Radiographic cervical spine injury patterns in admitted blunt trauma patients with and without prehospital spinal motion restriction

Introduction

Over 2% of blunt trauma patients sustain injury to the cervical spine.1 Universal application of prehospital cervical spine motion restriction (C-SMR) (ie, immobilization) by emergency medical service (EMS) personnel in trauma patients had been the mainstay of trauma care to prevent cervical SCI in patients whose injuries were yet to be determined, even in the absence of the signs or symptoms.2–4 This practice was based on the rationale that excessive C-spine motion or manipulation during transport and resuscitation could exacerbate spinal injuries.5 6

However, recent protocols suggest that a more selective prehospital motion restriction algorithm based on specific patient criteria may reduce adverse outcomes associated with C-SMR such as: (1) reduced cerebral venous drainage and resultant elevated intracranial pressure in patients with intracranial injuries; (2) compromised airway access; (3) derangement in oxygenation and ventilation parameters; (4) dysphagia and attendant increased risk of aspiration; (5) pressure ulcers on the occiput, chin, shoulders, and back; (6) potential for unnecessary radiation exposure to ‘clear the cervical spine’; (7) potential for harm in patients with spinal conditions such as ankylosing spondylitis.7–13 Furthermore, there is debate regarding their efficacy in terms of restriction of motion; one study demonstrated that four commonly used rigid collars decreased neck mobility by only about one-half to two-thirds of baseline.14 Additionally, a cadaver study indicated that several collars were ineffective in neck immobilization in an unstable C-spine.15

As such, there has been a national trend toward selective prehospital C-SMR. National guidelines have indicated that EMS personnel can be trained to apply criteria for selective C-SMR in adults.16 Indeed, recent consensus guidelines from American College of Emergency Physicians/American College of Surgeons Committee on Trauma/National Association of Emergency Medical Services Providers state that indications for C-SMR in blunt trauma are: acute altered level of consciousness, midline neck or back pain and/or tenderness, focal neurological signs and/or symptoms, anatomic spinal deformity and/or distracting circumstances or injury that impedes the patient’s ability to contribute to a reliable examination.17 The Eastern Association for the Surgery of Trauma guidelines for C-SMR in penetrating trauma go further by conditionally recommending against prehospital spinal motion restriction as it did not mitigate against neurological deficits and was associated with increased mortality.18 While several studies have suggested the validity of field protocols that selectively immobilize general trauma patients, these studies have also emphasized that special caution is needed when applying these protocols to the geriatric population.19 To this end, a substantial incidence of vertebral fractures among the elderly admitted following falls from low height was demonstrated in one county.20 This increased fracture frequency in the geriatric population is largely attributed to age-related changes, which make the spinal column more susceptible to injury.21 We determined the rates and types of C-spine injuries sustained by patients who did or did not receive prehospital C-SMR, hypothesizing that admitted blunt trauma patients without prehospital C-SMR would have different injury patterns.

Methods

A retrospective review of a single institution’s trauma registry (maintained on Trauma One, V.4.1, Lancet Technologies, Boston, Massachusetts, USA) was performed. All patients 18 years of age and older discharged from the hospital between January 2011 and August 2019 after sustaining blunt trauma and transported by EMS from the field to the regional/level 1 trauma center for a suburban county of approximately 1.5 million in New York State were included. Exclusion criteria included burns, drowning/diving injuries, hangings, penetrating trauma, death in the emergency department (ED), interfacility transfers, and non-EMS mode of arrival. Interfacility transfers were excluded as they may or may not reflect EMS decision making, for example, if the cervical collar was applied by the referring facility, then it is improbable that EMS would remove it for transfer. Of note, as this query is based on discharge criteria, three patients who were admitted with cervical spine injury in 2010 but discharged in 2011 were included.

