J Neurosurg / Volume 115 / September 2011

J Neurosurg 115:541–549, 2011

541

Large prospective trials have established and vali-dated guidelines for the clearance of cervical spine pathology in alert, cooperative patients.6,18,38 Under

these circumstances, injury may confidently be excluded by subjective symptom assessment and physical exami-nation of the patient. There is, however, no consensus on what equates to adequate evaluation of the obtunded or intubated patient unable to communicate deficit or dis-comfort.

Current standard practice in trauma care for clear-ing the cervical spine in obtunded or intubated patients involves a primary screening with CT and adjuvant as-sessment, most commonly with MR imaging, or delay-

ing clearance until a clinical examination can reliably be performed and NEXUS criteria applied.6,18 These prac-tices incur the prolonged use of cervical orthoses with a substantial risk of associated morbidities, including de-cubitus ulcers and skin breakdown, elevated intracranial pressure, deep venous thrombosis, aspiration pneumoni-tis, delay of surgical procedures, and pulmonary- or cath-eter-associated infections.1,10,32,37 Additionally, transport of severely injured trauma patients to the MR imaging suite is often risky, if not contraindicated. Thus, there is a need for expedited cervical clearance in this critically ill population.

Modern multislice helical CT has become the test of choice for primary screening of cervical spine injuries be-cause of its accuracy, cost-effectiveness, and efficiency.6 A growing body of evidence suggests that CT alone is suffi-cient to exclude cervical injury definitively. Nevertheless, standard practice continues to require adjuvant imaging,

Comparative effectiveness of using computed tomography alone to exclude cervical spine injuries in obtunded or intubated patients: meta-analysis of 14,327 patients with blunt trauma

A reviewDaviD M. Panczykowski, M.D., nestor D. toMycz, M.D., anD DaviD o. okonkwo, M.D., Ph.D.Department of Neurological Surgery, University of Pittsburgh Medical Center, Pittsburgh, Pennsylvania

Object. The current standard of practice for clearance of the cervical spine in obtunded patients suffering blunt trauma is to use CT and an adjuvant imaging modality (such as MR imaging). The objective of this study was to de-termine the comparative effectiveness of multislice helical CT alone to diagnose acute unstable cervical spine injury following blunt trauma.

Methods. The authors performed a meta-analysis of studies comparing modern CT with adjunctive imaging modalities and required that studies present acute traumatic findings as well as treatment for unstable injuries. Study quality, population characteristics, diagnostic protocols, and outcome data were extracted. Positive disease status included all injuries necessitating surgical or orthotic stabilization identified on imaging and/or clinical follow-up.

Results. Seventeen studies encompassing 14,327 patients met the inclusion criteria. Overall, the sensitivity and specificity for modern CT were both > 99.9% (95% CI 0.99–1.00 and 0.99–1.00, respectively). The negative likeli-hood ratio of an unstable cervical injury after a CT scan negative for acute injury was < 0.001 (95% CI 0.00–0.01), while the negative predictive value of a normal CT scan was 100% (95% CI 0.96–1.00). Global severity of injury, CT slice thickness, and study quality did not significantly affect accuracy estimates.

Conclusions. Modern CT alone is sufficient to detect unstable cervical spine injuries in trauma patients. Adju-vant imaging is unnecessary when the CT scan is negative for acute injury. Results of this meta-analysis strongly show that the cervical collar may be removed from obtunded or intubated trauma patients if a modern CT scan is negative for acute injury. (DOI: 10.3171/2011.4.JNS101672)

key worDs      •      cervical spine      •      blunt trauma      •      computed tomography      • meta-analysis

541

Abbreviations used in this paper: GCS = Glasgow Coma Scale; ISS = Injury Severity Score; NEXUS = National Emergency X-Radiography Utilization Study; QUADAS = Quality Assessment of Diagnostic Accuracy Studies.

See the corresponding editorial in this issue, pp 536–540.

D. Panczykowski, N. D. Tomycz, and D. O. Okonkwo

542 J Neurosurg / Volume 115 / September 2011

such as MR imaging or flexion-extension radiographs when clinically appropriate, to rule out unstable injury and permit removal of the cervical collar in obtunded or comatose trauma patients. It remains unclear what benefit is derived from further adjuvant imaging in the modern CT era, be it in the obtunded/comatose population or oth-erwise. Limitations inherent to adjuvant modalities pre-clude prospective multicenter validation trials of modern CT alone compared with plain radiography, dynamic flu-oroscopy, or MR imaging. Thus, the most effective means of establishing the accuracy of modern CT is to evaluate the rate of occult fracture or bone misalignment requiring subsequent surgical or orthotic stabilization.

Cervical spine clearance in the awake, cooperative patient is well established under NEXUS criteria. It was our hypothesis that, for obtunded or intubated patients, current literature provides support for the use of modern CT alone to exclude unstable cervical spine injuries. The purpose of this systematic review and meta-analysis was to determine the diagnostic performance of modern CT alone for detecting unstable cervical spine injuries in the population with blunt trauma.

MethodsWe performed a systematic review and meta-anal-

ysis in compliance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement;25 in addition, this study complies with the Meta-analysis Of Observational Studies in Epidemiology (MOOSE) group recommendations (Table 1) for the re-porting of meta-analyses of observational studies.25,39

Search StrategyAfter a priori development of our search strategy,

MEDLINE (January 1966–September 2009) and EM-BASE (January 1980–September 2009) were searched for relevant articles published in the English language. We used Medical Subject Headings (MeSH) terms and accompanying entry terms for the patient group (patients with suspected cervical spine injury), the diagnostic test (multislice helical CT), comparison tests (radiography and MR imaging), and outcome (unstable cervical spine injury necessitating orthotic or operative treatment; Table 2). The syntax used for the EMBASE search was as fol-lows: “spinal cord injury” or “blunt trauma” and “com-puted tomography” and/or “magnetic resonance imaging” and/or “radiography” and/or “fluoroscopy” and “english/lim.” We hand-searched the reference lists of all papers selected for full-text assessment and relevant reviews, as well as corresponded with investigators and experts in the field. Two independent reviewers (D.M.P. and N.D.T.) screened titles, abstracts, and full-text articles, while a third party (D.O.O.) arbitrated discrepancies. Studies that met inclusion criteria were those comparing modern CT with plain radiography, MR imaging, or dynamic fluoros-copy that included at least 30 patients who had suffered blunt trauma. Inclusion also required that modern CT slice thickness be < 3 mm and that studies present acute traumatic findings for all tested imaging modalities. Con-sensus on optimal slice thickness has not been reached; as

such the parameters used in this study represent the opti-mization of detection ability, low radiation exposure, and generalizability.6,27 Reviews and editorials were excluded from analysis.

Quality Assessment and Data ExtractionThe QUADAS tool was used to evaluate the method-

ological quality of the included studies, and was performed independently by 2 reviewers (D.M.P and N.D.T.).42 The representative patient spectrum was defined as those suf-fering blunt trauma undergoing radiographic cervical spine evaluation. The optimal image acquisition time interval between modern CT and the reference modality was defined as 72 hours.26 We also assessed the clarity of methodological descriptions and presence of blinding to both radiographic and clinical data during interpretation of all tested modalities.

Data extracted consisted of study population charac-teristics, methodological details of diagnostic protocols and imaging modalities, and outcome data, including injury and treatment descriptions. Positive disease status encompassed all unstable injuries identified by imag-ing modality or clinical follow-up. These were defined according to the system of White and Panjabi41 and the 3-column model of Denis.11 Although the 3-column the-ory was originally proposed for thoracolumbar injuries, cadaveric and retrospective clinical data have supported its applications in diagnosis and management of unstable cervical injuries.4,8,29 Cervical instability requires either a fracture or fractures involving contiguous columns or levels, bone misalignment (subluxations, listhesis, or splaying), or single-level ligamentous injury involving all 3 columns; as such necessitating orthotic or operative sta-bilization.

Statistical AnalysisDecisions on statistical method selection were made

a priori. Bivariate random-effects modeling of sensitiv-ity, specificity, positive predictive value, negative predic-tive value, and negative likelihood ratio were performed. Random-effects model estimates were obtained accord-ing to the DerSimonian-Laird method.12 Data synthesis and estimation were performed in a bivariate regression framework. Failure of this model occurs when a false positive (FPi) proportion is 0%, therefore adjusted false positive rates were obtained using the validated transfor-mation max(FPi, 1).12,21,26 Statistical heterogeneity was de-fined as an I2 value ≥ 50%. Summary receiver operating characteristic curves were not constructed because the majority of studies possessed sensitivities and specifici-ties approaching 100%.

