a b s t r a c t

Introduction: The National Emergency X-Radiography Utilization Study (NEXUS) clinical decision rule is extremely sensitive for clearance of cervical spine (C-spine) injury in blunt trauma patients with distracting injuries.

Objectives: We sought to determine whether the Nexus criteria would maintain sensitivity for blunt trauma patients when femur fractures were not considered a distracting injury and an absolute indication for diagnostic imaging.

Methods: We retrospectively analyzed blunt trauma patients with at least 1 femur fracture who presented to our emergency department as Trauma activations from 2009 to 2011 and underwent C-spine injury evaluation. Presence of C-spine injury requiring surgical intervention was evaluated.

Results: Of 566 trauma patients included, 77 (13.6%) were younger than 18 years. cervical spine injury was diagnosed in 53 (9.4%) of 566. A total of 241 patients (42.6%) had positive NEXUS findings in addition to distracting injury; 51 (21.2%) of these had C-spine injuries. Of 325 patients (57.4%) with femur fractures who were otherwise NEXUS negative, only 2 (0.6%) had C-spine injuries (95% confidence interval [CI], 0.2%-2.2%); both were stable and required no operative inter- vention. Use of NEXUS criteria, excluding femur fracture as an indication for imaging, detected all Significant injuries with a sensitivity for any C-spine injury of 96.2% (95% CI, 85.9%-99.3%) and negative predictive value of 99.4% (95% CI, 97.6%-99.9%).

Conclusions: In our patient population, all significant C-spine injuries were identified by NEXUS criteria without considering the femur fracture a distracting injury and indication for Computed tomographic imaging. Reconsidering femur fracture in this context may decrease radiation exposure and health care expenditure with little risk of missed diagnoses.

(C) 2015

Introduction

Cervical spine (C-spine) injury after trauma carries great morbidity, with a subsequent lifetime cost of care often in excess of US $1000000 per affected patient [1]. Furthermore, a missed or delayed diagnosis of C-spine injury results in up to 10 times the rate of neurologic injury, with 29.4% of these cases resulting in permanent neurologic deficit [2]. Missed or delayed diagnoses have been attributed commonly to inadequate or misinterpreted radiographic evaluation [3]. A 2006 review of 367 spinal injuries described a 4.9% incidence of delayed or missed diagnosis [4]. Ample literature highlights why C-spine injuries are addressed in a conservative manner.

The cost and radiation effect associated with computed tomographic (CT) imaging has called for a more judicious use of the technology.

? The authors declare no source of funding for this study.

* Corresponding author at: Carolinas Medical Center, 1000 Blythe Blvd, MEB 6th Floor, Charlotte, NC 28204. Tel.: +1 704 355 6904; fax: +1 704 355 5619.

E-mail address: [email protected] (R.F. Sing).

A recent analysis noted a cost of greater than US $50000 per quality- adjusted life-year for populations with a fracture incidence of less than 2.8% and called into question the cost-effectiveness of “blanket” CT scanning [5]. In addition, recent studies have brought radiation consequences into consideration. Muchow et al [6] described in 2012 an estimated median excess relative risk of thyroid cancer after 1 CT scan of the C-spine in pediatric patients at 13% for men and 25% for women. Fortunately, a large body of literature suggests a high degree of utility of clinical examination for C-spine injury [7-10], mitigating the need for radiographic analysis.

The National Emergency X-Radiography Utilization Study (NEXUS) clinical decision rule is widely used to exclude C-spine injury in blunt trauma patients and thereby avoid unnecessary imaging and the associated expense and radiation risk. As originally described, the NEXUS guidelines suggest cervical radiography for blunt trauma pa- tients with any of the following high-risk criteria: (1) a focal neurologic deficit, (2) midline C-spine tenderness on examination, (3) Altered level of consciousness, (4) intoxication, and (5) presence of distracting injury. According to the criteria, “distracting injury” includes any or all of the following: (1) a long bone fracture; (2) a visceral injury requiring surgi- cal consultation; (3) a large laceration, degloving injury, or Crush injury;

http://dx.doi.org/10.1016/j.ajem.2015.08.009

0735-6757/(C) 2015

(4) large burns; or (5) any other injury producing acute functional impairment or another injury determined based on clinician gestalt to be potentially distracting. Despite the vague elements of the distracting injury criteria, the interobserver reliability for the rule was acceptable in the trial (?, 0.73), and the NEXUS CDR has subsequently grown into widespread use [11,12].

