Abstract
-
Objective
Traumatic vertebral artery injuries (tVAIs) are uncommon but potentially devastating if missed. While computed tomography angiography (CTA) is routinely used for diagnosis, data on the number needed to image (NNI) remain limited. We hence analyzed tVAI epidemiology and imaging practices at a major Scandinavian level 1 trauma center.
-
Methods
A retrospective study (2013–2020) was performed based on a single-center trauma registry. Patients were grouped based on CTA imaging protocol used; selective screening (2013–2017) and universal screening (2018–2020). Imaging protocols, treatment strategies, and outcomes were analyzed.
-
Results
Among 2,843 patients admitted with level 1 trauma and receiving CTA imaging, 62 had a tVAI (2.2%) yielding a NNI of 46 patients to diagnose 1 tVAI. Twenty-five of these patients (40.3%) were found to have a posterior circulation stroke, resulting in an incidence of 0.9%, and a NNI of 114 to diagnose 1 stroke on CTA. NNIs for both tVAI and stroke detection increased with adoption of universal screening (tVAI: 35→65; stroke: 90→149). However, the detection rate of tVAI during the universal screening period was not significantly higher than during the selective screening period (p=0.261).
-
Conclusion
In our level 1 trauma cohort, the incidence of tVAI was 2.2% and stroke rate 0.9%. The NNI rose with universal screening, yet detection rates did not improve. These findings suggest that selective screening based on risk factors may be more efficient than a universal approach. Further research is needed to balance diagnostic accuracy with resource use in trauma care.
-
Keywords: Trauma, Cervical spine, Blunt cervical vascular injury, Vertebral artery, Computed tomography, Angiography
INTRODUCTION
Blunt cerebrovascular injuries (BCVIs) are relatively uncommon vascular traumas affecting carotid and vertebral arteries, with traumatic vertebral artery injuries (tVAIs) in cervical spine trauma now recognized as critical for prompt diagnosis and early intervention to prevent neurological complications [
1]. While the reported incidence of tVAI in general trauma is approximately 1%, it can reach up to 11% in patients with specific blunt trauma criteria such as traumatic brain injury (TBI) and cervical spine injuries, with patients having cervical transverse foramen fractures or facet dislocations showing an associated tVAI rate as high as 27.5% [
1-
4]. The substantial variability in reported incidence rates can be attributed to differences in study populations, inconsistent trauma management approaches, and diverse BCVI screening protocols implemented across healthcare institutions [
5-
7].
The Denver criteria are screening guidelines that help identify BCVIs in trauma patients. Originally developed in 1996 and updated in 2005 and 2012, these criteria have been progressively refined to determine which patients need computed tomography (CT) angiography (CTA), improving early detection and reducing stroke risk [
8]. BCVI are classified according to the Biffl scale, which categorizes vascular injuries into 5 grades [
9]. The 5 tier system scores vascular injuries in ascending order of severity: dissections (I–II), pseudoaneurysms (III), occlusions (IV), and complete transections (V) [
5,
9]. According to recent systematic reviews, stroke risk varies significantly by BCVI grade, with higher-grade injuries (III–IV) associated with a 10% stroke frequency compared to 2%–3% for lower-grade injuries (I–II), while grade V injuries remain too rare in the literature to support conclusive statistical analysis [
1]. Despite medical treatment, tVAI carry a substantial stroke risk of approximately 5%–9% with evidence suggesting that early initiation of antithrombotic therapy may reduce these rates; however, an optimal treatment protocol remains undefined, requiring careful consideration of bleeding risks and potential surgical interventions in polytraumatized patients [
1,
4]. Heightened clinical awareness has resulted in increased screening for tVAI [
10]. However, the low incidence of tVAI in general trauma screening results in a large number needed to image (NNI) values raising questions about cost-effectiveness and clinical utility. Improved screening strategies targeting subgroups at risk may alleviate these concerns [
3]. Subgroups that are more likely to have sustained tVAI include patients with cervical fractures, especially C1–3 and fractures involving the transverse foramen, as well as subluxations and injuries to the facet joints [
1,
11-
13]. The likelihood of BCVI also increases with severe TBI, craniofacial injuries, thoracic aperture rib fractures, thoracic vascular injuries, unexplained neurological deficits, and high-energy trauma [
12,
14].
International guidelines are often combined with local institutional guidelines resulting in a variety of different screening strategies [
5-
7]. Digital subtraction angiography (DSA) has historically been the gold standard for tVAI detection and characterization. However, DSA is nowadays primarily reserved for patients at elevated risk, whose CTA results are inconclusive or negative [
4]. Over the past 2 decades, CTA has replaced DSA as the preferred diagnostic method, owing to its cost-effectiveness, broader accessibility, quicker acquisition time, and fewer complications [
5]. Subsequently, several trauma guidelines now recommend CTA as the primary screening tool [
6,
7]. As a result of increased availability, there has been a significant rise in CTA imaging and consequently a rising concern for overutilization. While many studies and guidelines have focused on traumatic carotid artery injuries (tCAIs), this study aims to evaluate the frequency of tVAI in a cohort of level 1 trauma patients with emphasis on the NNI to find clinically relevant injuries, describing also our screening protocols.
