Article Text
Abstract
Traumatic aortic rupture (TAR) is a rare but serious injury, leading to death at the scene in most cases. Between 1990 and 2003, all consecutive patients and victims with TAR were retrospectively analysed by reviewing hospital and autopsy records. Univariate and multivariate Cox regression analyses were performed to define determinants of mortality. During the study period, a total of 85 patients (70 men, mean age 47±18.8 years) with TAR were observed in the greater area of Zurich giving a population-based rate of 0.6 cases per 100 000 persons per year. Prehospital, in-hospital and overall mortalities were 40.0%, 31.4% and 58.8%, respectively, with a median survival time of 2 days (IQR 1–3617 days). In the univariate analysis, significant determinants of prehospital and overall mortality were age (HR 1.05, p=0.006), complete aortic transection (HR 7.17, p=0.003), number (HR 1.35, p=0.009) and associated injuries to chest (HR 3.41, p=0.03), liver (HR 6.00, p=0.002) and spine (HR 5.19, p=0.01). By comparison, risk factors for in-hospital mortality included haemodynamic instability upon arrival in the emergency room (HR 16.11, p<0.001) and open surgical repair (HR 14.29, p=0.02). In the multivariate model, only age (p=0.02) and complete aortic transection (p=0.001) were significant determinants of mortality. Therefore, with the exception of complete aortic transection, risk factors of prehospital and in-hospital death in patients with TAR differ greatly. The in-hospital mortality was not affected by the number or localisation of associated injuries, whereas haemodynamic instability and open aortic repair seem to predict in-hospital mortality after TAR.
Clinical Trial Registration NCT01632774.
- Trauma, chest
- Trauma, epidemiology
- Trauma, majot trauma management
- vascular-Arterial
Statistics from Altmetric.com
Introduction
Aortic injuries after blunt thoracic trauma are relatively rare but serious, and can lead to death at the scene in most cases.1–4 Despite major advances in prehospital life support and the introduction of sophisticated preventive measures, prehospital mortality of patients presenting traumatic rupture of the thoracic aorta is still 75%–90%.3–5 However, the introduction of endovascular repair for patients who reach the hospital alive may result in an improvement of outcomes. In this observational study on 85 consecutive patients with traumatic aortic rupture (TAR), we compared the portion of patients who died at the scene with those transferred to the hospital alive in order to identify possible risk factors for death at the scene. Furthermore, we investigated the risk factors for death in those patients who reached the hospital alive.
Methods
During a 14-year period (January 1990–December 2003), all reported cases of traumatic rupture of the thoracic aorta in the greater area of Zurich with about one million inhabitants were enrolled in the present study. Based on the final outcome, patient data were drawn from either medical records (surgical reports, discharge files), from the University Hospital Zurich (UHZ), Triemli City Hospital of Zurich (TCHZ) or from autopsy records from the Institute of Forensic Medicine of the University of Zurich. Both medical and autopsy records of all subjects were reviewed for formal confirmation of diagnosis of blunt TAR. The diagnosis of TAR was identified with an electronic database applying several disease-specific search criteria (traumatic rupture of thoracic aorta, TAR, traumatic aortic laceration, traumatic aortic lesion and blunt aortic rupture). UHZ and TCHZ are tertiary care teaching hospitals and level I trauma centres for the greater area of Zurich. In the text, both hospitals are summarised as ‘tertiary centres’. Both of these hospitals have attained a superior level of expertise for treatment of TAR in the greater area of Zurich. The Institute of Forensic Medicine at the University of Zurich analyses all fatalities in this region. In Switzerland, criteria to perform a full autopsy include all cases in which events leading to death are unclear or the cause of death is the result of suicide or extrinsic factors. The factors considered in this study include death at scene, death during ambulance transport and in-hospital death. The variables studied were baseline characteristics, mechanism of trauma, haemodynamic instability (defined as systolic blood pressure below 90 mmHg), extent of aortic injury, surgical procedure and number and types of associated injuries, which were quoted in cases of parenchymal lesion or fractured bones. Differences in demographic and clinical characteristics between survivors of 30 days, and non-survivors were estimated using the unpaired Student's t test for continuous variables and the χ2 test or Fisher's exact test for categorical variables. Data were reported as mean±SD or percentages. The survival is displayed as median (IQR) using the Kaplan–Meier curve. Follow-up information to complete survival analysis was performed by repeated clinical examinations or by telephone. Determinants of mortality were tested with the univariate und multivariate Cox regression analysis. Three separate multivariate models were constructed according to prehospital, in-hospital and overall mortality. Only those characteristics which reached a significant univariate association were included in the multivariate model. p Values of all outcomes were two-sided; values less than 0.05 were considered to indicate statistical significance and were defined as a cut-off for entering into the multivariate model. For statistical significant variables, the HR with a 95% CI was calculated applying the Cox regression method. All statistical analyses were performed using IBM SPSS Statistics for Windows, V.20.0 (IBM Corporation, Armonk, New York, USA).
