Computed tomography angiography (CTA) is used to rapidly screen for blunt vertebral artery injuries (BCVI),1 which have been associated with devastating and deadly posterior circulation strokes.2 While the expanded Denver Criteria3 is widely used as the benchmark for institutional CTA screening criteria, minimal high-quality evidence supports its use.4 Multiple works argue that the risk of stroke and the complex relationship between BCVI, patient, and injury factors justifies the universal screening of blunt cervical trauma patients to identify all BCVIs.5–7 However, the low sensitivity of CTAs (< 80% in pooled analyses8,9) leads to a high false-positive rate and potentially harmful treatment with anticoagulation if used indiscriminately.10–13 Optimal CTA efficacy therefore requires judicious patient selection and standardized BCVI diagnostic thresholds.
Existing BCVI screening criteria are largely based on the landmark study by Biffl et al14 that led to the original Denver Criteria. The American College of Surgeons Committee on Trauma Advanced Trauma and Life Support (ATLS) guidelines mandate CTA screening for any C1–C3 fracture, cervical spine fractures that enter the foramen transversarium, and cervical fracture subluxations,15,16 without considering mechanism of injury. In contrast, the expanded 2011 Denver Criteria includes nonfracture indications for BCVI screening, such as mandibular fractures and focal neurologic deficits, in addition to the same cervical spine fracture criteria as the ACS standard.3
As long-term deficits from BCVIs that do not lead to a stroke are rare,17 it is critical to identify those lesions likely to cause the posterior circulation strokes that commonly have poor long-term functional outcomes and 25% to 50% mortality rates.4,10,13 While the goals of current screening criteria are to identify every BCVI, a better goal may be to effectively prevent posterior circulation strokes.18 Many of the current screening algorithms for BCVI are driven by retrospective studies of high-energy trauma patients. Indeed, in reviewing the works of Biffl and Scott19,20 no low-energy BCVI progressed to a posterior circulation stroke in either work.
The purpose of the present study is to evaluate a novel two-step algorithm for the use of a CTA by nonspine specialists in the acute trauma setting. Detection rates of BCVI with this novel instrument were compared to the gold standard American College of Surgeons (ACS) or the expanded Denver Criteria. We hypothesized that combining patient factors with fracture characteristics would narrow the indications for a CTA, without missing potentially devastating strokes.
MATERIALS AND METHODS
In an IRB-approved protocol, the institutional trauma database at a large academic tertiary referral level 1 trauma center was retrospectively analyzed for all cervical spine fractures evaluated by the general surgery trauma service from 2012 to 2016. Orthopaedic spine or neurosurgery was consulted to evaluate all cervical spine fractures identified by CT scan. Institutional trauma team protocols based on the Denver Criteria were used to mandate CTA for selected patients, and additional CTAs were ordered at the discretion of the consulting spine team. Excluded were patients with fractures that were more than 1-day old, those with a known vertebral artery injury or intracranial bleed that resulted in prolonged or permanently altered sensorium.
Our two-step CTA screening criteria are highlighted in Figure 1A and B. Major criteria for a CTA were: 1) A high-energy trauma, defined as a fall of > 10 feet, > 12 stairs, or any motor vehicle injury; 2) loss of consciousness at the time of trauma; and 3) an altered mental state or inability to protect the airway during the initial trauma evaluation. Fracture criteria were derived from the existing literature that identified injury types most closely associated with either BCVI or posterior circulation stroke.10,12,21–23
CTA was “always indicated” in patients meeting more than one major criterion, and “never indicated” in patients who did not meet any major criteria. In those patients who only met one major criterion, those who met any fracture criteria were indicated for a CTA. The ACS standard and expanded Denver Criteria were implemented as originally published.3,15
Glasgow coma score (GCS), Injury Severity Score (ISS), Trauma and Injury Severity Score (TRISS), demographic factors including age and body mass index (BMI), and Age-Adjusted Charlson Comorbidity Index (ACCI) were also recorded and analyzed. Estimated costs were derived from published Medicare-fee schedules.5
Statistical analysis was performed using Prism 7.0 (GraphPad, LaJolla, CA). BCVI detection with the novel proposed criteria was compared to gold standard ACS and Denver Criteria. Normality was confirmed using the Kolmogorov–Smirnov test. Continuous variables were compared using Student t test, while categorical variables were compared using Fisher exact test. All values are expressed as mean ± standard deviation unless otherwise noted.
