Keywords

These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

1 Key Points

  • The NEXUS and Canadian Cervical Spine prediction rules can be used to identify subjects in whom imaging is appropriate (strong evidence).

  • Cervical spine CT should be employed in high-risk patients (moderate evidence).

  • In low-risk victims not undergoing head CT, radiography may be the preferred cervical spine imaging approach (limited evidence).

  • Selection of subjects for thoracolumbar spine imaging can be made based on clinical criteria (moderate evidence).

  • CT, including reformations from CT scans performed of the abdomen and pelvis, is more accurate than radiographs in the thoracic and lumbar spine, but radiography may still be appropriate in low-risk subjects (limited evidence).

2 Introduction

There is a high risk of significant and permanent neurologic damage associated with spinal trauma. Although spinal cord injury is relatively uncommon, spinal imaging is utilized liberally across the United States to identify suspected and occult fractures. Spinal trauma and the sequela of spinal cord injury have broad-reaching ramifications beyond the obvious neurologic deficit for those affected. This includes a percipitous decline in probability of employment, educational achievement, and intact marriage [1]. As a result of widespread utilization, the positive yield of spine imaging is estimated to be only 2.4 % in the cervical spine when all patient populations are included [2]. Using the best available data, this chapter addresses diagnostic imaging of the spine in trauma including clinical prediction rules and cost-effectiveness.

3 Definition and Pathophysiology

Spinal fractures are estimated to account for 3–6 % of all skeletal injuries in the United States. A Canadian study in 2006 estimated that 56 % of spinal fractures are associated with spinal cord injuries and there is a general mortality rate of 8 % [3]. Although no recent epidemiologic studies have been performed, the annual incidence of cervical spine fracture was estimated at 10,000 per year in the United States in 1992 [4]. Better statistics are maintained for spinal cord injury of all causes and available from the National Spinal Cord Injury Statistical Center, Birmingham, Alabama. From this database, the annual incidence of spinal cord injury is estimated at 40 cases per million in the United States or 12,000–20,000 per year when on-scene fatalities are excluded [1].

The typical patient suffering from spinal cord injury is male (80.8 %) with an average age of injury of 33.7 years. The most common causes are traffic accidents, falls, and violence in decreasing frequency [1]. The hospital mortality for acute spinal injuries is high, up to 17 %, reflecting the presence of other severe injuries.

The cervical spine is both the most commonly fractured region in spinal trauma as well as the area where risk of cord injury is greatest compared to that of thoracic, lumbar, or sacral fractures [5]. Such fractures maybe clinically occult or patients unexaminable when obtunded or otherwise altered. In patients suffering from blunt trauma resulting in trauma team activation, the prevalence of cervical fracture is greater, 3.7 %, and up to 7.7 % in unexaminable patients. Once detected, between 42 % and 57 % of all cervical spine injuries are potentially unstable [6, 7].

The elderly population is a subset of patients with increased risk of significant injury resulting from relatively low-energy mechanisms of injury. The elderly spine has altered biomechanics, including decreased range of motion, lower muscular strength, and increased rigidity from degenerative changes, including ankylosis. In addition, degenerative changes may contribute to narrowing of the spinal canal with associated increased risk of cord injury [8].

4 Overall Cost to Society

Cervical spine injuries cause an estimated 6,000 deaths and 5,000 new cases of quadriplegia each year [1]. The total number of people with spinal cord injuries in the United States is estimated to be 265,000 persons, with a range of 232,000–316,000 persons [1]. The cost of care is dependent on severity of injury and is highest during the first year following injury. In 2010 dollars, the average annual expense for cervical spine injury resulting in incomplete motor function at any level was $321,720 in the first year and $39,077 for each subsequent year of life. In cases of high tetraplegia (C1-4), the first year cost of care averages $985,774 and $171,183 for each subsequent year of life [1]. The most recent comprehensive analysis of spinal cord injuries performed in 1996 concluded that the estimated total annual cost of all cervical spinal cord injuries was $9.7 billion per year [9].

5 Goals of Imaging

The primary goals of imaging are to (1) detect potentially unstable injuries to enable immobilization or stabilization and prevent development or progression of neurologic injury and (2) inform prognosis and guide surgical intervention for unstable fractures.

