Standard Plain X-ray, Three Views

Figures 3-1 through 3-7 demonstrate three-dimensional CT models of the cervical spine. Review these, as they will assist you in understanding cervical spine anatomy and the information to be gleaned from the two-dimensional projections obtained in plain x-ray. An adequate cervical spine plain x-ray series consists of three images: a lateral, an anterior – posterior (AP), and an open-mouth odontoid view (Figure 3-8). It is extremely important that adequate x-rays are obtained to maximize the sensitivity of this technique.

Figure 3-1 Lateral and oblique three-dimensional views of the normal cervical spine.

This three-dimensional reconstruction of the cervical spine was generated from standard computed tomography axial images using 1.25-mm slice thickness. Images like this are not routinely reviewed by emergency physicians, but they provide a good starting point for our discussion of cervical spine imaging. A, A lateral view, showing the normal relationship of C1 through T1 vertebrae. Compare this with the lateral x-ray in Figures 3-8 and 3-9. Note the overlap of the cervical facet joints, like shingles on a roof. B, An anterior oblique view, similar in perspective to an oblique x-ray view sometimes obtained to inspect for narrowing of the spinal foramen through which spinal nerves exit. C, A posterior oblique view. This has no routine x-ray equivalent but demonstrates the spinal lamina and posterior processes.

Figure 3-2 Anterior three-dimensional view of the normal cervical spine.

This three-dimensional reconstruction from axial cervical spine images shares the same perspective with an anterior–posterior x-ray or coronal computed tomography (CT) views. The C1 vertebrae is cropped from this image—we look at C1 in detail in Figure 3-3. C7 and T1 are also not in this image—as we discuss later, these must be seen in an x-ray series to allow adequate evaluation of the cervical spine. Note how the facet joints are not horizontal but slope caudad from anterior to posterior. As a result, these joints are not fully seen in a single axial CT image, as you will see in future figures. Also note how the spacing of vertebral bodies is symmetrical.

The lower portion of this reconstruction shows artifacts from a nasogastric (NG) tube.

Figure 3-3 C1 and C2 view of the normal cervical spine.

This three-dimensional reconstruction from computed tomography (CT) axial images focuses on the occipital–cervical junction and the C1-2 (atlantoaxial) junction. Note how the occipital condyles articulate with the superior surface of C1. The tip of the dens is hidden from view behind the anterior arch of C1. Two-dimensional slices can be more useful in identifying injuries because superficial structures do not obscure the view of deeper structures. The lateral margins of C1 and C2 align. These features will prove important in our discussions of fractures of this region in later figures.

The author generated this figure from a standard cervical CT scan using open-source software called OsiriX.

Figure 3-4 C1 and C2 view of the normal cervical spine.

Again, this three-dimensional reconstruction from computed tomography (CT) axial images focuses on the occipital–cervical junction and the C1-2 (atlantoaxial) junction. Compared with Figure 3-3, the view has been shifted to look down slightly upon C1. The tip of the dens is visible behind the anterior arch of C1. The occipital condyles articulate with the superior surface of C1. The skull has been cut away in this model, allowing the posterior ring of C1 to be seen through the foramen magnum. Later in this chapter we examine injuries to this region using multiplanar cross-sectional CT images. Refer to this figure for orientation if the two-dimensional images appear confusing.

Figure 3-5 C1 and C2 view of the normal cervical spine.

The skull base has been cut away nearly completely in this CT model, allowing the ring of C1 and its relationship to the dens of C2 to be seen in detail. Portions of the occipital condyles are still seen articulating with the superior surface of C1. Note that C1 does not have a true vertebral body—instead, the dens occupies the space immediately posterior to the anterior arch of C1. The posterior arch of C1 is in full view, and this perspective gives good views of the spinal canal, lamina of C2, and posterior spinous processes of C2 and more caudad vertebral bodies. C1 does not have a posterior spinous process.

Figure 3-6 C1 and C2 view of the normal cervical spine.

As in Figure 3-5, this three dimensional CT model has been rotated and the skull base cut away to allow the ring of C1 and its relationship to the dens of C2 to be seen in detail. The posterior arch of C1 is in full view. The transverse ligament, which holds the dens in position against the anterior arch of C1, is partially calcified in this patient and is therefore visible.

Figure 3-7 Transverse foramen (foramen transversarium).

This three-dimensional CT model is oriented with the observer looking cephalad along the anterior surface of the cervical spine. A series of holes perforating the transverse processes of each vertebra can be seen—the transverse foramen. These canals house the vertebral arteries as they ascend to the brain. Fractures through the transverse processes can result in arterial injuries, as can rotational injuries such as unilateral facet dislocation and subluxation injuries such as bilateral jumped facets. These injuries are discussed in detail in later figures, illustrated with the two-dimensional multiplanar computed tomography images routinely available to emergency physicians. Use this figure to understand how fractures and malalignment of vertebral bodies can disrupt the normal canal containing the vertebral arteries.

Figure 3-8 Lateral, anterior – posterior (AP) and odontoid view x-rays of the normal cervical spine.

A plain x-ray cervical spine series consists of three standard views: lateral (A), AP (B), and odontoid (C). An adequate series requires visualization of the entire cervical spine and its cephalad and caudad borders: the skull base and the first thoracic vertebral body (T1). An adequate odontoid x-ray must include the entire dens and the lateral borders of C1 and C2. We examine each of these views in more detail in the next several figures.

Normal Features of the Lateral X-ray

The lateral x-ray (Figures 3-8 through 3-13; see also 3-19 and 3-20) provides information about gross fractures, as well as alignment of vertebral bodies relative to one another. In addition, prevertebral soft tissue widening may suggest soft-tissue hematoma or swelling, a surrogate marker of fractures that may themselves be invisible on plain film. The x-ray must visualize the cervical spine from the skull base to the C7-T1 (cervical seventh vertebra–thoracic first vertebra) junction. This is not an arbitrary definition of an adequate plain film. An x-ray that fails to visualize the entire cervical spine and its junctions with the adjacent skull and thoracic spine may miss injuries. Widening of the space between the C1 vertebra and the skull base may be a subtle indication of a devastating occipital–cervical dissociation—although this injury is usually clinically evident with a moribund patient with quadriplegia. At the caudad end of the cervical spine, it is essential to visualize the C7-T1 junction to avoid missing subluxation of these vertebral bodies relative to one another. This injury is made more likely by the sudden change in the mobility of the spine, from the highly mobile cervical spine to the relative immobile thoracic spine, which is supported by the buttresses of the ribs.

Figure 3-9 Adequate and inadequate lateral cervical spine x-rays.

A, An adequate lateral cervical spine x-ray, showing the entire cervical spine and its cephalad border (the skull base) and caudad border (T1). B, The image is inadequate because C1 and T1 are not seen fully. Fractures or subluxation at these locations would be missed. The x-ray should be repeated, perhaps using special views such as the swimmer’s view to visualize the C7-T1 junction, or computed tomography could be used.

Figure 3-10 Interpretation of the lateral cervical spine x-ray.

A normal lateral cervical spine x-ray, with (A) and without (B) labels.

The spine should be inspected for four curved, roughly parallel lines:

1 Anterior longitudinal ligament line (follows the anterior border of the vertebral bodies)

2. Posterior longitudinal ligament line (follows the posterior border of the vertebral bodies)

3. Spinolaminar line (follows the anterior border of the posterior spinous processes, where the lamina converge)

4. Spinous processes line (follows the posterior border of the posterior spinous processes and does not fully parallel the other lines due to the increasing length of the spinous processes from C1 to C7)

The spinal canal, which houses the spinal cord, lies between lines 2 and 3.

If all lines are intact, the vertebral bodies should be properly aligned and the spinal canal should be preserved.

Figure 3-11 Evaluation of predental space.

On lateral cervical spine x-ray, the predental space should be evaluated. Several injury patterns can widen this space. Fracture of the ring of C1 can allow the anterior portion of the ring to migrate anteriorly with respect to the dens. Dens fracture can allow posterior migration of the dens relative to the ring of C1. Subluxation of the transverse ligament can allow widening of this space. Patients with rheumatoid arthritis are at particular risk of this latter abnormality.

Figure 3-12 Prevertebral soft tissues.

The lateral cervical spine x-ray should be examined for the width of prevertebral soft tissues. Widening suggests cervical spine injury with prevertebral soft-tissue swelling or hematoma, even in the absence of visible fracture. Lack of widening is relatively reassuring, as a fracture or ligamentous injury would be expected to be accompanied by some degree of soft-tissue swelling or hemorrhage.

In adults, soft tissues should be less than 7 mm in width at C2 and less than 5 mm at C3 and C4.

A rule of thumb in children is that the prevertebral soft tissues should be less than half the width of the adjacent vertebral body. False widening may be seen on expiratory films in children younger than 2 years, particularly if images are obtained during crying, which creates forced expiration.

A normal lateral cervical spine image shows a gentle cervical lordosis—a curve with its convex border oriented anteriorly (see Figure 3-10). This lordosis may be straightened slightly due to the presence of the cervical collar. When all cervical vertebrae are intact and in normal alignment with one another, this cervical lordosis results in four parallel curved lines, for which the lateral x-ray should be inspected (the numbered lines in Figure 3-10):

1. The anterior longitudinal ligament line, which follows the anterior border of the vertebral bodies.

2. The posterior longitudinal ligament line, which follows the posterior border of the vertebral bodies.

3. The spinolaminar line, which follows the anterior border of the posterior spinous processes, where the lamina converge.

4. The spinous processes line, which follows the posterior border of the posterior spinous processes. It does not fully parallel the other lines because of the increasing length of the spinous processes from C1 to C7. The space between the posterior longitudinal ligament line and the spinolaminar line is the spinal canal. If these lines are intact and parallel, the spinal canal should be intact, and the spinal cord should be safely contained within it.

In addition, the lateral film should be examined with attention to the following:

5. The predental space (see Figure 3-11), which separates the dens from the anterior ring of C1. Widening suggests dens or C1 ring fracture, or subluxation of the transverse ligament, which binds the dens to the anterior ring of C1. In adults, this space should be no more than 3 mm, and in children it should be no more than 4 to 5 mm.

6. Prevertebral soft tissues (see Figure 3-12). Widening suggests cervical spine injury with soft-tissue swelling or hematoma. In adults, this distance should be less than 7 mm at C2 and less than 5 mm at C3 and C4. In children, a commonly cited distance is less than half the width of the adjacent vertebral body. False widening may occur in children under the age of 2 years if the image is obtained during expiration. Below C4, the airway moves anteriorly, the esophagus occupies the prevertebral space, and soft-tissue widths vary considerably, usually between 10 and 20 mm in adults. Soft-tissue width exceeding the width of the adjacent vertebral body suggests injury.

7. The spinal facet joints (Figure 3-13), which should overlap like shingles on a roof. Decreased overlap can occur in the case of a jumped or perched facet joint.

8. Widening of the distance between posterior spinous processes, which can indicate fracture or ligamentous injury.

Figure 3-13 Assessment of facet joints.

A, Lateral cervical spine x-ray. B, Close-up. The facets should overlap like shingles on a roof. In the case of jumped or perched facet joints, the degree of facet overlap at the level of injury will be less than the overlap at uninjured levels. The facets can be recognized by their rhomboid shape in profile. Fracture through the facets and lamina is also common and should be looked for carefully.

Normal Features of the Anterior-posterior X-ray

The AP x-ray (Figure 3-14) provides information about lateral alignment of vertebral bodies with one another. A small number of additional fractures (particularly oblique fractures) may be noted on the AP x-ray, and subtle abnormalities from the lateral film may be confirmed. The posterior spinous processes should be centered in the midline and aligned with one another. The cortex of each vertebral body should be intact. The height of and spacing between adjacent vertebral bodies should be uniform.

Figure 3-14 Anterior–posterior (AP) normal cervical spine x-ray.

The AP cervical spine x-ray should be inspected for gross fractures. The intervertebral spacing (double arrows) should be symmetrical if the intervertebral discs (not visible on x-ray) are intact. The posterior spinous processes (black arrowheads) should project directly backward out of the plane of the x-ray and should be aligned with one another. In cases of unilateral jumped facet joint, rotation of one vertebral body relative to the adjacent bodies causes the posterior spinous process to project at an angle and thus not be aligned with the processes of other vertebrae.

Open-Mouth Odontoid X-ray

The open-mouth odontoid view (Figures 3-15 and 3-16) is essential for evaluation of two important injury patterns:

• Burst fractures of the C1 ring, also called Jefferson fractures

• Fractures of the odontoid process or dens, which is a structure of C2

Figure 3-15 Adequate and inadequate open-mouth odontoid view x-ray.

An adequate open-mouth odontoid view must visualize the entire dens and the lateral borders of C1 and C2. An incompletely visualized dens could result in an undiagnosed dens fracture. The lateral borders of C1 and C2 must be seen to assess for misalignment, which can occur in the setting of C1 burst fracture, in which the lateral masses of C1 are displaced laterally.

A, An adequate view. The lateral borders of C1 and C2 are visible and align normally. The spaces between the dens and the lateral masses of C1 are symmetrical. The entire dens is visible and appears intact. B, An inadequate view with the tip of the dens and the lateral masses of C1 and C2 cropped. Alignment and fracture cannot be assessed.