The distracting injury criteria, however, have been the subject of much debate. In the original series, these criteria were the indication for more than 30% of all cervical radiography tests. This CDR is based sci- entifically on the counterirritation phenomenon of pain, which suggests that the perception of pain can be altered by other noxious neurologic stimuli, if present simultaneously [13,14]. Studies have long shown that the counterirritation effect is correlated to the proximity as well as the amplitude of stimulus [15]. Recent published literature suggests that upper torso injuries may have a greater effect than lower extremity injuries on sensory inhibition of C-spine tenderness [11]; however, this effect has not been scientifically quantified nor fully explained, and studies have shown conflicting results, depending on the type of stimulus [14].

Research has been conducted to further qualify the need for imaging in the presence of distracting injury. Currently, conflicting data exist. A 2001 study that assessed the performance of each individual NEXUS cri- terion found that 39 patients with C-spine column injury met only the distracting injury criteria. This suggests an unacceptably low CDR sensi- tivity (93.5%) if the distracting injury criteria are removed [16]. In con- trast, a 2005 investigation reviewing 4698 patients found that only 2.4% of patients with only distracting injury as an indication for imaging had spinal fractures, with only 1 injury being cervical and none requir- ing operative intervention. The investigators also evaluated the type of distracting injury and found only bony fractures (such as femur frac- ture) to impact the sensitivity of clinical screening [17]. Further evi- dence has mounted to suggest minimal impact of distracting injury on cervical examination, with Rose et al [18] demonstrating a sensitivity and negative predictive value greater than 99% for the NEXUS criteria in patients with distracting injuries.

Insufficient literature exists to evaluate the ability of the NEXUS criteria to safely evaluate C-spine injury in the scenario of a lower ex- tremity fracture requiring operative intervention. In patients with a femur fracture, the rate of C-spine injury can be as high as 10%, and treatment of femur fractures generally requires endotracheal intubation and operative intervention [2]. Given the importance of bony fractures, in particular, on cervical neck examination, our objective was to deter- mine whether the NEXUS criteria would maintain sensitivity for blunt trauma patients with femur fractures if the fracture is not considered a distracting injury and an absolute indication for diagnostic imaging.

Methods

Study population

This retrospective study was conducted for consecutive adult and pediatric patients presenting to the emergency department (ED) of a large, level I trauma center in the southeastern United States between 2009 and 2011. All patients included in the study were consecutive trauma activations after blunt trauma who were evaluated for C-spine injury with imaging and who also had at least 1 femur fracture. Patients with additional potentially distracting injuries were included in the study.

Patients were included only if a complete documented examination was performed sufficiently to include all of the NEXUS criteria before any imaging obtained. Patients were excluded upon (1) death before imaging, (2) transferal from another hospital without documented examinations before imaging, or (3) involvement in low-mechanism trauma (falls from standing or injuries sustained from contact sports) with no apparent clinical or radiographic evaluation for cervical injury. Pediatric patients were included in this study as in the original NEXUS

validation study. To date, the proportion of pediatric patients included in such studies are low (2.5% age 8 years or younger in NEXUS). In addi- tion, spinal cord injury without radiographic abnormality can go unde- tected with CT or x-ray imaging alone. For this reason, although the NEXUS criteria can be assessed and used in this population, a conservative approach is taken at our institution that consists of C-spine immobiliza- tion and serial examinations as adjuncts to radiographic evaluation.

The primary end point was the presence of C-spine injury requiring operative intervention. The study population, methods, and protocol were reviewed and approved by the institutional review board of our institution. Because of the retrospective nature of the study and data handling and protections undertaken, patient consent was waived.

Evaluation

Patients arrived to the ED in spinal immobilization per prehospital provider protocols and subsequently underwent clinical evaluation. Pa- tient evaluations were documented by either surgical or emergency medicine residents or attending physicians. Results of these evaluations were examined on review of the electronic medical record (CERNER PowerNote). Pertinent data reviewed included all of the NEXUS criteria:

(1) any evidence of deficit on neurologic examination, (2) presence or absence of midline C-spine tenderness on examination, (3) evidence of altered level of consciousness, (4) evidence of drug or alcohol intoxication, and (5) presence of distracting injury. All subsequent ra- diographic images were also reviewed for each patient, with interpreta- tions provided by board-certified radiologists.

Outcome measures

The primary outcome was any C-spine injury requiring operative in- tervention. For patients who did not require operative intervention, other interventions that were prescribed (eg, cervical collar) were gathered and recorded.