In concordance with national trends, selective prehospital cervical SMR by EMS providers in New York State was mandated by the Department of Health in 2008 and guidelines were further updated in 2015.22 23 Briefly, the protocol indicated that for patients who do not meet major blunt trauma criteria, criteria for spinal motion restriction include altered mental status for any reason; complaint of neck and/or spine pain or tenderness; weakness, tingling, or numbness of the trunk or extremities at any time since the injury; deformity of the spine not present prior to this incident; distracting injury or circumstance; or high-risk mechanism of injury (MOI) such as axial load, high-speed motorized vehicle crashes or rollover, or falls greater than standing height. It further indicated that if any doubt, suspect spine injury. In 2015, the protocol updated high-risk mechanisms to include falls >3 feet/5 steps or greater than patient’s height and also included pedestrian or bicyclist collisions.23 Additional high-energy MOI include high-speed motor vehicle collision (MVC)/motorcycle collision, a height >3 times the patient’s body height, and axial loading forces. In contrast, low-energy MOI include falls from standing height or low surface and low speed MVC. The protocol indicated that a ‘positive MOI is not considered means to necessitate full motion restriction but should be used as a guide to heighten a provider’s suspicion for an SCI’.23 24 The corresponding presentation indicated that older age is a high risk factor for spinal injury.24

Patients with C-spine injuries meeting study criteria were identified from the state-mandated trauma registry. Demographics, injury characteristics, National Trauma Data Standard defined comorbidities and complications, as well as outcomes were extracted from the trauma registry.25 The determination of prehospital SMR was initially captured by a data field in the trauma registry indicating prehospital spinal immobilization, which refers to prehospital C-SMR. The registrars concurrently examine the prehospital patient care record (PCR) for documentation indicating the utilization of prehospital spinal motion restriction in the narrative and/or in a PCR field indicating ‘spine immobilization neck and back’. Additional sources for these data are ED records, including trauma team activation flowsheets, if any. Secondary chart review was conducted in those without prehospital spinal motion restriction but with C-spine injury. For patients to have been classified as having prehospital C-SMR, they could have had a rigid collar, straps/head blocks, etc. In cases of inconsistency in the medical record on secondary review, patients were classified as having prehospital C-SMR. In one patient with radiographic cervical spine injury, the presence/absence of prehospital C-SMR could not be confirmed and in another the C-spine injury was thought to have likely occurred at a prior event. Hence, these two patients were excluded from all analyses.

The medical records of all patients with cervical spine injury (bone, ligament, and/or SCI) in the trauma registry were examined to characterize imaging findings (CT scan and/or MRI) and management. Patients with C-spine injuries noted only on autopsy were not included, nor were the few patients with spinal cord injury without radiographic abnormality, for example, those with ‘stingers’ or clinical sprains. Patients with isolated disk herniation were also excluded, unless it was clear that this was an acute injury that resulted in neurological deficit from cord involvement. Bone injury (fracture or subluxation) was characterized based on CT reports. Of note, lateral mass and pars interarticularis fractures were coded as one. Body, endplate, and tubercle fractures were coded as vertebral body fractures. The term posterior elements fractures was also listed in the radiology reports, without further delineation, for example, pedicle, transverse processes, articular processes, lamina, spinous processes in seven patients.

MRI was only used for clarifying the acuity of a bony injury when this was not clear on CT. In contrast, ligamentous and cord injuries were based on MRI only. For ligamentous injuries, we included those where edema was noted in the ligaments as well as statements indicating possible/probable/suggestive/cannot exclude ligamentous injury in the setting of other factors. Hence, there is the potential for under coding of ligamentous injury, as those noted in the operating room would not be included, for example, if a patient went to the operating room for a jumped facet without MRI, likely they would have had a ligamentous injury, but given the absence of an MRI, it would not have been ascribed in this study. For cord injuries, we included cord compression, cord contusion, cord edema, cord hemorrhage, and cord injury only on MRI. In other words, if no MRI was performed, then no ligamentous or cord injury was counted in this study, regardless of CT or clinical findings. In the rare cases of discrepancy between radiology reports and clinical service, to maintain consistency, we used radiology reports.

The decision to perform spinal imaging in patients with or without prehospital cervical spinal motion restriction is at the discretion of the providers. Similarly, the decision to place or remove a collar on patients presenting to the ED is at the discretion of the treating provider in accordance with guidelines.26 27 CT scanning of the head/neck in the elderly is liberalized, regardless of symptomatology, in concordance with national trends and caveats of CT spine clearance rules.26 27 To this end, CT panscans in the elderly after ground level falls have noted significant new findings on imaging.28 Panscans in patients without obvious injury and significant mechanism has been previously shown to be worthwhile.29