Figure 3-16 Normal open-mouth odontoid view x-ray.

The open-mouth odontoid view is essential for evaluation of potential dens fractures and fractures of the C1 ring.

A, Routine open-mouth odontoid image. B, Close-up of the region of interest, which includes C2 with the dens, as well as the lateral masses of C1.

The open-mouth odontoid view is so called because it is obtained with the patient’s mouth opened as wide as possible, with the x-ray tube aimed into the open mouth. This gives a direct view of C1 and C2. Issues that may prevent adequate visualization include an obtunded or uncooperative patient who is unable to comply with this maneuver; a patient with trismus, temporomandibular joint disease, or other injuries (e.g., mandibular fractures) that prevent full opening of the mouth; and the presence of significant metal dental work. Sometimes, the cervical spine collar may restrict mouth opening too much to allow an adequate open-mouth odontoid view. When an adequate open-mouth odontoid view cannot be obtained, CT should be performed to avoid a missed injury of C1 or C2.

An adequate open-mouth odontoid view visualizes the entire dens and the lateral margins of both C1 and C2. A fully visualized dens is essential for evaluation of dens fractures. The cortex of the dens should be carefully inspected for defects. Shadows of overlying structures such as central incisors can simulate or hide fractures, so a wide-open view with no overlap of teeth with the dens is desirable. When the ring of C1 is intact, the lateral margins of C1 should align with the lateral margins of C2. When fractured, C1 typically “bursts,” with the fracture fragments spreading outward. Consequently, an indirect sign of C1 fracture is misalignment of the lateral margins of C1 with the lateral margins of C2. Frequently, the lateral margins of C1 appear displaced laterally compared with the lateral margins of C2, although medial displacement of C1 is also possible. This finding may be unilateral or bilateral. When C1 is intact and the patient is facing directly forward, the distance between the dens and the medial borders of the C1 ring should be bilaterally symmetrical. When C1 is fractured, displacement of fracture fragments typically renders this distance asymmetrical. The lateral margins of C1 and C2 may be hidden by radiopaque dental fillings, especially if the patient’s mouth is not fully opened.

The standard three-view series just described is sometimes augmented with additional plain x-ray views. Adding two oblique views allows visualization of the pedicles, lamina, and neural foramina. Although this technique may be useful in selected cases, this “five-view” technique has been compared with the standard three views and does not detect additional injuries at a rate high enough to make five views a useful standard. The five-view series was insensitive (44%) compared with CT in a high-risk group of patients with altered mental status and a high rate of cervical spine injury (>10%).⁶ CT should be obtained if subtle lamina, pedicle, or neural foramen injuries are suspected.

In cases in which the C7-T1 junction is not visualized on the lateral x-ray, “swimmer’s” (Figure 3-17) or “traction” views may allow visualization of this region. In a swimmer’s view, the patient is positioned with the arm closer to the x-ray detector raised above the head. This elevates the humeral head above the C7-T1 junction, preventing it from obscuring this area. The opposite arm is held at the patient’s side, resulting in that humeral head lying somewhat lower that the C7-T1 junction. Variations on this theme include having the patient raise both arms directly in front of the body or over the head. In a traction view, a health care worker pulls on the patient’s arms from a position at the foot of the bed, distracting the humeral heads down to reveal the C7-T1 junction – a method now discouraged because of the possibility of worsening spinal injury. When these maneuvers do not allow adequate visualization of C7-T1, CT is indicated. In many cases, these views are skipped in favor of CT. Flexion and extension views (Figure 3-18) are sometimes obtained to assess for ligamentous instability—the evidence for these views is discussed in detail later.

Figure 3-17 Swimmer’s view x-ray of the cervical spine.

A swimmer’s view is a lateral cervical spine x-ray in which the patient is positioned with one arm raised above the head. This position tilts the shoulder girdle, elevating one humeral head and depressing the other, thus revealing the C7-T1 junction. Other variations on the swimmer’s position are sometimes used, including raising both arms above the head. The swimmer’s view is only necessary when the C7-T1 junction is not visualized on a standard lateral x-ray. Today, computed tomography is more commonly used to visualize this region.

Figure 3-18 Flexion and extension x-rays of the cervical spine.

Flexion (A) and extension (B) views may be performed to evaluate for ligamentous injury if plain films in neutral position do not show fractures or subluxation but the patient has persistent pain. The patient is asked to flex and extend the cervical spine actively, stopping if pain or neurologic symptoms develop. Subluxation suggests a ligamentous injury. Passive flexion and extension (in which the patient is manipulated by the examiner) should never be used, because this may cause cervical spinal cord injury. The utility of flexion–extension views is an area of controversy, but generally these are not useful in the evaluation of acute trauma because patients with acute pain are often unable to flex or extend sufficiently to exclude ligamentous injury.

Interpreting Cervical Computed Tomography

Interpretation of cervical CT can be based on many of the same criteria used for plain x-ray. We make some direct comparisons of imaging findings on CT and x-ray, so skills you may already have from x-ray can be translated to CT. For those with little experience with either modality, starting with CT may prove easier, with CT findings then clarifying subtle findings on x-ray. We take a step-by-step approach, using the sagittal, coronal, and axial images for different purposes. Each image series should be reviewed, because fractures in the plane of a given image series are difficult to see in that series but are readily apparent in perpendicular planes. We spend relatively little time up front on normal findings; instead, we concentrate on pointing out abnormalities in the figures that follow, with comparison to normal findings.

CT has high resolution for bony injury and directly detects fractures and dislocations. Modern CT scanners allow multiplanar and three-dimensional reconstructions, facilitating injury characterization. Image data is helically acquired as the patient passes through the CT gantry, typically at 1-mm or submillimeter slice thickness. Reconstructions are performed in sagittal, coronal, and axial planes (Table 3-2). The sagittal plane yields information similar to that from the lateral plain x-ray. The coronal plane yields information similar to that from the AP and odontoid plain x-ray views. Axial views give detailed anatomy of individual vertebral bodies, including clear views of the lamina, pedicles, transverse foramen enclosing the vertebral arteries, and spinal canal. Although the spinal cord is not well visualized on CT, the preceding image sets allow the canal to be carefully inspected for fracture fragments or dislocations that might impinge on the spinal cord. CT is considered to have outstanding sensitivity for fractures and dislocations—a normal CT viewed on bone windows effectively rules out these injuries. Nevertheless, it remains common practice to continue spine immobilization following a normal CT if the patient cannot be clinically evaluated for neurologic complaints or continued pain, though recent studies suggest a low risk of unstable cervical injuries following normal CT (see later discussion).

TABLE 3-2 Information Provided by Various Computed Tomography Image Planes

Step 1: Examine the Sagittal Computed Tomography Reconstructions

The sagittal CT reconstructions provide detailed information about the AP alignment of the cervical spine, as well as fractures or subluxations that may impinge on the spinal canal. Start your evaluation by selecting bone windows, and then move to a slice in the middle of the sagittal series, corresponding to the midsagittal plane (Figure 3-19). If you are experienced in interpreting cervical spine x-rays, this image should look familiar—it strongly resembles the lateral cervical spine x-ray, without the confusing overlay of bones and soft tissues from other planes. Just as with the lateral x-ray view, the midsagittal CT image allows assessment of alignment, using the curves of the anterior and posterior longitudinal ligaments and the spinolaminar line. In this view, the spinal canal can be seen in detail and inspected for any fracture fragments or subluxation of vertebral bodies that could impinge on the spinal cord. Remember that you are only looking at a single plane—you need to scroll laterally to the patient’s right and then left to perform these same steps in planes moving away from the patient’s midline. In the midsagittal plane, notice how large the dens is—perhaps you did not appreciate this on x-ray, but look now at the lateral x-ray for comparison. Check the contour of the dens for fracture lines, and if fracture is present, inspect for retropulsion of the dens into the spinal canal. Inspect the predental space for widening. Look at prevertebral soft tissues for increased thickness suggesting soft-tissue injury. Inspect each vertebral body for fracture lines. Check the spinal lamina and posterior spinous processes for fracture. Notice that in the midsagittal plane, the facet joints are not visible. These are lateral structures and are seen in the far lateral parasagittal images (Figure 3-20). Inspect these joints for unilateral or bilateral dislocation (jumped facets, shown in a later figure). At the cephalad limit of the cervical spine, inspect the articulation of the occipital condyles with C1 (see Figure 3-20). It may take you some time to become accustomed to imagining the cervical spine in three dimensions as you scroll through it in this plane.

Figure 3-19 Normal mid-sagittal computed tomography (CT) image, compared with normal lateral cervical spine x-ray.

This 24-year-old female hydroplaned on a wet road at 45 mph and complained of midline cervical spine tenderness. Her CT is reviewed to demonstrate normal findings.

A, When reviewing CT sagittal images, follow the paradigm of the lateral x-ray. First, select bone windows. Next, select the midsagittal CT image and note the alignment of vertebral bodies, using the same four lines used for evaluating alignment on the lateral x-ray (see Figure 3-10). Then, inspect each vertebra for fractures (shown in subsequent figures). Notice how the facet joints are not visible in the midsagittal plane. Also notice how large the dens is and how small the ring of C1 is. Prevertebral soft tissues can be evaluated using the same criteria as for x-ray. B, Lateral x-ray for comparison.

Figure 3-20 Normal parasagittal computed tomography (CT) image, compared with lateral cervical spine x-ray.

A, In this far lateral parasagittal CT image, the facet joints can be seen articulating normally, like shingles on a roof. The vertebral bodies are not seen in this plane, as we are far from the midline. The occipital condyles are seen articulating with C1. B, Lateral x-ray for comparison. Notice how much clearer the occipital–cervical junction is on CT.

Step 2: Inspect the Coronal Images

Use the coronal CT images (Figure 3-21) to simulate the open-mouth odontoid and AP cervical x-rays. Scroll through the “stack” of coronal images until you see the odontoid process projecting between the lateral masses of C1. Check for the same features you would expect on the open-mouth odontoid view. The lateral masses of C1 and C2 should align along their lateral borders. The spaces between the odontoid process and the lateral masses of C1 should be symmetrical. The odontoid process itself should have a smooth contour, with no fracture lines present. Once you have completed this evaluation, scroll in anterior and posterior directions through the stack of images, inspecting for fractures of the vertebral bodies, facets, and transverse processes. Fractures in the sagittal plane are visible on coronal images, and may be missed on the sagittal images reviewed first. Look for lateral displacement of vertebral bodies with respect to one another. Spend some time reviewing the figures in this chapter to familiarize yourself with the most common fracture patterns.

Figure 3-21 Normal coronal computed tomography image, compared with odontoid view x-ray.

A, Same patient as Figures 3-19 and 3-20. A coronal image centered on the odontoid (dens) and C1, simulating the open-mouth odontoid x-ray. CT findings of a normal dens are the same as those seen on open-mouth odontoid x-ray, and deviation from normal should heighten suspicion for injury, even when a fracture line is not seen. Subluxation without fracture may be present. The lateral borders of the C1 and C2 vertebral bodies should align. Fractures of the ring of C1 (Jefferson fractures) commonly result in radial spread of C1 fragments, disturbing this normal alignment. The spaces between the lateral masses of C1 and the odontoid process should be bilaterally symmetrical. C1 fractures can result in asymmetrical spread of fragments, and subluxation of the transverse ligament can also result in asymmetry. The odontoid process should be smoothly contoured with no visible fracture lines. A slight notched appearance on each side of the base of the dens is normal. B, Open-mouth odontoid x-ray for comparison.

Step 3: Inspect the Axial Computed Tomography Images

The axial CT images (Figure 3-22) have no immediate analogue in the normal three-view x-ray series. They can provide additional information not readily seen on the sagittal and coronal views. Particularly, they demonstrate fractures of the canals housing the vertebral arteries (foramen transversarium), which are more difficult to assess on other views. They also provide good views of fractures in the sagittal and coronal planes, which may not be seen well on those image series as they lie parallel to the plane of the image. From the sagittal images, you should already have a good idea of the alignment of the cervical spine, although the axial images can also demonstrate jumped facet joints. The axial images provide an en face view of the spinal canal. As you scroll from the level of C1 to T1, imagine yourself traveling down the spinal canal, and watch for fracture fragments that narrow the canal and may impinge on the spinal cord. Check the body, pedicles, lamina, and transverse and posterior spinous processes for fractures. We review many abnormal findings on the axial images in the figures throughout this chapter (see list, Table 3-1).

Figure 3-22 Normal axial computed tomography image.

Same patient as Figures 3-19 through 3-21. An axial image shows a cervical vertebra in cross section. The vertebral body, spinal canal, pedicles, lamina, and base of the posterior spinous process are visible. It takes some practice to become accustomed to the normal appearance of vertebrae in axial section. They are irregular structures and never lie completely within a single axial plane. Apparent discontinuities in the ring may be visible on some slices but may not represent fractures—instead, they simply reflect the irregular contour of the vertebrae passing in and out of the image plane. A, B, Two sequential slices.

Which Criteria Should I Use: The National Emergency X-radiography Utilization Study or the Canadian Cervical Spine Rule?

