17
Imaging of the Spine in Victims of Trauma
C. Craig Blackmore and Gregory David Avey
Issues of Imaging of the Cervical Spine
I. Who should undergo cervical spine imaging?
A. NEXUS prediction rule
B. Canadian cervical spine prediction rule C. Applicability to children
II. What cervical spine imaging is appropriate in high-risk patients?
A. Cost-effectiveness analysis
III. Special case: defining patients at high fracture risk A. Applicability to children
IV. Special case: the unconscious patient
Issues of Imaging of the Thoracolumbar Spine
V. Who should undergo thoracolumbar spine imaging?
A. Applicability to children
VI. Which thoracolumbar imaging is appropriate in blunt trauma patients?
319
䊏
Cervical spine imaging is not necessary in subjects with all five of the following: (1) absence of posterior midline tenderness, (2) absence of focal neurologic deficit, (3) normal level of alertness, (4) no evidence of intoxication, and (5) absence of painful distracting injury (strong evidence).
䊏
Computed tomography (CT) scan of the cervical spine is cost- effective as the initial imaging strategy in patients at high probability of fracture (neurologic deficit, head injury, high energy mechanism) who are already to undergo head CT (moderate evidence).
Issues
Key Points
䊏
No adequate data exist on the appropriate cervical spine evaluation in subjects who cannot be examined due to a head injury (insufficient evidence).
䊏
Imaging of the thoracolumbar spine is not necessary in blunt trauma patients with all five of the following: (1) absence of thoracolumbar back pain, (2) absence of thoracolumbar spine tenderness on midline palpation, (3) normal level of alertness, (4) absence of distracting injury, and (5) no evidence of intoxication (moderate evidence).
Definition and Pathophysiology
The majority of spine fractures occur from high-energy trauma such as high-speed motor vehicle accidents and falls from heights (1,2). However, an important minority occur from relatively low-energy mechanisms such as falls from a standing height or low-velocity automobile accidents (3,4).
Epidemiology
Cervical spine fractures occur in approximately 10,000 individuals per year in the United States, most the result of blunt trauma (5,6). Among patients with a fracture, approximately one third will sustain severe neurologic injury (6,7). Unfortunately, fractures of the cervical spine may not be clinically obvious. Patients may be neurologically intact initially, but if not treated appropriately and promptly, progress to severe neurologic compromise (8). Delayed onset of paralysis occurs in up to 15% of missed fractures, and death due to unidentified cervical spine fracture is possible (9,10). Furthermore, the mechanism of injury is also not always useful for excluding cervical spine fracture.
Thoracolumbar spine injury has been estimated to occur in between 2%
and 4% of all blunt trauma patients (11,12). These injuries were judged to require treatment in approximately three fourths of those identified (13).
Much like cervical spine fractures, a resulting neurologic deficit is noted in approximately one third of those with thoracolumbar injury (14,15). Given the potentially serious consequences of these injuries, it is unsettling to find that studies have noted a significant delay in diagnosis in 11% to 22% of patients with spine fractures (9,16,17).
Overall Cost to Society
There is enormous variability in the practice of cervical spine imaging
(18,19), but in most centers, imaging is used liberally. As a result, the yield
from cervical spine imaging is low, with only 0.9% to 2.8% of such imaging
studies demonstrating injury (20,21). Overall, the total cost of the imaging,
evaluation, and care of patients with cervical spine trauma in the United
States is an estimated $3.4 billion per year (22). The yield of thoracolum-
bar imaging is somewhat higher than cervical spine imaging, with posi-
tive studies accounting for 7.6% to 9% of blunt trauma thoracolumbar
exams (23). The total societal cost of thoracolumbar spine injury has been
estimated at $1 billion per year (24).
Goals
The overall goal of initial spine imaging is to detect potentially unstable fractures to enable immobilization or stabilization and prevent develop- ment or progression of neurologic injury. Additional imaging studies may be performed to inform prognosis and guide surgical intervention for unstable injuries.
Methodology
A Medline search was performed using PubMed (National Library of Medicine, Bethesda, Maryland) for original research publications dis- cussing the diagnostic performance and effectiveness of imaging strategies in the cervical and thoracolumbar spine. Clinical predictors of cervical and thoracolumbar spine fracture were also included in the literature search.
The search for cervical spine–related publications covered the period 1966 to March 2002. The search strategy employed different combinations of the following terms: (1) cervical spine, (2) radiography or imaging or computed tomography, and (3) fracture or injury. The search for thoracolumbar spine–
related publications covered the period 1980 to March 2004. The search strategy included the MESH headings (1) spine and diagnosis, and (2) imaging and trauma. Additional articles were identified by reviewing the reference lists of relevant papers. This review was limited to human studies and the English-language literature. The authors performed an initial review of the titles and abstracts of the identified articles followed by review of the full text in articles that were relevant.
I. Who Should Undergo Cervical Spine Imaging?
Summary of Evidence: Determination of which blunt trauma subjects should undergo cervical spine imaging, and which should not undergo imaging, is a question that has been studied in detail in literally tens of thousands of subjects. The two major level I (strong evidence) studies, the NEXUS trial (Table 17.1), and the Canadian C-spine rule (Table 17.2) were comprehensive multicenter investigations of this topic. The NEXUS rule (Table 17.1) has undergone extensive validation and demonstrates high sensitivity for detec- tion of fractures. The Canadian C-spine rule (Table 17-2) also has high sensitivity, and potentially higher specificity than the NEXUS. However, neither of these rules has been tested in an implementation trial to deter- mine their impact outside the research setting.