Data collection

Methodological strategies were used in accordance with the recom- mendations of Gilbert et al [19] to enhance validity, reproducibility, and overall quality of data collected from the ED medical records and the institution-based trauma database by 2 abstractors (HD and AR). The abstractors were trained in data collection and supervised by the pri- mary investigator (RS) to ensure accuracy of data collection. Precisely defined variables were used to collect data; these included patient de- mographics, including age and sex; prehospital and ED Glasgow Coma Scale score; presence or absence of intoxication, including blood ethanol levels; ED vital signs; method of prehospital spine immobilization; re- sult of physical examination that included the C-spine; type of radiogra- phy performed and results; ultimate disposition; and neurosurgical or orthopedic spine treatment (if applicable). Interrater reliability was 100% as determined by comparison of a subsample (10%) of charts ab- stracted by both researchers.

Statistical analysis

Data were compiled into a spreadsheet (Microsoft Office Excel 2003; Microsoft Corporation, Redmond, WA) and were subsequently analyzed using standard statistical methods; P b .05 was considered to be statisti- cally significant. Descriptive statistics including mean +- SDs, counts, and percentages were used to describe the study population on all var- iables, and 95% confidence intervals (CIs) were calculated to further de- scribe sensitivities, specificities, and all predictive value calculations. Comparisons of statistical performance were made between our study and the original NEXUS validation trial using Fisher exact test. The SAS System version 8.02 (Cary, NC) was used to complete all statistical analyses.

Results

The study population included 566 trauma patients with at least 1 femur fracture. Of these 566 patients, 77 (13.6%) were younger than 18 years, and the age range was 2 months to 99 years (median age, 34 years). The population was 65.4% male. The mechanisms of injury were as follows: 315 Motor vehicle collisions (55.7%), 85 Motorcycle collisions (15.0%), 71 falls (12.5%), 41 pedestrians struck by motor vehicle, (7.2%), and 58 (10.2%) other mechanism (eg, all-terrain vehicle accidents, industrial accidents, assault). The average Injury Severity Score (ISS) for the cohort was 18.5, with a major trauma commonly de- fined as a score of 15 or greater [20].

All patients received C-spine imaging after clinical evaluation. In

almost all cases, the imaging modality was a C-spine CT scan, with the exception of 9 pediatric patients, 7 of whom received plain films of the C-spine and 2 of whom receivED magnetic resonance imaging. There were no cases of spinal cord injury without radiographic abnor- mality in our study population. Overall, 53 patients (9.4%) were diag- nosed with a C-spine injury. Ultimately, 8 of these patients (15.1%) went to the operating room for spinal fixation, but all of these had mul- tiple positive NEXUS criteria in addition to distracting injury.

Of the initial 566 patients, 241 patients (42.6%) had positive NEXUS findings on examination, in addition to distracting injury, that is, posterior C-spine tenderness, Glasgow Coma Scale score less than 15, intoxication, or neurologic deficit (Figure), and of these, 51 (21.2%) had C-spine injuries. Of the 325 patients (57.4%) who were negative by NEXUS criteria with the exception of their distracting injury (femur

fracture), only 2 (0.6%) had C-spine injuries (95% CI, 0.2%-2.2%); both in- juries were stable and did not require operative intervention (Table 1). The application of the NEXUS criteria in this study population, disregarding femur fracture as a distracting injury, had a sensitivity for C-spine injury of 96.2% (95% CI, 85.9%-99.3%) and negative predictive value of 99.4% (95% CI, 97.5%-99.9%) (Table 2). Compared with the perfor- mance of the criteria in the original NEXUS study, sensitivity and negative predictive value for detecting fracture did not differ significantly different.

Discussion

These data support our hypothesis that femur fracture should not qualify as a distracting injury by NEXUS criteria and should not be an absolute indication for imaging of the C-spine. In this population, disregarding the femur fracture as a distracting injury and imaging only those patients with other positive NEXUS criteria would result in a 57% decrease in C- spine CT usage with missed detection of 2 nonoperative C-spine fractures. Statistically, this approach produced a sensitivity and negative predictive value (96.2% and 99.4%, respectively) that performed well and were not sta- tistically different from the sensitivity and negative predictive value of the NEXUS validation study when compared with Fisher exact test (Table 2).

This patient population showed a higher incidence of C-spine injury detected on radiography (9.4% overall) than that of the 34069-patient NEXUS cohort, where the incidence of fracture was 2.4%. This elevated incidence is explained by the fact that trauma patients with femur frac- tures have higher ISSs than those associated with any other fracture.