Of the 745 patients screened as part of the trauma work-up for cervical spine fractures, 721 met the inclusion criteria (Figure 2). Patients with and without BCVI and with and without a stroke are compared in Tables 1 and 2. Patients with strokes were significantly younger (42.9 ± 24.2 vs. 62.0 ± 22.4 yrs, P = 0.03), and those with BCVIs had significantly lower ACCIs (2.4 ± 2.4 vs. 3.4 ± 2.9, P = 0.008). Both BCVIs and strokes were associated with a lower GCS and higher ISS.
A total of 185 patients met two major criteria, while 76 met one major criterion and one fracture criterion and were thus indicated for CTA. The remaining 460 patients were not indicated for CTAs by the study algorithm. A total of seven trauma-associated strokes (six posterior circulation, one anterior circulation) were diagnosed during the study period. No strokes were missed with the study protocol.
A stroke incidence of 2.7% (5/185) and a BCVI incidence of 17.3% (32/185) was calculated in patients with two major risk factors, while a stoke incidence of 2.6% (2/76) and a BCVI incidence of 26.3% (20/76) was observed in patients with one major risk factor and at least one fracture criterion. A total of 16 BCVIs (3.5%, 16/406) and no strokes were seen in patients who would not have been indicated for a CTA by the study criteria.
Both the ACS and Denver Criteria indicated a significantly larger number of patients for a CTA versus the study criteria (413/721, 57.3% with ACS and 417/721, 57.8% with Denver vs. 261/721, 36.2% for study criteria, P < = 0.0002 for both) (Table 3). Application of the ACS or Denver Criteria to the study cohort would detect two more BCVIs (54/68 vs. 52/68, P = 0.84), indicate for two fewer endovascular procedures, but miss one stroke (6/7, 85.7% of total). The missed stroke occurred in a 15-year-old female with C6–C7 fractures; she would have been indicated for a CTA by the study criteria given the high-energy mechanism of her injury and her loss of consciousness on scene (she was neurovascularly intact in the trauma bay). The overall proportion of CTAs that yielded a positive BCVI diagnosis was significantly greater when applying the study criteria (52/261, 19.9%) versus either the ACS (54/413, 13.1%, P = 0.02) or Denver Criteria (54/417, 12.9%, P = 0.02).
A total of 417 CTAs were performed on the study group (57.8% of all patients). Of these, 223 (53.5%) were not indicated for a CTA by the proposed criteria (Figure 3). Eleven endovascular procedures were performed on patients who would not have been indicated to undergo further imaging, with “possible BCVI without dissection” noted in all cases. The nonindicated population based on CTA findings accounted for 20.3% (11/54) of all procedures performed. Based on the current non-negotiated Medicare reimbursement of $708 per CTA, and an estimated procedure cost of $2674 per endovascular procedure,24 direct estimated costs of $187,298 would have been saved with the novel protocol during the study period. The comparative benefits of strict adherence to the ACS and Denver Criteria yielded more limited savings, and the additional cost and potential liability of the single missed stroke was estimated to be $609,483.24
As CTAs are a high specificity, low sensitivity screening modality for BCVI in the trauma population, criteria for their use that maximize stroke prevention while minimizing the potential for patient harm and unnecessary cost are critical. A novel algorithm for identifying patients with BCVI at a high risk of progression to a posterior circulation stroke was evaluated and compared to gold standard ACS and expanded Denver criteria in a 4-year retrospective cohort of cervical spine fractures managed at a tertiary-level 1 trauma center.
When compared to the 2012 ACS guidelines and the expanded Denver Criteria, the proposed novel algorithm detected an equivalent number of BCVIs, missed no strokes, and indicated markedly fewer overall patients for CTA. These findings suggest that patient and injury factors may be useful qualifying/disqualifying criteria for CTAs in the setting of cervical spine fractures, necessitating their consideration in future high-quality, prospective analyses.