Meta-regression was performed to assess the effect of covarying factors on between-study sensitivity and specificity heterogeneity. Such factors included modern CT slice thickness (0.75–2.5 mm), injury severity (GCS and ISS), and study quality. Quality was scored by assign-ing 0, 1, or 2 points for each QUADAS component de-pending on whether the criterion was not met, unclear, or clearly met, respectively. Publication bias was examined by construction of a funnel plot and statistically assessed

J Neurosurg / Volume 115 / September 2011

Use of CT alone to exclude cervical spine injuries

543

TABLE 1: Checklist of MOOSE criteria for reporting of meta-analyses of observational studies in epidemiology

MOOSE Criteria* How Criteria Were Handled in This Meta-Analysis

reporting of background should include  problem definition current standard of practice for clearance of the cervical spine in obtunded pa-

tients suffering blunt trauma is CT and an adjuvant imaging modality, but the requirement of an additional imaging modality beyond CT leads to prolonged cervical collar use and increased complications; the necessity of additional im- aging in the modern CT era has not been established

hypothesis statement current literature provides support for the use of modern CT alone to exclude un- stable cervical spine injuries in obtunded patients

description of study outcomes unstable cervical spine injury necessitating orthotic or operative treatment surgi-   cally identified on imaging and/or clinical follow-up

type of exposure or intervention used multislice helical CT type of study designs used all studies were of diagnostic cohort design, 7 of which were prospective study population adult patients suffering blunt trauma w/ suspected cervical spine injuryreporting of search strategy should include  qualifications of searchers credentials of the 3 investigators indicated on title page search strategy, including time period included in the synthesis and keywords

PubMed 1966–September 2009EMBASE 1980–September 2009 (see Table 2)

databases and registries searched PubMed and EMBASE search software used, name and version, including special features

no search software employed; EndNote was used to merge retrieved citations and eliminate duplications

use of hand-searching hand-searched reference sections of retrieved papers  list of citations located and those excluded, including justifica- tions

details of the literature search process are outlined in the flow chart (Fig. 1); cita- tion list available upon request

method of addressing articles published in languages other than English

restricted search to articles published in English; hand-searching did not reveal additional articles in other languages

method of handling abstracts and unpublished studies included only peer-reviewed, published articles  description of any contact w/ authors none to reportreporting of methods should include description of relevance or appropriateness of studies as- sembled for assessing the hypothesis to be tested

inclusion and exclusion criteria detailed in Methods section

rationale for selection and coding of data data extracted from each of the studies were relevant to the population charac- teristics, study design, imaging modalities employed, and outcome (see Meth- ods section)

assessment of confounding meta-regression was performed to assess effect of covarying factors on between-   study sensitivity and specificity heterogeneity; such factors included modern CT slice thickness, injury severity, and study quality

assessment of study quality, including blinding of quality asses-     sors; stratification or regression on possible predictors of study results

QUADAS tool was utilized to evaluate the methodological quality of the included studies (see Methods section)

assessment of heterogeneity heterogeneity of studies was explored by the I2 statistic; statistical heterogeneity   was defined as an I2 value ≥ 50%

  description of statistical methods in sufficient detail to be rep- licated

decisions on statistical method selection were made a priori; bivariate random-   effects modeling of sensitivity, specificity, positive predictive value, negative predictive value, and negative likelihood ratio were performed; random-effects model estimates were obtained according to the DerSimonian-Laird method; data synthesis and estimation were performed in a bivariate regression   framework; failure of this model occurs when a false positive (FPi) proportion is   0%, therefore adjusted FP rates were obtained using the validated transforma-   tion max(FPi, 1); statistical heterogeneity was defined as an I2 value ≥ 50%; meta-regression was performed to assess the effect of covarying factors on   between-study sensitivity and specificity heterogeneity; publication bias was examined by construction of a funnel plot and statistically assessed based on the Egger method;15,34 acceptable Type I error was set a priori at α = 0.05 for all statistical tests

(continued)

D. Panczykowski, N. D. Tomycz, and D. O. Okonkwo

544 J Neurosurg / Volume 115 / September 2011

based on the Egger method.15,34 Acceptable Type I error was set a priori at α = 0.05 for all statistical tests.

ResultsSearch Strategy and Study Selection

Our initial electronic database search yielded 861 ar-ticles (Fig. 1). A total of 816 articles were excluded based on review of article titles, abstracts, or both. Forty-five full-text publications were reviewed with 28 subsequently being excluded. The most common reasons for full-text exclusion were either failure to describe imaging parame-ters (12 articles) or modern CT slice thickness ≥ 3 mm (10 articles). The remaining studies were excluded either due to small sample size (4 articles) or failure to report patient

outcomes (1 article). Another study was excluded because the patient population was later reprocessed as a subset of a separate study from the same institution already in-cluded in our analysis.9 Agreement over study selection was excellent between reviewers (k = 0.95).

Study CharacteristicsSeventeen articles met the inclusion criteria stated

above. All studies were of diagnostic cohort design, and 7 were prospective.5,7,13,14,23,30,31 The median sample size was 381 patients (range 59–6558 patients) and overall 14,327 patients were evaluated.

Table 3 displays the diagnostic protocols used by each study. The median CT slice thickness was 2.0 mm (range 0.75–2.5 mm). Static, multislice helical CT was

TABLE 1: Checklist of MOOSE criteria for reporting of meta-analyses of observational studies in epidemiology (continued)

MOOSE Criteria* How Criteria Were Handled in This Meta-Analysis

provision of appropriate tables and graphics Table 1, search terms employed; Table 2, articles analyzed; Table 3, outcomes   explained; Fig. 1, study selection flow diagram; Fig. 2, nomogram demonstrat- ing negative predictive value and posttest probability

reporting of results should include graph summarizing individual study estimates and overall estimate

reported as text

table giving descriptive information for each study included Table 2 results of sensitivity testing reported as text  indication of statistical uncertainty of findings 95% CIs were presented w/ all summary estimates, I2 values, and results of sen-

sitivity analysesreporting of discussion should include quantitative assessment of bias statistical heterogeneity was present only for the sensitivity summary estimate, not

  specificity  justification for exclusion studies that meet inclusion criteria were those comparing modern CT w/ plain ra-

  diography, MR imaging, or dynamic fluoroscopy that included at least 30 patients who had suffered blunt trauma; inclusion also required that modern CT slice thickness be < 3 mm and that studies present acute traumatic findings for all tested imaging modalities; consensus on optimal slice thickness has not been reached, and as such the parameters used in this study represent the op- timization of detection ability, low radiation exposure, and generalizability

assessment of quality of included studies QUADAS tool results are discussed in Results and Discussion sectionsreporting of conclusions should include consideration of alternative explanations for observed results in 14,327 patients only 7 injuries were missed, although only 3 met criteria for be-

  ing classified as acutely unstable and all of which were identified on plain radi- ography; the superior accuracy of modern CT compared to plain radiography has already been validated, and as such these examples are most likely ex- plained by individual scan inadequacy or human error in radiological interpre- tation

generalization of the conclusions modern CT alone is sufficient to detect unstable cervical spine injuries in trauma patients; adjuvant imaging is unnecessary when the CT scan is negative for acute injury; results of this meta-analysis emphatically show that the cervical collar may be removed from obtunded or intubated trauma patients if a modern CT scan is negative for acute injury

guidelines for future research the nature of the study question precludes validation of diagnostic accuracy through randomized control trials because of modality limitations and ethical concerns

disclosure of funding source in Disclosure section

*  Based on data from the study by Stroup et al., 2000.

J Neurosurg / Volume 115 / September 2011

Use of CT alone to exclude cervical spine injuries

545

compared with plain radiographs in 4 studies,14,17,23,33 dy-namic fluoroscopy in 1,35 MR imaging in 10,22,36 and a combination of modalities in 2.5,30 Seven studies used more than 1 observer;13,14,16,19,30,33,35 however, only 1 pro-vided interobserver reliability for unstable cervical injury detection. Harris et al.16 reported an interrater reliability k statistic of 1.0 for both CT and MR imaging.