6 Methodology

PubMed (National Library of Medicine, Bethesda, Maryland) was used to search for original research publications discussing diagnostic performance and clinical predictors of cervical and thoracic spine injury. This includes publications from 1966 to August 31, 2011. The search strategy employed different combinations of the following terms: (1) spine, (2) radiography or imaging or computed tomography or magnetic resonance imaging, and (3) fracture or injury. MeSH headings included (1) spine and diagnosis, (2) imaging and spine, and (3) magnetic resonance imaging. Article was limited to human studies published in the English language. An initial review of the titles and abstracts of identified articles is followed by review of the full text in articles that were relevant.

7 Discussion of Issues

7.1 Who Should Undergo Cervical Spine Imaging?

Summary

The NEXUS [2] and Canadian C-spine [10] rules are two clinical prediction rules that have undergone multicenter validation, with the intent of determining which patients should undergo cervical spine imaging in blunt trauma patients. Both clinical prediction rules report a sensitivity greater than 99 %. Specificity is 42.5 % for the Canadian C-spine rule and 12.9 % for NEXUS (Table 31.1). A single randomized trial was implemented applying the Canadian C-spine rule which found that adherence to the decision rule demonstrated efficacy at reducing imaging of the cervical spine (strong evidence).

Table 31.1 Diagnostic performance
Full size table

Supporting Evidence

Low yield of cervical spine imaging has prompted a number of investigations to attempt to identify clinical factors that can be used to predict cervical spine fractures.

7.1.1 Nexus Prediction Rule

The National Emergency X-Radiography Utilization Study (NEXUS) was a multicenter observational study involving 23 diverse emergency departments throughout the United States. The NEXUS study was designed to assess the validity of four predetermined clinical criteria for prediction of cervical spine injury. According to the NEXUS criteria (Table 31.2), imaging is indicated if any of the following four determinations are met: (1) altered neurologic function, (2) intoxication, (3) midline posterior bony cervical spine tenderness, or (4) distracting injury (meaning an injury of sufficient pain to potentially distract the patient from noticing a cervical spine injury). In NEXUS, 34,069 patients were prospectively enrolled and underwent radiography of the cervical spine following blunt trauma. The above clinical predictors had a sensitivity of 99.6 % and specificity of 12.9 % for clinically significant injury [2]. In the participant population, 818 (2.4 % of total) had a cervical spine injury. It was estimated that adherence to the NEXUS criteria would reduce utilization of radiographs by 12.6 % (strong evidence).

Table 31.2 NEXUS criteria. Imaging of the cervical spine is not necessary if all five of the NEXUS criteria are met
Full size table

Though validated in multiple different emergency departments, the NEXUS may not be appropriate in high-energy trauma patients in whom the trauma team is activated. There is limited evidence in the trauma literature indicating that the clinical exam performed on a patient with a normal Glascow Coma Scale cannot be used to exclude cervical spine fracture in victims of major trauma. In a 2007 study, Duane et al. prospectively evaluated 534 blunt trauma patients followed by cervical spine CT, and the performance of clinical exam was compared against that of CT [11]. In evaluable patients with GCS of 15 or greater who were not intoxicated and did not have a distracting injury, 17 patients had cervical spine fractures, seven of which had a negative clinical exam. Of the seven fractures undetected clinically, three were transverse process fractures requiring no further intervention, and four required treatment with extended use of a rigid cervical collar. In 2011, Duane et al. performed a second study involving 2,606 trauma team activations, which also demonstrated that the NEXUS criteria where insufficient to exclude fracture in trauma team activation patients [12]. It is also not clear what was considered a distracting injury in the Duane studies as they report that over 60 % of the trauma team activation patients lacked distracting injuries.

There are no implementation studies documenting the efficacy of NEXUS for reducing utilization in the clinical setting.

7.1.2 Canadian Cervical Spine Prediction Rule

The Canadian C-spine rule for radiography was published subsequent to the NEXUS trial but had a similar goal of validating a prediction rule which is highly sensitive for detecting acute cervical spine injury. The Canadian C-spine study was a prospective cohort study performed at 10 community and university hospitals across Canada and included 8,924 subjects. The Canadian C-spine study was derived from an observational study which evaluated 20 potential predictive factors but concluded on three determinations. According to the Canadian C-spine rule (Table 31.3), imaging is not indicated if all of the following three determinations are made: (1) absence of high-risk factor (age >65, 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 tenderness), or (3) patient is able to actively rotate neck 45 ° to left and right. The group reported sensitivity of 100 % and specificity of 42.5 % with the rate of ordering radiography projected to be reduced by 58.2 % [10]. A 14-day follow-up was performed on all patients who did not undergo imaging in an attempt to discover all individuals with missed fractures (strong evidence).