Both NEXUS and the CCR are the result of well-conducted prospective clinical trials—level I evidence under the hierarchy used by the American College of Emergency Physicians (Table 3-3), which does not have a current clinical policy recommending the best CDR for evaluation of the cervical spine. The American College of Radiology (ACR) cites both rules as appropriate for evaluation of the need for cervical spine imaging in its appropriateness criteria.¹ Application of either rule is appropriate.

TABLE 3-3 Levels of Scientific Evidence

Can I Mix and Match the Rules?

It may be tempting to mix two well-validated rules in an attempt to capture the best features of both. Unfortunately, it is uncertain what the result of this strategy

Box 3-5 The Canadian Cervical Spine Rule

• Prospective

• 10 centers

• 8924 patients

• Stable vital signs

• Glasgow Coma Scale = 15

• Blunt head and neck trauma

151 injuries (1.7%)

• Derived rule for cervical radiography

Sensitivity = 100%

Specificity = 42.5%

Reduces x-ray by 41.8%

Do the Rules Apply to Geriatric Patients?

Geriatric patients raise special concerns for cervical spine imaging. Do NEXUS and the CCR apply to these populations? NEXUS included 2943 patients age 65 or older, accounting for 8.6% of the study group. Cervical spine injuries occurred in 4.59% of this group, a rate

Box 3-6 The Canadian Cervical Spine Rule versus the National Emergency X-radiography Utilization Study

• Prospective

• 9 Canadian centers

• 8283 patients

• Cervical spine injuries

169 injuries (2%)

845 (10.2%) could not be analyzed (no range of motion performed)

• Sensitivity = 99.4% (CCR) vs. 90.7% (NEXUS) (p < 0.001)

• Specificity = 45.1% (CCR) vs. 36.8% (NEXUS) (p < 0.001)

• Radiography = 55.9% (CCR) vs. (NEXUS) 66.6% (p < 0.001)

• CCR misses 1 of 169 injuries

• NEXUS misses 16 of 169 injuries

more than twice that in younger patients. Notably, 20% of injuries were odontoid fractures, compared with only 5% in younger patients.⁵ Certainly elderly patients in this study were a high-risk group—could the NEXUS CDR actually reduce use in this group? While popular conception might hold that it would be fruitless to apply a CDR to geriatric patients, in actuality a greater percentage of the elderly met the NEXUS low-risk criteria and avoided imaging: 14% of geriatric patients compared with 12.5% of younger patients. Following the NEXUS rule, only two cervical spine injuries were missed in older patients, both fitting into the predefined “insignificant” category. By this definition, NEXUS maintained 100% sensitivity in a geriatric population. When considering geriatric patients with blunt trauma, NEXUS can be applied safely. However, many elderly patients will require cervical CT, if they require any form of imaging, due to inadequate or abnormal x-rays demonstrating degenerative changes.

The CCR found age of 65 years or older to be a high-risk factor mandating imaging in its derivation set,⁷ and this was maintained as a mandatory criterion for cervical spine imaging in the multiple validation studies that have followed. Thus, for geriatric patients, application of NEXUS results in imaging of fewer patients than does the CCR.

How Sensitive Are X-ray, CT, and Magnetic Resonance Imaging for Cervical Spine Injury?

Once a patient is determined to require cervical spine imaging, the best test for evaluation must be selected. Let’s examine the sensitivity and specificity of plain x-ray, CT, and MRI for acute cervical spine injury.

How Sensitive Is X-ray for Cervical Spine Injury?

Studies of x-ray sensitivity find markedly different results, depending on whether inadequate films are considered or whether these are excluded from analysis and only “adequate” x-rays are considered in calculating the false-negative rate. In addition, if x-rays are credited as “positive” for detecting any fracture, they appear to have relatively good sensitivity, whereas when x-ray sensitivity is discounted for second and third fractures found on CT, their sensitivity appears quite poor. The consensus among multiple studies is that cervical spine plain x-rays miss some fractures—with a pooled sensitivity of 52%, compared with 98% for CT, according to a meta-analysis of seven studies.¹⁰ As a consequence, the ACR has now advocated CT as the initial screening test for suspected cervical spine injury in adult patients, in place of x-ray. The ACR endorses no cervical spine imaging in patients meeting either the NEXUS or the CCR low-risk criteria. Certainly when high clinical concern exists for fracture, CT should be performed, often as the initial imaging study rather than x-ray. In a low-pretest probability patient with a normal cervical spine plain x-ray series, the risk of fracture is very low. Let’s examine some of the evidence behind these conclusions.

Holmes et al. performed a meta-analysis of cervical spine imaging studies to determine the sensitivity of plain x-ray and CT, and their results are largely the basis of the current ACR recommendation. The authors imposed fairly strict inclusion and exclusion criteria for studies (Table 3-4). Studies were considered if they were either a randomized, controlled trial comparing x-ray and CT or a cohort study of patients undergoing both x-ray and CT. Studies were excluded if the x-ray series did not include the AP, lateral, and open-mouth odontoid views; if the CT scan did not extend from the occiput to the superior aspect of the first thoracic vertebra; or if distance between CT scan slices exceeded 5 mm. The authors also assessed the methodologic quality of the studies, using the rating scale in Table 3-2. Among 712 studies identified by the authors in Medline, only 7 met the inclusion criteria, and none of those were rated as methodologic quality I or II. The authors used the published raw data from these 7 studies and calculated a pooled sensitivity of 52% for x-ray (95% CI = 47%-56%) and 98% for CT (95% CI = 96%-99%). The meta-analysis authors acknowledge the methodologic limitations of the available data and state that CT should be used as the primary tool for screening high-risk patients, although further study is needed to determine whether x-ray is adequate for evaluation of low-risk patients. Despite this, the ACR has adopted an all-or-none approach to cervical spine imaging in the adult acute trauma patient and does not differentiate between patients at higher and those at lower risk once the decision has been made to perform cervical imaging.

TABLE 3-4 Criteria Used to Select Studies for Inclusion in Meta-analysis of Cervical Spine Imaging

Before accepting the results of this meta-analysis, let’s consider whether the authors’ criteria are reasonable and achievable. First, for cervical spine injury, what independent gold standard test should be used to confirm or rule out an injury, outside of CT? Clinical follow-up could be used but would only be useful for identifying injuries resulting in neurologic sequelae or surgical interventions, because stable fractures might heal uneventfully and not be appreciated on clinical grounds. Other imaging tests, such as MRI, could be used, but it is not clear that their results are a stronger gold standard than CT. Nonetheless, demanding an independent gold standard is essential when comparing x-ray and CT, as CT would otherwise be guaranteed the apparent better result. None of the studies included in this meta-analysis used a gold standard independent of CT results. Their use of CT as the gold standard represents incorporation bias and limits the validity of the studies and of the meta-analysis. Moreover, because no false-positive CT results are possible under this definition (because the CT result is considered correct in every instance), specificity and positive and negative likelihood ratios for CT cannot be determined. In addition to problems with the gold standard in the included studies, the studies suffer from spectrum bias as a consequence of inclusion of mostly badly injured patients at high risk of multilevel cervical spine injury. Holmes et al acknowledged these limitations and suggested that CT might not be necessary in all patients.

Were the authors too stringent in their inclusion criteria for studies to be considered? Note that the sentinel NEXUS and CCR studies, with more than 40,000 patients between them, were not included in this meta-analysis because neither study compared x-ray to CT scan in all patients. Both studies used a combination of imaging studies and clinical follow-up to determine if a patient had a clinically important cervical spine injury, rather than asking whether any fracture was present.

Let’s examine several additional studies, including some of those included in the meta-analysis. In a single-center study of 936 trauma registry patients over a 9-month period, 58 patients with cervical spine injuries (6.2%) were found. A substantial number of inadequate plain x-rays occurred. When considering only those cases in which a technically adequate plain x-ray series (three views of the cervical spine) was obtained, three false negatives occurred, resulting in a sensitivity of 90.3% and a specificity of 96.3% for x-ray. The positive predictive value of an adequate three-view series was 54.9%, with a negative predictive value of 99.5%.¹¹ In a second retrospective single-center study of 3018 blunt trauma patients, 1199 patients underwent both x-ray and CT scan. In this study, 116 patients with injury were detected (9.5%), and 41 (3.2%) had false-negative x-rays. All required treatment—the authors concluded that CT rather than plain x-ray should be used to screen trauma patients.^12-13 However, two limits of this study bear consideration. First, the average Glasgow Coma Scale score in patients with missed injuries was 12 and the average injury severity score (Box 3-7) was 14.6 (on a 0- to 75-point scale), so this was a population of moderately injured patients with altered mental status, not a population of alert and otherwise uninjured patients, which is a common scenario in which cervical spine evaluation is required. In addition, as in many studies, CT scan was the gold standard for diagnosis, and it automatically was credited with a sensitivity of 100%.

In a third, prospective study of 1356 blunt trauma patients with altered mental status, 95 injuries from C0 (skull base) to C3 were found in 70 patients (5.2%).

Plain x-ray found injuries in 38 patients (54%), while CT found injuries in 67 (96%).¹⁴ The remaining 3 injuries (4%) not detected on CT were recognized by clinical neurologic deficits, and represented spinal cord injury without radiographic abnormality (SCIWORA) (see later discussion).

A fourth, prospective study of 324 blunt trauma patients compared two protocols for assessment of the cervical spine. By protocol, patients requiring head CT underwent a single lateral cervical spine x-ray and cervical CT scan. If head CT was not clinically indicated, three views of the cervical spine were performed, with selective CT performed to evaluate areas not well seen on plain x-ray. Fifteen patients (4.6%) had cervical spine injuries, and x-ray had only 54% sensitivity compared with the gold standard of CT.¹⁵ The authors concluded that CT is the appropriate initial cervical imaging test in patients undergoing head CT.

Contrary to these studies is the largest dedicated prospective study of cervical spine injury and imaging—NEXUS. This study of 34,069 patients included 818 patients with injuries (2.4%). The gold standard in this study was review of all final radiology reports, including CT and MRI when available, and clinical follow-up. X-ray revealed 932 injuries in 498 patients and missed 564 injuries in 320 patients. But 436 missed injuries in 237 patients occurred in the context of an inadequate plain x-ray, or an x-ray that was in fact abnormal and revealed at least one cervical spine injury, though not all of the injuries that were ultimately discovered. When the definition of a false-negative x-ray was limited to a normal and adequate x-ray in the presence of cervical spine injury, 23 patients with 35 injuries were found, including three unstable cervical injuries. By this more limited definition of false negatives, x-ray had a sensitivity of 97%. The authors concluded that (1) inadequate x-rays mandate additional imaging and (2) rarely, normal x-rays miss injury.¹⁶

Should we accept the NEXUS authors’ definitions and believe their conclusions, or are they manipulating the data to defend x-ray? From a practical clinical standpoint, an x-ray that shows one fracture but misses a second fracture is true positive, not false negative. This conclusion is based on the premise that all abnormal plain x-rays will be followed by CT, which presumably will detect the second injury. So, if the purpose of x-ray is to screen for the presence of one or more injuries, this definition is fair. Excluding inadequate plain x-rays from analysis also makes sense from a clinical perspective, because an indeterminate test should not be relied upon to exclude an important diagnosis. The NEXUS authors’ take-home point is likely accurate—when normal and adequate x-rays are successfully obtained in a patient with a low pretest probability of injury, missed injuries are rare. The emergency physician should recognize patients with a high pretest probability of injury and those in whom inadequate plain x-rays are likely to be obtained and should choose CT as the first imaging test. In addition, the emergency physician must reject inadequate plain x-rays, recognizing the high risk of missed injury, and perform additional imaging (usually with CT) when these occur.

What Constitutes an Adequate Plain X-ray Series? Are Five Views Better than Three?

Traditionally, five views of the cervical spine were performed at many medical centers: the standard three views (AP, lateral, and odontoid) and bilateral oblique views. Do these additional two views improve the sensitivity of x-ray? In a single-center study of 1006 blunt trauma patients with altered mental status, 116 patients with 172 injuries were enrolled. All underwent five views of the cervical spine followed by CT. The five-view plain x-ray series missed 52.3% of injuries in 56% of patients, including 93% of occipital fractures and 47.2% of fractures from C1 to C3. The overall sensitivity of the five-view series was only 44%, with 100% specificity. The positive predictive value was 100%, with negative predictive value of 99.7%. The authors concluded that five views offer inadequate sensitivity and that CT is more appropriate in patients with altered mental status.⁶

How Sensitive Is CT for Cervical Spine Injury?

CT scan is believed to be nearly 100% sensitive for bony abnormalities or dislocation, although this conclusion is limited by the lack of a clear alternative gold standard. Because CT is generally considered superior to plain x-ray and even MRI for evaluation of bony abnormalities, a finding on CT is considered to be real—CT is 100% sensitive for fractures. Missed injuries on CT are generally restricted to isolated soft-tissue injuries, included cervical spinal cord injuries without fractures or subluxations. These injuries are called spinal cord injury without radiographic abnormality (SCIWORA).