In the studies providing an index of concomitant brain and/or multisystem injuries, the presenting GCS score ranged from 5 to 13 (median 11) and the ISS ranged from 15 to 29 (median 24). Eleven studies contained samples consisting solely of patients who were either obtunded or who failed to provide a persistently reliable neurological examination. Motor vehicle accident was the predominant vector of injury (mean prevalence 56%).

The median study quality assessed using the QUA-DAS tool was 24 points (range 23–28 points). The major-ity of studies (82%) were either unclear about or altogeth-er lacked blinded interpretation of imaging modalities. Descriptions of the intervals between imaging were also unclear or absent in a substantial number of studies (29% unclear, 24% lacking).

Data Synthesis and AnalysisSummary estimates for sensitivity and specificity

were both > 99.9% (95% CI 0.99–1.00 and 0.99–1.00, re-spectively). The overall negative likelihood ratio of an un-stable cervical injury after a modern CT scan negative for acute injury was < 0.001 (95% CI 0.00–0.01), while the negative predictive value of a normal CT scan was 100% (95% CI 0.96–1.00). Statistical heterogeneity was present only for the sensitivity summary estimate, not specificity (I2 sensitivity 75%, I2 specificity 38%).

To assess possible sources of heterogeneity, meta-regression was performed using severity of injury, CT slice thickness, and study quality. The overall severity of injuries, as delineated by GCS or ISS, was not signifi-cantly associated with modern CT sensitivity or specific-

ity estimates (GCS, p = 0.98 and p = 0.87; ISS, p = 0.90 and p = 0.53, respectively). Likewise, CT slice thickness failed to show a statistically significant association with modern CT accuracy (sensitivity, p = 0.38; specificity, p = 0.52). Finally, regression also revealed that accuracy did not vary significantly as a function of study quality (sen-sitivity, p = 0.94; specificity, p = 0.80).

After plotting study size against study accuracy, the Egger test revealed symmetric funnel plot distribution, thus indicating that publication bias was unlikely to have influenced these results (p = 0.81).

DiscussionThe pooled results of 17 studies and 14,327 obtund-

ed or intubated trauma patients undergoing radiological evaluation of the cervical spine indicate that modern multislice helical CT alone is sufficient to rule out trau-matic, unstable cervical spine injuries. The overall nega-tive predictive value approached 100% for a modern CT scan without evidence of an acute injury. Construction of a Bayesian (Fagan) nomogram produced an estimated posttest probability approaching 0% of an unstable cervi-cal injury after negative modern CT (Fig. 2). This post-test probability is directly related to the prevalence of unstable cervical injuries across included studies (14%), which compared favorably to reported rates in blunt trau-ma (2%–6%) and obtunded patients (4%–34%).5,28 The results of this meta-analysis emphatically show that the cervical collar may be removed from obtunded or intu-bated trauma patients if a modern CT scan is negative for acute injury.

The primary criticism of multislice helical CT has been of its inability to accurately detect soft tissue and ligamentous injury. However, as yet there does not exist a validated pathological description of an acute unstable injury due to ligamentous disruption without bone or neurological correlates. Seven injuries treated by surgery or external fixation with normal modern CT scans were described in the included studies (Table 4). However, of these injuries, only 3 met the criteria for classification as acutely unstable, all of which were identified on plain ra-diographs, and 1 of these patients was quadriplegic on presentation.30 The superior accuracy of modern CT com-pared with plain radiography has already been validated,

TABLE 2: Details of search strategy

Category MEDLINE* EMBASE†

patient wounds, nonpenetrating spinal cord injury, blunt trauma

intervention tomography, x-ray computed tomography, x-ray

comparison magnetic resonance imag- ing, radiography

magnetic resonance imag-   ing, radiography, fluo- roscopy

outcome spinal injuries cervical spine injuriescombination patient AND (interven-

tion or comparison) AND outcome

patient AND (interven- tion or comparison) AND outcome

limitations English language English language

* Medical Subject Headings (MeSH) terms and accompanying entry terms searched from January 1966–September 2009. † Terms searched using “explode” function from January 1980–Sep-tember 2009.

Fig. 1.  Flow diagram of articles included in the meta-analysis. 

D. Panczykowski, N. D. Tomycz, and D. O. Okonkwo

546 J Neurosurg / Volume 115 / September 2011

and as such these examples are most likely explained by individual scan inadequacy or human error in radiologi-cal interpretation.6

Nonetheless, the incidence of unstable cervical spine injury among these studies of negative modern

CT approximates to 1 in every 4776 patients not able to be cleared by clinical examination. In a typical Level I trauma center in the US, this translates into 1 patient every 14 years.40 These rates, though, are far exceeded by the 6.8%–67% incidence of complications directly at-

TABLE 3: Characteristics of included studies*

Authors & YearNo. of 

Patients

Injury Mechanism

(% MVA) Inclusion Criteria Diagnostic Process

Nega-tive CT Scans

Unstable Injuries After Negative CT

Scan

Schenarts et al., 2001 1,286 89 blunt trauma; AMS CT & 5-view plain radiograph series in all meeting entry criteria

1,216 3

Brohi et al., 2005 381 NA blunt trauma; unconscious, intubated

lat plain radiograph, CT, MRI performed if CT or plain radiography suggested liga-   mentous injury/instability, neuro deficits, or   contradictory findings btwn CT & radiog- raphy

348 0

Diaz et al., 2005 1,577 72 blunt trauma; AMS CT, plain radiography if pain or distracting in-   juries, MRI if CT abnormality w/o fracture/   neuro deficit/pain/obtunded

1,299 0

Hogan et al., 2005 366 46 blunt trauma; obtunded or inadequate exam

3-view plain radiographs, CT, MRI in those   w/ prolonged (> 72 hrs) AMS or surgical candidates

366 0

Schuster et al., 2005 585 NA blunt trauma; GCS score < 9 & moving all limbs,   awake w/ deficits or pain 

CT, MRI if neuro deficits/pain/GCS score < 9 (at discretion of attending neurosurgeon)

500 0

Adams et al., 2006 86 45 blunt trauma; pain, neuro   deficits, obtunded

CT, MRI if neuro deficit, pain, obtunded 44 0

Spiteri et al., 2006 434 80 blunt trauma; unconscious   &/or intubated

2-view plain radiographs, CT, dynamic radio- graphs if anticipated unconscious for >   24 hrs

347 0

Como et al., 2007 115 41 blunt trauma; obtunded, negative CT

CT, MRI if persistently unevaluable 115 0

Harris et al., 2008 563 33 blunt trauma; AMS CT in those at risk or unconscious/noncoop- erative, upright lat & bilat oblique plain ra- diographs if persistently unevaluable

367 0

Menaker et al., 2008 203 49 blunt trauma; no deficit, persistently unreliable exam (GCS score < 14)

CT, MRI if persistently unevaluable 201 2

Steigelman et al., 2008 120 53 blunt trauma; no deficit, persistently unreliable exam (GCS score < 14)

CT in those at risk or unconscious/noncoop- erative, MRI if persistently unevaluable

120 0

Tomycz et al., 2008 180 35 blunt trauma; obtunded CT, MRI if could not be clinically cleared   because of neuro deficits, pain, distracting injuries, or depressed mental status

180 0

Mathen et al., 2007 667 49 blunt trauma; not meeting NEXUS low-risk guide- lines

CT, 3-view plain radiographs in all patients not meeting NEXUS low-risk criteria

640 0

Kihiczak et al., 2001 59 NA blunt trauma; obtunded CT, MRI if neuro deficit, pain, obtunded 58 2Hashem et al., 2009 141 59 blunt trauma all meeting entry criteria had CT & 3-view

plain radiograph series20 0

Sekula et al., 2008 6,558 NA blunt trauma all meeting entry criteria had CT & lat plain radiograph

6,099 0

Diaz et al., 2003 1,006 74 blunt trauma; AMS, or dis- tracting injury

CT, plain radiography in hemodynamically   stable blunt trauma w/ AMS or distracting injuries

834 0

total 14,327 12,754 7

*  AMS = altered mental status; MVA = motor vehicle accident; NA = not available; neuro = neurological.