Table 31.3 The Canadian C-spine rule. If the following three determinations are made, then imaging is not indicated
Full size table

The implementation of the Canadian C-spine rule has also been investigated through a cluster randomized trial involving 12 Canadian emergency departments. A total of 11,824 alert and stable adults were included. The intervention group showed a relative reduction in cervical spine imaging of 12.8 % and the control group a relative increase of 12.5 % of cervical spine imaging [13].

There is no head-to-head trial supporting the adoption of either cervical spine prediction rule over the other. A retrospective analysis comparing Canadian C-spine and NEXUS prediction rules was attempted. However, for this analysis, altered level of consciousness was effectively eliminated as a criteria [14, 15]. This negatively affects the accuracy of NEXUS as this is included in the NEXUS criteria. In addition, the Canadian C-spine rule requires the active evaluation of cervical spine rotational range of motion, a criterion which may not be acceptable in many US emergency departments.

7.2 What Imaging Modality Should Be Used for the Cervical Spine in Blunt Trauma?

Summary

Cervical spine CT is both more sensitive and specific than radiography for all patients (Table 31.1). In addition, cost-effectiveness analysis supports the use of CT as the initial modality in patient populations at high and moderate risk of cervical fracture. This strategy has been shown to reduce repeat imaging and identify the rare fractures which may have been missed from radiography with the potential to lead to severe neurologic deficit (moderate evidence). In patient populations with low probability for cervical fracture, properly performed cervical spine radiography remains the initial imaging modality of choice (limited evidence). MRI is not recommended in the acute setting as the initial evaluation of the cervical spine.

Supporting Evidence

Accuracy of Imaging

  • There are no randomized clinical trials comparing the efficacy of computed tomography with that of cervical spine radiography. Historically, the sensitivity of cervical spine radiography has been reported in the 89–94 %, when adequate three-view radiographs were obtained on all patients [2, 16–18]. Weighted pooling of the larger studies using a clinical gold standard suggests that radiography is relatively accurate with a sensitivity of 94 % and a specificity of 95 % when all trauma patients are included (Table 31.1) [18].

    However, more recently performed observational studies suggest that standard cervical spine radiography is less sensitive than previously reported. The discrepancy varies widely based on choice of reference standard and adequacy of cervical spine radiographs. A representative 2003 study performed by Griffen et al. in a level I trauma center concluded that the sensitivity of plain radiographs was 65 %, using CT follow-up as the reference standard [19]. In a 2005 meta-analysis, the pooled sensitivity of cervical spine radiography for fractures was estimated to be 54 % versus 98 % for computed tomography [20]. As with all diagnostic accuracy studies, any modality fares worse than the reference standard (in this case CT) and biases against the use of radiography. Studies using fractures that become apparent clinically as the reference standard are probably more relevant for clinical practice. In addition, these recent studies are biased by low percentage of cervical spine radiography examinations including adequate views, related to reluctance to perform repeat imaging for nonvisualized portions of the cervical spine. Furthermore, inadequate visualization is often seen as rational for proceeding to CT imaging increasing bias against radiography. In a 2009 study, Bailitz et al. included 1,583 consecutive major trauma patients that were evaluated with both cervical spine CT and three-view cervical radiography [21]. In this particular study, the final diagnosis in the medical record at discharge was used as the gold standard for cervical spine injury, and a clinically significant injury was one defined as requiring either an operative procedure, halo application, and/or rigid cervical collar application. Of the 78 patients with radiographic evidence of fracture, 50 (3.3 %) were determined to have clinically significant injuries, and 42 % of the 50 required operative intervention or halo application. Using the risk stratification criteria defined by Blackmore et al. [22], 16 clinically significant cervical fractures were present in the low-risk patients of which only 4 were identified by cervical spine radiography (25 % sensitivity). It should be noted however that of the 32 clinically significant injuries “missed” by cervical spine radiography, only 6 had adequate radiography.