How commonly does CT miss spinal cord injuries? In one study of trauma patients with altered mental status, CT identified injuries in 67 of 70 patients with spinal injury (96%). The remaining 3 patients had neurologic deficits attributable to C0 to C3 spinal cord injury.¹⁴ In a similar study of trauma patients with abnormal mental status, CT

sensitivity for cervical spine injury was 97.4% and specificity was 100%, with positive and negative predictive values of 100% and 99.7%, respectively. Again, by definition, CT only missed SCIWORA—in 2 of 116 patients (1.7%).⁶

Does a Normal CT Rule Out Soft-Tissue Injuries Such as Spinal Ligament Tears, Disc Protrusions, and, Most Importantly, Spinal Cord Injuries? Should Patients Remain Immobilized After a Normal Cervical Spine CT? Is Magnetic Resonance Imaging or Other Imaging, Such as Flexion–Extension X-rays, Required to Rule Out Additional Injuries?

CT scan is extremely sensitive for cervical spine injuries such as fracture and subluxation. The presence of prevertebral soft-tissue swelling implies possible ligamentous injury, and subluxation of vertebral bodies virtually guarantees ligamentous injury, although the spinal ligaments themselves are not well seen on CT. When sufficient injury has occurred to tear spinal ligaments, associated hemorrhage and soft-tissue swelling are expected to occur, increasing the thickness of prevertebral soft tissues. On CT, the width of prevertebral soft tissues is easily measured. Theoretically, if this tissue thickness is normal, ligamentous injury is quite unlikely, even if the ligament itself cannot be visualized. MRI is the test of choice if ligamentous injury is highly suspected, because the injured tissue can be directly visualized. By adjusting the window level on cervical spine CT, other soft-tissue detail such as spinal epidural hematomas and disc protrusions can sometimes be seen, although MRI remains the gold standard.

When no fractures, subluxations, or soft-tissue injuries are seen on CT scan, are soft-tissue injuries ruled out, or is MRI needed, as has been the standard for the past decade? Two common clinical scenarios may raise this issue in emergency medicine practice. First, in the obtunded or sedated trauma patient, in whom a reliable neurologic examination cannot be performed, is MRI required to rule out cervical soft-tissue injuries? While this is an important question, for most emergency physicians this is a decision deferred to trauma teams following admission. Several recent studies have addressed this question, and their results may help us to understand the second common clinical scenario, that of the awake and alert emergency department patient who continues to complain of neck pain following a normal cervical spine CT. Does this patient require further imaging, either with MRI or with flexion–extension x-rays, to assess for ligamentous injury? Let’s examine the results of some recent studies that may shed light on this topic.

Menaker et al.²⁴ reviewed the Maryland trauma registry and identified 234 obtunded blunt trauma patients with normal cervical spine CT who underwent MRI for further assessment, following a clinical guideline. Of these patients, 18 (8.9%) had an abnormal MRI, including 2 who required surgical repair and 14 in whom extended cervical collar use was employed.

Stelfox et al.²⁵ retrospectively studied 140 consecutive intubated patients before and 75 consecutive intubated patients after a change in protocol for cervical spine clearance in obtunded trauma patients at Massachusetts General Hospital. Before the protocol change, cervical spine immobilization in these patients was discontinued after normal findings on cervical spine CT with multiplanar reconstructions and one of the following: a reliable normal clinical neurologic examination (implying normalization of mental status), normal passive flexion–extension x-rays (which have been criticized because of risk of injury when the cervical spine of an obtunded patient is moved by a physician), or normal cervical spine MRI within 48 hours of admission. After the protocol change, normal CT findings alone were considered sufficient for cervical spine clearance. A normal CT was defined as one with no evidence of fracture, dislocation, subluxation, or indirect signs such as soft-tissue edema, inappropriate lordosis, or widening of the atlantodental distance. The study authors concluded that no missed cervical spine injuries were documented in either group. Does this prove that no injuries were present? No; in many of these patients, the neurologic examination may have remained uninterpretable because of concurrent injuries, such as traumatic intracranial injuries or anoxic brain injury. Cervical spine injuries could have been present but not recognized, and cervical spinal cord injuries resulting from discontinuation of cervical spine immobilization could have occurred but not been recognized or documented in the patient medical record. A stronger gold standard would have been helpful in proving that discontinuing cervical spine immobilization on the basis of CT findings alone is a safe practice. Assume for a moment that the authors are correct that no injuries were missed in the second group, of whom 70 had cervical spine clearance before death or discharge. If no injuries were missed in this group, the upper 95% CI limit of the actual missed injury rate can be estimated as 3 of 70 (a well-described estimation method for series with zero outcomes), or 4%. This would mean that a normal CT in this scenario predicts better than 96% chance of no cervical spine injury. We should be careful in accepting the authors’ contention, however. The possibility of unrecognized missed injuries casts doubt on the results of this study.

A third, similar study examined the results of cervical spine MRI performed in obtunded trauma patients following a normal cervical spine CT with a 16-slice scanner. Como et al.²⁶ prospectively studied 115 obtunded blunt trauma patients at MetroHealth Medical Center in Cleveland, Ohio. They found that 6 of 115 patients (5.2%) had acute injuries identified on MRI that had not been recognized on CT, including microtrabecular bony injuries, intraspinous ligament injuries, a spinal cord signal abnormality, and a cervical spine epidural hematoma. The authors contend that none of these injuries changed patient management or required continued spine immobilization, though they admit limitations, including lack of long-term follow-up.¹⁹ The authors did not report the lower limit of their 95% CI, but using their reported data, the negative predictive value of CT could be as low as 89%. Tomycz et al.²⁷ studied a similar population of 180 patients with normal cervical CT and found that 21.1% had abnormalities on MRI, none of which was unstable or required surgery. Similar limitations apply.

Sekula et al.²⁸ reviewed the charts of 6558 patients admitted for blunt trauma. They identified 447 patients with cervical fractures. Among the remaining 6111 patients without fracture, they identified only 12 (0.2%) with injuries requiring surgical fixation or halo placement. The authors contend that all were diagnosable by findings on multidetector CT, although interestingly, 3 of 12 patients had injuries not recognized prospectively before MRI. The authors discount these as “misinterpretations” of CT. A more strict methodology would require that these 3 cases be counted as false-negative results, giving a sensitivity of only 75% (9 of 12) for CT in detection of unstable nonfracture injuries requiring stabilization. Moreover, the authors note that they had no follow-up on discharged patients who may have developed neurologic symptoms because of missed injuries.

Hogan et al.²⁹ retrospectively reviewed the records of 366 obtunded patients at Maryland’s R. Adams Cowley Shock Trauma Center who had undergone normal cervical CT followed by MRI. MRI identified seven cervical cord contusions, four ligamentous injuries, and three intervertebral disc contusions. In this study, CT with a 4- or 16-slice scanner had a negative predictive value of 98.9% for ligamentous injury and 100% for unstable cervical spine injury. Although negative predictive values are generally discouraged as a means of presenting study results, as they are subject to change depending on the prevalence of injury in the population, these figures are impressive given the high-risk population in this study. The study did not report CT sensitivity, as the denominator of all patients with cervical spine injuries was not studied, only those with negative CT.

Smaller studies provide conflicting results. Stassen et al.³⁰ found that up to 30% of patients had cervical spine ligamentous injuries on MRI, when no fractures were detected on eight-slice cervical CT. The significance of these injuries is uncertain. Adams et al.³¹ found no additional cervical spine injuries in 20 obtunded patients with normal CT.

How do these studies affect the management of the awake and alert patient with continued neck pain after a normal cervical CT? In general, these studies imply a very low rate of unstable cervical spine injury following a normal cervical spine CT scan. The populations studied are very high risk for injury, as the patients are intubated and obtunded, multiply injured patients at level I trauma centers. It is likely that cervical spine clearance following a normal CT would be even safer in an awake and alert patient with no other significant injuries. The current practice of immobilization until resolution of symptoms or clearance with flexion–extension x-rays has little evidence basis, although it carries the weight of long-held practice. Future higher-quality studies with strict gold standards (CT and MRI in all patients) and adequate follow-up would bolster the evidence at hand. Ideally, these studies would report sensitivity and specificity for CT by including in the studies not just patients with normal CT but all patients being evaluated with CT and MRI for cervical spine injury.

The studies we have discussed do demonstrate that rarely CT misses soft-tissue cervical spine injuries. Consequently, patients who complain of neurologic abnormalities such as weakness, numbness, or parethesias after a normal CT scan should continue to be evaluated with MRI.

What Is the Value of Flexion–Extension Views of the Cervical Spine? Do They Assist in Identifying Ligamentous Instability?

Flexion–extension views (see Figure 3-18) historically have been performed to evaluate for ligamentous instability once fracture has been “ruled out” by x-ray or CT. Flexion–extension views are contraindicated when an unstable fracture pattern or subluxation is evident on prior imaging (plain x-ray or CT), or when a patient has neurologic signs or symptoms, as motion of an unstable cervical spine may result in cervical spinal cord injury. Flexion-extension views consist of lateral x-rays of the patient’s cervical spine in active flexion and active extension. Active in this case means the patient must consciously and without assistance flex and extend the cervical spine. It is assumed that the patient will not worsen spinal cord injury, because increasing pain or neurologic symptoms will be warning signs to the patient to halt further flexion or extension. Passive flexion and extension, in which the patient’s cervical spine is physically manipulated into flexion and extension by a second person, should never be used; this may cause cervical spinal cord injury, because warning signs of pain or neurologic dysfunction might not be heeded quickly enough by the manipulator.

How might flexion–extension views detect ligamentous injuries? The spinal ligaments are not visible on plain x-ray, but torn, avulsed, or stretched spinal ligaments would presumably allow an abnormal degree of subluxation of one vertebral body relative to adjacent bodies. The acceptable degree of normal subluxation may vary with the age of the patient. In children under age 8, pseudosubluxation of up to 40% occurs in up to 40% of patients at the C2-3 level and in up to 14% of patients at the C3-4 level.^32-33 Pseudosubluxation should be visible in flexion but should resolve or diminish in extension views. Subluxation in extension likely represents real pathology. Swischuk’s line, a line from the anterior portion of the C1 spinous process to the same point on the C3 spinous process, can assist with differentiation of pseudosubluxation from subluxation.³³ The anterior portion of the C2 spinous process should be within 2 mm of this line. In addition, no significant prevertebral soft-tissue swelling should be seen with pseudosubluxation.

In an adequate normal flexion–extension series, the patient flexes and extends and no abnormal subluxation is seen. Normally, the disc spaces should widen by no more than 1.7 mm, and the vertebral bodies should not translate by more than 1 mm. The inferior endplates of two adjacent vertebral bodies should not display greater than 11 degrees of angulation. If abnormal subluxation is noted or if neurologic symptoms develop, MRI is indicated (Box 3-9). If pain prevents adequate flexion and extension, MRI may be performed, or the patient may be left immobilized in a cervical collar for several days, with flexion–extension views repeated when the patient’s pain has subsided to a sufficient degree. Caution should be exercised when obtaining flexion–extension views long after the injury; theoretically, it would be possible for the pain and swelling of an unstable ligamentous injury to have resolved completely, and the patient could experience a dangerous degree of

Does evidence support the practice of obtaining flexion–extension views? Subgroup analysis from NEXUS found that flexion–extension x-rays revealed only two stable fractures and four subluxations in 86 patients undergoing these additional views. The subluxation injuries were found in patients with other abnormalities already noted on neutrally positioned x-ray. The authors concluded that flexion–extension x-rays add little to the standard three views of the cervical spine—although this subgroup analysis includes too few patients to provide strong evidence.³⁴

A retrospective review of 106 consecutive, awake, blunt trauma patients evaluated with flexion–extension x-rays found that 32 patients (30%) had nondiagnostic studies due to inadequate flexion and extension and 4 of these patients (12.5%) had cervical spine injuries subsequently diagnosed by CT or MRI. In contrast, only 5 of the 74 patients (6.75%) achieving adequate range of motion on flexion–extension views were found to have cervical injuries, all detected with flexion–extension. This study is quite limited by its small numbers and the lack of a uniform gold standard. Although the authors asserted that adequate flexion–extension x-rays missed no injuries, many patients did not undergo further definitive imaging, a methodologic flaw called workup or verification bias.³⁵ The authors suggested that limited ability to perform flexion and extension on physical examination should indicate that flexion–extension x-rays not be performed, as they are likely to be nondiagnostic. They also concluded that additional cross-sectional imaging with either CT or MRI should be performed in patients who cannot perform flexion and extension after normal neutral x-rays, because this may be a marker of undetected injuries. Larger studies are required to confirm or refute this recommendation. The ACR states that the limited sensitivity and specificity of flexion-extension x-rays and the rarity of technically adequate x-rays make them almost useless in trauma patients. The ACR does suggest a role for flexion-extension x-rays in evaluation of potential ligamentous injury in patients with equivocal MRI, with abnormal ligament signal but no clear disruption.¹ As described earlier in the discussion of CT for cervical spine clearance, flexion–extension x-rays are also likely unnecessary after a completely normal cervical CT with no soft tissue abnormalities noted.

Is There a Role for Fluoroscopy in Assessing for Unstable Cervical Spine Injuries?

