Background: There are many clinical decision rules for adults with minor
head injury, but it is unclear how they compare in terms of diagnostic I t is estimated that at least 1.7 million people sustain head
injury every year in the United States.1 Most head injuries
(90%) will be minor (Glasgow Coma Scale [GCS] score,
accuracy. This study aimed to systematically identify clinical decision rules
for adults with minor head injury and compare the estimated diagnostic 13–15) and do not require immediate neurosurgical interven-
accuracies for any intracranial injury and injury requiring neurosurgical tion or inpatient care. These patients have a small but important
intervention. risk of subsequent deterioration because of intracranial bleeding.
Methods: Several electronic bibliographic databases covering biomedical, If these cases are recognized and treated early, then a full
scientific, and gray literature were searched from inception to March 2010. recovery is likely; if not, severe disability or death may ensue.
At least two independent reviewers determined the eligibility of cohort Computed tomography (CT) of the head is the diagnos-
studies that described a clinical decision rule to identify adults with minor tic standard for identifying intracranial injury. Routine CT of
head injury (Glasgow Coma Scale score, 13–15) at risk of intracranial injury
all minor head injury patients would result in a large number
or injury requiring neurosurgical intervention.
Results: Twenty-two relevant studies were identified. Differences existed in
of normal CT scans being performed with associated risks of
patient selection, outcome definition, and reference standards used. Nine radiation exposure and waste of health care resources. There-
rules stratified patients into high- and moderate-risk categories (to identify fore, researchers have attempted to derive clinical decision
neurosurgical or nonsurgical intracranial lesions). The Canadian Computed rules to identify those at risk of intracranial injury based on
Tomography Head Rule (CCHR) high-risk criteria have sensitivity of 99% clinical characteristics at presentation to select them for
to 100% with specificity of 48% to 77% for injury requiring neurosurgical imaging. There are many such clinical decision rules for
intervention. Other rules such as New Orleans criteria, National Emer- adults with minor head injury, all of which vary in the
gency X-Radiography Utilization Study II, Neurotraumatology Commit- clinical criteria they use to identify those at risk of intra-
tee of the World Federation of Neurosurgical Societies, Scandinavian, cranial injury. Table 1 (see Table, Supplemental Digital
and Scottish Intercollegiate Guidelines Network produce similar sensi-
Content 1, http://links.lww.com/TA/A66) outlines some com-
tivities for injury requiring neurosurgical intervention but with lower and
more variable specificity values.
mon rules.
Discussion: The most widely researched decision rule is the CCHR, which It is currently unclear how existing rules compare in
has consistently shown high sensitivity for identifying injury requiring terms of diagnostic accuracy. This study aims to systemati-
neurosurgical intervention with an acceptable specificity to allow considered cally identify clinical decision rules for adults with minor
use of cranial computed tomography. No other decision rule has been as head injury and compare the decision or prediction rules in
widely validated or demonstrated as acceptable results, but its exclusion terms of estimated diagnostic accuracy for any intracranial
criteria make it difficult to apply universally. injury and injury requiring neurosurgical intervention.
Key Words: Craniocerebral trauma, Decision support techniques, System-
atic review, Clinical decision rules, Sensitivity and specificity. METHODS
(J Trauma. 2011;71: 245–251)
Search Strategy
Potentially relevant studies were identified through
Submitted for publication June 23, 2010. electronic searches of key databases including MEDLINE,
Accepted for publication December 23, 2010.
Copyright © 2011 by Lippincott Williams & Wilkins
EMBASE, CINAHL, and CENTRAL up to April 2009,
From the School of Health and Related Research, University of Sheffield, supplemented with an update from MEDLINE to March
Sheffield, South Yorkshire, United Kingdom. 2010. Search terms used free text and thesaurus terms (where
Supported by the NIHR Health Technology Assessment Programme (project available) with Boolean operators and database-specific syn-
number 07/37/08) and will be published in full in the monograph series Health
Technology Assessment. See the HTA programme website for further project tax. Manual searches of reference lists in key publications
information. were used to identify addition studies. Experts in the field
The views and opinions expressed therein are those of the authors and do not were contacted for any additional literature.
necessarily reflect those of the Department of Health.