MATERIALS AND METHODS
1. Overview
This retrospective study was performed at a large Scandinavian urban level 1 Trauma Center functioning as the primary referral institution with a catchment area of approximately 2.5 million inhabitants [
15]. All adult level 1 trauma patients between January 2013 and July 2020 were included. Patients were identified from the Swedish Trauma Register (SweTrau), with any traumatic injuries to the head, neck, thorax, abdomen, lower, and upper extremities. Exclusion criteria included missing patient identification numbers or misclassified trauma in the registry. Detailed case record reviews were performed based on the patients’ identification numbers, including radiological assessment by 3 raters. After verifying patients with actual cerebrovascular injuries as graded by the Biffl scale, patients with tVAI were identified. Ethical permit was granted by the Swedish National Review Authority (Dnr: 2020-00900), which waived the need for consent per Swedish laws. This report was compiled according to the STROBE (Strengthening the Reporting of Observational Studies in Epidemiology) guidelines.
Data were collected retrospectively from the medical records, including clinical parameters and findings upon arrival, treatment strategies, and a 1-year follow-up based on patient journal documentation and radiological findings to assess stroke occurrence. Retrieved variables included but were not limited to patient demographics, injury characteristics, time to CT, CTA, and magnetic resonance imaging (MRI), time-to-surgery, laboratory tests, pharmacological treatment, in-hospital outcomes, neurological assessment, and 1-year follow-up, if available. Along with the injuries classified according to the Denver criteria [
14], all other concomitant traumatic injuries were noted. Posterior stroke was classified based on radiological verification from the first available imaging modality, either CT or MRI.
On arrival, all level 1 trauma patients were imaged with wholebody CT (WBCT). CTA was incorporated into the WBCT imaging protocol for patients with suspected cerebrovascular injuries.
From 2013 to 2017, the addition of CTA was at the discretion of the trauma surgeon, based on local hospital imaging guidelines in cervical spine trauma. CTA was performed in cases of confirmed cervical spine injury, skull base fracture, as well as in complicated midface fractures and mandibular fractures, and where the head CT examination did not explain central neurological symptoms. Beginning in 2018, CTA became part of the standard protocol for all high-energy level 1 trauma, irrespective of clinically suspected cervical spine trauma.
WBCT and CTA imaging was performed with the Revolution multidetector CT (GE Healthcare, USA) a 256-detector row scanner. Tube current and voltage was 150 mA and 100 kV in the arterial phase. A higher tube current and voltage was used for large patients, 300 mA and 100 kV. The region of interest was set in the thoracic aorta, with 150 HU (Hounsfield unit) as the scan timing value. The timing of contrast was automatic in all imaging exams, using SmartPrep software (GE Healthcare). Rotation time was 0.5 second and with a pitch of 0.992:1. The intravenous contrast injection speed was 4.5–6.0 mL/sec. The arms were positioned either up in hemodynamically stable patients or down, crossed over the abdomen, for a faster scan in hemodynamically unstable patients. All patients with suspected cervical spine injuries had additional imaging with MRI of the cervical spine. In selected cases, additional imaging such as MRI of the brain or repeat CT was performed depending on the patient’s clinical course during hospitalization, at the discretion of the referring surgeon. Posterior circulation ischemia was assessed on either head CT or MRI using T2‑weighted and diffusion-weighted sequences. Grading of vascular injuries according to Biffl was performed independently by 3 radiologists (MB, DT, PC), of whom 2 are specialists in emergency and trauma radiology with 25 years and 5 years of experience, and 1 resident with 3 years of experience. Assessments were made independently at a standard clinical workstation.
4. Injury Severity and Outcome Scales
The New Injury Severity Score (NISS) and Glasgow Outcome Scale (GOS) were retrieved from the SweTrau registry. NISS assesses the overall impact of multiple injuries, where a high NISS score (typically ≥15) indicates severe trauma and is associated with increased risk of complications, intensive care needs, and mortality [
16]. To evaluate neurological and functional outcomes at discharge, the GOS values were retrieved, providing a standardized measure of recovery commonly applied in patients with TBI. The GOS is a 5-point scale ranging from 1 (death) to 5 (good recovery), categorizing patient outcomes as death, vegetative state, severe disability, moderate disability, or good recovery based on functional independence and neurological status [
17]. To assess neurological function in cervical spine injuries, patient records of the American Spinal Injury Association Impairment Scale (AIS) were retrieved. The scale is a standardized 5-grade system, where grade A indicates complete motor and sensory loss, grades B to D represent varying degrees of preserved function, and grade E signifies normal neurological status [
18].
Continuous variables are reported as means with standard deviations, while categorical variables are presented as counts with corresponding percentages. For variables with skewed distributions or ordinal scales, data is reported using median and ranges. The incidence of tVAI was calculated as the number of patients with tVAI according to the Biffl classification divided by the total number of trauma patients imaged with CTA. Confidence intervals (95% CI) were calculated using the Wilson score interval method. The NNI was calculated as 1 divided by the incidence proportion. Descriptive analytics are provided for all patients with tVAI. Missing data was minimal and assumed to be missing at random and was thus not imputed. Interobserver reliability was evaluated using intraclass correlation coefficients (ICCs) with corresponding (95% CI). The analysis employed a 2-way random-effects model (ICC(2,k)) to measure consistency among multiple raters, permitting broader population inferences [
19]. Following the classification system proposed by Koo and Li [
20], agreement levels were categorized as: poor (below 0.50), moderate (0.50–0.75), good (0.75–0.90), and excellent (exceeding 0.90) [
20]. Two-tailed tests were considered statistically significant at p≤0.05. Analyses were carried out in R ver. 4.4.2 (R Foundation for Statistical Computing, Vienna, Austria).