J Neurosurg / Volume 115 / September 2011

Use of CT alone to exclude cervical spine injuries

547

tributable to cervical immobilization.1,10,32,37 Thus, while 1 patient every 14 years might be missed by a CT-only protocol, between 325 and 3200 patients would sustain a complication from prolonged cervical collar use during the same time frame. Although the personal and financial costs of delayed deterioration due to missed diagnosis are theoretically high, it is unclear what, if any, sequelae would have resulted from the missed injuries described above. In this series of studies, not 1 of the 12,754 patients with negative modern CT scans experienced delayed de-cline in neurological function.

The diagnostic accuracy of modern CT for detect-ing unstable injuries calculated here is comparable to that published for MR imaging.26 However, MR imaging is associated with a substantial false-positive rate (25%–40%), thus leading to unwarranted treatments and proce-dures, each with their inherent risks.3,6,20,24 Magnetic reso-nance imaging also exposes the patient to both the risks of transport and inaccessibility while in the scanner itself. Modern CT is superior to MR imaging in the detection of osseous pathology and equivalent to MR imaging in the detection of unstable injury; thus, performing MR imag-ing after a negative CT scan is redundant and exposes patients to unnecessary risks.

The current report makes recommendations germane to the obtunded or comatose trauma patient. However, in light of the superior diagnostic accuracy of modern CT for unstable cervical injuries demonstrated, these results might warrant extrapolation to the blunt trauma popula-tion at large (alert, cooperative patients). However, the literature comprising this systematic review cannot sub-stantiate that conclusion on its own.

A limitation of the search strategy applied in our sys-tematic review is that only articles published in the Eng-lish language were searched and used for meta-analysis, therefore studies may have been missed. However, hand-searching the references of 45 relevant articles did not dis-close additional foreign-language publications, therefore

TABLE 4: Injuries missed by multislice helical CT in 14,327 patients studied*

Authors & YearMultislice Helical CT

Specifications Description Clinical Evaluation TreatmentMeets Criteria for Instability†

Schenarts et al., 2001 2-mm cuts atlantooccipital dislocation quadriplegia halo yesC2–3 subluxation no deficits, non-

tendercollar yes

C3–4 subluxation unexaminable collar yesMenaker et al., 2008 16-slice, 2-mm cuts w/ 1-mm

overlapC1–2 PLL rupture, LF rupture unexaminable surgical stabili-

zationno

C4–5 spinal cord contusion, C2–5 prevertebral edema

unexaminable surgical stabili- zation

no

Kihiczak et al., 2001 2.5-mm slice thickness; 2.5-mm cuts; reconstruction to   1.25-mm thick, 2.5-mm sagit-   tal/coronals

abnormal MR signal of C5–6 disc, possible hyperextension injury

C5–7 interspinous ligamentous injury

tenderness

tenderness

collar

surgical stabili- zation

no

no

*  LF = ligamentum flavum; PLL = posterior longitudinal ligament.†  Fracture of contiguous columns or levels, bone misalignment, or ligamentous injury involving all 3 columns.

Fig. 2. Nomogram for CT evaluation of the cervical spine using literature-based estimates of injury incidence (4%–34%) and likelihood ratios to predict posttest probability of unstable injury after negative modern CT.5,28

D. Panczykowski, N. D. Tomycz, and D. O. Okonkwo

548 J Neurosurg / Volume 115 / September 2011

a significant impact on the results of this study would not be expected. Unfortunately, only 2 of the included articles specifically described the incidence of false-positives on modern CT. The transformation method employed to overcome this has been previously validated and, in fact, underestimates the true sensitivity.12,21,26 Other limitations include the proportion of retrospective studies analyzed (59%) and that all 17 articles were cohort trials. These limitations are, however, insurmountable given the sub-stantial differences between imaging modalities that make it both impractical and unethical to conduct ran-domized control trials comparing CT to adjuvant imag-ing techniques. Additionally, the use of treatment as an outcome measure may underestimate the negative predic-tive value of modern CT scans depending on the treat-ing physician’s or surgeon’s adherence to the validated and most widely accepted definition of what constitutes an unstable injury.11,41 Despite this underestimation, the negative predictive value of a modern CT scan without evidence of acute injury still exceeded 99.9%.

Overall, our results support the contention that mod-ern CT is able to rule out acute, unstable cervical injury following blunt trauma in obtunded or intubated patients. Adjuvant imaging is therefore unnecessary for cervical clearance following a negative modern CT scan in pa-tients unable to be cleared by NEXUS criteria. These re-sults also provide categorical evidence that the cervical collar may be safely and expeditiously removed after a negative modern CT scan, thus minimizing the risk of complications in this critically ill population.

Disclosure

This work was supported by a grant from the Doris Duke Chari-table Foundation to the University of Pittsburgh to fund Clinical Research Fellow David Panczykowski.

Author contributions to the study and manuscript prepara-tion include the following. Conception and design: all authors. Ac quisition of data: Panczykowski, Tomycz. Analysis and interpre-tation of data: all authors. Drafting the article: all authors. Critically revising the article: all authors. Statistical analysis: Panczykowski, Tomycz. Study supervision: all authors.

References

1. Ackland HM, Cooper DJ, Malham GM, Stuckey SL: Magnetic resonance imaging for clearing the cervical spine in uncon-scious intensive care trauma patients. J Trauma 60:668–673, 2006

2. Adams JM, Cockburn MI, Difazio LT, Garcia FA, Siegel BK, Bilaniuk JW: Spinal clearance in the difficult trauma patient: a role for screening MRI of the spine. Am Surg 72:101–105, 2006

3. Anonymous: Radiographic assessment of the cervical spine in symptomatic trauma patients. Neurosurgery 50 (3 Suppl): S36–S43, 2002

4. Bozkus H, Ames CP, Chamberlain RH, Nottmeier EW, Sonntag VK, Papadopoulos SM, et al: Biomechanical analy-sis of rigid stabilization techniques for three-column injury in the lower cervical spine. Spine (Phila Pa 1976) 30:915–922, 2005

5. Brohi K, Healy M, Fotheringham T, Chan O, Aylwin C, Whit-ley S, et al: Helical computed tomographic scanning for the evaluation of the cervical spine in the unconscious, intubated trauma patient. J Trauma 58:897–901, 2005

6. Como JJ, Diaz JJ, Dunham CM, Chiu WC, Duane TM, Capel-la JM, et al: Practice management guidelines for identification of cervical spine injuries following trauma: update from the eastern association for the surgery of trauma practice man-agement guidelines committee. J Trauma 67:651–659, 2009

7. Como JJ, Thompson MA, Anderson JS, Shah RR, Claridge JA, Yowler CJ, et al: Is magnetic resonance imaging essential in clearing the cervical spine in obtunded patients with blunt trauma? J Trauma 63:544–549, 2007

8. Cybulski GR, Douglas RA, Meyer PR Jr, Rovin RA: Com-plications in three-column cervical spine injuries requiring anterior-posterior stabilization. Spine  (Phila  Pa  1976)  17: 253–256, 1992

9. Daffner RH, Sciulli RL, Rodriguez A, Protetch J: Imaging for evaluation of suspected cervical spine trauma: a 2-year analy-sis. Injury 37:652–658, 2006

10. Davis JW, Parks SN, Detlefs CL, Williams GG, Williams JL, Smith RW: Clearing the cervical spine in obtunded patients: the use of dynamic fluoroscopy. J Trauma 39:435–438, 1995

11. Denis F: The three column spine and its significance in the classification of acute thoracolumbar spinal injuries. Spine (Phila Pa 1976) 8:817–831, 1983

12. DerSimonian R, Laird N: Meta-analysis in clinical trials. Control Clin Trials 7:177–188, 1986

13. Diaz JJ Jr, Aulino JM, Collier B, Roman C, May AK, Miller RS, et al: The early work-up for isolated ligamentous injury of the cervical spine: does computed tomography scan have a role? J Trauma 59:897–904, 2005

14. Diaz JJ Jr, Gillman C, Morris JA Jr, May AK, Carrillo YM, Guy J: Are five-view plain films of the cervical spine unreli-able? A prospective evaluation in blunt trauma patients with altered mental status. J Trauma 55:658–664, 2003