566 trauma activations with femur fractures (all NEXUS positive

by distracting injury)

325 (57.4%)

otherwise NEXUS negative

241 (42.6%) with

positive NEXUS criteria in addition to distracting injury

325 (100%) received

cervical imaging, 3 pediatric X-ray only

241 (100%) received

cervical imaging, 4 pediatric X-rays and 2 pediatric MRI

2 (0.6%) cervical spine injury, non-

operatively treated

323 (99.4%) no

cervical spine injury

51 (21.2%) cervical

spine injury

190 (78.8%) no

cervical spine injury

8 (15.7%) requiring

operative fixation

Figure. Distribution of patients by NEXUS finding and cervical injury status.

Table 1

Cervical spine injuries in patients with distracting injury but otherwise NEXUS negative

Demographics	Mechanism of injury	Cervical injury	Other injuries	Management	ISS
Female, 42 y Male, 49 y	motorcycle crash	Type I C2 pedicle fracture (stable) Left C2 superior articular surface fracture	Right femur fracture Right tibia/fibula fracture Fifth finger fracture Middle finger fracture Clavicle fracture T2 vertebral body fracture	Aspen collar x 6 wk Aspen collar x 6 wk	13 22
	Motorcycle crash	Left C6 superior facet fracture Small amount of possible epidural hemorrhage C2-C3 without mass effect	L1 compression fracture Left Humerus fracture Left femur fracture Right fibular fracture Right rib fracture
			Right scapular fracture right hemothorax

Abbreviations: C, cervical; T, thoracic; L, lumbar.

This rate of C-spine injury in this study population is similar to the rates of 5% to 11% of C-spine injury evidenced in admitted trauma patients at other centers as previously described [21,22]. As originally described with its distracting injury criteria, the NEXUS CDR demonstrated a sen- sitivity of 99.6% (95% CI, 98.6%-100%) and negative predictive value of 99.9% (95% CI, 99.8%-100%). Our approach, disregarding femur fracture as a distracting injury, produced sensitivity and negative predictive values that were not significantly different (P N .05) in performance from the NEXUS validation study (Table 2). Obviously, all fractures must be properly immobilized to allow for an appropriate clinical evaluation of the neck and C-spine.

Furthermore, a relative strength of this study, in terms of ability to demonstrate sensitivity and negative predictive value, is the higher incidence of disease in this population. The difference in the positive predictive value and specificity of our criteria compared to the NEXUS validation study is explained by the substantial difference in disease prevalence in the 2 populations.

Limitations

This study has several limitations. First, a sample size of 566 patients is relatively small when compared with large trials such as the original NEXUS study. This study was conducted at a large level I trauma center, where ISSs and disease prevalence are higher than at smaller rural or community EDs. The relative experience of the providers in our institu- tion in diagnosis and management of trauma is, therefore, a limitation and may call into question the applicability of this study to patients and providers in centers who less commonly manage injured patients. Finally, this study is subject to the weaknesses inherent in any retro- spective investigation.

Table 2

Performance of statistical variables

Conclusions

These data support the assertion that if femur fracture were not con- sidered a distracting injury, the NEXUS CDR would maintain acceptable sensitivity and negative predictive value in excluding C-spine injury. In this study population, the NEXUS CDR would have identified all unsta- ble C-spine injuries requiring operative intervention and would have re- duced the use of C-spine CT scans substantially. The NEXUS low-risk criteria may allow for avoidance of C-spine radiography, even in pa- tients with femur fracture, which would significantly decrease radiation exposure and unnecessary health care expenditure.

Funding source

The authors declare no source of funding for this study.

Acknowledgments

The authors thank Jennifer C. Barnes, PhD, ELS, for editorial assistance with this manuscript.