Supplemental digital content is available for this article. Direct URL citations Inclusion Criteria
appear in the printed text, and links to the digital files are provided in the
HTML text of this article on the journal’s Web site (www.jtrauma.com). Studies were included if they met the following criteria:
Address for reprints: Sue E. Harnan, MSc, School of Health and Related Research, (a) a cohort study with minimum 20 patients at least half of
The University of Sheffield, Regent Court, 30 Regent Street, Sheffield S1 whom had GCS score of 1 to 15 at presentation, (b) they
4DA, United Kingdom; email: s.harnan@sheffield.ac.uk.
evaluated a decision rule that used three or more clinical
DOI: 10.1097/TA.0b013e31820d090f criteria (such as history, physical examination, or a simple
The Journal of TRAUMA® Injury, Infection, and Critical Care • Volume 71, Number 1, July 2011 245
Harnan et al. The Journal of TRAUMA® Injury, Infection, and Critical Care • Volume 71, Number 1, July 2011
Identification
Records identified through database Additional records identified
searching through other sources
(n = 7987) (n =16)
Excluded by title
(n = 4578)
Record screened by
abstract
(n = 3425)
Excluded by abstract
(n = 3203)
for eligibility
(n = 222)
Full-text articles excluded,
with reasons
(n =200)
selection of patients on these criteria may result in a cohort dard for injuries requiring neurosurgical intervention were
with more severe injury. usually because not all patients were followed-up or the
Only three studies scored well for the intracranial length of follow-up was sooner than 30 days after injury.
injury reference standard.12,17,19 Problems with the refer- Both of these factors are likely to affect estimates of
ence standard for intracranial injury included variation in diagnostic accuracy. For intracranial injury, partial verifi-
methods, a lack of clarity about whether CT was per- cation bias and differential verification bias were generally
formed within 24 hours (rather than CT not being used at avoided but may be a problem in two large cohorts6,23 as
all),4,7–11,13–16,18 and use of telephone follow-up and/or participants received different reference standards depen-
review of hospital records where CT was not performed for dent on their initial presentation. Less than half the studies
all. It is unlikely that follow-up would accurately identify scored well for blinding, and studies were of mixed quality
all intracranial injuries and could affect estimates of sen- for clinical review bias, which may limit external validity
sitivity and specificity.25 Problems with the reference stan- of results.
Figure 2. Methodological quality graph: review authors’ judgements about each quality item presented as percentages across
all included studies.
Figure 3. Sensitivity and specificity of decision rules for identifying injuries that require neurosurgical intervention: rules with
multiple cohorts of data.
requiring neurosurgical intervention and any injury but variable intervention (94 –99%)20,32 and specificity for injuries requir-
specificity and limited validation. The Neurotraumatology Com- ing neurosurgical intervention (20 –50%) or any injury (21–
mittee of the World Federation of Neurosurgical Societies31 60%). Other rules have not been validated in sufficient
guidelines and Scottish Intercollegiate Guidelines Network cohorts and settings to draw meaningful conclusions.
guidelines both have sensitivities in a similar range to the CCHR
when lenient criteria are used, but results for specificity are
variable and generally much lower. The Scandinavian lenient DISCUSSION
criteria27 have diagnostic parameters in the same range but with There are many clinical decision rules for minor head
more variation in sensitivity for injuries requiring neurosurgical injury but few have been adequately validated. The CCHR
has been shown to be consistent across a number of studies scanning and missed intracranial injury, as reflected in the
and to have diagnostic parameters that would allow CT sensitivity and specificity estimates reported. The substantial
scanning to be avoided in a worthwhile proportion of patients benefit of neurosurgical interventions for certain lesions
without an excessive risk of significant intracranial injury means that specificity should be sacrificed to optimize sensi-
being missed. The NOC seems to have higher sensitivity for tivity for injuries requiring neurosurgical intervention. How-
intracranial injury but at the expense of specificity. Other ever, the benefit of identifying and treating injuries that do
rules may have comparable diagnostic performance with the not require surgical intervention is less certain, and the
CCHR, when applied appropriately but have not been as balance between unnecessary investigation and missed pa-
widely validated. thology is less clear. This uncertainty is reflected in variations
The CCHR has high sensitivity for detecting injuries in the definition of what constitutes a clinically significant
requiring neurosurgical intervention, whether high-risk or intracranial lesion; most studies considered any acute lesion
high-and medium-risk criteria are used. This is a consistent as a positive outcome. Consistently listed injuries include
finding in the available data, and so clinicians can be reason- extradural, subdural, subarachnoid, intracerebral, and intra-
ably assured that selecting patients for CT scanning on the parenchymal hemorrhage as well as cerebral contusion. Only
basis of the CCHR will carry a low risk of missing injuries two authors13,24 predefined their clinically significant findings
that require neurosurgical intervention. The sensitivity of the on CT (which have been adopted for other cohorts). Stiell et
CCHR medium-risk criteria for detecting any intracranial al.,24 more specifically, defines what was considered not
lesion is less consistent, although the lower reported sensi- significant, after extensive consultation with expert contrib-
tivity in some studies may reflect failure to detect injuries that utors, but disagreement still exists between these definitions.