The study was performed according to the local legality concerning retrospective and previously anonymised data in concordance with the Ethics committee of the Canton Zurich, Switzerland.
Results
During the 14-year study period, a total of 85 patients (70 men and 15 women) with TAR were observed in the greater area of Zurich giving a population-based rate of 0.6 cases per 100 000 persons per year. The mean age of all patients at the time of accident was 46.8±18.8 years (maximum 86, minimum 13 years). In most cases, car accidents were the cause of injury (N=42, 49.4%) and drivers were affected threefold as much in comparison with non-drivers. Other causes of injury were falls from heights (N=12, 14.1%), motorcycle accidents (N=12, 14.1%), pedestrians struck by cars (N=9, 10.6%), bicycle accidents (N=4, 4.7%), gliding flight accidents (N=3, 3.5%) and in two cases, aeroplane crashes (2.4%). In one case, a snowboarding accident led to aortic transection. Relevant associated injuries were present in 82 patients (96.5%), whereas a median of five associated injuries were observed per patient. External chest trauma (fractures of ribs and sternum, or haemato-/pneumothorax) was the most frequent associated injury (N=57, 67.1%), followed by medium-to-severe brain injury (N=45, 52.9%), and in 44 cases (51.8%), intrathoracic injuries (lung contusion or laceration, myocardial contusion or laceration, bronchus rupture, diaphragm rupture).The aorta was completely transected in 42 patients (49.4%), whereas a partial rupture was found in 50.6%. The overall mortality of the 85 patients was 58.8% (50/85) with a median survival time of 2 days (IQR 1–3617 days) (figure 1). Thirty patients (35.3%) died on the scene and four patients (4.7%) died during or immediately before patient transfer to the tertiary centres. Fifty-one patients (60%) were transferred alive. In these patients, the median Injury Severity Score (ISS) was 32 (IQR 26–41). A total of 16 out of these 51 patients (31.4%) died during hospitalisation. A flow chart illustrating the timing of death is shown in figure 2. Demographic and clinical characteristics of all patients are displayed in table 1.
Prehospital mortality
Thirty-four patients (40.0%) with TAR died before admission (death at the scene, during or immediately before patient transfer). The mean age of the patients being transferred alive to the tertiary centres was 41.0 years, compared with those of 55.5 years who died before reaching our emergency room (p=0.01). Complete aortic transections compared with partial ruptures and the number of additional injuries were statistically significant associations of mortality (73.5% vs 33.3%, p=0.015 and 7.1 vs 4.2, p=0.009, respectively). However, not all additional injuries seem to represent risk factors for dying before reaching hospital care. Injuries to head, kidneys, intestines, bladder, pelvis or limbs were equally distributed between survivors and non-survivors, whereas additional chest trauma (73.5% vs 37.3%, p=0.01) and injuries to liver (61.8% vs 17.6%, p=0.008) and spine (55.9% vs 19.6%, p=0.01) were significantly more prevalent in those patients who did not reach the hospital alive compared with those who were admitted alive. In the multivariate model, only a complete aortic transection was significant in determining an adverse outcome (p=0.03). Gender and the mechanism of trauma were equally distributed between the groups. A marginally significant risk factor for prehospital mortality was found in pedestrians struck by cars, as 7/9 (77.8%) of these subjects died at the scene compared with 27/76 (35.5%) by another mechanism of trauma (p=0.05). All significant determinants of prehospital mortality are summarised in table 2.