Fluoroscopy has been suggested as a means of detecting ligamentous injury of the cervical spine in patients with altered mental status,³⁶ but critics strongly discourage the use of fluoroscopy for this purpose. Passive manipulation of the cervical spine during fluoroscopy can result in cervical spine cord injury and should never be used. In a study of 301 obtunded trauma patients, only 2 patients (0.7%) had ligamentous injury detected by fluoroscopy, and 1 patient developed quadriplegia.³⁷ Fluoroscopy shows bones but not soft tissues. It can detect ligamentous instability by revealing subluxation in real time, giving indirect evidence of ligamentous injury. Unfortunately, as the ligamentous injury is revealed on fluoroscopy, cervical spinal cord injury may be occurring, because the obtunded patient cannot complain of neurologic symptoms or pain. MRI should be used to evaluate for ligamentous injuries, because it does not carry this risk.

Do Cervical Spine Injuries Predict Arterial Injury? When Should Imaging of Cervical Arteries Be Performed, Following Detection of Cervical Spine Injury?

Sometimes cervical CT demonstrates fractures involving the transverse foramen, the bony channels housing the vertebral arteries. In other instances, subluxations or rotatory injuries of the cervical spine may shear or stretch these vessels, potentially causing vascular tears. Do these injuries predict an increased risk of arterial injury? In a retrospective study, 92 of 605 patients (15%) evaluated with angiography for cervical arterial injury after blunt trauma were found to have arterial injuries. Of these 92 patients, 71 (77%) had an associated cervical spine injury. Of these spine injuries, 55% were subluxations, and 26% were fractures involving the transverse foramen. Most of the remaining spinal injuries associated with cervical arterial injury were injuries to C1 through C3. Does this study indicate that all high cervical spine fractures, subluxations of the cervical spine, or transverse foramen fractures should be followed with evaluation of cervical arteries? Blunt vertebral artery dissection is associated with stroke, which is perhaps prevented by anticoagulation therapy. When these types of cervical spine injury are present, arterial imaging appears warranted based on current evidence.³⁸ Arterial injuries are discussed in more detail in Chapter 4.

When Should CT Be the Initial Cervical Spine Imaging Modality, Rather Than X-ray?

The ACR now recommends a CT-first strategy in all adult patients with suspected cervical spine injury, as discussed earlier. If a technically adequate x-ray series can be obtained in a low-risk patient, x-ray can be used to screen for fracture, as it has been used for the past century. CT is the clear preferred first-line imaging technique in patients with high likelihood of cervical spine injury, in patients with multiple other injuries requiring CT for evaluation, in patients in whom rapid cervical spine clearance is especially desirable (e.g., to allow unusual positioning in the operating suite for treatment of other injuries), and in patients with a low likelihood of adequate plain x-ray, such as the elderly, obese patients, and patients with short necks, extensive cervical spine degenerative joint disease, prior injuries to the cervical spine that might be confused with acute injuries on plain film, and bone disorders such as ankylosing spondylitis (see Box 3-2).

Can Computed Tomography Detect Spinal Cord Injury and Nontraumatic Spinal Cord Abnormalities?

MRI, not CT, is considered the test of choice for spinal cord injury. CT shows secondary findings suggesting cord impingement, such as narrowing of the spinal canal from fracture fragments or subluxation of vertebral bodies. In some cases, the cord can be seen, but its tissue density is similar to that of cerebrospinal fluid, making detailed inspection difficult. A special technique called CT myelography (CT myelogram) is an alternative to MRI for visualization of the spinal cord. In CT myelography, contrast material is injected into the spinal canal through a spinal needle in a technique similar to diagnostic lumbar puncture. CT is then performed, and the spinal cord and nerve roots can be seen in relief against the contrast material outlining them. This technique is not commonly used but can be used in patients with contraindications to MRI. CT myelography provides information about spinal cord compression but does not provide information about intrinsic spinal cord pathology such as transverse myelitis, spinal cord infarction or edema, and multiple sclerosis. This type of pathology requires the soft tissue contrast of MRI.

How Long Does Cervical Computed Tomography Take to Perform? Which Is Faster, X-ray or Computed Tomography?

CT is the most time-efficient method of imaging the cervical spine. Daffner³⁹ measured the time required to perform six views of the cervical spine (AP, lateral, open-mouth odontoid, bilateral obliques, and swimmer’s view) and found the average to be 22 minutes (range of 5 to 46 minutes). Many patients required at least one view to be repeated to achieve a satisfactory image. In the same study, and again in a separate study, Daffner measured the time required for cervical spine CT and found the average to be between 11 and 12 minutes (range of 3 to 35 minutes), depending on whether the cervical CT was performed in conjunction with a head CT.^39-40 These results were achieved using a 4-slice scanner, and imaging times with newer scanners are even faster. A 64-slice CT can perform sixty-four 0.625-mm slices per second, meaning that in each second of scanning, a territory 40 mm (64 × 0.625 mm) in length can be imaged. If the cervical spine is 20 cm (200 mm) in length, CT could be completed in 5 seconds. In addition, CT has the advantage of performing cervical spine imaging in the same location that diagnostic imaging for other injuries is being performed. Even if only three x-ray views of the cervical spine were performed, halving the time for x-ray, CT is faster with modern scanners.

Radiation

The radiation doses for cervical spine x-rays and CT differ by an order of magnitude.

A three-view cervical x-ray series (consisting of an AP, a lateral, and an open-mouth odontoid view) generates about 0.1 mSv, although the dose may rise if additional views or repeated attempts at the same view are needed to obtain diagnostic-quality x-rays. CT of the skull base and upper cervical spine for odontoid evaluation creates an effective dose of about 4.4 mSv using a single-slice scanner and about 2.3 mSv using a 16-slice scanner. CT of C0 to C3 and the cervicothoracic junction creates a dose of about 7.1 mSv using a single-slice scanner and 4.3 mSv for a 16-slice scanner. CT of the entire cervical spine generates a dose of 8.2 mSv using a single-slice scanner and 5.4 mSv for a 16-slice scanner.⁵⁴ Substantial thyroid radiation exposures occur with cervical spine CT. Estimates of lifetime attributable mortality from complete cervical CT are in the range of 1 in 2400 for a single-slice scanner and 1 in 3700 using a 16-slice CT scanner. Attributable mortality estimates may vary with age at exposure, with younger patients incurring greater risk. Improvements in CT will likely reduce the effective dose, but significant radiation doses will likely remain using CT.⁴¹

Cost of Cervical Imaging

Cost-effectiveness of cervical radiography and CT have been compared in models with estimates based on patient pretest probability of injury, assumptions about the cost of missed injuries and resulting health care costs to society, and the direct costs of imaging.^42-43 Blackmore⁴² found that high- and moderate-risk patients should be imaged with CT as the most cost-effective strategy. In low-risk patients, CT screening would be predicted to identify additional potentially unstable injuries, but the cost per quality-adjusted life-year would be greater than $80,000—making x-ray the more cost-effective strategy in this group. Like all cost-effectiveness analyses, this report may be subject to many inaccuracies due to potentially false assumptions about the sensitivity of x-ray and CT, the incidence of injury, the direct costs of imaging, and the costs associated with missed injuries. In a second cost-effectiveness analysis with similar suppositions about sensitivity, likelihood of injury, and costs, Grogan⁴³ found CT to be more cost-effective than x-ray. According to that analysis, CT would have an institutional cost of $554 per patient, compared with $2142 per patient for x-ray, when the costs of missed injury settlements are included. Sensitivity analysis showed that x-ray could be more cost-effective if CT costs more than $1918 or if x-ray has a sensitivity greater than 90%.

How Are X-rays of the Thoracolumbar Spine Interpreted?

X-rays of the lumbar and thoracic spine are interpreted with many of the same considerations as those for cervical spine radiographs. The standard views of the thoracic spine are AP and lateral, whereas the standard lumbar views are AP, lateral, and a spot view of L5 to S1 to allow adequate penetration while limiting radiation exposure.

On the AP view (see Figure 3-86), the thoracic spine should be assessed for alignment of adjacent vertebral bodies. The pedicles are usually visible like the “eyes of an owl,” with the posterior spinous process present in the midline as the owl’s beak. The posterior spinous processes may be displaced from the midline when fracture or subluxation is present. Pedicles may appear asymmetrical because of fracture or may be eroded by lytic processes such as spinal metastases and infection. The transverse processes should be inspected for fractures. Paraspinous soft-tissue masses or hematomas may be evident as asymmetrical densities, and their presence can suggest fracture, even when a fracture is not directly visualized. The heights of vertebral bodies should be equal on the AP projection, and the space between vertebral bodies should be even. Compression fractures of vertebral bodies can decrease their height, and disc injuries can decrease the space between adjacent bodies.

Figure 3-86 T2 corner avulsion fracture (extension teardrop) with T3 and T4 compression fracture.

A, Lateral thoracic spine x-ray. B, Anterior–posterior thoracic spine x-ray. C, D, Close-ups from A and B, respectively.

On a lateral view, each vertebral body should appear nearly rectangular, with similar height along its anterior and posterior borders. Anterior loss of height or “wedging” is common with compression fractures. Each body should appear similar in height to adjacent vertebra. The alignment of vertebral bodies should be assessed and should follow a slight thoracic kyphosis. Inspect carefully for any subluxation indicating injury. On the AP view, each vertebral body has two visible pedicles laterally (resembling the eyes of an owl) and a posterior spinous process in the midline (resembling an owl’s beak). Fractures, dislocations, or malignant lytic lesions can disrupt this normal symmetrical radiographic anatomy. The alignment of vertebral bodies should be smooth, and the height of each body should be similar to that of the adjacent vertebrae. Fractures in the high thoracic spine can be extremely difficult to detect on x-ray because of the overlap with other thoracic structures, including scapula, ribs, soft tissues of the chest, and upper extremities. CT is more sensitive and is indicated when significant injuries are suspected. Low energy compression fractures such as from falls from standing may not require CT confirmation.

Figure 3-87 T2 corner avulsion fracture (extension teardrop).

A, Sagittal CT image of the cervical spine in the same patient as in Figure 3-86 revealed anterior corner fractures of T2 and T3 (incompletely seen at the lower margin of the cervical spine CT). B, Close-up. He has previously undergone anterior fusion of C6, C7, and T1, perhaps contributing to this injury. The typical findings are triangular (teardrop) avulsion fragments, usually described as being due to the anterior longitudinal spinal ligament pulling away bone at its attachment points during hyperextension of the cervical spine. Whether this is the true mechanism is uncertain—in this patient, hyperflexion appears more likely, as the subsequent figures show the affected vertebrae to have findings of compression. More frequently, corner avulsion injuries occur in the cervical spine and result in teardrop fragments at the inferior corner of the vertebral body, rather than at the superior margin, as in this patient. Additional fractures are visible, including a T3 facet fracture and C7 lamina fracture. These are discussed in detail in other figures.

Figure 3-88 Thoracolumbar spine compression fracture.

A, This patient has an L4 compression fracture, seen well on this lateral x-ray. B, Close-up.

The anterior–posterior x-ray is less revealing (Figure 3-89). Computed tomography (CT) (Figures 3-90 through 3-92) revealed an unstable fracture. Note the normal L3 vertebra.

On the lateral x-ray (see Figure 3-86), the normal thoracic spine has a slight kyphosis, which should be a gradual and continuous curve. If the anterior margins of adjacent vertebral bodies do not align, subluxation is present; fractures may also be present. The upper thoracic spine is usually difficult to see on the lateral x-ray due to overlying shoulders. Swimmer’s views can assist in revealing upper thoracic spine pathology. The spinal level involved can be determined by counting up from the 12th rib on the AP projection. Children often have a normal variant that may be mistaken for an avulsion fracture on the lateral view. These are apophyses, small densities on the anterior superior and anterior inferior margins of adjacent vertebral bodies, which have the same appearance as anterior avulsion fractures in adults. Wedge compression fractures of the thoracic spine are common and result in decrease in the height of the anterior portion of a vertebral body relative to the posterior portion. Burst fractures can result in retropulsion of fragments into the spinal canal, and fragments may be seen on the lateral view posterior to the vertebral bodies.

The normal lumbar spine has a slight lordosis (see Figure 3-89), which should be smooth and continuous. Other features of the lumbar spine are generally similar to those described earlier regarding the thoracic spine.

Figure 3-89 Thoracolumbar spine compression fracture.

Same patient as in Figure 3-88. This patient has an L4 compression fracture, seen well on lateral x-ray (B). The anterior–posterior x-ray is less revealing (A). CT (Figures 3-90 through 3-92) revealed an unstable fracture.

What Findings Suggest an Unstable Thoracolumbar Injury? What Is the Three-Column Concept?