References

1. Wyndaele M, Wyndaele JJ. Incidence, prevalence and epidemiology of spinal cord in- jury: what learns a worldwide literature survey? Spinal Cord 2006;44:523-9.
2. Reid DC, Henderson R, Saboe L, Miller JD. Etiology and clinical course of missed spine fractures. J Trauma 1987;27:980-6.
3. Davis JW, Phreaner DL, Hoyt DB, Mackersie RC. The etiology of missed cervical spine injuries. J Trauma 1993;34:342-6.
4. Platzer P, Hauswirth N, Jaindl M, Chatwani S, Vecsei V, Gaebler C. Delayed or missed diagnosis of cervical spine injuries. J Trauma 2006;61:150-5.
5. Blackmore CC, Ramsey SD, Mann FA, Deyo RA. Cervical spine screening with CT in trauma patients: a cost-effectiveness analysis. Radiology 1999;212:117-25.
6. Muchow RD, Egan KR, Peppler WW, Anderson PA. Theoretical increase of thyroid cancer induction from cervical spine Multidetector computed tomography in pediat- ric trauma patients. J Trauma Acute Care Surg 2012;72:403-9.
7. Gonzalez RP, Fried PO, Bukhalo M, Holevar MR, Falimirski ME. Role of clinical exam- ination in screening for blunt cervical spine injury. J Am Coll Surg 1999;189:152-7.
8. Kreipke DL, Gillespie KR, McCarthy MC, Mail JT, Lappas JC, Broadie TA. Reliability of indications for cervical spine films in trauma patients. J Trauma 1989;29:1438-9.
9. Roberge RJ, Wears RC. Evaluation of neck discomfort, Neck tenderness, and neuro-

   NEXUS performance without distracting injury criteria in this cohort

   NEXUS performance in original validation study

   logic deficits as indicators for radiography in blunt Trauma victims. J Emerg Med 1992;10:539-44.

   Velmahos GC, Theodorou D, Tatevossian R, Belzberg H, Cornwell III EE, Berne TV, et al. Radiographic cervical spine evaluation in the alert asymptomatic blunt trauma victim: much ado about nothing. J Trauma 1996;40:768-74.

   Variable	Performance	95% CI		Performance	95% CI	P
   Sensitivity	96.2%	85.9%-99.3%		99.0%	98.0%-99.6%	.1195
   Specificity	63.0%	58.6%-67.1%		12.9%	12.8%-13.0%	b.0001
   PPV	21.2%	16.3%-27.0%		2.7%	2.6%-2.8%	b.0001
   NPV	99.4%	97.5%-99.9%		99.8%	99.6%-100%	.1521
   Positive LR	2.60	2.30-2.94		1.14	1.13-1.15	–
   Negative LR	0.06	0.01-0.25		0.08	0.04-0.15	–

   Heffernan DS, Schermer CR, Lu SW. What defines a distracting injury in cervical spine assessment? J Trauma 2005;59:1396-9.

10. Hoffman JR, Wolfson AB, Todd K, Mower WR. Selective Cervical spine radiography in blunt trauma: methodology of the National Emergency X-Radiography Utilization Study (NEXUS). Ann Emerg Med 1998;32:461-9.
11. Price DD, McHaffie JG. Effects of heterotopic conditioning stimuli on first and second pain: a psychophysical evaluation in humans. Pain 1988;34:245-52.
12. Bouhassira D, Danziger N, Attal N, Guirimand F. Comparison of the pain suppressive

    Abbreviations: PPV, positive predictive value; NPV, negative predictive value; LR, likelihood ratio.

    effects of clinical and experimental painful conditioning stimuli. Brain 2003;126: 1068-78.

    Bouhassira D, Le Bars D, Villanueva L. Heterotopic activation of A delta and C fibres triggers inhibition of trigeminal and spinal convergent neurones in the rat. J Physiol 1987;389:301-17.

13. Panacek EA, Mower WR, Holmes JF, Hoffman JR. Test performance of the individual NEXUS low-risk clinical screening criteria for cervical spine injury. Ann Emerg Med 2001;38:22-5.
14. Chang CH, Holmes JF, Mower WR, Panacek EA. Distracting injuries in patients with vertebral injuries. J Emerg Med 2005;28:147-52.
15. Rose MK, Rosal LM, Gonzalez RP, Rostas JW, Baker JA, Simmons JD, et al. Clinical clearance of the cervical spine in patients with distracting injuries: it is time to dispel the myth. J Trauma Acute Care Surg 2012;73:498-502.
16. Gilbert EH, Lowenstein SR, Koziol-McLain J, Barta DC, Steiner J. Chart reviews in emergency medicine research: where are the methods? Ann Emerg Med 1996; 27:305-8.
17. Baker SP, O’Neill B, Haddon Jr W, Long WB. The injury severity score: a method for describing patients with multiple injuries and evaluating emergency care. J Trauma 1974;14:187-96.
18. Grossman MD, Reilly PM, Gillett T, Gillett D. National survey of the incidence of cervical spine injury and approach to cervical spine clearance in U.S. trauma centers. J Trauma 1999;47:684-90.
19. Robertson A, Branfoot T, Barlow IF, Giannoudis PV. Spinal injury patterns resulting from car and Motorcycle accidents. Spine (Phila Pa 1976) 2002;27:2825-30.