In-hospital mortality
In-hospital mortality (within 30 days) of the patients being transferred alive to the tertiary centres was 31.4% (16/51). One patient (2.0%) died immediately after admission in the emergency room. Fifty patients underwent surgical repair of the aorta. Intraoperative and postoperative mortalities were 30.0% (15/50), of which, six patients died during surgery and nine patients died postoperatively. Three patients died on the first day after surgery, five patients died between the second and seventh day and one patient died between the 8th and 30th day after surgery. The median survival time of those patients who were admitted alive was 2112.5 days (3–5511 days). Primary repair of the aorta by end-to-end anastomosis was performed in 22 patients, synthetic graft interposition in 17 and endovascular stenting in 11 patients. Notably, in the study centres, aortic stenting for TAR was first introduced in 1998. In 28 patients, the aorta was repaired prior to treatment of associated injuries (13 end-to-end anastomoses, 10 grafts and 5 stents). Delayed aortic repair was performed in 22 patients (9 end-to-end anastomoses, 7 grafts and 6 stents), five patients were treated from the second to the seventh day after accident and nine patients on the 8th to the 30th day. In the univariate statistical analysis, there were three significant risk factors for in-hospital mortality (table 3). The risk for in-hospital death was significantly higher in those patients who were haemodynamic unstable upon arrival to the emergency room in comparison with those who were haemodynamically stable (83.3% vs 16.7%, p<0.001). Moreover, the extent of aortic injury was a significant risk factor for in-hospital mortality; 52.9% (10/12) of the patients with complete aortic transection died during hospitalisation compared with those with a partial rupture (7/34, 20.6%, p=0.03). Furthermore, the surgical procedure used had significant impact on in-hospital mortality (p=0.02); 58.8% of patients (10/17) with open graft interpositions died postoperatively. The perioperative mortality rate after direct suture and endovascular stenting was 18.2% (4/22) and 9.1% (1/11). In the multivariate model, haemodynamic instability upon arrival was the sole determinant of in-hospital mortality. The variables, age, gender, mechanism of trauma, ISS, the timing of surgery and pretreatment in a peripheral, admitting hospital were equally distributed between 30-day survivors and non-survivors.
Overall mortality
The overall mortality rate of the 85 patients was 58.8% (50/85). Non-survivors were significantly older than survivors (53.0 vs 37.9 years, p=0.006). The two factors of gender and the mechanism of trauma were equally distributed between survivors and non-survivors. Patients with complete aortic transection had 81.0% (34/42) risk of dying compared with 37.2% (16/43) of patients with a partial rupture (p=0.003). Further injuries that led to a higher death rate were additional chest trauma and liver injuries. These patients had 72.7% (32/44) and 83.3% (25/30) risk of death compared with those patients without chest and liver injuries (43.9%, 18/41, p=0.03 and 45.5%, 25/55, p=0.02, respectively). Associated injuries of head, spleen, kidneys, intestines, spine, pelvis and limbs were equally distributed between overall survivors and non-survivors. In the multivariate model, the age at time of accident and a complete aortic transection were significant determinants of overall mortality (p=0.02 and p=0.001, respectively). All significant determinants of overall mortality are summarised in table 4.
Discussion
Injuries of the thoracic aorta with its reported incidence of 20–30 cases per one million inhabitants per year occur in one third of blunt traumatic fatalities leading to death at the scene in majority of the cases.3–6 The aim of the present retrospective study was to investigate mortality and its risk factors in patients with traumatic rupture of the thoracic aorta who died at the scene compared with patients who were transferred alive to the tertiary centres. Moreover, risk factors for in-hospital mortality in these patients may help in identifying those at risk for an adverse outcome. Based on our results, we identified associations with mortality after blunt aortic rupture on three different levels (prehospital, in-hospital and overall).
The overall mortality rate in our study was 58.8% with a median survival time of 2 days. This noticeably short survival time reflects the high rate of prehospital deaths included in the study. Thirty-four patients (40.0%) died before reaching hospital care, which is proportionally lower compared with 75%–90% reported by others,3–5 and may reflect the lower road fatality rate in Switzerland compared with the USA (83 vs 152 deaths per one million inhabitants in the year 2000). Furthermore, the driving or flight distances in Switzerland between ambulance home bases to the scene of accident and from there to the next trauma centre are considerably shorter than in American or Australian landscapes. However, transfer times were not investigated in the present study. Another issue might be the unequal introduction to seat belt laws. In Switzerland, the seat belt law was introduced in 1981, whereas up to the present it still varies greatly from state to state in America. Nonetheless, based on our data, we are unable to prove this hypothesis. Regardless, prevention needs to be the highest priority in future efforts.