RESULTS
1. Incidence
A total of 5,002 patients with identifiable records and available imaging data were included from the SweTrau registry (
Fig. 1). Of these, 2,159 patients were excluded due to the lack of CTA, leaving 2,843 patients (56.8%) eligible for further analysis. Among these, 119 patients (2.4%) were reported as having cervical vascular injuries. After expert radiological review, 26 cases were excluded due to misclassification (non-traumatic cause or venous injury), resulting in 93 confirmed cases (1.9%) of cervical arterial injury. Of these, 31 had isolated tCAIs and were excluded. Isolated tVAI was identified in 55 patients (1.1%), and 7 patients (0.1%) had combined tVAI and tCAI. This resulted in a total of 62 tVAI cases among 2,843 patients, corresponding to an incidence of 2.2% (95% CI, 1.7%–2.8%). When comparing the 2 periods, before and after universal screening, we found that in 2013–2017, 39 tVAI were detected on 1,354 CTAs, corresponding to an NNI of 35 (95% CI, 25–48). During that same period, 15 posterior strokes were detected (NNI 90; 95% CI, 54–155). Of these, 14 (93.3%) were verified on MRI and 2 (13.3%) with CT. By contrast, in 2018–2020, the incidence of tVAI fell to 23 events among 1,489 CTAs (NNI 65; 95% CI, 42–100), while posterior circulation strokes were detected in 10 cases (NNI 149; 95% CI, 78–292), all verified on MRI.
Over the entire 2013–2020 period, 62 tVAI and 25 strokes occurred on 2,843 CTAs, yielding overall NNIs of 46 (95% CI, 36–59) and 114 (95% CI, 76–172), respectively (
Tables 1 and
2).
Including patients who did not receive any CTA, the first screening protocol used during the 2013–2017 period resulted in the detection of 39 tVAI among 3,473 level 1 trauma patients (1.1%), while the second screening protocol used during the 2018–2020 period detected 23 tVAI among 1,529 patients admitted (1.5%). There was no significant difference in the rate of tVAI detection between the 2 periods and using the 2 different screening protocols (1.1% vs. 1.5%, p=0.261).
2. Follow-up Imaging
Overall, 49 of 62 patients with tVAI underwent a follow‑up head CT, yielding a rate of 79.0%. When the cohort was stratified by time period, a notable decline in follow-up head CT utilization was observed: 87.2% (34 of 39) of patients treated during 2013–2017 received repeat imaging, compared to only 65.2% (15 of 23) of patients managed during 2018–2020. The rate of follow-up brain MRI among tVAI patients was 37.1% (23 of 62). When analyzed by time, a marked increase in follow-up imaging was observed: between 2013 and 2017, 28.2% (11 of 39) of cases received repeat MRI, compared to 52.2% (12 of 23) imaged during 2018–2020 (
Table 1).
The mean age was 53.8±18.0 years. The majority were male (n=44, 69.8%), and the mean body mass index (BMI) was 24.7±3.6 kg/m2. Almost all patients suffered blunt trauma (n=61, 98.4%), while only 1 patient (1.6%) had a penetrating injury, which did not require coiling. The median NISS was 27 (16–41), indicating severe trauma.
The most common injury mechanism was fall from a height (>2 m) in 22 patients (35.5%), followed by car accidents in 12 (19.4%), bicycle accidents in 8 (12.9%), and pedestrian injuries in 4 (6.5%). Active smokers accounted for 4 patients (6.5%). The median Glasgow Coma Scale (GCS) at admission was 14 (3–15), however 37% cases had missing data in the registry. Median time from admission to initial CT was 29 (11–78) minutes (
Table 1). The median intensive care unit time was 8 (1–52) days, and hospital stay 20 (2–166) days (
Table 3).
The most frequent Biffl grade was grade IV (n=38, 61.3%), followed by grade II (n=13, 21.0%), grade I (n=7, 11.3%), and grades III and V (n=2 each, 3.2%). Bilateral vertebral artery injury occurred in 10 (16.1%). The most affected vertebral segment was V2 (n=32, 51.6%), followed by V3 (n=13, 21.0%), V1 (n=10, 16.1%), and V4 (n=7, 11.3%). tVAI occurred on the dominant arterial side in 5 cases (8.1%). Concomitant tCAI was present in 7 (11.3%). Infratentorial stroke was observed in 25 patients (40.3%). Of those, 3 (12.0%) had Biffl grade I injuries, 8 (32.0%) grade II, none (0%) grade III, 13 (52.0%) grade IV, and 1 (4.0%) grade V; additionally, 5 of these 25 strokes (20.0%) were bilateral tVAI (
Table 4,
Fig. 2).
Antithrombotic treatment included low molecular weight heparin (LMWH; n=28, 45.2%), acetylsalicylic acid (ASA; n=10, 16.1%), LMWH and ASA combined (n=8, 12.9%), non-vitamin K-dependent oral anticoagulants (n=1, 1.6%), and Warfarin (n=1, 1.6%). Vertebral artery coiling procedures were performed in only 1 patient (1.6%), while 13 patients (21.0%) were managed conservatively (
Table 4). No additional strokes were observed in the 1-year follow-up of patients receiving antithrombotic treatment. Postdischarge complications related to clinical management were observed in 2 cases, one gastrointestinal bleed from a gastric ulcer, resulting in a change of therapy from LMWH to ASA and one case of endovascular coiling experienced progression of ischemic stroke after discharge, which led to the initiation of antithrombotic treatment.
