Secondary Logo

Journal Logo

Classification of Lower Cervical Spine Injuries

Moore, Timothy A., MD*; Vaccaro, Alexander R., MD; Anderson, Paul A., MD

doi: 10.1097/01.brs.0000217942.93428.f7
Cervical Spine

Study Design. Blinded assessment by multiple observers of consecutive case series.

Objectives. Measure the reliability of a new system of determining stability in subaxial cervical spine injuries.

Summary of Background Data. Classification is fundamental to allow communication, determine prognosis, and direct treatment. Current systems have many limitations, including difficultly of use, lack of proven reliability and validity, and no assessment of stability. A new system to assess instability is proposed.

Methods. A literature review of the most commonly described classification systems is reported. The Cervical Spine Injury Severity Score was tested for reliability by 10 examiners who graded 35 consecutive cases of cervical trauma. Plain radiographs and CT were saved as read using Efilm Lite in random order. Each was scored and intraobserver and interobserver agreement was measured using intraclass correlation coefficients (ICC).

Results. Intraobserver agreement was excellent with ICC ranging from 0.97 to 0.99. Interobserver agreement was also excellent with mean 0.80 ranging from 0.75 to 0.98.

Conclusion. A new cervical spine classification system of injury is paramount to treatment and outcomes. A new system may increase reliability and therefore allow more accurate determination of stability and dictate treatment.

Cervical spine injury classification is paramount to management. Current systems are not validated nor in wide agreement. The new cervical spine injury severity score has excellent reliability.

From the *MetroHealth Medical Center, Department of Orthopaedic Surgery, Cleveland, OH; †Departments of Orthopedic Surgery and Rehabilitation and Neurologic Surgery, University of Wisconsin, Madison, WI; and ‡Department of Neurosurgery and Orthopaedic Surgery; Thomas Jefferson University, Rothman Institute, and Delaware Valley Regional Spinal Cord Injury Center, Philadelphia, PA.

The manuscript submitted does not contain information about medical device(s)/drug(s).

No funds were received in support of this work. No benefits in any form have been or will be received from a commercial party related directly or indirectly to the subject of this manuscript.

Address correspondence and reprint requests to Paul A. Anderson, MD, University of Wisconsin, 600 Highland Ave., K4/726, Madison, WI 53593.

Cervical spine injuries are common occurring in up to 3% of all trauma patients. These injuries are often associated with neurologic deficits ranging from radiculopathy to incomplete or to complete spinal cord injuries. Before the advances in spinal instrumentation, these injuries were treated with traction and external bracing, most often without decompression of the neural elements. Advancements in fixation techniques and our understanding of the pathophysiologic basis of spinal cord injury have led surgeons to treat these injuries surgically with the aims of decompression of the neural elements and maintenance of spinal stability for long-term function. The degree of instability imparted on the spine by the injury can predict prognosis and treatment options. A method to describe instability using a well-designed classification scheme is lacking.

The purpose of a classification system is multifactorial. The system should be descriptive to allow communication, designate gradation based on the severity of injury, provide insight and prognostication to a clinical scenario, and help direct treatment decisions. No single cervical spine classification scheme has been widely accepted. Mirza et al presented six major expectations of an ideal classification system for thoracolumbar fractures, including identification and terminology, injury and treatment, characterization, neurologic factors, grading, and prognosis.1 It is thought that no current system met all expectations. It is our opinion that to fill all of these criteria, the cervical spine classification system would need to be divided into two components: a morphologic description and a quantification of stability.

Through modern imaging methods, our ability to detect cervical spinal injuries continues to improve. Despite the advancements in detecting these injuries, we lack an accepted and reproducible classification scheme for subaxial cervical injuries. We hypothesize that using a quantifiable system would help to determine whether a patient should have surgery, nonoperative treatment, or may be treated by either method. The purpose of this manuscript is to review the new classification system and report its validation reliability.

Back to Top | Article Outline


New Classification Proposal.

The classification system is based on morphologic descriptions and, secondly, by stability based on a quantifiable value.

Back to Top | Article Outline
Morphologic Description of Subaxial Cervical Injuries.

A morphologic system similar to that described by Bohlman is used to describe fracture types (Table 1).2 These are categorized into relatively broad groups as the secondary quantitative scale will allow further division into clinically relevant subdivisions. The following principles were used when choosing this morphologic based system: identification of the primary injury location, use of common nomenclature, and avoidance of mechanistic terminology.