Univariate statistical analyses of demographic and outcome variables were performed using SPSS V.26 (IBM, Armonk, New York, USA), Tableau V.2020.4.13 (Seattle, Washington, USA), and R statistical software (Vienna, Austria). No data were imputed; missing variable details are described in the text and/or tables. Median values or percentages were reported as appropriate. Non-parametric tests for independent samples were used to examine the statistical significance of differences in continuous and categorical variables, respectively. Multivariate logistic regression models were also generated using R to determine predictors for prehospital C-SMR. In these analyses, we considered the following demographic factors: age, sex, year of discharge, MOI, Glasgow Coma Score in the ED (ED GCS), and severity of injuries (AIS) in the face, chest, abdomen, and extremity. ED GCS, as opposed to head/neck AIS, was used as it indicates mental status in the ED, regardless of underlying head or C-spine injury. A p value <0.05 was considered statistically significant.

Results

From 2011 to 2019, 7977 admitted adult blunt trauma patients were transported to our facility via EMS. Of these, 41.9% received prehospital C-SMR. Of admitted patients, 658 (8.2%) patients sustained cervical spine injury on imaging and met study inclusion criteria. Of these 658 patients, 117 (17.8%) did not receive prehospital spinal mobilization restriction and 541 (82.2%) received C-SMR (table 1). Of these 117 no prehospital C-SMR patients, 6 refused or removed C-SMR and in 3 patients C-SMR could not be completed for anatomic/physiological/other reasons. While the percentage of patients with C-spine injuries per year remained relatively constant at 7.0%–9.5%, the rate of without prehospital C-SMR (non-SMR) ranged from 5.0% to 34.5%. It generally increased in the later years, with a caveat that 2016 had a 27.8% rate (figure 1). This trend had an r² value of 0.74 for linear regression. Concomitantly, the non-SMR rate among patients who did not have a C-spine injury increased from 49.5% to 73.2% with an r² value of 0.93, reflecting improving specificity (figure 2).

Request permission

Request permission

Of 658 patients with cervical spine injury, the median age in patients with C-SMR was 54 years vs 76 years in those without C-SMR. Patients without C-SMR were less often male (55.5% vs 65.4%), less often had multisystem injury (11.9% vs 30.6%), with a median Injury Severity Score of 10 vs 17 (table 1). They had thoracic spine injuries less often (12.8% vs 25.8%). The median ED GCS in non-SMR patients was 15 (IQR 15–15) vs 15 (IQR 14–15) in C-SMR patients (p<0.001). Patients without C-SMR most often had low fall as MOI (59.8% vs 15.9%). Patients without C-SMR were more often white (91.4% vs 82.1%). Patients without C-SMR were more often Medicare enrollees (62.4% vs 22.0%).

With regard to injuries, 22/658 patients had unclear traumatic spine findings on imaging. They were included in the study because they were treated with a cervical collar during hospitalization as follows: age indeterminate cervical spine fracture by CT but acute by MRI (n=1); subacute/healing/evolving fracture (n=3); cervical spine body versus osteophyte fractures (n=4); age indeterminate (n=8); and unclear if non-union/chronic fracture/or other anatomic abnormality such as a vascular groove (n=6).

However, these 22 patients were not included in analyses of acute cervical spine fracture patterns, leaving 489 patients who had an acute cervical spine fracture; the remaining had injuries other than fractures (vide infra). Approximately 24.7% of patients with or without C-SMR had two cervical vertebrae fractured; three or more vertebral level fractures occurred in 9.1% of those without C-SMR and 11.2% of those with C-SMR (figure 3). The incidence of C1, C3, C4, C5, and C6 fractures were not significantly different between groups. While C2 fracture incidence did not reach statistical significance (36.4% vs 26.2%, p=0.057), fractures of the dens (26.1% vs 11.2%, p=0.001) occurred significantly more often in non-SMR group. Meanwhile, C7 (22.7% vs 37.6%, p=0.02) fractures occurred significantly less often in non-SMR patients. Indeed, C2 was the most frequently fractured vertebra in the non-SMR group and C7 was the most frequently fractured vertebra in the C-SMR group. Subluxation, including acute vertebral anterolisthesis/posterolisthesis and perched facets, occurred significantly less often in the non-SMR group (11.4% vs 21.9%, p=0.04).