CTA was increasingly added as part of the imaging protocol over the examined period from 2013 to 2020. In 2013, CTA accounted for 20.1% (139 of 691) of all WBCT examinations. This proportion rose to 26.6% (186 of 699) in 2014, 34.0% (226 of 665) in 2015, and 44.5% (311 of 699) in 2016. By 2017, CTA made up 68.4% (492 of 719) of WBCTs, and in 2018 it had increased further to 93.8% (606 of 646). From 2019 onward, CTA represented 100% of the examinations, indicating a complete transition to CTA-based imaging in level 1 trauma cases (
Fig. 3).
Interrater reliability for Biffl grading among the 3 radiologists, was 0.97 (95% CI, 0.95–0.98), indicating excellent agreement, according to the criteria established by Koo and Li [
20], with a statistically significant result (p<0.001).
Patients were most commonly discharged to a rehabilitation facility (n=32, 51.6%). The median GOS at discharge was 3 (1–4), and 30-day mortality occurred in 10 (16.1%), all in-hospital, with a mean time to death of 2 (1–12) days. The 30-day mortality was strongly associated with several critical factors: severe polytrauma involving multiple anatomical regions, median NISS 34 (27–41), and markedly reduced consciousness on arrival, median GCS 3 (3–15), a high incidence of upper cervical spine injuries at levels C0–3 (n=8, 80%), and a history of pre-admission cardiac arrest with successful spontaneous resuscitation (n=6, 55%), as outlined in
Table 3. Among the patients with cervical spine injuries who underwent detailed neurologic assessment, nearly one‐third (n=18, 29.0%) remained neurologically intact. Of those with impairment, the majority had incomplete deficits (AIS B: n=3 [4.8%]; AIS C: 5 [8.1%]; AIS D: 9 [14.5%]), while 8 (12.9%) suffered complete spinal cord injury (AIS A), data was missing for 10 (16.1%) of cases (
Table 1).
DISCUSSION
In this large retrospective study of level 1 trauma patients, the incidence of tVAI during the study period was found to be 2.2%, with a posterior circulation stroke rate of 0.9%. The NNI to detect one tVAI was 46, and for one posterior circulation stroke, 114. Notably, these NNI values increased over time, from 35 to 65 for tVAI, and from 90 to 149 for posterior stroke between 2013–2017 and 2018–2020, upon implementation of a universal CTA screening approach for all patients with high-energy level 1 trauma. These findings highlight the challenges in balancing diagnostic yield and imaging efficiency in modern trauma imaging and care.
The most frequent tVAI types were Biffl grade IV (61.3%) and grade II (21.0%). Previous meta-analytic data suggest that grade III and IV injuries carry a significantly elevated stroke risk, while grade I and II injuries are associated with lower risk [
1]. Of the 25 infratentorial strokes, 12.0% occurred in Biffl grade I, 32.0% in grade II, none in grade III, 52.0% in grade IV, and 4.0% in grade V. However, given the small subgroup sizes—especially in grades III and V—and the complete lack of events in the 2 grade III patients, the statistical analysis lacked sufficient precision to draw meaningful conclusions. Larger, multicenter cohorts will therefore be required to robustly assess the relationship between Biffl grade and stroke risk. Furthermore, while the Biffl grading system has been a cornerstone in the evaluation of BCVI, the recent classification of traumatic vascular injuries from the European Society for Vascular Surgery (ESVS) offers a potentially more nuanced framework [
21]. The ESVS classification differs from the Biffl system in its stratification, e.g., Biffl grades I and II are consolidated into ESVS grade I, which may allow for a more streamlined yet clinically meaningful assessment of vascular trauma, potentially offering improved alignment with therapeutic decision-making and outcome prediction. Future studies could benefit from incorporating or comparing this updated grading system, potentially leading to improved diagnostic accuracy, risk stratification, and management pathways for patients with tVAI.
The incidence of bilateral tVAI (16.1%) was slightly higher than the 12.3% reported by a meta-analysis by Michalopoulos et al. [
1], suggesting a potentially greater severity of injury or improved detection in our population. Conversely, the rate of concurrent tCAI was lower in our study (11%) compared to 19.2% in the meta-analysis. This discrepancy may reflect differences in injury mechanisms, imaging protocols, or inclusion criteria between studies.
The most common antithrombotic treatments were LMWH alone, used in 45% of cases, followed by ASA alone (16.1%), and a combination of LMWH and ASA (12.9%). According to the ESVS guidelines, there is currently insufficient high-level evidence to support the use of one specific antithrombotic agent over another in the treatment of tVAI [
21]. Nonetheless, the evidence available tends to favor single-agent antiplatelet therapy, which has been associated with a lower risk of bleeding in the trauma setting. Given its safety and efficacy, low-dose ASA is recommended as first-line antithrombotic therapy for tVAI without active hemorrhage. Postdischarge complications related to clinical management were observed in 2 cases. One patient developed a gastrointestinal bleed from a gastric ulcer, resulting in a change of therapy from LMWH to ASA. In the other case, a patient who had undergone endovascular coiling experienced progression of ischemic stroke after discharge, which led to the initiation of antithrombotic treatment. A notable 21% of patients with tVAI were managed expectantly without specific pharmacological treatment (2 Biffl grade II and 9 Biffl grade IV). This rate slightly exceeds the 18% reported by Goyal et al. [
4] and underscores the individual treatments in clinical scenarios where bleeding risk, planned surgery, or overall injury burden precluded immediate antithrombotic therapy.