Radiologists and spine surgeons (orthopedists and neurosurgeons) use a three-column concept to describe thoracolumbar spine injuries and to predict their stability. Although described first for the thoracic and lumbar spine, the three-column system can also be applied to cervical spine injuries. For emergency physicians, it may often be sufficient to recognize that an injury is present. However, understanding the potential for an unstable fracture can be helpful in protecting the patient from neurologic injury, in ordering follow-up studies, and in engaging a consultant. Denis described a three-column system of classification in 1983, based on retrospective review of 412 thoracolumbar injuries (see Figure 3-85). Previously, an anterior column (consisting of the anterior longitudinal ligament, marked on x-ray by the anterior borders of vertebral bodies) and a posterior column (consisting of posterior spinal ligaments) had been described. Denis⁵⁹ added a middle column, formed by the posterior border of the vertebral body, the posterior longitudinal ligament, and the posterior annulus fibrosis of the intervertebral disc. The same year, a CT description of the middle column was published, and CT findings have become the basis for operative treatment decisions, though these are outside the scope of this text.⁶⁰

When two of three columns are disrupted, the spine is considered unstable. Whether reviewing x-ray or CT images, inspect the three columns for fractures or malalignment. Injuries to more than one column should be considered unstable; however, x-ray is relatively insensitive for thoracolumbar fracture (see later discussion) and spinal precautions should be maintained even when only one column appears injured on x-ray. CT should be obtained for further evaluation. Injuries to more than one column usually involve failure of the middle and anterior columns, the middle and posterior columns, or all three columns.⁶¹ Injuries to anterior and posterior columns sparing the middle column are rare. Injuries to the anterior and posterior columns are often obvious. Middle-column injuries can be more subtle, but findings include widening of the pedicles, loss of posterior wall height greater than 25%, or obvious fracture of the cortex of the posterior vertebral body. For subluxation injuries without apparent fracture, translation greater than 3.5 mm or angulation greater than 11 degrees suggests column failure due to ligamentous injury.⁶¹

How Is Computed Tomography of the Thoracolumbar Spine Interpreted?

Interpretation of CT of the thoracolumbar spine follows the principles we described for cervical spine CT. Bone windows should be inspected in sagittal, coronal, and axial planes (see Table 3-2). Inspection of the sagittal images is extremely helpful for detection of anterior–posterior subluxation, as well as loss of height from compression fractures. The sagittal and axial images are useful for identifying retropulsion of fracture fragments into the spinal canal (see Figures 3-86 through 3-101). The coronal images are useful for assessing loss of height of vertebral bodies and disc spaces, lateral subluxation injuries, burst fractures with spread of fragments, and transverse process fractures. The sagittal images are particularly useful for assessing the three columns described earlier. All three series should be inspected to maximize detection of injuries, because fractures parallel to a given image plane will not be seen in that plane but are readily identified in images reconstructed in perpendicular planes. Anterior–posterior subluxation is best seen in sagittal images, whereas lateral subluxation is best seen in coronal views.

Figure 3-90 Thoracolumbar spine compression–burst fracture.

Same patient as in Figures 3-88 and 3-89. A, Sagittal CT shows L4 compression fracture, with loss of height of L4 vertebral body and slight radial spread of fragments. B, Close-up.

The anterior margin of L4 is not aligned with the bodies above and below. The posterior margin is displaced slightly into the spinal canal.

Figure 3-91 Thoracolumbar spine burst–compression fracture.

Same patient as in Figures 3-88 through 3-90. Axial computed tomography views of the same lumbar (L4) burst fracture. Note the radial dispersion of the fracture fragments. A to C are arranged from caudad to cephalad. The central void in C is due to the cupped cephalad surface of the vertebral body (see diagram). The center of the vertebral body lies below the plane of C, while the raised rim lies in the plane of C. This is an unstable fracture because it involves all three columns of the spine.

Figure 3-92 Thoracolumbar spine burst–compression fracture.

Same patient as in Figures 3-88 through 3-91. These coronal CT views demonstrate the significant loss of height and comminution of L4.

Figure 3-93 Thoracolumbar spine trauma: L5 burst fracture with cord compression.

This 39-year-old female was ejected from a motor vehicle and sustained an L5 burst fracture. Due to retropulsion of bone fragments into the spinal canal, she had no motor function and minimal sensation after this injury.

A, Sagittal CT view showing significant narrowing of the spinal canal at L5 due to retropulsion of fragments of the burst L5 vertebral body. B, Close-up. The L5 vertebral body is also shortened substantially compared with L3 and L4 due to a compression injury mechanism. Compare with the axial and coronal CT images in Figures 3-94 and 3-95.

Figure 3-94 Thoracolumbar spine trauma: L5 burst fracture with cord compression.

Same patient as in Figure 3-93. Axial CT views (A, cephalad portion of vertebral body, B, middle of vertebral body, C, caudad portion of vertebral body) show significant narrowing of the spinal canal at L5 due to retropulsion of fragments of the burst L5 vertebral body. D, The normal L4 vertebra for comparison.

Figure 3-95 Thoracolumbar spine trauma: L5 burst fracture with cord compression.

Same patient as in Figures 3-93 and 3-94. A, Coronal CT view shows radial spread of fracture fragments and loss of height of the L5 vertebral body. B, Close-up.

Figure 3-96 T2, T3, and T4 compression fractures.

In this patient, thoracic spine CT shows spinal fixation screws at C6, C7, and T1, which provide convenient landmarks. A to C, Sagittal CT images (C, close-up from B) reveal significant loss of height of T3, consistent with a compression-type fracture. T4 also has loss of height and evidence of an anterior wedge compression fracture. T4 nicely demonstrates the “wedge” shape typical of flexion injuries of the thoracic and lumbar spine, with the anterior height being significantly less than the posterior height. T3, in comparison, has more uniform loss of height anteriorly and posteriorly. Compression fractures often lead to radial spread of fracture fragments, and retropulsion of fragments into the spinal canal can occur, leading to spinal cord injury. The spinal canal in this patient is preserved, with no retropulsion of fracture fragments.

Figure 3-97 T2, T3, and T4 compression fractures.

Same patient as in Figure 3-96. On these sequential axial CT images, the extent of comminution of the fractures becomes more apparent. The T3 vertebral body is seen with comminuted fragments of the anterior vertebral body.

Figure 3-98 T4 wedge compression fracture.

Same patient as in Figures 3-96 and 3-97. This wedge compression fracture of T4 extends through the right pedicle and base of the transverse process, but the spinal canal is not compromised. A to C, Sequential axial CT images. D, Close-up from B.

Figure 3-99 Thoracolumbar spine trauma: Chance fracture.

A, Sagittal computed tomography image. B, C, Close-ups from A.

A Chance fracture is a flexion injury, usually of the lower thoracic or upper lumbar spine. Unlike simple compression fractures, a Chance fracture involves fractures through the anterior, middle, and posterior columns of the spine. With lumbar Chance fractures, intraabdominal injuries are commonly found. A common mechanism is hyperflexion in a patient wearing a lap belt without a shoulder harness in a motor vehicle collision.

This patient has a multilevel thoracic spine flexion injury, with wedge compression fractures of T7 and T8, as well as a fracture of T10. The T8 injury follows the Chance pattern, with a comminuted fracture through the pars, lamina, and spinous process and extending through the vertebral body, with an obliquely oriented fracture through the posterior third of the middle column extending to the inferior endplate. The posterior elements are splayed with 6 mm of craniocaudal distraction. This results in focal kyphosis at T8 and approximately 33% height loss of the anterior vertebral body.

This fracture is explored in more detail in Figures 3-100 and 3-101.

Figure 3-100 Thoracolumbar spine trauma: T8 Chance fracture.

Same patient as Figure 3-99. Axial computed tomography (CT) reconstructions of the spine (A to F), performed from a dataset from CT of the chest, abdomen, and pelvis. The T8 vertebra shows a Chance fracture pattern, with injuries to the vertebral body including the lamina, pars interarticularis (a region at the junction of the pedicle and lamina), and vertebral body. The anterior body of T8 shows compression fracture findings.

Figure 3-101 Thoracolumbar spine trauma: T8 Chance fracture.

Same patient as Figures 3-99 and 3-100. These coronal reconstructions were created from the dataset from the computed tomography (CT) of the chest, abdomen, and pelvis performed for evaluation of visceral injury in this patient. They reveal a T8 Chance fracture, as well as injuries of T7 and T10. T8 and T7 have not only anterior wedge compression (seen best on the sagittal reconstructions in Figure 3-99) but also left-sided compression, creating a scoliosis toward the patient’s left side. Spinal fractures often have complex patterns that do not strictly adhere to a single textbook description. A, Coronal view through the mid portion of the thoracic vertebral bodies. B, A more posterior coronal plane than A, showing the pars interarticularis. C and D, Close-ups from A and B, respectively.

Figure 3-102 Thoracic spine metastases with cord compression.

CT with IV contrast was performed due to concern for malignancy. In comparison, dedicated spinal CT for evaluation of trauma can be performed without IV contrast. In practice today, multisystem trauma patients usually have spinal CT images reconstructed from IV contrast enhanced CT datasets of the chest, abdomen and pelvis. IV contrast in that scenario is used to evaluate solid organ and vascular injuries, not spinal injuries. This patient has a right lung mass lesion that has now invaded the T2 thoracic vertebral body. The mass has eroded the vertebral body and extends posteriorly into the paravertebral musculature. The images show the same slice on soft tissue (A) and bone windows (B). A close-up shows erosion of the right pedicle of the vertebral body (C).

Magnetic resonance imaging (MRI) demonstrates this lesion in greater detail in Figure 3-103.

Figure 3-103 Thoracic spine metastases with cord compression.

Same patient as in Figure 3-102. The intervertebral discs show abnormal alignment because of collapse of the vertebral body. A to C, The sagittal T2-weighted images (not to be confused with the T2 vertebral body) show a lesion obliterating the cerebrospinal fluid (CSF) space surrounding the spinal cord. With T2-weighted magnetic resonance imaging (MRI), fluid appears white and fat appears darker. The spinal cord appears gray due to its myelin (fat) content. The vertebral bodies are visible due to lipid-rich marrow. The bone of the vertebral bodies is black due to a lack of resonating protons to provide a radio signal. Computed tomography (CT) is generally better than MRI for evaluation of calcified bony pathology.

Figure 3-104 Thoracic spine metastasis with cord compression.

A, In this T2-weighted axial magnetic resonance image from the same patient as Figures 3-102 and 3-103, a soft-tissue mass is seen involving the T2 vertebral body. B, Close-up. The spinal cord is visible, surrounded by cerebrospinal fluid (CSF). The cord is not impinged upon in this image.

Figure 3-105 Osteomyelitis and discitis of L3 and L4.

A, T2-weighted fast spin–echo magnetic resonance imaging without contrast demonstrates typical findings of vertebral osteomyelitis and discitis. B, Close-up. The inferior endplate of L3 and the superior endplate of L4 are abnormal and ill-defined. The L3 and L4 vertebral bodies show a high T2 signal, indicating an abnormally high fluid content. In comparison, L2 shows a normal marrow signal and appears dark gray due to a normal fat signal. CSF, cerebrospinal fluid.

Figure 3-106 Discitis of L3 and L4.

Same patient as in Figure 3-105. T2-weighted fast spin–echo magnetic resonance imaging without contrast demonstrates typical findings of vertebral discitis. A, The L3-4 disc is shown and has a high T2 signal, indicating elevated fluid content consistent with an inflammatory process. B, The normal L2-3 disc in the same patient is shown, with its normal dark appearance (a low T2 signal) due to low fluid content. A, B, A high T2 signal is seen in the paraspinal muscles, suggesting inflammation.

Figure 3-107 Osteomyelitis of L3 and L4.

Same patient as in Figures 3-105 and 3-106. T2-weighted fast spin–echo magnetic resonance imaging without contrast demonstrates typical findings of vertebral osteomyelitis. A, The L4 vertebral body is shown and has a high T2 signal, indicating elevated fluid content consistent with an inflammatory process. B, The normal L2 vertebral body in the same patient is shown, with its normal dark appearance (a low T2 signal) due to low fluid content. A, B, A high T2 signal is seen in the paraspinal muscles, suggesting inflammation. Compare Figure 3-107B and Figure 3-106B. Normal intervertebral discs are darker in appearance than normal vertebral bodies on this MRI sequence.

Figure 3-108 Spinal epidural abscess.

This 67-year-old female presented with delirium and fever. Three months prior, she had undergone lumbar laminectomy, and her wound had been treated with a wound VAC dressing. Magnetic resonance imaging was not initially available, so noncontrast computed tomography (CT) was performed. Noncontrast CT is excellent at delineating air, which appears black on bone windows.

A, The midsagittal view demonstrates air (black) in the spinal canal at the L2 and L3 levels. On the axial views (B, C), air is visible in the spinal canal, in paraspinal soft tissues, and within the vertebral body. These findings are concerning for a paraspinal infection that has developed into an epidural abscess with vertebral osteomyelitis. Compare with the MRI in Figure 3-109.

Figure 3-109 Spinal epidural abscess.

Same patient as in Figure 3-108, in whom noncontrast computed tomography showed air in the spinal canal, concerning for epidural abscess. Magnetic resonance imaging (MRI) of the lumbar spine without contrast was performed, as the patient was in acute renal failure.

A, This T2-weighted sagittal MR image provides useful information even without gadolinium contrast. B, Close-up.

On T2-weighted MRI sequences, fluid including cerebrospinal fluid appears white. Fat-containing tissues such as bone marrow and the spinal cord or cauda equina appear dark gray. Calcified bone appears nearly black due to an absence of resonating protons. Air appears completely black for the same reason.

The midline sagittal image shows the cauda equina to be impinged upon by an epidural fluid collection containing air—an epidural abscess. The dura mater is visible as a thin, dark gray line parallel to the spinal cord. It is indented in the region of the epidural abscess. Compare with the axial MRI images in Figure 3-110.

Figure 3-110 Spinal epidural abscess.