Request permission

The parts of a vertebra that were acutely fractured were not significantly different between those without and with prehospital C-SMR (figure 4). The most common fracture locations in both groups were transverse process (25.0% without C-SMR vs 29.7% with C-SMR) and vertebral body (29.5% without C-SMR vs 24.9% with C-SMR). Notably, patients may have had involvement of more than one part of a single cervical vertebra via extension of the fracture and/or as a separate fracture, approximately 39.8% without C-SMR and 30.4% with C-SMR had more than one fracture of a single vertebra on CT scan (p=0.13). Furthermore, ligamentous injuries by MRI occurred less often (5.1% vs 12.6%, p=0.001) in those without C-SMR versus with C-SMR. Spinal cord injuries by MRI were also not infrequent in both groups, but occurred less often (8.6% vs 19.6%, p=0.008) in those without C-SMR versus with C-SMR.

Request permission

With regard to hospitalization outcomes in patients confirmed to have a spinal injury on imaging, a cervical collar was placed in the ED in all patients that had not received prehospital motion restriction. In non-operative non-SMR patients, the collar was cleared during hospitalization in nine patients. Cervical spine surgery, including treatment with a halo, was performed less often in non-SMR patients than C-SMR patients (8.5% vs 20.0%, p=0.004) during the hospitalization (table 1). In the non-SMR group, there were six anterior cervical diskectomy fusions (ACDF), three posterior cervical fusions, and one open reduction internal fixation of the dens; three of these procedures were performed in patients without fractures (two ACDF and one posterior fusion). Those without prehospital C-SMR had a significantly lower intensive care unit (ICU) admission rate (29.1% vs 55.2%) and median length of hospital stay (6 days vs 8 days), but not ICU length of stay (LOS). The rate of cervical collar-associated decubitus was 1.2% (one without prehospital C-SMR and seven with prehospital C-SMR). These patients had prolonged cervical collar immobilization, with hospital LOS of weeks to months. In-hospital mortality was not significantly different between those without prehospital spinal motion restriction and those with (4.3% vs 7.4%). Hospital discharge disposition was also not significantly different between groups.

Of the factors considered in multivariate analysis (table 2), only age (OR 1.02), discharge year (OR 1.21), and low fall MOI (OR 3.12) were significant predictors for non-SMR prehospital in those with spinal injuries. ED GCS and sex were not. Severity of injury in the defined body regions (ie, face AIS, chest AIS, abdomen AIS, extremity AIS) were also not significant predictors. The model had an area under the curve of 0.81.

Discussion

Principal findings from this study are: (1) rate of prehospital cervical spinal motion restriction decreased; (2) patients without C-SMR had similar bony vertebral components injured as those with prehospital C-SMR, although they less often had spinal cord or ligamentous injuries on MRI; (3) dens fractures occurred more often in those without C-SMR while C7 fractures occurred more often in those with C-SMR; and (4) those without prehospital C-SMR were older and more often sustained a low fall as the MOI.

We observed a general increase in no prehospital C-SMR rates in the latter part of the study, with a notable exception in 2016, where the rate was a large uptick in patients with C-spine injuries. We hypothesize that this uptick was related to the adoption of the guideline update at the end of 2015. Concomitantly, the rate of C-SMR in patients without C-spine injury also decreased from 2011 to 2019 (50.6% to 26.8%). Our state’s guidelines emphasized selective spinal motion restriction based on recommendations in the guidelines and the clinical judgment of EMS providers as opposed to universal application of C-SMR in trauma patients. To this end, as all EMS agencies were mandated by the state Department of Health to complete protocol training within the same calendar years, namely 2008 and 2015, the general decline in cervical SMR over the study period would be congruent with protocol adherence.22 23 This is consistent with recent trends, whereby an increasing number of emergency medical service agencies have implemented selective C-SMR.30–32 Reasons for the decreasing use of prehospital C-SMR across the country include questions on the efficacy of cervical collars in injury prevention, potential for spinal manipulation during C-collar application, optimal use of finite resources, time spent (and therefore cost) by prehospital providers in C-SMR application and maintenance throughout the patient transport process, among others previously mentioned.33 34

A further notable finding of this study is that patients who did not receive prehospital C-SMR had many injury patterns that were similar to those with prehospital C-SMR. Both groups had the various components of a vertebrae (ie, lamina, facet, body, etc) injured in similar frequencies, with transverse process and vertebral body fractures being the most common. The number of cervical vertebrae fractures was also not significantly different between groups; nearly 25% of patients in both groups had more than two fractured cervical vertebrae. By vertebral level, the groups also sustained similar frequencies of injuries to C1, C3, C4, C5, and C6.