To date, the trauma literature discussing BCVI screening has largely focused on preventing “misses” rather than avoiding unnecessary procedures.6 While the direct and indirect costs of a missed stroke are substantial,24,25 the substantial cost of these scans in the setting of the unclear benefits of treating low-grade BCVIs26 magnifies the non-negligible risk of complications from CTAs and endovascular procedures. IV contrast-associated nephropathy may occur in 5% to 38% of patients, particularly those who are critically ill.27 Endovascular procedures for false-positive indications expose patients to possible neurologic and vascular sequela. Kaufmann et al's28 retrospective analysis of 19,826 patients who underwent cerebral angiograms over a 22-year period noted a neurologic complication rate of 2.63%, a stroke rate of.14%, and 12 deaths. More recent work suggests that the current complication risk is now between 0.3 and 1.3%,29,30 with a iatrogenic dissection rate of 0.4%.31 Iatrogenic vasospam is thought to occur in up to 20% of patients who undergo an endovascular procedure.32 Age-related atherosclerotic disease was identified as a risk factor for a cerebral angiography complication,29 an important finding as our older population largely sustained low-energy trauma and rarely qualified for a CTA by our metric. It is possible that judicious scan selection may lead to a higher proportion of useful endovascular procedures, minimizing risk where there ought not to be any.
It remains unclear if older patients are more vulnerable to BCVI-related strokes. An inverse relationship between an age over 60 and the risk of a BCVI was noted by Weber et al33 in their large trauma database analysis, a finding that is likely secondary to the increased prevalence of low-energy trauma with age. Odontoid fractures in the setting of low-energy trauma have been found to have a very low risk of BCVI by Durand et al,11 suggesting that the mere presence of a C1–C3 injury does not mandate screening. Thus, age itself was not a consideration for screening with the study criteria, also potentially making it a valuable addition for pediatric applications where the ACS and Denver Criteria are less helpful.
Of note, Anto et al34 cautioned against exempting older patients from BCVI screening, reporting an institutional incidence of BCVI over a 7-year period in the ≥65-year-old population of 37%. Of particular importance is the authors’ finding that many of these BCVI were grade 3–4 (37.5%).34 However, their assertion that screening for BCVIs lowered mortality versus unscreened patients is by the authors’ own admission a product of the selection bias inherent to retrospective studies of broad CTA protocols. Unscanned patients exempted from scans were significantly older, sicker, more likely to have pre-existing renal disease, and had lower admission Glasgow Coma Scores. It is therefore difficult to ascertain if the elevated mortality rate of unscreened older patients is due to their injuries, or their comorbidities.
While suggested as “discretionary” indications for a CTA in the 2012 ACS screening criteria, few patient factors have been studied for their association with BCVI. Berne et al23 do note a small but significant (OR 0.93, P < 0.05) increase in the risk of BCVI in patients with reduced GCS. Biffl et al's14 criteria indicate an immediate CTA for patients with a GCS ≤ 6 or a focal neurologic deficit, although loss of consciousness and generalized altered sensorium were not included. High-energy trauma has been noted in several sources as a risk factor for BCVI.10,33 However, to the authors’ knowledge no prior criteria have considered a loss of consciousness to be an indication for BCVI screening. The success of this novel algorithm for the screening of BCVI is likely due in large part to the inclusion of loss of consciousness as a major criterion. In aggregate, the present study suggests that patient presentation and on-scene history is likely as important as cervical spine fracture characteristics. This is especially important in a level-one trauma setting, as the acuity of the patient presentation makes it difficult to differentiate between causes of altered mental status.35
The management of BCVI detected on CTA remains at the discretion of the treating physician. The efficacy of any single intervention is still unclear, and prior survey work has found that treatment selection is both individual and specialty dependent.36 This variability is likely to be related to the shortcomings of a retrospective assessment of the ability of prophylactic therapy to prevent an undesirable outcome. Stein et al37 note that some treatment modality, be it antiplatelet medications, systemic anticoagulation, or an endovascular procedure, in the setting of a BCVI can reduce the overall incidence of a stroke in the trauma population. However, the authors concede that it is impossible to retrospectively assess the relative efficacy of any individual therapy used in isolation. Griffin et al38 found that the blanket prescription of aspirin to all patients with a BCVI led to an increased risk of blood transfusions (RR 1.7, 1.32–2.20) within 24 hours of admission without an overt hemorrhagic event, particularly in older (> 50 yrs) patients. However, antiplatelet agents were found to be safer and more effective39 than systemic anticoagulation with heparin in the setting of carotid injuries. This is likely due to the artery–artery thromboembolic pathophysiology behind BCVI-related strokes.40 Endovascular treatment is generally safe but is far more expensive than other treatments and requires an OR and treatment team. It is therefore typically considered only in cases of vertebral artery dissection or a symptomatic BCVI.41 Further clinical analyses are necessary to establish a postdiagnostic BCVI management protocol, which may be challenging given the relatively low incidence and late presentation of symptomatic BCVI, and the continued debate as to the effectiveness of stroke prevention in those patients in whom they occur.37
There are several limitations to the present study. The proposed algorithm for the use of CTA to detect BCVI is derived from retrospective data and has not been clinically validated using prospective data. There is little doubt that large prospective analyses and external validation are necessary to confirm the predictive value of these criteria, as retrospectively derived single-center algorithms suffer from inevitable selection bias. Additionally, while the study was adequately powered to detect differences between the novel, ACS, and Denver Criteria, stroke remains a low occurrence event.