None of the tVAI cases experienced additional ischemic strokes after initiating antithrombotic therapy, during the 1-year follow-up. This finding suggests that routine follow-up CTA imaging may be unnecessary in the absence of risk factors, provided appropriate treatment is initiated, aligning with previous studies on follow-up strategies [
22]. However, from the small number of cases of tVAIs in our cohort, and that only half of the patients were followed up with CTA during the study period, conclusions regarding the necessity of follow-up CTA remain limited, and further studies are still warranted [
22,
23].
In contrast to our earlier work focusing on patients undergoing fixation for subaxial cervical spine injuries [
13], this study included a broader range of high-energy trauma patients, with severe trauma as indicated by high injury severity scores (NISS 27). In the current study of level 1 trauma, 85% of the tVAI cases had a concomitant cervical fracture. However, only 47% of them had an unstable fracture requiring fixation surgery. In our previous study on tVAI in patients undergoing fixation surgery for subaxial injuries, the cohort was a mix of all types of trauma severity, from level 1 trauma to same level falls in the elderly. This explains the difference in tVAI incidence of 14% in this study compared to the 7% in the previous work [
13]. This also underscores how trauma severity influences tVAI incidence and neurological outcomes.
The 40.3% stroke rate reported in this study surpasses most previously cited ranges of 0.5%–33% [
1,
4]. This elevated number may in part be due to an awareness of the risk for stroke in tVAI and the use of early MRI for ischemia detection, which was used more frequently in the 2018–2020 period. This enabled identification of both symptomatic and asymptomatic infarcts at a much earlier timeframe. Stroke resulting from tVAI in trauma is likely underreported. This underreporting can be attributed to challenges in early diagnosis, inconsistent screening protocols, and the often subtle or delayed presentation of symptoms. As shown in previous studies, many trauma-related strokes are radiologically apparent but clinically silent [
22,
24]. This distinction is especially relevant in posterior circulation strokes, which often mimic benign symptoms like dizziness or nausea and may be missed on early CT [
25]. Subsequently, the discrepancy between radiological and clinical stroke manifestations is clinically significant. While silent infarctions may go undetected, they can still impact long-term [
26,
27] cognitive and functional outcomes [
28,
29].
Screening criteria, initially largely developed for tCAI, have evolved to address the complexity of BCVI (including both tCAI and tVAI) and stroke. The original Denver criteria in 1996 were initially limited to a narrow subset of high-risk injuries. Subsequent revisions in 2005 and 2012 expanded to include craniofacial, thoracic, and broader cervical spine injuries [
26,
27]. Despite these updates, current protocols may still miss up to 30% of BCVI cases, underlining the need for continual reassessment of screening algorithms [
30]. There is a growing debate on screening strategies in trauma patients, particularly between universal and targeted approaches. In our study, CTA was increasingly adopted as part of the imaging protocol over the period examined from 2013 to 2020. Initially, CTA constituted only a fraction of the total WBCT examinations, accounting for 20% in 2013. However, a clear upward trend was observed, with the percentage of CTA steadily increasing each year. By 2017, CTA comprised 68.4% of the WBCT examinations, and from 2019 onward, it represented 100% of the cases. This shift reflects a transition in imaging strategy, driven by studies that have highlighted that a substantial number of BCVI could potentially be missed on initial WBCT when CTA was not included [
31]. Furthermore, incorporating CTA into the initial WBCT offers a time-efficient diagnostic pathway, as it allows for immediate assessment and management of BCVI, including anatomical description for interventional procedures. Without it, the patient may have already left the CT suite by the time the radiology report identifies injuries suspicious for the presence of BCVI, necessitating a return for additional imaging and potentially delaying treatment.
Universal screening, implemented at our level 1 trauma center since 2018, may maximize diagnostic yield but comes with drawbacks, including higher numbers NNI, increased radiation exposure, and elevated healthcare costs. Notably, despite this broad approach, the detection rate of tVAI during the universal screening period was not significantly higher than in the subsequent phase when selective screening based on expanded risk criteria was used. These findings suggest that a selective screening strategy may offer greater diagnostic efficiency while minimizing unnecessary interventions [
2,
12,
13,
26].
It has to be noted that, in Sweden, access to imaging has a low threshold, especially in the emergency setting, and universal screening approaches are thus possible without major effort. On the other hand, the generally lower in-hospital workload compared to other global regions, as well as the strong interdisciplinary focus of emergency care und multidisciplinary involvement in primary trauma care, makes implementing changes such as a sophisticated selective screening protocol possible.
It also has to be noted that, while generally safe, contrast-enhanced imaging carries a minor risk for adverse events, too. In a multicenter cohort of 196,081 iodinated contrast administrations, immediate hypersensitivity reactions occurred in 0.73% and severe events in 0.0087%, corresponding to numbers needed to harm (NNH) of ~137 (any immediate reaction) and ~11,500 (severe) [
32]. Among emergency department patients with chronic kidney disease, a propensity-matched analysis reported acute kidney injury rates of 13.2% with contrast versus 8.3% without (absolute risk increase 5.0%; 95% CI, 3.8–6.1), yielding an NNH of ~20 [
33].
Our data suggests that even within a universal screening protocol, all the tVAI cases occurred in patients presenting with known risk factors. This supports the potential feasibility of a more refined, risk-based screening approach in a general trauma population. This holds especially true given that smaller posterior circulation strokes, the most feared complication of tVAI, are often clinically silent and are typically not associated with increased risks of mortality [
13,
34]. While our findings are drawn from a level 1 trauma cohort—where the threshold for imaging is appropriately low to avoid missed injuries —the relatively high NNI also prompts consideration of whether a higher diagnostic accuracy could be achieved through more selective screening protocols. Although extrapolation to less severe trauma cases must be done with caution, our findings may serve as a basis for future investigation into whether selective screening strategies could be safely applied in lower-trauma settings, where the prevalence of tVAI is expected to be lower. Further prospective studies are needed to validate these approaches and to establish safe, evidence-based criteria for imaging in varying levels of trauma severity.