A, This axial image is taken at the same level as the air-fluid collection shown in the last figure. The air–fluid collection comprising the epidural abscess is seen posterior to the thecal sac (dura mater), which contains small, dark-gray circles representing a cross-section through the nerve roots of the cauda equina. These nerve roots are surrounded by white cerebrospinal fluid within the thecal sac. B, In another patient, the nerve roots of the cauda equina are more widely separated and are seen as discrete bundles with circular axial cross-sections.

Figure 3-111 Vertebral tuberculosis (Pott’s disease).

A, Anterior–posterior x-ray. B, Lateral x-ray. C, D, Close-ups from A and B, respectively.

This patient with known tuberculosis and medication noncompliance presented with worsening back pain and lower extremity weakness. Plain x-rays show extreme loss of height of the T9 vertebral body. The high-density material is cement from an earlier vertebroplasty intended to arrest loss of height of the vertebral body. Cultures from the vertebral body confirmed mycobacterium tuberculosis. Compare with the CT and MR images in Figures 3-112 through 3-116.

Figure 3-112 Vertebral tuberculosis (Pott’s disease).

Same patient as in Figure 3-111. Sagittal computed tomography reconstructions show extreme loss of height of the T9 vertebral body as a result of vertebral tuberculosis infection. Compression of the T9 body has resulted in kyphosis at this level. The high-density material is vertebroplasty cement from a prior procedure. A, A parasagittal section slightly to the right of midline. B, A more midline sagittal section. C, D, Close-ups from A and B, respectively.

Figure 3-113 Vertebral tuberculosis (Pott’s disease).

Same patient as in Figures 3-111 and 3-112. Two axial computed tomography reconstructions (A, B) show nearly complete erosion of the T9 vertebral body as a result of vertebral tuberculosis infection. The high-density material is vertebroplasty cement from a prior procedure. Residual bone shows a moth-eaten appearance.

Figure 3-114 Vertebral tuberculosis (Pott’s disease).

Same patient as in Figures 3-111 through 3-113. Two coronal computed tomography reconstructions (A, B) show nearly complete erosion of the T9 vertebral body as a result of vertebral tuberculosis infection. The high-density material is vertebroplasty cement from a prior procedure. Residual bone shows a moth-eaten appearance. C, D, Close-ups from A and B, respectively.

Figure 3-115 Vertebral tuberculosis (Pott’s disease) with epidural abscess.

A, Same patient as in Figures 3-111 through 3-114. Sagittal T2-weighted magnetic resonance reconstructions show nearly complete erosion of the T9 vertebral body as a result of vertebral tuberculosis infection. B, Close-up. The spinal cord is compressed from an epidural fluid collection. CSF, Cerebrospinal fluid.

Figure 3-116 Vertebral tuberculosis (Pott’s disease) with epidural abscess.

Same patient as in Figures 3-111 through 3-115. A, Axial T2-weighted magnetic resonance (MR) images show nearly complete erosion of the T9 vertebral body as a result of vertebral tuberculosis infection. The region normally occupied by the T9 vertebral body shows a high T2 signal (white), indicating the presence of abnormal fluid (which appears white on T2-weighted MRI). The cerebrospinal fluid (CSF) that normally surrounds the spinal cord at this level has been displaced. B, A normal adjacent vertebral body and spinal cord from the same patient for comparison. C, D, Close-ups from A and B, respectively.

Figure 3-117 Degenerative joint disease and lumbar disc herniation.

This patient has extensive degenerative joint disease of the lumbar spine. Note the normal spacing between the L1 and the L2 vertebral bodies and the loss of normal disc height at the L2-3, L3-4, and L4-5 levels. A, Lateral x-ray. B, Anterior–posterior x-ray. C, D, Close-ups from A and B, respectively. Compare with the MRI from the same patient in Figure 3-118 through 120.

Figure 3-118 Degenerative joint disease and lumbar disc herniation: Magnetic resonance without contrast, T2-weighted image.

Same patient in Figure 3-117. A, Sagittal reconstruction. B, Close-up. This patient has extensive degenerative joint disease of the lumbar spine. At the T12-L1 level, note the normal disc space. The spinal canal is patent, with cerebrospinal fluid (CSF) (white) surrounding the cord (gray). At the L2-3 level, the intervertebral disc is compressed and indents the thecal sac into the spinal canal, displacing the normal rim of CSF and resulting in severe spinal stenosis.

Figure 3-119 Normal cauda equina.

Same patient in Figures 3-117 and 118. A, This axial T2-weighted magnetic resonance image is taken through the L3-4 region. It illustrates that the spinal cord has given way to the cauda equina, with individual nerve roots visible in cross section (black circles). B, Close-up.

Incidentally, the distance from the skin surface to the thecal sac (cerebrospinal fluid space) is 50 mm, so lumbar puncture at this level would be feasible.

Figure 3-120 Disk herniation resulting in cauda equina compression: Magnetic resonance imaging (MRI) lumbar spine without contrast.

Same patient as Figure 3-119, but at a lower spinal level. A, This axial T2-weighted MRI image is taken through the L4-5 disc region. A paracentral disc herniation is compressing the thecal sac on the patient’s left, narrowing the left neural foramen. The spinal cord has already given way to the nerve roots of the cauda equina at this level. B, A normal section above the level of the disc herniation for comparison—note the triangular shape of the normal thecal sac at this level. C, Close-up from A.

Figure 3-121 Osteopetrosis with type II dens fracture.

A, Lateral x-ray. B, Anterior–posterior x-ray. C, Close-up from A.

Osteopetrosis is a pathologic condition of increased bone density but high predisposition to fracture. Note this patient’s high bone density, which is not simply a function of the exposure. This patient has a type II dens fracture. Compare with the CT in Figure 3-122.

Figure 3-122 Osteopetrosis with type II dens fracture.

Same patient as in Figure 3-121. A, Midsagittal computed tomography (CT) slice. B, Close-up. Note this patient’s high bone density. This is not simply a function of the window level, which is set to “bone,” as with most of the other CT images in this chapter. This patient has a type II dens fracture, with slight craniocaudal distraction of the fracture fragments. The dens is not retropulsed into the spinal canal.

Figure 3-123 Ankylosing spondylitis with type II dens fracture.

A, This lateral x-ray (B, broader view) shows typical features of ankylosing spondylitis with fusion of facet joints, straightening of the cervical spine, and diffuse osteopenia. A dens fracture is present with retropulsion of the dens into the spinal canal. The posterior longitudinal ligament line would intersect the dens rather than passing posterior to it. The patient’s computed tomography scan is explored in Figures 3-124 through 3-126.

Figure 3-124 Ankylosing spondylitis with type II dens fracture.

Same patient as in Figure 3-123. These sagittal computed tomography reconstructions show fusion of facets bilaterally (A, right, and C, left) and “bamboo” spine in the midline sagittal section (B). The spine shows diffuse osteopenia that is also typical of ankylosing spondylitis. The cervical spine is straightened, not due to the cervical collar but due to fusion of vertebrae. Worse still, the patient has a type II dens fracture. The dens and C1 vertebra are bound together by the transverse ligament and have moved posteriorly as a unit, intruding into the spinal canal and narrowing it by more than 50%.

Figure 3-125 Ankylosing spondylitis with type II dens fracture.

Same patient as in Figures 3-123 and 3-124. These axial computed tomography views may appear confusing at first. In the most cephalad slice (A), the dens is normally positioned relative to the ring of C1. This is because the dens has remained attached to the anterior ring of C1 by the transverse ligament. However, a few slices caudad (B), a cross section through the dens shows it to be in the center of the spinal canal of the C2 vertebra, not positioned anteriorly. Further caudad (C), the bottom of the dens fragment is still seen in the spinal canal. Bone in the spinal canal threatens the spinal cord (as seen in magnetic resonance imaging in Figure 3-127).

Figure 3-126 Ankylosing spondylitis with type II dens fracture.

Same patient as in Figures 3-123 through 3-125. This badly displaced type II dens fracture is depicted in coronal CT views. Because the dens has been displaced significantly from the body of C2, the two are not seen in the same coronal view. These two views are separated by 10 mm in an anterior–posterior direction. The dens remains in its normal position relative to C1. C, D, Close-ups from A and B, respectively.

Figure 3-127 Ankylosing spondylitis with type II dens fracture: Cervical magnetic resonance imaging (MRI) without contrast.

Same patient as in Figures 3-123 through 3-126. A, In this T2-weighted midsagittal MRI, a dens fracture with retropulsion of the dens into the spinal canal has compressed the spinal cord. B, Close-up.

Several features are evident. First, the thecal sac has been indented, displacing the rim of cerebrospinal fluid (CSF) that normally surrounds the spinal cord. In addition, the spinal cord shows an increased T2 signal, indicating traumatic edema of the cord. Compare this abnormal, whiter appearance of the cord with the normal, darker gray appearance of the cord above and below the level of injury.

Figure 3-128 Penetrating spinal trauma: Bullet in spinal canal.

Scout computed tomography (CT) images show a bullet overlying the L1-2 disc space on both frontal (A) and lateral (B) reconstructions. Multiplanar CT reconstructions of the spine were performed (Figures 3-129 through 3-131).

Figure 3-129 Penetrating spinal trauma: Bullet in spinal canal.

Same patient as in Figure 3-128. A, B, Axial computed tomography (CT) slices. C, D, Close-ups from A and B, respectively.

Axial CT reconstructions show a bullet located in the right lateral aspect of the disc space, neural foramina, and central canal at the L1-2 level. Metallic streak artifact limits detection of fractures in this CT. The patient is also morbidly obese, limiting the image quality.

Figure 3-130 Penetrating spinal trauma: Bullet in spinal canal.

Same patient as in Figures 3-128 and 3-129. A, B, Sagittal computed tomography (CT) slices. C, D, Close-ups from A and B, respectively.

Sagittal reconstructions show the bullet in the right neural foramen and central spinal canal. Streak artifact makes it impossible to discern the exact margins of the bullet.

Figure 3-131 Penetrating spinal trauma: Bullet in spinal canal.

Same patient as in Figures 3-128 through 3-130. These coronal CT reconstructions show the bullet to lie just inferior to L1, in the L1-2 interspace. Again, streak artifact makes finding the bullet margins and recognizing adjacent fractures impossible. A, Coronal view through the midportion of the affected vertebral bodies. B, A more posterior coronal plane shows posterior facets. C and D, Close-ups from A and B, respectively.

Figure 3-132 Epidural hematoma after lumbar puncture with motion artifact.

A, T2-weighted sagittal image. B, Close-up.

This patient underwent magnetic resonance imaging (MRI) to assess for epidural hematoma after a lumbar puncture. The patient had difficulty remaining still during the exam, which lasted approximately 45 minutes. As a result, significant motion artifact limits the diagnostic quality of the MRI.

Figure 3-133 Syrinx.

A, Sagittal T2-weighted magnetic resonance image. B, Axial T2-weighted image.

A syrinx occurs when spinal fluid dissects within the spinal cord. A risk factor is Chiari malformation, as this can increase the pressure of cerebrospinal fluid (CSF) within the spinal cord. In this 10-year-old female presenting with leg numbness, a dilated cervical central canal gave way to a small syrinx at the T7 level.

Figure 3-134 Pseudosubluxation.

A, Lateral CT scout image. B, Close-up.

This 3-year-old male presented with torticollis and was suspected of having a retropharyngeal abscess. The computed tomography (CT) was normal, but it demonstrates some important pediatric findings that can be mistaken for pathology by an inexperienced reader, especially one more accustomed to adult CT. The scout CT shows pseudosubluxation of C1 on C2.

Figure 3-135 Pseudosubluxation.

Same patient as in Figure 3-134. A, Sagittal computed tomography (CT) slice. B, Close-up.

This 3-year-old male wmistaken for pathology by an inexperienced reader. The midsagittal view shows a slightly widened predental space (normal for age) and incomplete fusion of the dens to the body of C2 (normal for age), which should not be confused with type II dens fracture.

Who Needs Thoracic and Lumbar Imaging? Is There a Clinical Decision Rule for These Regions?

Unlike the case with injuries to the cervical spine, no single large study has produced a well-validated CDR for imaging of the thoracic and lumbar spine that both detects all injuries and avoids unnecessary imaging. First, we lay out the best-evidence indications for imaging the thoracic and lumbar spine. Then, we examine the evidence behind these recommendations.

The ACR recommends thoracic and lumbar spine imaging for any of the following after major blunt trauma^1,62-63:

Notice the similarity to NEXUS—the items are identical, with the addition of local examination findings and coexisting cervical spine fracture.