    The discord between historical estimates of radiography sensitivity of 89–94 % and current estimates of 54–65 % confound determination of appropriate imaging. It is likely that the methodological limitations in the more recent literature, including consideration of inadequate radiographs as normal, use of an imaging rather than a clinical reference standard, and inclusion of only high-risk trauma patients explain much of this difference. Historical data indicating that missed cervical spine injuries were in fact rare prior to widespread use of CT also calls into question recent low estimates of radiograph sensitivity. However, with decreased utilization of cervical spine radiographs comes decreased proficiency at performance and interpretation, and sensitivity may actually have decreased as a consequence.

High and Moderate Risk Patients

  • Cervical spine radiography performs significantly worse in compared to patient populations at moderate and high risk of cervical fracture (probability >4 %) [18]. These patients are commonly immobilized on backboards, have poly-trauma, and are unable to cooperate. These factors result in lower specificity, more inadequate radiographs and repeat imaging, greater utilization of hospital resources, and ultimately higher cost [23]. Additionally, CT evaluation has been shown to be more time efficient when compared to radiography, allowing for faster disposition of patients from the emergency department [24, 25]. This is particularly true when evaluation of the cervical spine follows CT scan of the head [26]. The decreased sensitivity of radiography in the major trauma population, time efficiency, and increased prevalence of cervical fracture support initial evaluation of the cervical spine utilizing CT in moderate and high-risk patients. Cost-effectiveness analysis supports use of CT in this population. In a 1999 study, Blackmore performed a cost-effectiveness analysis from the societal perspective comparing cervical radiography to that of CT and found that CT was cost-effective in high and moderate risk [18]. This was confirmed by Grogan et al. in 2005 [27] (moderate evidence).

Low-Risk Patients

  • There is neither strong evidence nor consensus on the appropriate approach to cervical spine imaging in trauma victims in whom some imaging is indicated through use of NEXUS or the Canadian C-spine rule, but who are at low risk of injury. The standard has been radiography, but more recently, CT has been promoted as an initial imaging strategy, even in low-risk individuals. Recent societal consensus guidelines in the United States, including the ACR Appropriateness Criteria [28] and Eastern Association for the Surgery of Trauma [29], have advocated for use of CT for all patients who undergo cervical spine imaging in trauma. However, guidelines supporting use of CT in low-risk patients generally rely on recent estimates of accuracy, despite the methodological limitations. In addition, such guidelines do not consider the fact that use of CT carries much greater radiation risk and societal cost.

    Radiography may be most appropriate in the evaluation of patients who cannot be cleared clinically but have low-risk factors for significant cervical trauma such as young age, low-impact trauma, and no distracting injuries [18, 22, 30]. Inability to obtain technically adequate radiographs due to incomplete visualization or suboptimal quality (low specificity) is the single biggest limitation of radiography (Table 31.1) [20]. In the very low-risk patient population, adequate films are more easily obtained. CT is indicated when adequate radiographs cannot be obtained.

    Radiation risks are difficult to estimate with any precision due to the need for extrapolation of radiation effects from higher administered doses to the very low doses found in diagnostic imaging. However, use of CT rather than radiography for evaluation of the cervical spine comes with an estimated 14-fold greater patient exposure to ionizing radiation. The organ-specific dose to the thyroid gland with cervical spine CT has been estimated at 26 mGy compared to 1.8 mGy for radiography [31], resulting in increased risk of radiation-induced malignancy [32].

    Reconciliation of the higher sensitivity of CT versus the lower cost and radiation dose of radiography is challenging. From 2002 to 2007, there was a significant increase in the use of CT and plain radiographs in the management of trauma patients, leading to significantly higher radiation exposure with no demonstrable improvements in the diagnosis of missed injuries, mortality, or length of stay [33].

    Table 31.4 makes the trade-offs explicit through a crude estimation of the number needed to treat and the number needed to harm when substituting CT for radiography in low-risk patients. There is substantial uncertainty in the estimates of both benefits and harms from CT. However, it is likely that the rate of cancer mortality is at least an order of magnitude greater than the probability of preventing paralysis through use of CT in low-risk trauma patients. Accordingly, radiography, when adequately performed, should be considered as the initial imaging approach in patients at low risk (limited evidence).

    Cost-effectiveness analysis also supports radiography as initial imaging strategy in low-risk patients. The threshold for when CT becomes cost-effective is somewhat uncertain. In the original cost-effectiveness analysis, Blackmore found a risk threshold of 4 % to be the criterion for use of CT. However, subsequent investigators have proposed lower thresholds. Grogan suggested 0.9 %, though this was based on extremely low estimates of radiograph sensitivity (64 %) found in severely injured patients. Likely however, the appropriate threshold is lower than the original 4 % estimate, due to lower current estimates of performance of radiography detailed above.