Table 1

Table 1

Isolated bony and ligamentous injuries are described based on location. Common examples are fractures of spinous processes, lamina, transverse processes, lateral masses, facets, compression fractures, and posterior ligamentous complex. Complex injuries involve more than one column or bone and ligament together. These are described based on their most obvious or commonly recognized location, that is: anterior; lateral mass pillar; and posterior. Additionally, there are several special cases that are described separately that do not fit easily into these groupings.

Anterior Column Injuries. Isolated anterior column injuries include compression fractures, disc distraction injuries without subluxation, and traumatic disc herniations. Complex injuries are distraction injuries with posterior subluxation with or without an avulsion fracture, burst fracture with retropulsion of bone, and the flexion axial loading injury or tear drop injury as described by Schneider et al (Figure 1).3 In this latter injury, there is a shearing in the coronal plane through the vertebral body with posterior rotation of the vertebral body into the spinal canal (Figure 3). There may be a significant posterior ligamentous injury or bilateral lamina fractures.

Figure 1

Figure 1

Figure 3

Figure 3

Lateral Column Injuries. Isolated lateral column injuries include superior or inferior facet fractures without subluxation and pedicle fractures (Figure 4). Complex injuries are fracture separation of the lateral mass, unilateral facet dislocations with or without fracture, and bilateral facet dislocations with or without fractures. A fracture separation of the lateral mass occurs when a free-floating lateral mass is created with a pedicle fracture and a fracture at the junction of the lamina and lateral mass (Figure 5).4 This allows rotation of the lateral mass and loss of rotational and anterior translational stability. Often, patients will have one or even two levels of mild (0%–25%) subluxation.

Figure 4

Figure 4

Figure 5

Figure 5

Unilateral facet dislocations result in 0% to 25% subluxation and may be pure dislocations without fractures or more commonly associated with a facet fracture.

Bilateral facet dislocations result in 25% to 50% or even greater subluxation, and in some cases there may be significant distraction across the intervertebral disc space (Figure 1). These may be associated with facet fractures and often bilateral lamina or spinous process fractures. In all cases, the disc anulus is significantly disrupted and there may be associated disc herniation.

Posterior Column Injuries. Isolated posterior column injuries include spinous process fractures, lamina fractures, and posterior ligamentous injuries. Complex injuries of the posterior column include posterior ligamentous injuries with or without fracture of the spinous processes or lamina. As noted by Nicoll5 and Holdsworth,6 one of the major factors critical to stability is the posterior ligamentous complex; therefore, this structure is injured in many of the other injury types described above. We believe that the use of other common names, such as bilateral facet dislocation, more reliably describes the injury and would be used by the majority of physicians even though a significant posterior ligamentous injury is present.

Special Cases. Several patterns of injuries are complex and do not fit well into the previously described sections or are associated with preexisting disease. Special cases include the bilateral pedicle fracture with traumatic spondylolisthesis, fractures in the ankylosed spine, and spinal cord injury without radiographic abnormality.

Back to Top | Article Outline
Quantification of Stability: Cervical Spine Injury Severity Score.

The Cervical Spine Injury Severity Score was developed to measure stability after cervical spine trauma. The system is based on bony and ligamentous disruptions and does not include neurologic function. Neurologic function has been adequately addressed with standardized measurements published by the American Spinal Injury Association.7

The Cervical Spine Injury Severity Score applies to all areas of the subaxial spine from the caudal aspect of C2 to T1. It is easy to learn, reliable, and applies to all fracture types. It is a continuous variable so that discriminate statistical analysis can be performed. The score correlates to increasing instability and, we hypothesize, will relate to treatment decisions and ultimate prognosis. The score is based on skeletal injury and does not take into account neurologic function or deficits. Preexisting diseases such as ankylosis or congenital stenosis can modify the score based on decisions made by the observer.

Four-Column Model. The injury severity score is based on evaluation of the four columns of the cervical spine independently using a standard visual analog scale. The four columns are modification of the three pillars described by Louis.8 In his model, the spine is structurally supported by three legs: the anterior column consisting of the disc, body, and ligaments; and by the paired lateral pillars, including the lateral masses with facet articulations and capsules. In addition, the posterior osteoligamentous complex is added creating the fourth column (Figure 2A).