are of little clinical significance. Clinicians using the CCHR Further research is required to validate existing rules
should be aware that it may miss some nonneurosurgical and compare the performance of different rules in the same
lesions of questionable clinical significance. population. Decision analysis and economic modeling is
Some limitations of the available data should be con- required to explore the trade-off between sensitivity and
sidered when using the CCHR in practice. Patients with specificity in terms of the costs and benefits of missed
coagulopathy, age ⬍16 years, pregnancy, seizure postinjury, pathology and unnecessary investigation. Research into the
focal neurologic deficit, or injuries considered minimal were prognosis of nonneurosurgical intracranial injuries and the
excluded from developmental work, and so the rule may not effect of identification and treatment on prognosis is required
be applicable to these patients. However, diagnostic accuracy to allow better definition of what constitutes a clinically
significant injury and what benefits can be derived from
was maintained in subsequent studies that included these
accurate identification. In addition, researchers should con-
patients (see CCHR high and medium risk adapted to cohort,
sider how best to minimize bias and sources of inaccuracy in
Figs. 3 and 4).19 Whenever rules have been directly compared
their study design. In the face of practical and ethical diffi-
in the same patient cohort, only marginal differences in
culties, researchers have to choose between a selected cohort
sensitivity have been identified, translating to little clinical
where all patients have CT (with the attendant risk of spec-
difference in injury detection. The primary advantage of the
trum bias) and a relatively unselected cohort that uses
CCHR over other decision rules is in its improved specificity, follow-up without CT in the reference standard (with the
leading to a reduction in the number of scans required to attendant risk of verification bias). Where follow-up is used,
identify the same number of injuries. careful consideration of adequacy should be made.
To the best of our knowledge, this study constitutes the The heterogeneity of these rules has highlighted the dif-
most extensive search of diagnostic test accuracy data relat- ficulties that exist with any research into minor head injury.
ing to minor head injury decision rules performed to date. Definitions of injury and outcomes vary while patient selection
However, as with all systematic reviews, our search strategy and reference standards are not uniform. To take minor head
may have missed some eligible cohorts and others may have injury research forward, agreed definitions and standard out-
been excluded on the basis of non-English language. There is come assessments must be adopted, along the lines of the
a high degree of heterogeneity in minor head injury defini- approach by Utstein37 in cardiopulmonary resuscitation re-
tion, patient spectrum, outcome definition (for both intracra- search. The call for unification of mild brain injury definitions
nial injuries and injuries requiring neurosurgical intervention) for the future development of patient management is not new.38
and reference standards used. These variations lead to signif- The current evidence base suggests that the CCHR has the
icant differences in prevalence, selection bias, and quality most consistent and acceptable sensitivity and specificity when
assessment and, as such, limited direct comparability of rules compared with other decision rules for adults with minor head
with each other, and precluded meta-analysis. However, the injury.