    Determination of appropriate imaging therefore requires stratification of patients in to low- and higher-risk cohorts. Blackmore [22] and Hanson [34] developed and validated a clinical prediction rule to identify subjects at low risk (Table 31.4). In the validation cohort, subjects lacking any of the high-risk factors had a risk of cervical spine fracture of only 0.2 %, indicating that radiography was the preferred imaging approach. In the NEXUS study, the probability of fracture was 2.4 % overall but 0.4 % in the low-risk patients [2], again confirming that a group can be identified where adequate cervical spine radiography is appropriate as the initial screening tool.

Table 31.4 Number needed to treat and harm for cervical spine imaging in low-risk patients
Full size table

7.2.1 Special Cases

7.2.1.1 Obtunded Patient

Summary

A normal cervical CT in obtunded patients with blunt trauma essentially excludes unstable cervical spine injuries. MRI is unlikely to change management when there is no neurologic deficit or abnormality by cervical spine CT and is therefore not routinely recommended given risks and benefits (limited evidence).

Supporting Evidence

There are several valid cohort studies of the accuracy of cervical spine CT in excluding unstable injuries in obtunded or clinically unexaminable patients. Hennessy in 2010 reported a prospective cohort study of 402 intubated, unexaminable blunt trauma patients with normal CT. Using flexion extension radiography and clinical follow-up as a reference standard, one patient was found to have an unstable injury missed by the CT (negative predictive value 99.7 %) [35]. Hogan et al. retrospectively examined 366 patients with negative CT, using MR and clinical follow-up as the reference standard. The authors concluded that the negative predictive value of CT for ligamentous injury was 98.9 % and 100 % for unstable CS injury [36]. Harris and colleagues evaluated a retrospective cohort of 367 obtunded patients using a clinical and radiographic reference standard. A normal multi-detector row CT scan of the cervical spine in obtunded patients with blunt trauma had a negative predictive value of 99.7 % [37]. Brohi and colleagues prospectively evaluated 442 consecutive unconscious trauma patients and defined the sensitivity of CT at 98.1 % (51/52), with a negative predictive value of 99.7 % [38]. In addition, a 2005 retrospective cohort study by Schuster et al. included 93 patients with a normal motor examination and a negative cervical spine CT with MR as the reference standard. In this study, all patients had negative MRI examinations unless there was a neurologic deficit or a positive CT [39]. Como evaluated 197 patients who were obtunded by moving all four extremities and reported no missed injuries on CT, with clinical or MRI follow-up [40] (moderate evidence).

However, it is also clear that CT is imperfect. As an example, Schoenfeld and colleagues culled from the medical literature multiple cases (particularly of ligamentous injuries) missed at CT but discovered on subsequent MRI [41]. However, in a common failing of the literature on this topic, the authors failed to report the number of true-negative CT scans, instead only reporting the number of false-negative CT scans among the group who went on to MRI. This verification bias, due to selection of the cohort based on performance of the reference standard, makes calculation of negative predictive value meaningless [42].

Finally, there are potential risks related to use of MRI in obtunded patients, related to the transfer of patients to the MRI suite, and limited ability to monitor patients while in the MRI scanner. In addition, delay in clearance of the cervical spine, with prolonged immobilization, may lead to complications including pressure ulcers, increased intracranial pressure, thromboembolism, and pulmonary aspiration [43–45].

7.2.1.2 Elderly

Summary

Elderly individuals are at higher risk of cervical spine injury from both high- and low-energy mechanisms. However, no prediction rules have been validated to specifically identify predictors of injury in the elderly. The same predictors in younger patients appear to work in the elderly [46]. Accordingly, the same approach to imaging may be applied in the elderly as in younger patients, but with a lower threshold for use of CT (limited evidence).

7.2.1.3 Children

Summary

The NEXUS clinical prediction rule is a reasonable method of identifying which older children and adolescents should undergo cervical spine imaging after trauma. Imaging should be performed in subjects with (1) altered neurologic function, (2) intoxication, (3) midline posterior bony cervical spine tenderness, and (4) distracting injury (moderate evidence). Under the age of 3 years, cervical spine imaging may be limited to subjects with high-energy mechanism (motor vehicle crash) or Glascow Coma Score of less than 14 (limited evidence). Radiography can appropriately be used to exclude cervical spine fracture in children, though cervical spine CT may be useful in high-risk subjects. In younger children, CT should be limited to the upper cervical spine (limited evidence).