Although supported by a large level 1 hospital registry, our chart reviews were retrospective in nature, with the risks of missing data and potential for selection bias. In some parameters such as GCS there was substantial missing data. The selection of included cases was checked by a panel of 3 expert radiologists to verify presence of tVAI among all patients with reported diagnostic codes for cerebrovascular vascular injury. In addition, our study is single center. Thus, although our center exclusively covers a wide catchment area and thus enables an almost population-based study, our results are affected by inhouse treatment protocols. We were unable to access detailed information on concomitant injuries for patients without tVAI, which would have allowed an additional analysis of risk factors and risk stratification. Notably, our assessment of cerebrovascular ischemia was only based on radiological evidence of either CT findings or demarcated ischemia or clear diffusion restriction on MRI, as there were no structured neurological examinations such as NIHSS scores. This limits our findings regarding stroke rates to radiologically identified ischemic lesions in the posterior circulation area. Finally, interpretation of AIS and GOS scores in relation to tVAI is challenging in the context of multiple injuries, as the overall trauma burden may obscure the specific impact and outcomes related to tVAI. Future studies on tVAI should include validated quality of life instruments such as the Stroke-Specific Quality of Life Scale and EuroQol-5D to assess patient-centered outcomes comprehensively and long-term impact of posterior circulation strokes that are radiologically identified [
28,
29].
CONCLUSION
Our study found a tVAI incidence of 2.2% and posterior circulation stroke rate of 0.9% among level 1 trauma patients in a large Scandinavian trauma center. Along with detailed injury patterns and treatment regimens, the NNI was 46 for detecting one tVAI and 114 for one posterior circulation stroke, with both values increasing after implementation of universal screening in high-energy level 1 trauma patients (tVAI: 35→65; stroke: 90→149 when comparing 2013–2017 to 2018–2020, or selective screening to universal screening, respectively). Despite the adoption of the universal screening approach, the detection rate of tVAI did not improve compared to the earlier selective screening protocol. These findings suggest that a tailored screening strategy may offer a more efficient alternative to a universal scanall approach. Further studies are needed to define the optimal balance between diagnostic yield, resource use, and patient outcomes across different trauma levels. These findings underscore the ongoing challenge of optimizing diagnostic yield versus imaging resource utilization in trauma care.
NOTES
-
Conflict of Interest
The authors have nothing to disclose.
-
Funding/Support
This study received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.
-
Author Contribution
Conceptualization: PC, VES, VGEH, IAAR, BM, JMR, DT, MB, MO, PJ, AET, EE; Data curation: PC, VES, VGEH, IAAR, AET, EE; Formal analysis: PC, VES, VGEH, MB, AET, EE; Methodology: PC, VES, VGEH, IAAR, DT, MB, MO, PJ, AET, EE; Project administration: PC, VES, VGEH, IAAR, AET, EE; Visualization: PC, VES, VGEH, DT, MB, AET, EE; Writing – original draft: PC, VES, VGEH, AET, EE; Writing – review & editing: VES, IAAR, BM, JMR, DT, MB, MO, PJ.
Fig. 1.Flowchart of inclusion. WBCT, whole-body computed tomography; CTA, computed tomography angiography; tCAI, traumatic carotid artery injury; tVAI, traumatic vertebral artery injury.
Fig. 2.Demonstration of Biffl grade distribution. tVAI, traumatic vertebral artery injury.
Fig. 3.WBCT and CTA volume with CTA% (2013–2020). CTA, computed tomography angiography; WBCT, whole-body computed tomography.
Table 1.Demographic description of the patient cohort, including injury mechanism
Table 1.
|
Variable |
Value |
Missing |
|
Age (yr) |
53.8 ± 18.0 |
|
|
Male sex |
44 (69.8) |
|
|
BMI (kg/m2) |
24.7 ± 3.6 |
7 (11.3) |
|
AIS classification |
|
10 (16.1) |
|
AIS A |
8 (12.9) |
|
|
AIS B |
3 (4.8) |
|
|
AIS C |
5 (8.1) |
|
|
AIS D |
9 (14.5) |
|
|
Neurologically intact |
18 (29.0) |
|
|
Active smokers |
4 (6.5) |
|
|
Injury mechanism |
|
|
|
Assault |
1 (1.6) |
|
|
Car driver/passenger |
12 (19.4) |
|
|
Cyclist |
8 (12.9) |
|
|
Fall from height (> 2 m, including from horse) |
22 (35.5) |
|
|
Fall from same level plane |
3 (4.8) |
|
|
Hanging |
2 (3.2) |
|
|
Knife injury |
1 (1.6) |
|
|
Moped/motorcycle |
3 (4.8) |
|
|
Other motor vehicle (boat/scooter/4-wheeler) |
3 (4.8) |
|
|
Pedestrian hit by car |
4 (6.5) |
|
|
Train collision |
3 (4.8) |
|
|
Blunt trauma |
61 (98.4) |
|
|
GCS at admission |
14 (3–15) |
23 (37.1) |
|
Time from admission to initial CT (min) |
29 (11–78) |
12 (19.3) |
|
Imaging protocols |
|
|
|
WBCT |
5,002 |
|
|
2013–2017 |
3,473 |
|
|
2018–2020 |
1,529 |
|
|
CTA |
2,843/5,002 (56.8) |
|
|
2013–2017 |
1,354/3,473 (38.9) |
|
|
2018–2020 |
1,489/1,529 (97.4) |
|
|
tVAI |
62/5,002 (1.2) |
|
|
2013–2017 |
39/3,473 (1.1) |
|
|
2018–2020 |
23/1,529 (1.5) |
|
|
Stroke |
25/62 (40.3) |
|
|
2013–2017 |
16/39 (41.0) |
|
|
2018–2020 |
9/23 (39.1) |
|
|
Follow-up imaging |
|
|
|
Brain CT |
49/62 (79.0) |
|
|
2013–2017 |
34/39 (87) |
|
|
2018–2020 |
15/23 (65.2) |
|
|
CTA |
35/62 (56.4) |
|
|
2013–2017 |
23/39 (58.9) |
|
|
2018–2020 |
12/23 (52.2) |
|
|
Brain MRI |
23/62 (37.0) |
|
|
2013–2017 |
11/39 (28.0) |
|
|
2018–2020 |
12/23 (52.2) |
|
Table 2.NNI for tVAI and posterior stroke
Table 2.