main purpose of data synthesis is to increase the precision of
estimates. The primary data included some large studies with REFERENCES
reasonably precise estimates of sensitivity and especially 1. Faul M, Xu L, Wald MM, Coronado VG. Traumatic Brain Injury in
specificity. The main limitation of the available primary data the United States: Emergency Department Visits, Hospitalizations
are thus likely to be inaccuracy rather than imprecision. and Deaths 2002–2006. Atlanta, GA: Centers for Disease Control
and Prevention, National Center for Injury Prevention and Control;
Policy makers and clinicians must decide on their 2010.
underlying priorities. We feel that decision making in minor 2. Whiting P, Rutjes AW, Reitsma JB, Bossuyt PM, Kleijnen J. The
head injury involves a trade-off between unnecessary CT development of QUADAS: a tool for the quality assessment of studies of
diagnostic accuracy included in systematic reviews. BMC Med Res 22. Stiell IG, Clement CM, Rowe BH, et al. Comparison of the Canadian CT
Methodol. 2003;3:25. head rule and the New Orleans criteria in patients with minor head
3. Arienta C, Caroli M, Balby S. Management of head-injured patients in injury. JAMA. 2005;294:1511–1518.
the emergency department: a practical protocol. Surg Neurol. 1997;48: 23. Stein SC, Fabbri A, Servadei F, Glick HA. A critical comparison of
213–219. clinical decision instruments for computed tomographic scanning in
4. Borczuk P. Predictors of intracranial injury in patients with mild head mild closed traumatic brain injury in adolescents and adults. Ann Emerg
trauma. Ann Emerg Med. 1995;25:731–736. Med. 2009;53:180 –188.
5. Duus BR, Lind B, Christensen H, Nielsen OA. The role of neuroimaging 24. Stiell IG, Wells GA, Vandemheen K, et al. The Canadian CT Head Rule
in the initial management of patients with minor head injury. Ann Emerg for patients with minor head injury. Lancet. 2001;357:1319 –1396.
Med. 1994;23:1279 –1283. 25. Whiting P, Rutjes AW, Reitsma JB, Glas AS, Bossuyt PM, Kleijnen J.
6. Fabbri A, Servadei F, Marchesini G, et al. Clinical performance of NICE Sources of variation and bias in studies of diagnostic accuracy: a
recommendations versus NCWFNS proposal in patients with mild head systematic review. Ann Intern Med. 2004;140:189 –202.
injury. J Neurotrauma. 2005;22:1419 –1427. 26. Teasdale G, Gentleman D, Andrews P, et al. Early Management of
7. Falimirski ME, Gonzalez R, Rodriguez A, Wilberger J. The need for Patients With Head Injury. Edinburgh, United Kingdom: Scottish Inter-
head computed tomography in patients sustaining loss of consciousness collegiate Guidelines Network; 2000.
after mild head injury. J Trauma. 2003;55:1– 6. 27. Ingebrigtsen T, Romner B, Kock-Jensen C. Scandinavian guidelines for
8. Haydel MJ, Preston CA, Mills TJ, Luber S, Blaudeau E, DeBlieux PM. initial management of minimal, mild, and moderate head injuries. The
Indications for computed tomography in patients with minor head injury. Scandinavian Neurotrauma Committee. J Trauma. 2000;48:760 –766.
N Engl J Med. 2000;343:100 –105. 28. Lapierre F. Guide-lines for head injured patients management in adult
9. Holmes JF, Baier ME, Derlet RW. Failure of the Miller criteria to predict age. Neurosurgical Society of France [in French]. Neurochirurgie. 1998;
significant intracranial injury in patients with a Glasgow Coma Score of 44:55–56.
14 after minor head trauma. Acad Emerg Med. 1997;4:788 –792. 29. National Institute for Health and Clinical Excellence. Head Injury:
10. Ibanez J, Arikan F, Pedraza S, et al. Reliability of clinical guidelines in Triage, Assessment, Investigation and Early Management of Head
the detection of patients at risk following mild head injury: results of a Injury in Infants, Children and Adults (NICE Clinical Guideline No. 4).
prospective study. J Neurosurg. 2004;100:825– 834. London, United Kingdom: National Institute for Health and Clinical
11. Madden C, Witzkc DB, Arthur, Sanders B, Valente J, Fritz M. High- Excellence; 2003.
yield selection criteria for cranial computed tomography after acute 30. National Institute for Health and Clinical Excellence. Head Injury:
trauma. Acad Emerg Med. 1995;2:248 –253. Triage, Assessment, Investigation and Early Management of Head
12. Miller EC, Holmes JF, Derlet RW. Utilizing clinical factors to reduce Injury in Infants, Children and Adults (NICE Clinical Guideline No. 56).
head CT scan ordering for minor head trauma patients. J Emerg Med. London, United Kingdom: National Institute for Health and Clinical
1997;15:453– 457. Excellence; 2007.