Supporting Evidence

Evidence for who should undergo imaging is less complete in children than in adults. Determination of clinical predictors of injury in pediatric subjects is complicated by the decreased incidence of injury in children, requiring larger sample size for adequate study [47–49]. In addition, children may sustain serious cervical cord injuries that are not radiographically apparent [47, 48]. Among adult clinical prediction rules, the Canadian Clinical Prediction Rule development study excluded children [10]. The NEXUS trial included children, but there were only 30 injuries in subjects under age 18, and only four in subjects under age 9 [2]. Although no pediatric injuries were missed in the NEXUS study, the sample size was too small to adequately assess the sensitivity of the prediction rule in this group. Further validation of a pediatric version of the NEXUS was performed at a single academic pediatric trauma center in the United States. In 647 trauma victims age 3 or older, injuries were found in approximately 2 %, of whom four required operative fixation. No missed injuries were reported [50].

A pediatric adaptation of the NEXUS is a thus reasonable approach in children over age 3, suggesting that imaging is only indicated when subjects have any of the following: (1) altered neurologic function, (2) intoxication, (3) midline posterior bony cervical spine tenderness, and (4) distracting injury (moderate evidence) [50].

Pieretti-Vanmarcke and colleagues performed a retrospective analysis of trauma registry data from multiple institutions, including 12,537 patients under the age of 3. They found that limiting imaging to subjects with decreased level of consciousness manifest by pediatric Glascow Coma Score of less than 14 or high-energy mechanism (motor vehicle crash) identified 78 of 83 (94 %) clinically important injuries with a negative predictive value of 99.9 %. The low negative predictive value was driven largely by the extremely low incidence of injury in this population (0.66 %) even in subjects evaluated at major trauma centers [49]. This study has not yet been validated prospectively (limited evidence).

Comparison of CT versus radiography has not been well explored in children. Radiography has accuracy for cervical spine fracture of approximately 94 %, [51] similar to adults [18]. The odontoid view and flexion extension radiographs contribute little in young children [52–55]. CT is likely more accurate than radiography but does encompass higher radiation doses and higher costs [56]. The cost-effectiveness analysis of Blackmore and colleagues excluded children, [18, 22, 34] as did the studies of the Harborview high-risk cervical spine criteria (Table 31.5) [22, 34]. Further, the lower frequency of injury in children [47, 57] and the increased radiosensitivity of pediatric subjects [58] suggest that cost-effectiveness results from adults may not be relevant.

Table 31.5 Harborview high-risk cervical spine criteria. Presence of any of the following criteria indicates a subject at sufficiently high risk to warrant initial use of CT to evaluate the cervical spine
Full size table

A reasonable approach to pediatric cervical spine imaging is the Harborview protocol (Fig. 31.1). Overall, radiography is adequate to exclude cervical spine fracture in most younger children [56, 59] (limited evidence). However, use of upper cervical CT in high-risk younger children [60] who are getting head CT is probably reasonable, as the time and cost is minimal, and the thyroid can be spared in the CT radiation dose if imaging is limited to the upper cervical spine (insufficient evidence). In addition, upper cervical spine injuries are more common than lower cervical injuries in younger children (Fig. 31.2a, b) [57, 61, 62].

Fig. 31.1
figure 1figure 1

Pediatric imaging protocol for blunt trauma from Harborview Medical Center (Reprinted with kind permission of Springer Science+Business Media from Blackmore CC. Imaging of the spine for traumatic and nontraumatic etiologies. In: Medina LS, Applegate KE, Blackmore CC, editors. Evidence-based imaging in pediatrics: optimizing imaging in pediatric patient care. New York: Springer; 2010)

Full size image
Fig. 31.2
figure 2figure 2

Atlantooccipital subluxation with occipital condyle fracture in a 9-year-old boy. (a) Axial CT demonstrates right occipital condyle fracture (arrow). (b) Coronal reformation demonstrates the right occipital condyle fracture (arrow) as well as widening at the left atlantooccipital joint (arrowheads) (Reprinted with kind permission of Springer Science+Business Media from Blackmore CC. Imaging of the spine for traumatic and nontraumtic etiologies. In: Medina LS, Applegate KE, Blackmore CC, editors. Evidence-based imaging in pediatrics: optimizing imaging in pediatric patient care. New York: Springer; 2010)

Full size image

7.3 Who Should Undergo Imaging of the Thoracic and Lumbar Spine Following Trauma?

Summary

Clinical prediction rules to determine which patients should undergo thoracolumbar spine imaging have been developed, but not validated. Although these prediction rules have high sensitivities for detecting thoracolumbar fractures, their low specificities and low positive predictive values would require imaging a large number of patients without thoracolumbar injuries. This drawback limits the clinical utility of these prediction rules (moderate evidence).