15. Egger M, Davey Smith G, Schneider M, Minder C: Bias in meta-analysis detected by a simple, graphical test. BMJ 315: 629–634, 1997

16. Harris TJ, Blackmore CC, Mirza SK, Jurkovich GJ: Clear-ing the cervical spine in obtunded patients. Spine (Phila Pa 1976) 33:1547–1553, 2008

17. Hashem R, Evans CC, Farrokhyar F, Kahnamoui K: Plain ra-diography does not add any clinically significant advantage to multidetector row computed tomography in diagnosing cer-vical spine injuries in blunt trauma patients. J Trauma 66: 423–428, 2009

18. Hoffman JR, Mower WR, Wolfson AB, Todd KH, Zucker MI: Validity of a set of clinical criteria to rule out injury to the cervical spine in patients with blunt trauma. N Engl J Med 343:94–99, 2000

19. Hogan GJ, Mirvis SE, Shanmuganathan K, Scalea TM: Exclu-sion of unstable cervical spine injury in obtunded patients with blunt trauma: is MR imaging needed when multi-detector row CT findings are normal? Radiology 237:106–113, 2005

20. Holmes JF, Mirvis SE, Panacek EA, Hoffman JR, Mower WR, Velmahos GC: Variability in computed tomography and magnetic resonance imaging in patients with cervical spine injuries. J Trauma 53:524–530, 2002

21. Ihaka R, Gentleman R: R: A language for data analysis and graphics. J Comput Graph Statist 5:299–314, 1996

22. Kihiczak D, Novelline RA, Lawrason JN, Ptak T, Rhea JT, Sacknoff R: Should an MR scan be performed routinely after a normal clearance CT scan in the trauma patient? Experience with 59 cases. Emerg Radiol 8:276–278, 2001

23. Mathen R, Inaba K, Munera F, Teixeira PG, Rivas L, McKen-ney M, et al: Prospective evaluation of multislice computed tomography versus plain radiographic cervical spine clear-ance in trauma patients. J Trauma 62:1427–1431, 2007

24. Menaker J, Philp A, Boswell S, Scalea TM: Computed tomog-raphy alone for cervical spine clearance in the unreliable pa-tient—are we there yet? J Trauma 64:898–904, 2008

25. Moher D, Liberati A, Tetzlaff J, Altman DG; PRISMA Group: Preferred reporting items for systematic reviews and meta-

J Neurosurg / Volume 115 / September 2011

Use of CT alone to exclude cervical spine injuries

549

analyses: the PRISMA statement. Ann Intern Med 151:264–269, W64, 2009

26. Muchow RD, Resnick DK, Abdel MP, Munoz A, Anderson PA: Magnetic resonance imaging (MRI) in the clearance of the cervical spine in blunt trauma: a meta-analysis. J Trauma 64:179–189, 2008

27. Phal PM, Riccelli LP, Wang P, Nesbit GM, Anderson JC: Frac-ture detection in the cervical spine with multidetector CT: 1-mm versus 3-mm axial images. AJNR Am J Neuroradiol 29:1446–1449, 2008

28. Richards PJ: Cervical spine clearance: a review. Injury 36: 248–270, 2005

29. Samartzis D, Wein SM, Shen FH, Beazell J, Francke EI, An-derson DG: A revisitation of distractive-extension injuries of the subaxial cervical spine: a cadaveric and radiographic soft tissue analysis. Spine (Phila Pa 1976) 35:395–402, 2010

30. Schenarts PJ, Diaz J, Kaiser C, Carrillo Y, Eddy V, Morris JA Jr: Prospective comparison of admission computed to-mographic scan and plain films of the upper cervical spine in trauma patients with altered mental status. J Trauma 51: 663–669, 2001

31. Schuster R, Waxman K, Sanchez B, Becerra S, Chung R, Conner S, et al: Magnetic resonance imaging is not needed to clear cervical spines in blunt trauma patients with normal computed tomographic results and no motor deficits. Arch Surg 140:762–766, 2005

32. Sees DW, Rodriguez Cruz LR, Flaherty SF, Ciceri DP: The use of bedside fluoroscopy to evaluate the cervical spine in obtunded trauma patients. J Trauma 45:768–771, 1998

33. Sekula RF Jr, Daffner RH, Quigley MR, Rodriguez A, Wil-berger JE, Oh MY, et al: Exclusion of cervical spine instabil-ity in patients with blunt trauma with normal multidetector CT (MDCT) and radiography. Br J Neurosurg 22:669–674, 2008

34. Song F, Khan KS, Dinnes J, Sutton AJ: Asymmetric funnel plots and publication bias in meta-analyses of diagnostic ac-curacy. Int J Epidemiol 31:88–95, 2002

35. Spiteri V, Kotnis R, Singh P, Elzein R, Madhu R, Brooks A, et al: Cervical dynamic screening in spinal clearance: now redundant. J Trauma 61:1171–1177, 2006

36. Steigelman M, Lopez P, Dent D, Myers J, Corneille M, Stew-art R, et al: Screening cervical spine MRI after normal cervi-cal spine CT scans in patients in whom cervical spine injury cannot be excluded by physical examination. Am J Surg 196: 857–863, 2008

37. Stelfox HT, Velmahos GC, Gettings E, Bigatello LM, Schmidt U: Computed tomography for early and safe discontinuation of cervical spine immobilization in obtunded multiply injured patients. J Trauma 63:630–636, 2007

38. Stiell IG, Clement CM, McKnight RD, Brison R, Schull MJ, Rowe BH, et al: The Canadian C-spine rule versus the NEX-US low-risk criteria in patients with trauma. N Engl J Med 349:2510–2518, 2003

39. Stroup DF, Berlin JA, Morton SC, Olkin I, Williamson GD, Rennie D, et al: Meta-analysis of observational studies in epi-demiology: a proposal for reporting. Meta-analysis Of Obser-vational Studies in Epidemiology (MOOSE) group. JAMA 283:2008–2012, 2000

40. Tomycz ND, Chew BG, Chang YF, Darby JM, Gunn SR, Nicholas DH, et al: MRI is unnecessary to clear the cervical spine in obtunded/comatose trauma patients: the four-year ex-perience of a level I trauma center. J Trauma 64:1258–1263, 2008

41. White AA III, Panjabi MM: Clinical Biomechanics  of  the Spine,  ed  2. Philadelphia: Lippincott Williams & Wilkins, 1990

42. Whiting P, Rutjes AW, Reitsma JB, Bossuyt PM, Kleijnen J: The development of QUADAS: a tool for the quality assess-ment of studies of diagnostic accuracy included in systematic reviews. BMC Med Res Methodol 3:25, 2003

Manuscript submitted September 29, 2010.Accepted April 13, 2011.Please include this information when citing this paper: published

online May 27, 2011; DOI: 10.3171/2011.4.JNS101672.Address correspondence to: David O. Okonkwo, M.D., Ph.D.,

Department of Neurological Surgery, University of Pittsburgh Medi-cal Center, 200 Lothrop Street, Suite B-400, Pittsburgh, Pennsylva-nia 15213. email: okonkwodo@upmc.edu.

J Neurosurg 115:536–540, 2011

536 J Neurosurg / Volume 115 / September 2011

Cervical clearance

Vincent c. traynelis, M.D., anD Manish K. Kasliwal, M.D., M.ch.

Department of Neurosurgery, Rush University Medical Center, Chi cago, Illinois

A cervical spine clearance after trauma is defined as accurately confirming the absence of a significant cervi-cal spine injury. Although such injuries occur in only 2%–6% of blunt trauma victims, the assessment and clearance of the cervical spine is a priority in the management of these patients due to the potential for significant neuro-logical injury. The National Emergency X-Radiography Utilization Study (NEXUS) criteria have been accepted as a standard for cervical spine clearance in the awake, cooperative patient with 99.9% sensitivity for detection of significant cervical injuries.12

Definitive assessment of the cervical spine in ob-tunded trauma patients is more difficult. Multidetector CT has become the imaging modality of choice for evalu-ation of cervical spine trauma, but the need for adjunc-tive studies following a negative CT scan in cognitively impaired patients, due to concern that an important soft tissue injury may be missed, has been a topic of debate. Usually important ligamentous or discal injuries are as-sociated with neurological deficits, but there are excep-tions. An acute rupture of the transverse atlantal ligament produces significant instability despite no neurological deficits and normal alignment while the patient is supine in a CT scanner. The obtunded patient may have a mild to moderate central cord syndrome that may be difficult to discern in the setting of a concomitant supratentorial injury. The management options for the obtunded trauma patient with a negative CT scan include collar removal without additional imaging, collar removal after negative dynamic fluoroscopy, collar removal after negative MR imaging, and rigid orthotic immobilization until the pa-tient is competent and reliable enough to obtain a nega-tive clinical examination of his or her cervical spine.

