 Hypothermia, Induced/6 4 and 57 limit 6 to “all child (0 to 18 yrs)”8 (2001$ or 2002$ or 2003$ or 2004$ or

2005$ or 2006$ or 2007$ or 2008$ or2009$).ed.

9 8 and 7

Line Search

1 Exp Craniocerebral Trauma/2 ((brain$ or cereb$ or cerebell$ or

head) adj3 (traum$ or damag$ orinjur$ or wound$)).mp. [mp � title,original title, abstract, name of sub-stance word, subject heading word,unique identifier]

3 1 or 24 exp Fever/5 Fever$.mp.6 4 or 5

7 3 and 68 limit 7 to “all child (0 to 18 yrs)”9 limit 8 to English language

10 hypertherm$.mp.11 1 and 1012 limit 11 to “all child (0 to 18 yrs)”13 limit 12 to English language14 9 or 13

Cerebrospinal Fluid DrainageDatabase: Ovid Medline �1996 to 2010�Search Strategy

Line Search




4 1 or 2 or 35 lumbar drain$.mp.6 lumbar shunt$.mp.7 exp Cerebrospinal Fluid Shunts/8 *Drainage/9 5 or 6 or 7 or 8

10 4 and 911 limit 10 to “all child (0 to 18 yrs)”12 (2001$ or 2002$ or 2003$ or 2004$ or

2005$ or 2006$ or 2007$ or 2008$ or2009$).ed.

Decompressive CraniotomyDatabase: Ovid Medline �1996 to 2010�Search Strategy

Line Search




4 1 or 2 or 35 intracranial hypertension.mp. or Exp

Intracranial Hypertension/6 4 and 57 limit 6 to “all child (0 to 18 yrs)”8 limit 7 to English language9 su.fs.

10 drain$.mp.11 cerebrospinal fluid shunts.mp. or exp

Cerebrospinal Fluid Shunts/12 neurosurgery.mp. Or Neurosurgery/13 shunt$.mp.14 9 or 10 or 11 or 12 or 13


15 8 and 1416 (2001$ or 2002$ or 2003$ or 2004$ or

2005$ or 2006$ or 2007$ or 2008$ or2009$).ed.

17 16 and 15

HyperventilationDatabase: Ovid Medline �1950 to 2010�Search Strategy

Line Search

1 exp Craniocerebral Trauma/2 exp ISCHEMIA/3 exp Jugular Veins/4 exp Regional Blood Flow/5 exp PERFUSION/6 Exp HYPERVENTILATION/7 2 or 3 or 4 or 5 or 68 1 and 79 limit 8 to “all child (0 to 18 yrs)”


CorticosteroidsDatabase: Ovid Medline �1996 to 2010�Search Strategy

Line Search




4 1 or 2 or 35 exp Steroids/or steroids.mp.6 synthetic glucocorticoids.mp.7 5 or 68 4 and 79 limit 8 to “all child (0 to 18 yrs)”


11 10 and 9

Analgesics, Sedatives, and Neuromus-cular Blockade

Database: Ovid Medline �1950 to 2010�Search Strategy

Line Search

1 exp Analgesics/2 exp “Hypnotics and Sedatives”/3 propofol.mp. [mp � title, original title,

abstract, name of substance word, sub-ject heading word]

4 exp phenothiazines/5 exp central nervous system depressants/6 1 or 2 or 4 or 57 exp Craniocerebral Trauma/8 6 and 79 Limit 8 to (English language and hu-

mans)10 limit 9 to “all child (0 to 18 yrs)”

Glucose and NutritionDatabase: Ovid Medline �1950 to 2010�Search Strategy

Line Search

1 exp Craniocerebral Trauma/2 ((head or brain$ or cereb$ or cerebell$)

adj3 (wound$ or traum$ or injur$ ordamag$)).mp. [mp � title, original ti-tle, abstract, name of substance word,subject heading word]

3 1 or 24 exp Glucose/5 exp hyperglycemia/or exp hypoglyce-

mia/6 exp Insulin/7 exp diet/8 exp Nutrition Therapy/9 exp nutritional status/

10 exp nutritional requirements/11 exp Enteral Nutrition/12 exp Intubation, Gastrointestinal/13 exp Feeding Methods/14 exp Gastrostomy/15 exp Energy Metabolism/16 Exp Energy Intake/17 harris-benedict equation.mp.18 exp Nutritional Requirements/19 intralipid.mp. or exp Fat Emulsions,

Intravenous/20 (metaboli$ adj3 (cart or carts)).mp.

