A multivariate logistic regression analysis of risk factors for blunt cerebrovascular injury

J Vasc Surg. 2010 Jan;51(1):57-64. doi: 10.1016/j.jvs.2009.08.071. Epub 2009 Dec 2.


Introduction: The diagnosis of blunt cerebrovascular injuries (BCVI) has improved with widespread adaptation of screening protocols and more accurate multi-detector computed tomography (MDCT-A) angiography. The population at risk and for whom screening is indicated is still controversial. To help determine which blunt trauma patients would best benefit from screening we performed a comprehensive analysis of risk factors associated with BCVI.

Methods: All patients with BCVI from June 12, 2000 (the date at which our institution began screening for these injuries) to June 30, 2009 were identified by the primary author (JDB) and recorded in a prospective database. Associated injuries were identified retrospectively by International Classification of Diseases, Ninth Revision (ICD-9) code and compared with similar patients without BCVI. Demographic information was also compared from data obtained from the trauma registry. Univariate analyses exploring associations between individual risk factors and BCVI were performed using Fisher's exact test for dichotomous variables and Student's t test for continuous variables. Additionally, relative risk (RR) was calculated for dichotomous variables to describe the strength of the relationship between the categorical risk factors and BCVI. Multivariate logistic regression models for BCVI, BCAI (blunt internal carotid artery injury), and BVAI (blunt vertebral artery injury) were developed to explore the relative contributions of the various risk factors.

Results: One hundred two patients with BCVI were identified out of 9935 blunt trauma patients admitted during this time period (1.03% incidence). Fifty-nine patients (0.59% incidence) had a BVAI and 43 patients (0.43% incidence) had a BCAI. Univariate analysis found cervical spine fracture (CSI) (RR = 10.4), basilar skull fracture (RR = 3.60), and mandible fracture (RR = 2.51) to be most predictive of the presence of BCVI (P < .005). Independent predictors of BCVI on multivariate logistic regression were CSI (OR = 7.46), mandible fracture (OR = 2.59), basilar skull fracture (OR = 1.76), injury severity score (ISS) (OR = 1.05), and emergency department Glasgow Coma Scale (ED-GCS) (OR = 0.93): all P < .05.

Conclusions: Blunt trauma patients with a high risk mechanism and a low GCS, high injury severity score, mandible fracture, basilar skull fracture, or cervical spine injury are at high risk for BCVI should be screened with MDCT-A.

MeSH terms

  • Adult
  • Carotid Artery Injuries / diagnostic imaging*
  • Carotid Artery Injuries / epidemiology*
  • Cervical Vertebrae / injuries
  • Early Diagnosis
  • Female
  • Glasgow Coma Scale
  • Head Injuries, Closed / diagnostic imaging*
  • Head Injuries, Closed / epidemiology*
  • Humans
  • Incidence
  • Injury Severity Score
  • Logistic Models
  • Male
  • Mandibular Fractures / epidemiology
  • Mass Screening / methods*
  • Odds Ratio
  • Patient Selection*
  • Predictive Value of Tests
  • Registries
  • Retrospective Studies
  • Risk Assessment
  • Risk Factors
  • Skull Fracture, Basilar / epidemiology
  • Spinal Fractures / epidemiology
  • Tomography, X-Ray Computed*
  • Trauma Centers
  • Vertebral Artery / diagnostic imaging*
  • Vertebral Artery / injuries*