Supporting Evidence

Given the relative lack of clarity regarding which blunt trauma patients require thoracolumbar imaging, several observational (limited evidence) studies have examined potential risks for thoracolumbar fracture. These limited studies have identified associations between the risk of thoracolumbar injury and high-speed motor vehicle crash [63, 64], fall from a significant height [65–67], complaint of back pain, [65–69], elevated injury score [65, 66], decreased level of consciousness [66–68, 70], and abnormal neurological exam [67, 68].

Two separate clinical prediction rules to guide thoracolumbar spine imaging decisions have been developed, although neither prediction rule has been validated. The smaller study, conducted by Hsu et al., examined the effect of six clinical criteria on two retrospective groups [71]. The first group consisted of a cohort of 100 patients with known thoracolumbar fracture, while the second group consisted of 100 randomly selected multi-trauma patients. The criteria evaluated were (1) back pain/midline tenderness, (2) local signs of injury, (3) neurological deficit, (4) cervical spine fracture, (5) distracting injury, and (6) intoxication. The results of this small scale, retrospective trial found that 100 % of the patients in the known thoracolumbar fracture group would have been imaged appropriately using the proposed criteria. This proposed pathway was then tested retrospectively in the group of randomly selected blunt trauma patients and was found to have a sensitivity of 100 %, a specificity of 11.3 %, and a negative predictive value of 100 %. Implementing these criteria would still require imaging the thoracolumbar spine in 92 % of the selected multi-trauma patients.

A much larger prospective, single-center study by Holmes et al. evaluated similar criteria in 2,404 consecutive blunt trauma patients who underwent thoracolumbar imaging [72]. These clinical criteria (Table 31.6) were (1) complaints of thoracolumbar spine pain, (2) thoracolumbar spine pain on midline palpation, (3) decreased level of consciousness, (4) abnormal peripheral nerve examination, (5) distracting injury, and (6) intoxication. This prediction rule was successful in achieving 100 % sensitivity for detecting thoracolumbar fracture; however, the specificity was only 3.9 %. Due to this low specificity, implementing this prediction rule in this patient population would have decreased the rate of thoracolumbar imaging by merely 4 % (Table 31.1) (moderate evidence).

Table 31.6 Thoracolumbar spine imaging criteria
Full size table

Though not specifically evaluating a clinical prediction rule, Sava and colleagues did identify that clinical exam may not be sufficiently reliable to exclude fracture in subjects with substantial blunt trauma and altered sensorium [73].

7.4 What Imaging Modality Should Be Used to Evaluate the Thoracic and Lumbar Spine in Blunt Trauma?

Summary

Multiple studies have shown that some CT protocols used for imaging the chest and abdominal visceral organs, when performed with sagittal reformations, are more sensitive and specific for detecting thoracolumbar spine fracture than conventional radiography. In patients undergoing such scans, conventional radiography may be eliminated (limited evidence). The effect of primary screening with CT scan on cost and radiation exposure has not been thoroughly studied for the thoracolumbar spine.

Supporting Evidence

Multiple limited evidence studies examine the possibility of eliminating conventional radiography in those patients who are candidates for both conventional thoracolumbar radiographs and CT evaluation of the chest or abdominal viscera; however, many of these trials are hampered by small sample sizes and/or verification bias [74–78]. Studies that combine the results of both CT and conventional radiography as the reference standard suggest that CT has a sensitivity of 78.1–100 %, while conventional radiographs have a sensitivity of 32.0–74 % for detecting thoracolumbar fracture (Table 31.1) [75–77, 79]. The clinical importance of thoracolumbar fractures not found with conventional radiography is unknown, as no studies with clinically based outcome measures were located.