Figure 2

Figure 2

The four columns are the anterior, right pillar, left pillar, and posterior osseous ligamentous complex. The anterior column includes the body, intervertebral disc and anulus, and anterior-posterior longitudinal ligaments. The pillars include the pedicles starting at their junction with the vertebral body, the superior and inferior facets, the lateral mass, and facet capsules. Each pillar is scored separately. The posterior column includes the lamina, spinous processes, nuchal ligaments (supraspinous, infraspinous, the ligament of nuchae), and ligamentum flavum.

Analog Score. A visual analog score from 0 to 5 is applied to each column and summed. Thus, the injury severity score ranges from 0 to 20, with 0 being no injury and 20 the most severe. Fractional scores may be used.

The analog score is based on the degree of bony displacement and ligamentous disruption (Figure 2B). Standard computed tomography sagittal and axial reconstructions are used to evaluate the injuries. For example, a score of 1 is given for nondisplaced fractures, whereas for complete ligamentous disruption or displacement of greater than 5 mm, a score of 5 is given. A general concept is that add score of 5 is given to the most severe injury that can occur to that particular column. In patients with multiple levels of injury, only the most severe level is scored. Figure 2B outlines generally the analog scores based on degree of bony and ligamentous displacement. These are only general guidelines as each observer is tasked to rate the severity of injury based on his own criteria. Although this may increase variations among individuals, it has the advantage of allowing other factors not usually included in systems, such as results of ancillary studies, to be factored in as new information becomes available.

To test reliability, 35 consecutive cases of cervical spine injuries were scored by 10 reviewers. The reviewers’ clinical experience ranged from second year orthopedic surgery residents to an attending spine surgeon with 20 years of clinical experience. These reviewers were from different academic trauma centers. The study was performed with IRB approval. Additionally, to examine intraobserver reliability, 5 cases were repeated.

Plain radiographs and CT were downloaded to CDs as DICCOM images. Efilm lite software was used to open and scan images. Enhancement tools included brightness/contrast, zoom, measuring rulers, and synchronization. CT data included scout views and axial, coronal, and sagittal reconstructions. CTs were obtained in 1.5-mm slice thickness and included the entire cervical spine from the occiput to T1.

The cases were redacted of patient-identifying information and given a case number. The cases were scored in random order by each examiner without any clinical information. Interclass correlation coefficients (ICCs) were measured to determine both intra and interobserver reliability. A score of ≤0.4 was poor, 0.4 to 0.75 fair or moderate, and >0.75 excellent.9 A power analysis was done before initiation of the study. Using 10 observers and 35 cases was predicted to give greater than 85% power of showing reliability if present.10 All analyses were performed with SAS (SAS Institute, Inc., Cary, NC).

Back to Top | Article Outline


There was a broad range of stability in the case series with mean 8.2 and standard deviation of 6.6. Thus, the cases were distributed from very stable to highly unstable. Examples of the use of the system are given in Figures 1, 3, 4, and 5.

The interobserver ICCs ranged from 0.75 to 0.98 and averaged 0.88. Intraobserver reliability for the five duplicate cases ranged from 0.97 to 0.99. These data show excellent reliability in using the quantification injury severity score. Table 2 gives scores of observers and primary author for cases examples.

Table 2

Table 2

No differences in reliability were observed based on fracture morphology, between columns, or by experience of the observer.

Back to Top | Article Outline


Despite a long history and numerous approaches, no classification systems for subaxial cervical spine injuries have ever been universally accepted or clinically validated. All such systems are based on one of three factors: mechanism of injury, morphologic features, or degree of instability. In some cases, combinations of these factors are used.

Mechanistic classification systems attempt to predict the primary injury vector from patterns seen on radiographs or computed tomography scans.11 These injury vectors can be simulated in cadaveric models to allow confirmation and assist in injury prevention research. Many of the mechanisms accepted in describing injury mechanisms have entered the accepted nomenclature describing injuries. In some cases, the mechanistic terms imply a specific morphologic pattern, e.g., compression fracture, flexion-distraction injury.