The efficacy of dynamic fluoroscopy is limited due to the need for repeat examinations, the difficulty in iden-tifying specific ligamentous injuries, and inadequate vi-sualization of the lower cervical spine.2,6,21 Even in awake and cooperative blunt trauma patients the probability of obtaining adequate flexion-extension radiographs is just over 30%.14 The limitations are so significant that the 2009 Eastern Association for the Surgery of Trauma (EAST) guidelines recommended that dynamic fluoros-copy should no longer be an option for cervical spine clearance in the obtunded patient.4

Immobilization in a rigid cervical collar in this pa-tient population for more than 48–72 hrs has been associ-ated with an increased incidence of pressure sores, in-tracranial hypertension (in the setting of a simultaneous head injury), airway management challenges, compro-mised central venous access, infection due to suboptimal venous catheter site care, and difficult oral hygiene.13,19,24

Although important, the morbidity associated with these potential orthosis-related complications is almost never as significant as a permanent cervical spinal cord injury. Nonetheless, it is unreasonable to immobilize the cervi-cal spine indefinitely while waiting for cognitive function to improve.

The limitations of dynamic fluoroscopy and pro-longed immobilization have led to a number of studies attempting to assess the utility of MR imaging of cog-nitively dysfunctional trauma patients with a negative cervical CT scan. Magnetic resonance imaging is expen-sive, requires patient transport, has a relatively long study time, and is so sensitive that it can detect subtle nonstabi-lizing injuries. Muchow et al.20 first used the technique of meta-analysis to determine the efficacy of MR imaging in the management of these patients, and in the follow-ing communication, Panczykowski et al.22 present their meta-analysis of the literature comparing CT technology with plain radiography, dynamic fluoroscopy, or MR im-aging in obtunded blunt trauma patients. Panczykowski et al. report that 17 studies totaling 14,327 patients met their study inclusion criteria and overall sensitivity and specificity for CT were both > 99.9% (95% CI 0.99–1.00 and 0.99–1.00, respectively). Interestingly, the authors report that global severity of injury, CT slice thickness, and study quality did not significantly affect accuracy es-timates.

Although CT is an excellent imaging modality, the data as presented in this study do not support the authors’ emphatic declaration that a cervical CT scan alone is suf-ficient to detect unstable cervical spine injuries and that a cervical collar may be removed from obtunded or intu-bated trauma patients without need of any MR imaging, provided a radiology report shows the CT scan is negative for acute injury.

It is important to carefully examine the data the authors chose to include in their meta-analysis. Five of the studies compared CT to plain radiography and con-cluded that CT was more sensitive in terms of detecting injury.3,8,10,11,16 The data provided by these studies do not address the adequacy of CT to detect all significant trau-matic lesions but merely support its role as superior to the most basic imaging technique. Furthermore, Diaz et al.8

concluded that CT is not an effective modality for screen-ing for ligamentous injury and MR imaging is a better modality for this purpose. None of these works address the need or lack thereof for additional testing following

Editorial

See the corresponding article in this issue, pp 541–549.

J Neurosurg / Volume 115 / September 2011

Editorial

537

a normal CT scan in obtunded trauma patients. Schen-arts et al.25 performed CT scans of just the upper cervical spine (occiput to C-3) and MR images were only obtained in patients with neck tenderness. Despite these major lim-itations, 3 patients were found to have significant injuries despite a normal CT. Schenarts et al. only concluded that CT was superior to plain radiography for the detection of upper cervical spinal injuries. Spiteri et al.27 used dynam-ic radiographs to evaluate patients following CT. Despite the previously mentioned problems with this technique, 2 of 434 patients were found to have instability despite the original CT scan reading of normal. One patient had at-lantooccipital dislocation and the other atlantoaxial liga-mentous instability. It is difficult to understand why the authors failed to count these injuries in Table 3.

The most appropriate data come from those reports detailing the use of both CT and MR imaging. Almost all of these studies are compromised by either their ret-rospective design, failure to obtain an MR image in all patients, or lack of follow-up. Como et al.5 performed the only prospective study and also obtained MR imaging in all patients. Adams et al.1 reported that of 85 normal CT scans, 1 patient had extensive ligamentous injuries requir-ing treatment with “an extended course of cervical immo-bilization.” Only 29 patients in this study were obtunded and it is not clear whether the false negative CT scan was among that group. Steigelman et al.29 found that 5% of obtunded trauma patients had findings consistent with acute injuries on MR imaging that were not observed on CT. Two of these patients required prolonged immobili-zation with an orthosis. Kihiczak et al.15 only treated 19 obtunded patients and, similar to the work of Adams et al., this paper met the entrance criteria as listed by the authors but fell well short of the mark if one considers only obtunded patients. In the obtunded subgroup 10% of the patients with normal CT scans had abnormal MR images, and while these patients did not undergo surgery, it is not known whether they were treated with immo-bilization. Tomycz et al.30 reported that acute traumatic findings were detected by MR imaging in 21% of 180 ob-tunded patients with normal CT scans. Although none of these patients required acute surgery for instability, all with ligamentous injury per MR imaging were treated with a rigid cervical orthosis. Menaker et al.17 reported that 8.9% of obtunded trauma patients with a negative CT scan had an abnormal MR image. Two of these patients required surgery and 14 were treated with extended im-mobilization, leading these investigators to conclude that MR imaging was still a necessary part of the evaluation of this patient population.

In an article not included in this meta-analysis, Stas-sen et al.28 reviewed all obtunded adult trauma patients treated at their institution over a 1-year period (52 total) who were evaluated with both a CT scan (CTI Helical scanner, GE Medical Systems) and MR imaging and found that 30% of those with a negative CT scan (slice thickness 3 × 1.5 mm helical, 1:1 pitch from the skull base to the top of T-1 with sagittal reconstructions) had an MR image that demonstrated a ligamentous injury requiring treatment. The authors of the current meta-analysis re-ported in a prior publication30 that they detected acute

ligamentous injuries in 8.9% of obtunded trauma patients with a normal CT scan and within the discussion in that paper they attributed the discrepancy between their find-ings and those of Stassen et al. to a difference in the tim-ing of the MR images following trauma.

The determination of cervical instability has long been a topic of interest. While it is easy to agree on what constitutes absolute stability or instability, the gradations between these two extremes are more difficult to catego-rize. Denis7 proposed a strategy to determine instability for thoracolumbar fractures primarily using plain radio-graphs and reserving CT for only selected injuries. Three structural columns were defined and patients with injury to 2 or all of the columns were considered to have an un-stable condition. While the 3-column concept has been applied to the cervical region it has not been validated for this area of the spine. White and Panjabi32,33 intro-duced a more sophisticated approach to cervical instabil-ity. They defined the anterior structures as the posterior longitudinal ligament and everything anterior to it. The components dorsal to this ligament were labeled posterior structures. These investigators concluded from a care-fully performed series of experiments that if all of the ligamentous structures of one group were lost plus one additional supporting structure, then instability occurred. From this foundation these spine biomechanical pioneers proposed a means of determining clinical instability that was radically different because it included neurological status, the presence or absence of stenosis, and the activ-ity level of the individual. The importance of ligamentous loss and neurological status continue to be emphasized in the more recently formulated Subaxial Cervical Injury Classification System (SLICS).23,31

Panczykowski et al.22 chose to reassess the data with-in the publications comprising this meta-analysis and decide whether injuries missed by CT and detected by MR imaging were stable or not. The data provided in the original works are insufficient to make such judgments. Although the authors state that the criteria of White and Panjabi33 were used to make these determinations, none of the papers cited in their meta-analysis provide any information regarding the exact structures that were in-jured, clinical examination, canal stenosis, and others. It also appears that the authors consider the lack of need for surgery as synonymous with stability. This is not always true and immobilization is an important treatment mo-dality. For example, most spine surgeons would consider a hangman’s fracture as an unstable injury, yet many of these patients can be managed successfully with a rigid cervical orthosis.9

A final concern about the adequacy of the studies performed to date relates to patient follow-up, which is almost universally restricted to the acute hospitalization in the papers of this meta-analysis. Certainly the number of patients who present in a delayed fashion with a missed injury is small but it is not zero. Over the past 20 years the senior author of this editorial (V.C.T.) has cared for 2 such patients and these individuals may not choose to return to the original treating institution.