[mp � title, original title, abstract, nameof substance word, subject heading word]

21 4 and 322 3 and 523 6 and 324 3 and 7

25 3 and 826 3 and 927 3 and 1028 3 and 1129 12 and 330 3 and 1331 3 and 1432 15 and 333 3 and 1634 3 and 1735 3 and 1836 3 and 1937 3 and 2038 24 or 25 or 26 or 27 or 35 or 33 or 36

or 29 or 34 or 21 or 28 or 30 or 22 or32 or 23 or 31 or 37

39 limit 38 to English language40 limit 39 to humans41 limit 40 to “all child (0 to 18 yrs)”

Antiseizure ProphylaxisDatabase: Ovid Medline �1996 to 2010�Search Strategy

Line Search




4 1 or 2 or 35 exp Seizures/or Seizures.mp.6 exp Epilepsy/7 exp convulsions/or convulsions.mp.8 5 or 6 or 79 4 and 8

10 limit 9 to “all child (0 to 18 yrs)”11 exp seizures/dt, pc or exp epilepsy/dt,

pc or convulsions/dt, pc12 4 and 1113 limit 12 to “all child (0 to 18 yrs)”14 exp Clinical Trials as Topic/15 Exp Practice Guidelines as Topic/or

practice guidelines.mp.16 14 or 1517 10 and 1618 13 or 1719 (2001$ or 2002$ or 2003$ or 2004$ or

2005$ or 2006$ or 2007$ or 2008$ or2009$).ed.

20 18 and 19


APPENDIX C

Literature Search Yield

TopicSearchResults

AbstractsRead

PublicationsRead

Included FirstEdition Studies

IncludedNew Studies

Indications for intracranialpressure monitoring

756 422 35 9 7

Intracranial pressuretreatment threshold

756 422 60 6 5

Cerebral perfusion pressurethresholds

219 161 78 3 8

Advanced neuromonitoring 121 74 44 N/A 2Neuroimaging 344 161 89 N/A 1Hyperosmolar therapy 213 31 9 3 0Temperature control 228 53 17 2 2Cerebrospinal fluid drainage 136 32 6 3 1Barbiturates 212 87 47 2 0Decompressive craniectomy 160 83 18 1 7Hyperventilation 295 141 16 1 1Steroids 138 20 19 2 0Analgesics, sedatives

Neuromuscular blockade699 121 44 0 2

Nutrition 593 182 113 0 1Antiseizure prophylaxis 68 17 10 1 0

N/A, not applicable.


APPENDIX D

Mixed Patient Samples

Criteria for including a study in whichthe sample includes patients with TBIand patients with other pathologies, orpediatric and adult patients

If:

● the sample for a study includes pa-tients with TBI as well as patientswith other pathologies, pediatric aswell as adult patients, or mild/moderate as well as patients with se-vere TBI,

● and the data are not reported separately,● and there is an effect of the study,

then it cannot be known if the effectexisted for the TBI group or if it waslarge in the non-TBI group and small inthe TBI group. Similarly, it cannot beknown if the effect existed for the pedi-atric group or if it was large in the adultgroup and small in the pediatric group.Therefore, we cannot know with confi-dence that the intervention had an ef-fect for TBI in pediatric patients.

We have established the following crite-ria to minimize the uncertainty when in-cluding publications with mixed samples:

1. Sample size must be �25 patients.2. �85% of the patients must have se-

vere TBI.3. �85% of the patients must be �18 yrs

of age.

4. Such a study could never be used tosupport a level I recommendation.

5. Such a study can only support a levelII or III recommendation. It cannotbe used to support a level II recom-mendation if it is the only class IIstudy available.

6. If a publication mixes the results of pedi-atric patients with those of adults, and themean and standard deviation for age areprovided, the mean and standard devia-tion can be used to calculate the propor-tion of pediatric patients, and if the pro-portion is �85%, the study can be used asevidence.

7. If the study does not report the per-cent of patients with TBI, it cannot beused as evidence at any level.


APPENDIX E

Evidence Table Template

Source Design Study

Setting/Population CointerventionsConfounding

Variables

Length ofFollow-Up Measures Evidence

Level of

Sample Intervention

Analysis Results Caveats


american academy of pediatrics-section on neurological ... · guidelines for the acute medical...

Documents