Unfortunately, mechanistic systems may not always accurately describe injury patterns or the actual mechanism of injury. From cadaveric studies, it has been shown that application of the same injury vector can produce different fracture patterns. Shono et al created cervical spine injuries in cadavers using impulses applied to the skull vertex.12 Injuries created from the same magnitude and direction of forces included bursting fractures of the vertebral body and bilateral facet dislocations. Other limitations of mechanistic systems include the lack of knowledge of head position at the time of impact, the influence of musculature, and the effect of disease states such as age, ankylosis, and osteoporosis on injury patterns. Furthermore, the spine under acute loads has been shown to buckle, thereby producing variable injury vectors at different levels.13,14 For example, a flexion moment may be present at one level with an extension moment with rotation at a lower level (Figure 1). Once the spine is rendered unstable, its position can be in almost any location, making identification of the mechanism difficult. Mechanistic classification systems do not account for secondary vectors that act on the injured spine.

Morphologic classification systems describe what is observed on plain radiographs, CT, and in some cases MRI. Common terms such as burst fracture, avulsion fracture, and fracture dislocation are well accepted and are used to communicate injury patterns among surgeons. Too often, morphologic terms tend to describe mechanism of injuries but do not characterize instability. Other morphologic descriptors imply a mechanism but are accepted as a pathoanatomic entity. Terms such as hyperflexion, extension, and compression describe presumed direction of forces but also give a morphologic image of a specific injury pattern. Thus, when attempting to use pure morphologic systems, mechanistic terms are difficult to avoid.

The concept of stability was initially described in 1949 by Nicoll who reported 166 patients with thoracolumbar injuries.5 He identified radiographic factors that prevented return to work in Welsh injured miners. The factors that were associated with an “unstable” spine were disruption of the posterior osseous ligamentous complex and complete dislocations. Thus, Nicoll characterized stability and its prognosis on outcome; he concluded that an injured miner would function according to three characteristics: pain, mobility, and power and endurance.5

Holdsworth furthered Nicoll’s concepts in 1970.6 He specifically described the five types of “violence” or mechanisms that act on the spine during injury: pure flexion, flexion and rotation, extension, vertical compression, and direct shearing force. He was the first to describe the importance of the posterior osteoligamentous complex in terms of stability of the motion segment.

White and Panjabi have provided the most widely accepted definition of stability as “ability of the spine under physiologic loads to maintain an association between vertebral segments in such a way that there is neither damage nor subsequent irritation of the spinal cord or nerve roots, and, in addition, there is no development of incapacitating deformity or pain due to structural changes.”15 Using this definition and based on classic biomechanical experiments, they created a checklist of parameters that, when scored, produced an index of stability (Table 1). A score of 5 or greater results in clinical instability. This is an effective checklist to evaluate cervical injuries but has never been validated clinically and has poor reproducibility among observers.

Important characteristics of useful classification systems are that it is clinically relevant, easy to use, applicable in a variety of situations, teachable, reliable, and valid. The system must have the ability to be used readily in multiple settings. When dealing with traumatic cervical spine injuries, the most important setting is the initial patient contact. Once an injury is identified, the system must be applicable to determine treatment direction. The system should have relevance in other environments such as during various periods of healing and in the research laboratory.

The currently proposed system of morphologic descriptions is based on commonly used and conceptually recognized patterns. It does not divide into many subgroups such as Allen-Ferguson or the AO system since the quantification scale can provide the distinctive patterns that differentiate injuries even with similar patterns to one another.11,16 The proposed morphologic description is currently undergoing testing for reliability.

A classification system must be reliable and valid. Reliability is the measure of variability among observers and by repeated observations. Most commonly, this is measured by interobserver and intraobserver kappa values or ICC.9,17 These statistical values range from −1 to 1, with acceptable values being greater that 0.6. Validity is the accuracy of the system to measure the condition under assessment. The Cervical Spine Injury Severity Score had excellent reliability with intraobservor ICC >0.97 and interobservor ICC >0.88. Furthermore, the results were equal between experienced and inexperienced observers.

A limitation of this study is that MRI was not used. This could affect results as occult ligamentous injuries may be better appreciated and therefore scored appropriately. Aversely minor injuries of questionable significance may be identified which could inappropriately increase the quantifiable score. MRI was purposefully not used in this investigation as they were only available in the more severe cases. Further investigations testing the reliability using MRI are ongoing. Another important component of stability and treatment decisions is neurologic function. This was not included in the system as the authors believed that it can be scored separately using an accepted systems such as recommended by the American Spinal Injury Association or by the method described by Vaccaro et al.18,19 Further studies evaluating the reliabilty of the morphologic descriptions and validation of the quantification system are planned.