13. Mower WRH, Hoffman HJ, Herbert M, et al. Developing a decision 31. Servadei F, Teasdale G, Merry G; Neurotraumatology Committee of the
instrument to guide computed tomographic imaging of blunt head injury World Federation of Neurosurgical Societies. Defining acute mild head
patients. J Trauma. 2005;59:954 –959. injury in adults: a proposal based on prognostic factors, diagnosis, and
14. Ono K, Wada K, Takahara T, Shirotani T. Indications for computed management. J Neurotrauma. 2001;18:657– 664.
tomography in patients with mild head injury. Neurol Med Chir (Tokyo). 32. Stein SC. Management of minor closed head injury. Neurosurg Q.
2007;47:291–297. 1996;6:108.
15. Reinus WR, Wippold FJ II, Erickson KK, et al. Practical selection 33. Tomei G, Brambilla GL, Delfini R, et al. Guidelines for minor head injured
criteria for noncontrast cranial computed tomography in patients with patient’s management in adult age. The Study Group on Head Injury of the
head trauma. Ann Emerg Med. 1993;22:1148 –1155. Italian Society for Neurosurgery. J Neurosurg Sci. 1996;40:11–15.
16. Rosengren D, Rothwell S, Brown AF, Chu K. The application of North 34. Vos PE, Battistin L, Birbamer G, et al; European Federation of Neuro-
American CT scan criteria to an Australian population with minor head logical Societies. EFNS guideline on mild traumatic brain injury: report
injury. Emerg Med Australas. 2004;16:195–200. of an EFNS task force. Eur J Neurol. 2002;9:207–219.
17. Saadat S, Ghodsi SM, Naieni KH, et al. Prediction of intracranial 35. Murshid WR. Management of minor head injuries: admission criteria,
computed tomography findings in patients with minor head injury by radiological evaluation and treatment of complications. Acta Neurochir
using logistic regression. J Neurosurg. 2009;111:688 – 694. (Wien). 1998;140:56 – 64.
18. Saboori M, Ahmadi J, Farajzadegan Z. Indications for brain CT scan in 36. Pandor A, Goodacre S, Harnan S, et al. Diagnostic Management Strat-
patients with minor head injury. Clin Neurol Neurosurg. 2007;109:399 – egies for Adults and Children With Minor Head Injury: A Systematic
405. Review and an Economic Evaluation (Health Technology Assessment
19. Smits M, Dippel DW, de Haan GG, et al. External validation of the Reports). United Kingdom: National Institute for Health Research; 2011.
Canadian Head Rule and the New Orleans Criteria for CT scanning in 37. Recommended guidelines for uniform reporting of data from out-of-
patients with minor head injury. JAMA. 2005;294:1519 –1525. hospital cardiac arrest: the “Utstein style.” Prepared by a Task Force of
20. Smits M, Dippel DW, Steyerberg EW, et al. Predicting intracranial Representatives from the European Resuscitation Council, American
traumatic findings on computed tomography in patients with minor head Heart Association, Heart and Stroke Foundation of Canada, Australian
injury: the CHIP prediction rule. Ann Intern Med. 2007;146:397– 405. Resuscitation Council. Resuscitation. 1991;22:1–26.
21. Smits M, Dippel DW, de Haan GG, et al. Minor head injury: Guidelines 38. Jagoda AS, Bazarian JJ, Bruns JJ Jr, et al. Clinical policy: neuroimaging
for the use of CT—a multicenter validation study. Radiology. 2007;245: and decisionmaking in adult mild traumatic brain injury in the acute
831– 838. setting. J Emerg Nurs. 2009;35:e5– e40.