Several studies have contributed to these recommendations. A prospective study of 2404 patients evaluated seven high-risk criteria as a screening tool for identification of patients at risk for thoracolumbar injury.⁶⁴ The criteria included the five NEXUS criteria, with back pain and “severe mechanism of injury” as additional potential predictors of thoracolumbar fracture (Box 3-12). The authors also reported the predictive value of the presence of a cervical spine injury for detecting other spinal injuries, although this is not necessarily a data point available to the emergency physician at the initial assessment when many diagnostic imaging plans are formulated. The study identified 152 patients with spinal injuries. Use of the first six high-risk criteria resulted in 100% sensitivity (95% CI = 98%-100%) and 100% negative predictive value (95% CI = 97%-100%). Unfortunately, this rule is quite nonspecific (3.9%, 95% CI = 3.1%-4.8%) and has poor positive predictive value (6.6%, 95% CI = 97%-100%). As a result, application of this rule, while detecting all injuries, would result in imaging of virtually all major blunt trauma patients. Addition of “severe mechanism of injury” did not improve sensitivity (already 100%) but worsened specificity to only 2% (95% CI = 1.5%-2.7%). Of note, each of the first six high-risk predictors except intoxication was present as the sole predictor of injury in some cases—meaning that a simpler rule with fewer elements would miss injuries. Eliminating intoxication as a criterion would slightly improve specificity to 4.6% (95% CI = 3.7%-5.5%). The authors did calculate the sensitivity and specificity

Box 3-12 High-Risk Criteria for Thoracolumbar Injury^∗

^∗ Studies suggest that the presence of even a single criterion indicates imaging. The first 6 criteria above are 100% sensitive, without inclusion of severe mechanism.

• Complaints of thoracolumbar pain

• Thoracolumbar spine tenderness on midline palpation

• Decreased level of consciousness

• Abnormal peripheral neurologic examination

• Distracting painful injury

• Evidence of intoxication with ethanol or drugs

• Severe mechanism of injury

Car collision with speed > 45 mph or rollover

Car collision requiring extrication

Ejection from a motor vehicle

Motorcycle accident at speed > 20 mph

Automobile versus pedestrian at speed > 5 mph

Fall greater than 10 feet

Cervical spine injury

From Holmes JF, Panacek EA, Miller PQ, et al: Prospective evaluation of criteria for obtaining thoracolumbar radiographs in trauma patients, J Emerg Med 24(1): 1–7, 2003.

of other combinations of the high-risk elements, but minimal improvements in specificity were achieved at the price of loss of sensitivity. The presence of a cervical spine fracture was noted to have a stronger predictive value than any of the other risk factors assessed, increasing the risk of thoracolumbar injury by threefold when present. The authors recommended adding this to a CDR to enhance predictive value.

The same research group went on to delineate the types of distracting injuries most associated with thoracolumbar fractures. When “painful distracting injury” was the only high-risk predictor present, bony fractures were associated with thoracolumbar injury, while soft-tissue contusions, lacerations, head injuries, abrasions, visceral injuries, dental injuries, burns, ligamentous injuries, amputations, and compartment syndrome were not. Although the investigators suggest that this data can be used to refine the screening definition of “painful distracting injury,” only 13 thoracolumbar injuries were found in this group, limiting the ability to consider injury subtypes.⁶⁵

Hsu et al.⁶² conducted a retrospective chart review of 200 patients and reported the sensitivity of various physical exam findings for thoracolumbar fracture. In this study, a midline step-off was found to be poorly sensitive (13.8%) but highly specific (100%). Back bruising was only 6.9% sensitive but 98.6% specific. Back pain or midline tenderness had a sensitivity of 62.1% and a specificity of 91.5%. An abnormal neurologic examination was 41.4% sensitive and 95.8% specific.⁶² This study is limited by its small size, resulting in wide CIs. In addition, the study faces significant methodologic problems. Physical examination findings were abstracted from the emergency department chart, and it is possible that examination findings may have been present but not written in the chart, resulting in inaccurate calculations of sensitivity and specificity. However, the apparent specificity of local findings such as bruising or step-offs has led to the recommendation by professional societies that these be criteria for thoracolumbar imaging.^1,63

The case is less clear for low-energy mechanisms of trauma—for example, is any imaging required in the 30-year-old patient who complains of back pain after low-speed rear-end collision but is ambulatory in the emergency department? Thoracic and lumbar injuries are less common after minor trauma mechanisms, and many patients may require no imaging. Spectrum bias in studies such as those described earlier (biased toward moderately severely injured patients) may limit the application of these rules to less significantly injured patients.

What Is the Sensitivity of X-ray and Computed Tomography for Injuries of the Thoracolumbar Spine? Which Imaging Modality Should Be Used?

Studies comparing x-ray and CT for the diagnosis of thoracolumbar fracture suffer the same methodologic flaws as those for cervical spine injury, including the lack of a uniform and independent gold standard. The use of CT as the gold standard in studies is called incorporation bias and virtually guarantees that CT will appear to be the superior test. That said, CT is almost certainly far superior to x-ray for detection and characterization of fractures. Rhee et al.⁶⁶ reported the missed lumbar fracture rate for abdominal and pelvic CT to be 23.1%, versus 12.7% for AP and lateral x-ray. However, these figures are based on axial images from abdominal and pelvic CT viewed on bone windows without sagittal or coronal reformatted images, using a single-slice helical scan with slice thicknesses of 5 to 10 mm. Modern protocols routinely use thinner slices and incorporate multiplanar reformations, improving recognition of malalignment and fractures that may lie in the axial plane. In addition, this study had no uniform diagnostic standard, making the results uncertain. Multiple other studies have shown excellent test performance characteristics for images obtained from abdominal and pelvic CT, reformatted in the sagittal and coronal planes.^67-77 Some reports focused on use of axial CT images, supplemented by AP and lateral scout CT images. These were found to have sensitivity of 92% to 100% and specificity of 100% compared with AP and lateral lumbar spine x-ray.^68,78 A prospective study compared axial images from the chest–abdomen–pelvis CT with AP and lateral x-rays. The gold standard for the study was either separately acquired, dedicated thin-cut spine CT (1- to 2-mm cuts) or clinical exam of the patient once stable and awake. Axial images from the chest–abdomen–pelvis CT were found to have a sensitivity of 97% (95% CI = 86%-100%), with a specificity of 99% (95% CI = 96%-100%). In comparison, x-ray had a sensitivity of only 58% (95% CI = 41%-75%) and a specificity of 93% (95% CI = 89%-97%). From a methodology standpoint, this study suffers from its inconsistent gold standard (also called verification or workup bias) and a lack of blinding, as radiologists were aware of the results of other imaging studies when rendering their interpretation of the CT and x-rays.⁶⁹

Sheridan et al.⁷⁵ prospectively investigated reformatted sagittal and coronal images obtained from chest–abdomen–pelvis CT–the technique used in most trauma centers today and advocated by the ACR. They reported 97% sensitivity and 95% specificity of CT for lumbar fractures, compared with 62% sensitivity and 86% specificity of x-ray. This study is limited by the lack of a uniform diagnostic standard; the final discharge diagnosis was considered to be correct, though not all patients underwent the same final workup. In addition, although the authors report that the CT and lumbar spine x-rays were generally read blinded, they admit that some were read by the same radiologist without blinding. This practice likely artificially inflates the sensitivity of x-ray, because findings on CT might lead to recognition of additional fractures. Theoretically, x-ray findings might also have led to recognition of CT abnormalities.⁷⁵

Wintermark et al.⁷⁷ performed a prospective double-blinded study of 100 consecutive trauma patients undergoing both thoracolumbar x-ray and CT of the chest, abdomen, and pelvis with coronal and sagittal reformats. They reported the mean sensitivity for unstable fractures to be 97% for CT (95% CI = 85.5%-99.9%) and the mean sensitivity to be 33.3% for x-ray (95% CI = 21.7%-46.7%). In addition, the K value for interobserver agreement was 0.951 for CT, indicating near-perfect agreement of different interpreting radiologists, compared with poor agreement (K = 0.368) for x-ray.⁷⁷ For all fractures, including fractures predicted to be stable, agreement (K = 0.787), and sensitivity (78.1%) of CT were less perfect although arguably, stable fractures are less important to patient outcome. CT nonetheless was superior to x-ray in this category as well (sensitivity = 32%, K = 0.661). This study was well done, with its primary weakness being a mixed gold standard based on additional imaging, orthopedic interventions, and final discharge diagnosis or autopsy report.⁷⁷

Who Needs Magnetic Resonance Imaging of the Thoracic and Lumbar Spine?

The ACR does not recommend MRI for evaluation of ligamentous injuries of the thoracolumbar spine when CT is normal. Unlike the case for the cervical spine, in which some controversy exists and MRI is commonly performed for evaluation of ligamentous injuries even in the presence of a normal CT scan (see earlier discussion), ligamentous injuries of the thoracolumbar spine almost never occur in the absence of CT findings. MRI is recommended solely for patients with neurologic deficits localizing to the thoracolumbar spine.¹

Which Is Faster, X-ray or CT of the Thoracic and Lumbar Spine?

Wintermark et al.⁷⁷ reported thoracolumbar x-ray required a median of 23 minutes, while CT of the chest, abdomen, and pelvis required a median of 40 minutes, of which 7 minutes was required for postprocessing of images. In patients undergoing CT of the chest, abdomen, and pelvis for evaluation of other traumatic injuries, CT reformations are clearly the more rapid means of assessing the thoracolumbar spine.

What Is the Cost of X-ray and CT of the Thoracic and Lumbar Spine?

Wintermark et al.⁷⁷ reported costs of $145 per patient for x-ray and $880 for CT. They using reformats of the spine derived from chest–abdomen–pelvis CT. They discounted the cost of the CT spinal reformats, stating that these added nothing to the cost of CT. However, in some cases, additional fees for postprocessing may be charged to the patient. If the patient does not require chest, abdomen, and pelvis CT for evaluation of other injuries and dedicated spine CT is performed, CT is more costly, accounting for direct costs alone. Given the higher sensitivity of CT, economic models incorporating the cost of missed injuries would likely favor CT.

What Is the Radiation Exposure of Thoracolumbar X-ray and CT?

Wintermark et al.⁷⁷ reported mean effective radiation dose from thoracolumbar x-ray to be 6.36 mSv, compared with 19.42 mSv for CT reformations of the spine derived from chest–abdomen–pelvis CT. They note that the reformations do not require any additional imaging or radiation exposure, compared with CT of the chest, abdomen, and pelvis alone. Consequently, if the patient requires CT of these regions for evaluation of other traumatic injuries, use of CT reformations allows a reduction of 6.36 mSv (by eliminating the additional radiation exposure from x-ray), nearly 25% of the entire dose required to perform both abdominal CT and x-rays of the spine.⁷⁷ However, if the patient does not require visceral CT of the chest, abdomen, and pelvis for trauma evaluation, dedicated CT of the spine would result in a significant increase in radiation exposure compared with x-ray.

Should Minor Trauma Patients Undergo Lumbar CT or X-ray?

Most of the published studies comparing CT and x-ray involve patients with high trauma mechanisms and significant associated injuries. In these patients, CT is more sensitive, faster, more cost effective, and more radiation efficient than x-ray, because CT images of the spine can be reformatted from CT acquired for evaluation of other injuries to the chest, abdomen, and pelvis. Less clear is the best workup for patients with more minor trauma mechanisms. Some of these patients likely do not require CT evaluation of the chest, abdomen, and pelvis. For example, many blunt trauma patients with relatively low-energy mechanisms are probably not at risk for aortic trauma, a high-energy injury that is the strongest indication for chest CT. Despite this, CT is often ordered in this circumstance because of its value in imaging the thoracolumbar spine. Probably, patients with a low pretest probability of both major torso organ injury and thoracolumbar spinal injury should undergo thoracolumbar x-ray instead of CT. The ACR rates x-ray as less appropriate (three on a nine-point scale) than CT (nine on the same scale) for “blunt trauma meeting criteria for thoracic or lumbar imaging.”¹ However, given the lower incidence of thoracolumbar injuries in very-low-energy trauma, such as falls from standing, CT appears inappropriate given its cost and radiation exposure.¹ If evaluation of the chest or abdomen is anticipated, it is clearly more efficient to order chest–abdomen–pelvis CT with spine reformats as the initial imaging study, rather than potentially subjecting the patient to two CTs: a dedicated spine CT followed later by CT of the chest, abdomen, and pelvis.

Which Patients Require Thoracolumbar Imaging for Low Back Pain? Are So-Called Red Flags Useful?

Red flags, or clinical features suggesting serious causes of back pain, are frequently cited in medical texts as indications for imaging. A number of national guidelines have been published with recommendations for imaging only when red flags are present. Are these based on expert opinion or on high-quality research? A systematic review of Medline, the Cumulative Index to Nursing and Allied Health Literature, and Embase found 12 studies meeting sound methodologic quality criteria and investigating 51 clinical “red flags.” Five clinical features were identified as useful in raising or lowering the probability of vertebral fracture (Table 3-5).⁷⁹ Likelihood ratios for each finding are presented. A likelihood ratio represents a multiplier used to modify pretest probability of disease, based on a “test,” which in this case may be a clinical exam maneuver or historical feature. Of note, a positive likelihood ratio greater than or equal to 10, or a negative likelihood ratio less than or equal to 0.1, is generally accepted as having substantial value in clinical practice. Less extreme likelihood ratio values do not alter pretest probability to an extent that is likely to assist in clinical decisions. Even the five “best” red flags identified in this review generally fall short of these criteria. The authors intentionally set a lower threshold, choosing features for which the upper 95% CI of the likelihood ratio was substantially below 1.0 or for which the lower 95% CI of the likelihood ratio was substantially above 1.0. This is because a likelihood ratio approaching 1.0 results in no mathematical change in pretest probability.

TABLE 3-5 Red Flags for Lumbar Vertebral Fracture