Two important works have been published since the authors submitted their manuscript. Menaker et al.18 eval-

Editorial

538 J Neurosurg / Volume 115 / September 2011

uated 213 blunt trauma patients with a negative 40-slice CT sequence of the cervical spine. Overall, 24% of the study population had abnormal MR images and of these, 15 required surgery, 23 required extended rigid immo-bilization, and in 14 no treatment was recommended. Noteworthy is the subset of patients with an unreliable examination (96 total), 15 of whom had an unrecognized injury detected by MR imaging. Error in the interpreta-tion of the CT could account for some missed injuries that occurred in at least 1 of the studies incorporated in this meta-analysis.27 Simon et al.26 reported that spine surgeons more accurately evaluated cervical CT scans than radiologists identifying missed fractures and find-ings suspicious for instability.

Computed tomography is an excellent imaging mo-dality for trauma and it is adequate to detect most cervical spine injuries, but the report of a “normal” CT scan should not be blindly accepted as sufficient to discontinue exter-nal immobilization and halt the pursuit of other diagnos-tic studies or treatments in the obtunded trauma patient. Evaluation of these patients by a spine surgeon represents the true standard. Such a physician can clinically exam-ine the patient, appraise the mechanism or mechanisms of injury, and review the imaging studies personally while thoughtfully considering the official radiologist’s report. In many patients such an assessment will be enough to discontinue immobilization after a normal CT scan but in some further testing may be advised. Protocols are useful for optimizing care and resource utilization, but in the end all that matters is the patient. Cervical spine clear-ance of obtunded blunt injury patients by well-meaning trauma surgeons or intensivists who rely on a nonclini-cian’s report of an imaging study is a recipe for rare but real error. The most appropriate protocol for the patient with a potential spine injury is to consult a spine surgeon.

Disclosure

Dr. Traynelis serves as a consultant to Medtronic and United-Healthcare.

References

1. Adams JM, Cockburn MI, Difazio LT, Garcia FA, Siegel BK, Bilaniuk JW: Spinal clearance in the difficult trauma patient: a role for screening MRI of the spine. Am Surg 72:101–105, 2006

2. Bolinger B, Shartz M, Marion D: Bedside fluoroscopic flex-ion and extension cervical spine radiographs for clearance of the cervical spine in comatose trauma patients. J Trauma 56: 132–136, 2004

3. Brohi K, Healy M, Fotheringham T, Chan O, Aylwin C, Whit-ley S, et al: Helical computed tomographic scanning for the evaluation of the cervical spine in the unconscious, in tubated trauma patient. J Trauma 58:897–901, 2005

4. Como JJ, Diaz JJ, Dunham CM, Chiu WC, Duane TM, Capel-la JM, et al: Practice management guidelines for identification of cervical spine injuries following trauma: update from the eastern association for the surgery of trauma practice man-agement guidelines committee. J Trauma 67:651–659, 2009

5. Como JJ, Thompson MA, Anderson JS, Shah RR, Claridge JA, Yowler CJ, et al: Is magnetic resonance imaging essential in clearing the cervical spine in obtunded patients with blunt trauma? J Trauma 63:544–549, 2007

6. Davis JW, Kaups KL, Cunningham MA, Parks SN, Nowak TP, Bilello JF, et al: Routine evaluation of the cervical spine in head-injured patients with dynamic fluoroscopy: a reap-praisal. J Trauma 50:1044–1047, 2001

7. Denis F: Spinal instability as defined by the three-column spine concept in acute spinal trauma. Clin Orthop Relat Res (189):65–76, 1984

8. Diaz JJ Jr, Aulino JM, Collier B, Roman C, May AK, Miller RS, et al: The early work-up for isolated ligamentous injury of the cervical spine: does computed tomography scan have a role? J Trauma 59:897–904, 2005

9. Hadley MN, Walters BC, Grabb PA, Oyesiku NM, Przybyl-ski GJ, Resnick DK, et al: Guidelines for the management of acute cervical spine and spinal cord injuries. Clin Neurosurg 49:407–498, 2002

10. Harris TJ, Blackmore CC, Mirza SK, Jurkovich GJ: Clear-ing the cervical spine in obtunded patients. Spine (Phila Pa 1976) 33:1547–1553, 2008

11. Hashem R, Evans CC, Farrokhyar F, Kahnamoui K: Plain ra-diography does not add any clinically significant advantage to multidetector row computed tomography in diagnosing cer-vical spine injuries in blunt trauma patients. J Trauma 66: 423–428, 2009

12. Hoffman JR, Mower WR, Wolfson AB, Todd KH, Zucker MI: Validity of a set of clinical criteria to rule out injury to the cervical spine in patients with blunt trauma. N Engl J Med 343:94–99, 2000

13. Hunt K, Hallworth S, Smith M: The effects of rigid collar placement on intracranial and cerebral perfusion pressures. Anaesthesia 56:511–513, 2001

14. Khan SN, Erickson G, Sena MJ, Gupta MC: Use of flexion and extension radiographs of the cervical spine to rule out acute instability in patients with negative computed tomography scans. J Orthop Trauma 25:51–56, 2011

15. Kihiczak D, Novelline RA, Lawrason JN, Ptak T, Rhea JT, Sacknoff R: Should an MR scan be performed routinely after a normal clearance CT scan in the trauma patient? Experience with 59 cases. Emerg Radiol 8:276–278, 2001

16. Mathen R, Inaba K, Munera F, Teixeira PG, Rivas L, McKen-ney M, et al: Prospective evaluation of multislice computed tomography versus plain radiographic cervical spine clear-ance in trauma patients. J Trauma 62:1427–1431, 2007

17. Menaker J, Philp A, Boswell S, Scalea TM: Computed tomog-raphy alone for cervical spine clearance in the unreliable pa-tient—are we there yet? J Trauma 64:898–904, 2008

18. Menaker J, Stein DM, Philp AS, Scalea TM: 40-slice multide-tector CT: is MRI still necessary for cervical spine clearance after blunt trauma? Am Surg 76:157–163, 2010

19. Morris CG, McCoy EP, Lavery GG: Spinal immobilisation for unconscious patients with multiple injuries. BMJ 329:495–499, 2004 (Errata in BMJ 329:673, 2004; BMJ 329:773, 2004)

20. Muchow RD, Resnick DK, Abdel MP, Munoz A, Anderson PA: Magnetic resonance imaging (MRI) in the clearance of the cervical spine in blunt trauma: a meta-analysis. J Trauma 64:179–189, 2008

21. Padayachee L, Cooper DJ, Irons S, Ackland HM, Thomson K, Rosenfeld J, et al: Cervical spine clearance in unconscious traumatic brain injury patients: dynamic flexion-extension fluoroscopy versus computed tomography with three-dimen-sional reconstruction. J Trauma 60:341–345, 2006

22. Panczykowski D, Tomycz ND, Okonkwo DO: Comparative effectiveness of using computed tomography alone to exclude cervical spine injuries in obtunded or intubated patients: a meta-analysis of 14,327 patients with blunt trauma. A re-view. J Neurosurg [epub ahead of print May 27, 2011. DOI: 10.3171/2011.4.JNS101672]

23. Patel AA, Hurlbert RJ, Bono CM, Bessey JT, Yang N, Vaccaro AR: Classification and surgical decision making in acute sub-

J Neurosurg / Volume 115 / September 2011

Editorial

539

axial cervical spine trauma. Spine (Phila Pa 1976) 35:S228–S234, 2010

24. Richards PJ: Cervical spine clearance: a review. Injury 36: 248–270, 2005

25. Schenarts PJ, Diaz J, Kaiser C, Carrillo Y, Eddy V, Morris JA Jr: Prospective comparison of admission computed to-mographic scan and plain films of the upper cervical spine in trauma patients with altered mental status. J Trauma 51: 663–669, 2001