However, there were also important differences in injury patterns. The most commonly injured vertebra in non-SMR patients was C2; 26.1% of patients having dens fractures. Indeed, C2 type 2 dens fractures have been previously dubbed as geriatric odontoid fractures because of their increased frequency in the elderly low-level fall patients and are thought to be secondary to degenerative changes.35 In contrast, C7 was often injured in C-SMR patients; these patients also had a higher incidence of traumatic vertebral listhesis/subluxation. While C-SMR patients had higher frequencies of cervical cord injuries on MRI, ligamentous injuries on MRI as well as need for cervical spine surgery, rates in the non-SMR group were still substantial. Except for possibly patients who underwent surgery for central cord syndrome, the need for surgery implies a potentially unstable or at least serious spine injury in those not receiving prehospital C-SMR. In turn, delayed diagnosis and management of traumatic C-spine instability can lead to poor outcomes, and timely motion restriction substantially reduces the risk.33 The vast majority of cervical spine injury patients were treated non-operatively with a C-collar in the hospital and at discharge, that is, the cervical spine was only cleared early during hospitalization in nine patients who did not have prehospital C-SMR and did not undergo surgery. Hence, non-SMR patients did not only have ‘trivial’ injuries, but rather the whole gamut of injuries; a substantial number required cervical spine surgery.

The cervical spine injury pattern is also consistent with other findings on univariate analysis in that the non-SMR group had older patients, predominantly Medicare enrollees, with low-level fall as the most common MOI. This suggests potential underutilization of C-SMR in the elderly population and is consistent with previous reports.36 Of note, cervical spine pain or tenderness is not felt to be a reliable indicator of cervical spine fractures as a study of patients aged 55 years or older with cervical spine fractures found that 21% did not have either of these findings.37 Concomitantly, GCS <15 or altered level of consciousness have also not demonstrated a clear association with cervical spine injury.38 We also did not find an association between ED GCS and prehospital C-SMR.

As the geriatric population in the USA continues to grow, an increasing incidence of low falls and resultant injury would be expected.39 According to the 2016 American College of Surgeons National Trauma Databank findings, 39% of trauma victims were over the age of 65 years.40 The geriatric population is more likely to experience spinal injury even after low-energy MOI due to age-related physiological changes.41 Accurate prehospital and in-hospital trauma assessment of the geriatric population is challenged by physiological changes, medications, and comorbid conditions.42 These factors likely contribute to the exclusion of patients over the age of 65 years from C-spine clearance via the Nexus criteria or Canadian C-spine rule.35 36 Considering these factors together suggests the presence of a distinct patient population warranting further consideration of prehospital C-collar application, namely, elderly patients who sustain a low-level fall.

One commonly described complication/concern, namely C-collar-associated decubiti was rarely observed, with a prevalence of 1.1%; these patients had cervical collars for a prolonged period of time in the hospital. Prevalence of C-collar decubiti varies widely among reports, likely due to differences in decubitus prevention protocols or duration of the C-collar than prehospital C-SMR utilization.9 34 43 The increased vigilance for pressure ulcer injuries in recent years has further decreased the incidence. As all patients received a C-collar in the ED, no neurological complications are attributable to the absence of prehospital C-SMR. An additional reason for the lack of attributable neurological complications from no prehospital C-SMR is that various cervical spine injuries are generally inherently stable.13

Conclusions

The components of a vertebra that are fractured in patients with or without prehospital C-SMR are similar. A substantial proportion of patients with or without prehospital C-SMR will have fractures at more than one cervical vertebral level. Patients without prehospital C-SMR demonstrated a distinct injury pattern in the preponderance of dens fractures—fractures that are common in the elderly sustaining low-level falls. The vast majority of C-spine injuries in non-SMR patients were treated with a cervical collar during hospitalization; approximately 8.5% underwent C-spine surgery. These findings challenge the assumption that patients without prehospital C-SMR would have minor/not clinically significant fractures where further evaluation is not needed.