|
Year |
Trauma protocol
|
tVAI on CTA
|
Posterior stroke
|
|
WBCT (n) |
CTA (n) |
Events (n) |
NNI |
95% CI |
Events (n) |
NNI |
95% CI |
|
2013–2017 |
3,473 |
1,354 |
39 |
35 |
25–48 |
15 |
90 |
54–155 |
|
2018–2020 |
1,529 |
1,489 |
23 |
65 |
42–100 |
10 |
149 |
78–292 |
|
2013–2020 |
5,002 |
2,843 |
62 |
46 |
36–59 |
25 |
114 |
76–172 |
Table 3.Overview of injury severity, hospital stays, and outcome
Table 3.
|
Variable |
Value |
Missing |
|
NISS |
27 (16–41) |
2 (3.2) |
|
Days on ventilator |
8 (1–52) |
30 (48.4) |
|
Length of hospital stay (day) |
20 (2–166) |
11 (17.7) |
|
Discharge disposition |
|
11 (17.7) |
|
Home |
5 (8.1) |
|
|
Rehabilitation facility |
32 (51.6) |
|
|
Morgue |
11 (17.7) |
|
|
Surgical ward |
4 (6.5) |
|
|
Psychiatric ward |
2 (3.2) |
|
|
GOS at discharge |
3 (1–4) |
11 (17.7) |
|
30-Day mortality |
10 (16.1) |
|
|
Time to death (day) |
2 (1–12) |
|
|
NISS |
34 (27–41) |
|
|
GCS on arrival |
3 (3–15) |
|
|
Cardiac arrest with ROSC |
6 (9.7) |
|
Table 4.Radiological grading of tVAI and its specific treatment in our cohort
Table 4.
|
Variable |
Value |
|
Biffl grade |
|
|
I |
7 (11.3) |
|
II |
13 (21.0) |
|
III |
2 (3.2) |
|
IV |
38 (61.3) |
|
V |
2 (3.2) |
|
Bilateral tVAI |
10 (16.1) |
|
Segment |
|
|
V1 |
10 (16.1) |
|
V2 |
32 (51.6) |
|
V3 |
13 (21.0) |
|
V4 |
7 (11.3) |
|
Dominant side |
5 (8.1) |
|
Concomitant tCAI |
7 (11.3) |
|
Infratentorial stroke on imaging |
25 (40.3) |
|
Biffl grade I |
3 (12.0) |
|
Biffl grade II |
8 (32.0) |
|
Biffl grade III |
0 (0) |
|
Biffl grade IV |
13 (52.0) |
|
Biffl grade V |
1 (4.0) |
|
Bilateral tVAI |
5 (20.0) |
|
Specific treatment for tVAI |
|
|
Vertebral coiling |
1 (1.6) |
|
ASA |
10 (16.1) |
|
NOAK |
1 (1.6) |
|
Warfarin |
1 (1.6) |
|
LMWH |
28 (45.2) |
|
LMWH and ASA |
8 (12.9) |
|
No treatment/expectancy |
13 (21.0) |
REFERENCES
- 1. Michalopoulos GD, Pennington Z, Bambakidis P, et al. Traumatic vertebral artery injury: denver grade, bilaterality, and stroke risk. A systematic review and meta-analysis. J Neurosurg 2024;140:522-36.
- 2. Desouza RM, Crocker MJ, Haliasos N, et al. Blunt traumatic vertebral artery injury: a clinical review. Eur Spine J 2011;20:1405-16.
- 3. Miller PR, Fabian TC, Croce MA, et al. Prospective screening for blunt cerebrovascular injuries: analysis of diagnostic modalities and outcomes. Ann Surg 2002;236:386-93. discussion 393-5.
- 4. Goyal K, Sunny JT, Gillespie CS, et al. A Systematic review and meta-analysis of vertebral artery injury after cervical spine trauma. Global Spine J 2024;14:1356-68.
- 5. Chatterjee AR, Malhotra A, Curl P, et al. Traumatic cervical cerebrovascular injury and the role of CTA: AJR expert panel narrative review. AJR Am J Roentgenol 2024;223:e2329783.
- 6. Bromberg WJ, Collier BC, Diebel LN, et al. Blunt cerebrovascular injury practice management guidelines: the Eastern Association for the Surgery of Trauma. J Trauma 2010;68:471-7.