Table 17.1. NEXUS criteria: imaging of the cervi- cal spine is not necessary if all five of the NEXUS criteria are met
1. Absence of posterior midline tenderness 2. Absence of focal neurologic deficit 3. Normal level of alertness
4. No evidence of intoxication
5. Absence of painful distracting injury
Source: Adapted from Hoffman et al. (29).
Supporting Evidence: The low yield of cervical imaging has prompted a number of investigators to attempt to identify clinical factors that can be used to predict cervical spine fracture. Early studies of this question were largely level III (limited evidence) investigations consisting of unselected case series. For example, in 1988, Roberge and colleagues (25) studied 467 consecutive subjects who underwent cervical spine radiography and found that subjects with cervical discomfort or tenderness were more likely to have a fracture than those without such symptoms or signs. Additional investigators identified associations between cervical spine fracture and mechanism of injury (26,27), level of consciousness (20,21,27), and intoxi- cation (20,28). However, all of these investigations involved small numbers of subjects with fracture and a single or small number of centers.
A. NEXUS Prediction Rule
The first major cohort investigation of clinical indicators for cervical spine imaging was the National Emergency X-Radiography Utilization Study (NEXUS) (5,29). This was a large Level I study performed at 23 different emergency departments across the United States. The goal of the NEXUS study was to assess the validity of four predetermined clinical criteria for cervical spine injury (Table 17.1). These criteria were (1) altered neurologic function, (2) intoxication, (3) midline posterior bony cervical spine tender- ness, and (4) distracting injury. The NEXUS investigators prospectively enrolled over 34,000 patients who underwent radiography of the cervical spine following blunt trauma. Of these, 818 (2.4%) had cervical spine injury. These authors found that the clinical predictors had a sensitivity of 99.6% for clinically significant injury (Table 17.3) (5,29). The authors also reported high interobserver agreement ( k = 0.73) for the prediction rule (30), and reported that use of the rule would have decreased the overall ordering of cervical radiography by an estimated 12.6% (29).
Table 17.2. The Canadian C-spine rule
If the following three determinations are made, then imaging is not indicated 1. No high-risk factor, including:
Age >64 years
Dangerous mechanism, including:
Fall from >3 m/5 stairs Axial load to head (diving)
High-speed motor vehicle accident (60 mph, rollover, ejection) Bicycle collision
Motorized recreational vehicle Paresthesias in extremities 2. Low-risk factor is present
Simple rear-end vehicular crash, excluding:
Pushed into oncoming traffic Hit by bus/large truck Rollover
Hit by high-speed vehicle
Sitting position in emergency department Ambulatory at any time
Delayed onset of neck pain
Absence of midline cervical tenderness
3. Able to actively rotate neck (45 degrees left and right)
Source: Adapted from Dickinson et al. (33).
B. Canadian Cervical Spine Prediction Rule
A second level I clinical prediction rule, the Canadian C-spine rule for radi- ography (25) was published subsequent to the NEXUS trial, but with a similar objective: to derive a clinical decision rule that is highly sensitive for detecting acute cervical spine injury. The Canadian C-spine rule was a prospective cohort study of 8924 subjects from 10 community and univer- sity hospitals in Canada. Excluded were patients who had neurologic impairment, decreased mental status, or penetrating trauma. Like the NEXUS study, the Canadian C-Spine Study was an observational study performed without informed patient consent. However, patients who were eligible for the study but did not undergo radiography were followed up with a structured telephone interview 14 days following their discharge from the emergency department (ED). Thus, any patients who had not undergone radiography, and who had missed fracture would potentially be discovered during the investigation. The Canadian study investigated the predictive ability of 20 factors, and based on the reliability and pre- dictive properties of these factors, developed a prediction rule consisting of three questions. According to the Canadian C-spine rule (Table 17.2), the probability of cervical spine injury is extremely low, and imaging is not indicated if the following three determinations are made: (1) absence of high-risk factor (age >65 years, dangerous mechanism, paresthesias in extremities); (2) presence of a low-risk factor (simple rear-end motor vehicle collision, sitting position in ED, ambulatory at any time since injury, delayed onset of neck pain, or absence of midline cervical C-spine tender- ness); or (3) patient is able to actively rotate neck 45 degrees to left and right. The Canadian study group reported sensitivity of 100% and speci- ficity of 42.5% for this clinical prediction rule and also reported that the rate of ordering radiography would be 58.2% of the current rate (Table 17.3) (31).
The Canadian C-spine rule was validated using a prospective cohort study of 8283 patients presenting at the same 10 Canadian community and academic hospitals as the original study (32). The results of this verifica- tion trial noted a sensitivity of 99.4% and a specificity of 45.1%, very similar
Table 17.3. Diagnostic performance
Potential decrease Test (reference) Sensitivity Specificity in radiography C-spine prediction rules
NEXUS (29) 99.6 12.9 12.6
Canadian C-spine rule (31) 100 42.5 41.8
TL-spine prediction rules
Holmes et al. (11) 100 3.9 3.7
C-spine imaging
Radiography (43,45) Overall 93.9 95.3 N/A
Low risk 96.4 N/A
High risk 78.1–89.3 N/A
CT (39,41,42,46)
1Overall
TL-spine imaging 99.0 93.1 N/A
Radiography (60,64)
163.0 94.6 N/A
CT (60–64) 97.8 99.6 N/A
1