26. Simon JB, Schoenfeld AJ, Katz JN, Kamath A, Wood K, Bono CM, et al: Are “normal” multidetector computed tomographic scans sufficient to allow collar removal in the trauma patient? J Trauma 68:103–108, 2010 (Erratum in J Trauma 69:242, 2010)

27. Spiteri V, Kotnis R, Singh P, Elzein R, Madhu R, Brooks A, et al: Cervical dynamic screening in spinal clearance: now redundant. J Trauma 61:1171–1177, 2006

28. Stassen NA, Williams VA, Gestring ML, Cheng JD, Bankey PE: Magnetic resonance imaging in combination with helical computed tomography provides a safe and efficient method of cervical spine clearance in the obtunded trauma patient. J Trauma 60:171–177, 2006

29. Steigelman M, Lopez P, Dent D, Myers J, Corneille M, Stew-art R, et al: Screening cervical spine MRI after normal cer-vical spine CT scans in patients in whom cervical spine in-jury cannot be excluded by physical examination. Am J Surg 196:857–863, 2008

30. Tomycz ND, Chew BG, Chang YF, Darby JM, Gunn SR, Nicholas DH, et al: MRI is unnecessary to clear the cervical spine in obtunded/comatose trauma patients: the four-year ex-perience of a level I trauma center. J Trauma 64:1258–1263, 2008

31. Vaccaro AR, Hulbert RJ, Patel AA, Fisher C, Dvorak M, Lehman RA Jr, et al: The subaxial cervical spine injury classi-fication system: a novel approach to recognize the importance of morphology, neurology, and integrity of the disco-ligamen-tous complex. Spine (Phila Pa 1976) 32:2365–2374, 2007

32. White AA III, Johnson RM, Panjabi MM, Southwick WO: Biomechanical analysis of clinical stability in the cervical spine. Clin Orthop Relat Res 109:85–96, 1975

33. White AA III, Punjabi MM: Clinical Biomechanics of the Spine. Philadelphia: Lippincott, 1978

Response

DaViD M. PanczyKowsKi, M.D., nestor D. toMycz, M.D., anD DaViD o. oKonKwo, M.D., Ph.D.

Department of Neurological Surgery, University of Pittsburgh Medical Center, Pittsburgh, Pennsylvania

We greatly appreciate the thoughtful and diligent re-view of our article provided by Drs. Traynelis and Kasli-wal. We welcome the opportunity their editorial provides to further explicate and emphasize important elements of our study.

Allow us to begin with a very simple observation: the use of multislice helical CT alone to clear the cervical spine is already standard practice in trauma medicine in several centers throughout Europe, Australia, New Zea-land, and Canada.2,5 An algorithm for clearing the cer-vical spine after a negative CT scan in unconscious pa-tients appears as far back as 2004 in the British Medical Journal.4 It is principally the medicolegally plagued US

that has not already adopted this approach. Nevertheless, many US centers, including our own, remove the cervi-cal collar when the CT scan is negative. Yet, we are not aware of a single case of neurological deficit engendered by removing a cervical collar from a patient with a nor-mal multislice CT scan.

Perhaps much of the current debate with cervical clearance in the obtunded population has arisen from the fact that too much of the focus has been on the abstract concept of stability and not on the more clinically rel-evant question of preserving neurological function. Even if, theoretically, there is instability by some definition that is missed by multislice CT, there is no evidence that this instability will harm the patient’s neurological function. On the contrary, there is ample evidence that collars are associated with significant morbidity as the reviewers have cited. Bottom line, despite numerous centers nation-ally and internationally removing cervical collars after a negative CT scan, there is no evidence—not even an iso-lated case report—that this practice significantly threat-ens neurological function.

The preferred method of assessing the accuracy of a diagnostic test is, of course, prospective analysis against a criterion standard: a modality with superior negative pre-dictive value. However, significant inherent limitations of MR imaging (risk of transport, cost, and a substantial false positive rate) prevent direct comparisons with mul-tislice CT in determining the true diagnostic accuracy. The most practical solution to establishing the validity of modern CT alone as a screening tool for acute cervical instability is to do what we reported in our article: in-vestigate the rate of either subsequent surgical or orthotic stabilization. As opposed to “blind” use of the radiology reports, this method incorporates the clinical decision-making of both the radiologist and the treating surgeon.

Detailed review of individual study results, as pro-vided by both the 2009 EAST guidelines as well as the current reviewers, has to date failed at providing defini-tive recommendations for the need or timing of MR im-aging following a negative CT scan in the obtunded popu-lation.1 To remedy this, our study harnesses the power of meta-analyses to combine systematically the results of relevant studies and examine the effect of study hetero-geneity (design, population, and outcome) to arrive at our conclusion. To eliminate possible bias and further bolster this approach, we adhered to strict article inclusion cri-teria requiring pathophysiological reporting of all false negative CT scans.

Few tests in medicine have the sensitivity and nega-tive predictive value as high as multislice CT for cervical spine injury. As Drs. Traynelis and Kasliwal point out, even the NEXUS criteria are not 100% sensitive and are, in fact, significantly less sensitive and less specific than a negative CT scan, yet the NEXUS criteria have been almost universally adopted.3 If the magnitude of accu-racy of a CT scan as a diagnostic test fails to convince the neurosurgical community to alter practice patterns with regards to cervical spine clearance, then evidence-based medicine suffers a significant defeat. We believe that the pervasive fear of missing cervical spine injuries in obtunded patents is a relic of the pre-CT age in which

Editorial

540 J Neurosurg / Volume 115 / September 2011

bone fractures and dislocations missed by radiographs led to delayed neurological deficits. Anecdotal reports of delayed catastrophic paralysis after reportedly negative cervical radiographs must be tempered by the reality that modern CT with reconstructions has been shown to be vastly superior to radiography.

We recognize that it is difficult to take the side of less aggressive imaging within an alarmingly defensive and litigious medical environment in which more test-ing creates the illusion of better medical care. Although the complications of collar immobilization appear minor compared to the possibility of neurological deficit from a missed cervical injury, when summed across thousands of patients these “minor” morbidities are not negligible. As is frequently the case in medicine, less may be more.

We grant that Drs. Traynelis and Kasliwal have un-derscored an important point within our study. As they adeptly state, a radiology report of a “normal” CT (or MR image for that matter) cannot be blindly accepted, especially by those lacking dedicated spine or neurosur-gical training. Although evidence-based practice may incorporate powerful meta-analytic techniques such as we applied, the clinical decision-making of the evaluat-ing spine surgeon ultimately requires careful assessment of the actual CT scans integrated with the patient’s indi-vidual risk of acute cervical instability. It is under those circumstances that the spine surgeon may express to the family, fellow treating physicians, and themselves with

99.99% certainty after a negative modern CT scan that the cervical collar can come off.

References

1. Como JJ, Diaz JJ, Dunham CM, Chiu WC, Duane TM, Capel-la JM, et al: Practice management guidelines for identification of cervical spine injuries following trauma: update from the eastern association for the surgery of trauma practice man-agement guidelines committee. J Trauma 67:651–659, 2009

2. Hennessy D, Widder S, Zygun D, Hurlbert RJ, Burrowes P, Kortbeek JB: Cervical spine clearance in obtunded blunt trau-ma patients: a prospective study. J Trauma 68:576–582, 2010

3. Hoffman JR, Mower WR, Wolfson AB, Todd KH, Zucker MI: Validity of a set of clinical criteria to rule out injury to the cervical spine in patients with blunt trauma. National Emer-gency X-Radiography Utilization Study Group. N Engl J Med 343:94–99, 2000

4. Morris CG, McCoy EP, Lavery GG: Spinal immobilisation for unconscious patients with multiple injuries. BMJ 329:495–499, 2004 (Errata in BMJ 329:673, 2004; BMJ 329:773, 2004)

5. Padayachee L, Cooper DJ, Irons S, Ackland HM, Thomson K, Rosenfeld J, et al: Cervical spine clearance in unconscious traumatic brain injury patients: dynamic flexion-extension fluoroscopy versus computed tomography with three-dimen-sional reconstruction. J Trauma 60:341–345, 2006

Please include this information when citing this paper: published online May 27, 2011; DOI: 10.3171/2011.2.JNS102121.