Only 555 of 721 (77.0%) of the patients in the study group, regardless of their risk of BCVI as estimated by our algorithm, underwent a CTA. While this scan rate is much higher than what is reported in the literature,42 it is possible that BCVIs that did not receive scans were missed in the study group.
Finally, the purpose of the study was not necessarily to identify all BCVIs, but to identify those at risk of a stroke, which was successful. This emphasis also acknowledges the unclear value of BCVIs after low-energy trauma, making further analysis of this group critical. However, the difficulty in asserting that no high-risk BCVI is missed by the proposed criteria highlights the importance of more extensive validation of this work with larger sample sizes over a longer time period.
In conclusion, the present study proposes a novel metric for indicating BCVI screening after cervical spine fracture. These criteria were validated using 4 years of retrospective data at a large tertiary-level 1 trauma center and identified all strokes that occurred during the study period, while requiring significantly fewer screenings than currently accepted gold standard criteria. Prospective validation will be necessary to establish the clinical superiority of this metric.
1. Biffl WL, Egglin T, Benedetto B, et al. Sixteen-slice computed tomographic angiography is a reliable noninvasive screening test for clinically significant blunt cerebrovascular injuries
. J Trauma
2. Burlew CC, Biffl WL. Blunt cerebrovascular trauma. Curr Opin Crit Care
3. Geddes AE, Burlew CC, Wagenaar AE, et al. Expanded screening criteria for blunt cerebrovascular injury: a bigger impact than anticipated. Am J Surg
4. Brommeland T, Helseth E, Aarhus M, et al. Best practice guidelines for blunt cerebrovascular injury (BCVI
). Scand J Trauma Resusc Emerg Med
5. Cothren CC, Moore EE, Ray CE Jr, et al. Cervical spine fracture patterns mandating screening to rule out blunt cerebrovascular injury. Surgery
6. Jacobson LE, Ziemba-Davis M, Herrera AJ. The limitations of using risk factors to screen for blunt cerebrovascular injuries
: the harder you look, the more you find. World J Emerg Surg
7. Cook A, Osler T, Gaudet M, et al. Blunt cerebrovascular injury is poorly predicted by modeling with other injuries: analysis of NTDB data. J Trauma
8. Stengel D, Rademacher G, Hanson B, et al. Screening for blunt cerebrovascular injuries
: the essential role of computed tomography angiography
. Semin Ultrasound CT MR
9. Roberts DJ, Chaubey VP, Zygun DA, et al. Diagnostic accuracy of computed tomographic angiography for blunt cerebrovascular injury detection in trauma patients: a systematic review and meta-analysis. Ann Surg
10. Lockwood MM, Smith GA, Tanenbaum J, et al. Screening via CT angiogram after traumatic cervical spine fractures
: narrowing imaging to improve cost effectiveness. Experience of a Level I trauma center. J Neurosurg Spine
11. Durand D, Wu X, Kalra VB, et al. Predictors of vertebral artery injury
in isolated C2 fractures based on fracture morphology using CT angiography. Spine (Phila Pa 1976)
12. Fassett DR, Dailey AT, Vaccaro AR. Vertebral artery injuries associated with cervical spine injuries: a review of the literature. J Spinal Disord Tech
13. Dreger T, Place H, Mattingly T, et al. Analysis of cervical angiograms in cervical spine trauma patients, does it make a difference? Clin Spine Surg
14. Biffl WL, Moore EE, Offner PJ, et al. Optimizing screening for blunt cerebrovascular injuries
. Am J Surg
15. ATLS Subcommittee; American College of Surgeons’ Committee on Trauma; International ATLS working group. Advanced trauma life support (ATLS(R)): the ninth edition. J Trauma Acute Care Surg
16. Shafafy R, Suresh S, Afolayan JO, et al. Blunt vertebral vascular injury in trauma patients: ATLS((R)) recommendations and review of current evidence. J Spine Surg