- 7. Biffl WL, Cothren CC, Moore EE, et al. Western Trauma Association critical decisions in trauma: screening for and treatment of blunt cerebrovascular injuries. J Trauma 2009;67:1150-3.
- 8. Wagner MJ, Hussein I, Low G, et al. Comparing the Denver criteria sets for blunt trauma: a retrospective study of cases in Edmonton, Alberta. Br J Radiol 2023;96:20221116.
- 9. Biffl WL, Moore EE, Offner PJ, et al. Blunt carotid arterial injuries: implications of a new grading scale. J Trauma 1999;47:845-53.
- 10. Miller PR. Blunt cerebrovascular injury: contribution of Timothy C Fabian MD and investigators from the University of Tennessee at Memphis to our understanding of the injury. Trauma Surg Acute Care Open 2023;8:e001112.
- 11. Dunn CJ, Changoor S, Issa K, et al. Cervical computed tomography angiography rarely leads to intervention in patients with cervical spine fractures. Global Spine J 2020;10:992-7.
- 12. Geddes AE, Burlew CC, Wagenaar AE, et al. Expanded screening criteria for blunt cerebrovascular injury: a bigger impact than anticipated. Am J Surg 2016;212:1167-74.
- 13. El-Hajj VG, Habashy KJ, Cewe P, et al. Traumatic vertebral artery injury after subaxial cervical spine injuries: incidence, risk factors, and long-term outcomes: a population-based cohort study. Neurosurgery 2025;96:881-91.
- 14. Biffl WL, Moore EE, Offner PJ, et al. Optimizing screening for blunt cerebrovascular injuries. Am J Surg 1999;178:517-22.
- 15. Cewe P, Burström G, Drnasin I, et al. Evaluation of a novel teleradiology technology for image-based distant consultations: applications in neurosurgery. Diagnostics (Basel) 2021;11:1413.
- 16. Chun M, Zhang Y, Becnel C, et al. New injury severity score and trauma injury severity score are superior in predicting trauma mortality. J Trauma Acute Care Surg 2022;92:528-34.
- 17. McMillan T, Wilson L, Ponsford J, et al. The Glasgow Outcome Scale - 40 years of application and refinement. Nat Rev Neurol 2016;12:477-85.
- 18. Roberts TT, Leonard GR, Cepela DJ. Classifications in brief: American Spinal Injury Association (ASIA) Impairment Scale. Clin Orthop Relat Res 2017;475:1499-504.
- 19. Liljequist D, Elfving B, Skavberg Roaldsen K. Intraclass correlation - a discussion and demonstration of basic features. PLoS One 2019;14:e0219854.
- 20. Koo TK, Li MY. A guideline of selecting and reporting intraclass correlation coefficients for reliability research. J Chiropr Med 2016;15:155-63.
- 21. Wahlgren CM, Aylwin C, Davenport RA, et al. Editor’s choice — European Society for Vascular Surgery (ESVS) 2025 Clinical Practice Guidelines on the Management of Vascular Trauma. Eur J Vasc Endovasc Surg 2025;69:179-237.
- 22. Fang R, Duering M, Bode FJ, et al. Risk factors and clinical significance of post-stroke incident ischemic lesions. Alzheimers Dement 2024;20:8412-28.
- 23. Nally MC, Kling C, Hocking KM, et al. Follow-up imaging of traumatic vertebral artery dissections is unnecessary in asymptomatic patients. J Vasc Surg 2019;69:1704-9.
- 24. Saini M, Ikram K, Hilal S, et al. Silent stroke: not listened to rather than silent. Stroke 2012;43:3102-4.
- 25. Schmahmann JD, Macmore J, Vangel M. Cerebellar stroke without motor deficit: clinical evidence for motor and nonmotor domains within the human cerebellum. Neuroscience 2009;162:852-61.
- 26. Burlew CC, Biffl WL, Moore EE, et al. Blunt cerebrovascular injuries: redefining screening criteria in the era of noninvasive diagnosis. J Trauma Acute Care Surg 2012;72:330-5. discussion 336-7, quiz 539.
- 27. Kim DY, Biffl W, Bokhari F, et al. Evaluation and management of blunt cerebrovascular injury: a practice management guideline from the Eastern Association for the Surgery of Trauma. J Trauma Acute Care Surg 2020;88:875-87.
- 28. Czechowsky D, Hill MD. Neurological outcome and quality of life after stroke due to vertebral artery dissection. Cerebrovasc Dis 2002;13:192-7.
- 29. Strege RJ, Kiefer R, Herrmann M. Contributing factors to quality of life after vertebral artery dissection: a prospective comparative study. BMC Neurol 2019;19:312.
- 30. Schmidt JC, Huang DD, Fleming AM, et al. Missed blunt cerebrovascular injuries using current screening criteria - The time for liberalized screening is now. Injury 2023;54:1342-8.
- 31. 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 2017;3:217-25.
- 32. Cha MJ, Kang DY, Lee W, et al. Hypersensitivity reactions to iodinated contrast media: a multicenter study of 196 081 patients. Radiology 2019;293:117-24.
- 33. Kene M, Arasu VA, Mahapatra AK, et al. Acute kidney injury after CT in emergency patients with chronic kidney disease: a propensity score-matched analysis. West J Emerg Med 2021;22:614-22.
- 34. El-Hajj VG, Al-Rikabi IA, Staartjes VE, et al. In reply: traumatic vertebral artery injury after subaxial cervical spine injuries: incidence, risk factors, and long-term outcomes: a population-based cohort study. Neurosurgery 2025;97:e66.
, Victor E. Staartjes1,3,4