Back to Top | Article Outline


A name or description allows communication, distinguishes injuries from one another, may imply or directly state important biomechanical forces, and allow categorization into phylogenies. However, the name or description may not relate to prognosis, guide treatment, assess severity of skeletal injury, or predict natural history. Quantification of the severity of instability will overcome many of these descriptive limitations. Such a severity score will incorporate both bony and ligamentous pathologic changes and can be adapted to a variety of pathoanatomic conditions, provide numerical values so that discriminate statistical methods can be used, and hopefully direct treatment decisions. The results demonstrate excellent reliability of the injury severity score system to quantify stability in subaxial cervical spinal injuries. Quantifying stability based on fracture morphology along with accurate assessment of the neurologic injury will allow surgeons to better characterize these injuries and ultimately produce treatment algorithms that can be tested in clinical trials.

Back to Top | Article Outline

Key Points

  • Classification of subaxial cervical spine injuries is essential to allow communication, determine prognosis, and direct treatment.
  • Current systems based on mechanism and morphology are limited and do not reliably measure degree of instability.
  • The development of common morphologic names may aid in the management of these injuries.
  • The cervical spine injury severity score based on summations of analog scores for each of four spinal columns is reliable and easy to use. The development of common morphologic names may aid in the management of these injuries.
Back to Top | Article Outline


1. Mirza SK, Mirza AJ, Chapman JR, et al. Classifications of thoracic and lumbar fractures: rationale and supporting data. J Am Acad Orthop Surg 2002;10:364–77.
2. Bohlman HH. Acute fractures and dislocations of the cervical spine: an analysis of three hundred hospitalized patients and review of the literature. J Bone Joint Surg Am 1979;61:1119–42.
3. Schneider RC, Crosby EC, Russo RH, et al. Traumatic spinal cord syndromes and their management. Clin Neurosurg 1973;20:424–92.
4. Levine AM, Mazel C, Roy-Camille R. Management of fracture separations of the articular mass using posterior cervical plating. Spine 1992;17(suppl):447–54.
5. Nicoll EA. Fractures of the dorsolumbar spine. J Bone Joint Surg Br 1949;31:376–94.
6. Holdsworth F. Fractures, dislocations, and fracture-dislocations of the spine. J Bone Joint Surg Am 1970;52:1534–51.
7. American Spinal Injury Association. International Standards for Neurological and Functional Classification of Spinal Cord Injury. Chicago: American Spinal Injury Association, 1992.
8. Louis R. Spinal stability as defined by the three-column spine concept. Anat Clin 1985;7:33–42.
9. Shrout PE, Fleiss JL. Intraclass correlations: uses in assessing rater reliability. Psychol Bull 1979;86:420–8.
10. Bonett DG. Sample size requirements for estimating intraclass correlations with desired precision. Stat Med 2002;21:1331–5.
11. Ferguson RL, Allen BL Jr. A mechanistic classification of thoracolumbar spine fractures. Clin Orthop 1984;189:77–88.
12. Shono Y, McAfee PC, Cunningham BW. The pathomechanics of compression injuries in the cervical spine: nondestructive and destructive investigative methods. Spine 1993;18:2009–19.
13. Nightingale RW, McElhaney JH, Richardson WJ, et al. Experimental impact injury to the cervical spine: relating motion of the head and the mechanism of injury. J Bone Joint Surg Am 1996;78:412–21.
14. Nightingale RW, McElhaney JH, Richardson WJ, et al. Dynamic responses of the head and cervical spine to axial impact loading. J Biomech 1996;29:307–18.
15. White AA, Panjabi MM. Clinical Biomechanics of the Spine. Philadelphia: Lippincott, 1970.
16. Magerl F, Aebi M, Gertzbein SD, et al. A comprehensive classification of thoracic and lumbar injuries. Eur Spine J 1994;3:184–201.
17. Bartko JJ. The intraclass correlation coefficient as a measure of reliability. Psychol Rep 1966;19:3–11.
18. American Spinal Injury Association. American Spinal Injury Association: International Standards for Neurological Classification of Spinal Cord Injury. Chicago: American Spinal Injury Association, 2002.
19. Vaccaro AR, Zeiller SC, Hulbert RJ, et al. The thoracolumbar injury severity score: a proposed treatment algorithm. J Spinal Disord Tech 2005;18:209–15.

cervical spine injuries; spinal cord injuries; classification; reliability; intraclass correlation coefficient; mechanism of injury

© 2006 Lippincott Williams & Wilkins, Inc.