17. Alterman DM, Heidel RE, Daley BJ, et al. Contemporary outcomes of vertebral artery injury
. J Vasc Surg
18. Spaniolas K, Velmahos GC, Alam HB, et al. Does improved detection of blunt vertebral artery injuries lead to improved outcomes? Analysis of the National Trauma Data Bank. World J Surg
19. Biffl WL, Moore EE, Offner PJ, et al. Blunt carotid arterial injuries: implications of a new grading scale. J Trauma
20. Scott WW, Sharp S, Figueroa SA, et al. Clinical and radiological outcomes following traumatic Grade 1 and 2 vertebral artery injuries: a 10-year retrospective analysis from a Level 1 trauma center. J Neurosurg
21. Inamasu J, Guiot BH. Vertebral artery injury
after blunt cervical trauma: an update. Surg Neurol
22. Nakajima H, Nemoto M, Torio T, et al. Factors associated with blunt cerebrovascular injury in patients with cervical spine injury. Neurol Med Chir (Tokyo)
23. Berne JD, Cook A, Rowe SA, et al. A multivariate logistic regression analysis of risk factors for blunt cerebrovascular injury. J Vasc Surg
24. Cothren CC, Moore EE, Ray CE Jr, et al. Screening for blunt cerebrovascular injuries
is cost-effective. Am J Surg
25. Shafafy R, Suresh S, Afolayan JO, et al. Blunt vertebral vascular injury in trauma patients: ATLS(®
) recommendations and review of current evidence. J Spine Surg
26. Hagedorn JC 2nd, Emery SE, France JC, et al. Does CT angiography matter for patients with cervical spine injuries? J Bone Joint Surg Am
27. Trivedi M, Prabhahar T, Preller J. Prevention of contrast induced nephropathy in the critically ill, Conference Paper in Intensive Care Medicine, September 2009.
28. Kaufmann TJ, Huston J 3rd, Mandrekar JN, et al. Complications of diagnostic cerebral angiography: evaluation of 19,826 consecutive patients. Radiology
29. Willinsky RA, Taylor SM, TerBrugge K, et al. Neurologic complications of cerebral angiography: prospective analysis of 2,899 procedures and review of the literature. Radiology
30. Fifi JT, Meyers PM, Lavine SD, et al. Complications of modern diagnostic cerebral angiography in an academic medical center. J Vasc Interv Radiol
31. Cloft HJ, Jensen ME, Kallmes DF, et al. Arterial dissections complicating cerebral angiography and cerebrovascular interventions. AJNR Am J Neuroradiol
32. Akpinar SH, Yilmaz G. Periprocedural complications in endovascular stroke treatment. Br J Radiol
33. Weber CD, Lefering R, Kobbe P, et al. Blunt cerebrovascular artery injury and stroke in severely injured patients: an international multicenter analysis. World J Surg
34. Anto VP, Brown JB, Peitzman AB, et al. Blunt cerebrovascular injury in elderly fall patients: are we screening enough? World J Emerg Surg
35. Jang J-W, Lee J-K, Hur H, et al. Vertebral artery injury
after cervical spine trauma: a prospective study using computed tomographic angiography. Surg Neurol Int
36. Harrigan MR, Weinberg JA, Peaks Y-S, et al. Management of blunt extracranial traumatic cerebrovascular injury: a multidisciplinary survey of current practice. World J Emerg Surg
37. Stein DM, Boswell S, Sliker CW, et al. Blunt cerebrovascular injuries
: does treatment always matter? J Trauma
38. Griffin RL, Falatko SR, Aslibekyan S, et al. Aspirin for primary prevention of stroke in traumatic cerebrovascular injury: association with increased risk of transfusion. J Neurosurg
2018; 1–8. Epub ahead of print.
39. Wahl WL, Brandt MM, Thompson BG, et al. Antiplatelet therapy: an alternative to heparin for blunt carotid injury. J Trauma
40. Griessenauer CJ, Fleming JB, Richards BF, et al. Timing and mechanism of ischemic stroke due to extracranial blunt traumatic cerebrovascular injury. J Neurosurg
41. Mei Q, Sui M, Xiao W, et al. Individualized endovascular treatment of high-grade traumatic vertebral artery injury
. Acta Neurochir (Wien)
42. Lebl DR, Bono CM, Velmahos G, et al. Vertebral artery injury
associated with blunt cervical spine trauma: a multivariate regression analysis. Spine (Phila Pa 1976)