A single limited evidence trial examined the use of CT as an initial evaluation in patients for which a CT scan is not indicated for other reasons [76]. This prospective, single-center trial examined 222 trauma patients with both CT and conventional radiographs as initial screening exams. The reported sensitivity was 97 % for CT examination and 58 % for conventional radiographs. The results of this trial are limited in that only 36 patients were diagnosed with thoracolumbar fracture during the course of the trial.

7.4.1 Applicability to Children

Summary

There are no clinical prediction rules validated in children for the determination of when imaging is indicated. However, a reasonable approach in older children is to image when any of the following are present: (1) complaints of thoracolumbar spine pain, (2) thoracolumbar spine pain on midline palpation, (3) decreased level of consciousness, (4) abnormal peripheral nerve examination, (5) distracting injury, and (6) intoxication (limited evidence). No reliable data exists on when to image in younger children (insufficient evidence). Compared to adults, younger children are less likely to localize pain and may have pain referred to the spine from intra-abdominal causes, particularly renal (infection and obstruction).

Supporting Evidence

Data on appropriate indications for thoracolumbar spine imaging in children is limited. The adult clinical prediction rule from Holmes and colleagues did enroll children. However, the actual number of children in the study is not reported [72]. The youngest patient enrolled in the small clinical prediction rule validation trial by Hsu et al. was 14 years of age [71]. Given the 100 % sensitivity in adults, it is reasonable to employ the Holmes clinical prediction rule in older children (limited evidence). In younger children, the criteria would have to be modified ad hoc to meet the clinical perception of the child’s ability to provide reasonable responses and the clinical picture (insufficient evidence). The specificity of the Holmes prediction rule in adults was low (3.9 %), so it is not expected that the use of this prediction rule would decrease unnecessary imaging [72].

8 Take-Home Figure and Tables

Figure 31.1 shows a pediatric imaging protocol for blunt trauma from Harborview Medical Center.

Tables 31.1 through 31.6 highlight key recommendations and supporting evidence.

9 Imaging Case Studies

Case 1: Atlantooccipital subluxation with occipital condyle fracture in a 9-year-old boy (Fig. 31.2a, b)

Case 2: Victim of a motor vehicle crash who met criteria for initial cervical spine imaging with CT scan (Fig. 31.3a, b)

Fig. 31.3
figure 3figure 3

(a, b) Victim of a motor vehicle crash who met criteria for initial cervical spine imaging with CT scan. A potentially unstable C6–7 facet and pars interarticularis fracture is apparent on CT (a) but may be missed on contemporaneous radiography (b). CT has higher sensitivity for fracture than radiography (Reprinted with kind permission of Springer Science+Business Media from Blackmore CC, Avey GD. Imaging of the spine in victims of trauma. In: Medina LS, Blackmore CC, editors. Evidence-based imaging: optimizing imaging in patient care. New York: Springer; 2006)

Full size image

10 Recommended Imaging Protocol

10.1 Cervical Spine

CT protocol: Multi-detector CT with axial image reconstruction at 2.5 mm or less, in both bone and soft tissue algorithms, and with sagittal and coronal reformations in bone algorithm at 2-mm collimation.

Radiography protocol: AP, open mouth, lateral, and swimmers. Note that all images must be adequate for evaluation, and the entire region from skull base to T1 must be visible in both frontal and lateral projections. If adequate films cannot be obtained after repeat imaging, then CT should be performed.

10.2 Thoracic and Lumbar Spine

CT protocol: Axial images in bone algorithm through the area of concern, with 2.5-mm collimation. Must include sagittal reformations and preferable coronal, in bone algorithm, at 2-mm collimation.

Radiography protocol: AP and lateral views covering the entire area of interest.

11 Future Research

  • Studies in both cervical spine and thoracolumbar spine imaging indicate that CT is more sensitive than traditional radiography in detecting fractures. However, further clinical studies addressing the relevance of these fractures are needed.

  • The applicability of cervical spine injury clinical prediction rules in pediatric patients is unknown. In addition, the sensitivity, specificity, and cost-effectiveness of the various imaging exams in the pediatric population are not well established.

  • Clinical prediction rules for imaging of the thoracolumbar spine have been developed, but further research is necessary to validate such approaches. The effect of implementing these rules on cost, cost-effectiveness, and radiation exposure has not been determined.

  • Appropriate imaging to detect unstable ligamentous injury, particularly in clinically unexaminable subjects, remains unresolved.