In the statistical analysis, we identified age at time of accident; complete aortic transection; and additional injuries of chest, liver and spine as risk factors for death at the scene. In the retrospective analysis performed by Brundage et al,7 pedestrians struck by cars were shown to have a significantly higher mortality due to TAR. According to our research, we are unable to support this conclusion as the mechanism of trauma was not a significant determinant of mortality. A pedestrian struck by a car was a marginally significant determinant of prehospital mortality (p=0.05). However, the risk of aortic injury is greater in occupants of motor vehicle crashes, which is the leading cause of aortic rupture in our own (49.4%) and other studies with a prevalence between 50% and 57%.3 ,4 ,6 We found drivers to be affected threefold as much as non-drivers, but we are not able to compare safety belt users versus non-users, airbag versus non-airbag and side impact versus frontal collision. Based on several studies, it seems clear that the risk of aortic rupture is greater in non-belted or improperly belted occupants and in side impacts.8–12 Notably, even side impacts at low energy can lead to aortic injury compared with frontal collisions.10 The exact physical mechanism of aortic rupture is still being debated, but seems to correspond with a complex combination of torsional and shearing forces after chest deceleration/compression, resulting in transverse laceration and rupture of the aorta.13 ,14 Associated injuries in terms of a second intrathoracic and abdominal trauma were already found to be risk factors for mortality in the work of Sturm et al,15 which is supported by the present study. Moreover, an associated spinal injury turned out to be a significant determinant of prehospital mortality. Therefore, we emphasise proper diagnostic and therapeutic management in patients with TAR and additional injuries of chest, liver and spine as these patients would have an adverse outcome if an aortic rupture is detectable.
The in-hospital mortality in our study was 31.4%, which is comparable with the mortality observed by others.4 ,7 However, the surgical procedure contributed to improved survival after thoracic endovascular aortic repair (TEVAR) in comparison with the open procedure (58.8% vs 9.1%). The perioperative mortality after TEVAR for TAR reported in literature is between 4.2% and 19%.16–19 Furthermore, we were able to find the age on the date of accident as a significant risk factor for in-hospital and overall mortality. Data concerning age as a risk factor for death in TAR is conflicting in the literature.15 ,20 However, the HR for early death after TAR was only 1.05 suggesting that age has no major impact on the outcome. Moreover, the mean age in our cohort was evidently higher compared with others (47 vs 38 years) although the mortality rate in our study was lower.4 As with previous reports,5 ,19 ,21 we identified unstable haemodynamic condition on hospital admission as a risk factor for in-hospital mortality (HR 16.11, p<0.001). In haemodynamic stability, delayed aortic repair favouring surgical treatment of other life-threatening associated injuries is recommendable as the timing of the aortic intervention has had no significant impact on the outcome in our study. This was consistently stated in other studies.5 ,20 ,22 The sole significant determinant of mortality on all three levels (prehospital, in-hospital and overall) was complete aortic transection, which was also significant in the multivariate analysis. It seems logical that a circular disruption of the aorta is associated with a higher mortality than partial rupture, as the risk of hypovolemic shock is greatly enhanced. Thus, aortic repair of patients with circular disruption of the aorta should not be delayed. Curiously, the in-hospital mortality in our cohort was not affected by the number or localisation of associated injuries.
There are some limitations of this study including the nature of its retrospective design and a possible selection bias, associated with the criteria used to perform a postmortem examination. However, in Switzerland, all fatalities from accidents are subject to a legal autopsy. Furthermore, a lack of completion on the analysis of all patients may pose a source of uncertainty. However, due to the fact that the three study centres are the sole institutions caring for subjects (patients or victims) with TAR in the greater area of Zurich with a defined number of inhabitants, the calculation of incidence and mortality are quite reliable. Lastly, the sample size in our study was relatively small and the study was terminated before routine use of TEVAR, which may result in improved in-hospital and overall outcome. The strength of the study includes the population-based selection of patients identified with TAR, allowing a comparison of deaths at scene with those who survived to hospital in order to identify possible risk factors of mortality. The present study is the first to include both TAR-related deaths at the scene and those who survived to hospital.
Conclusion
In multivariate analysis, age and complete aortic transection were significant risk factors of overall mortality, whereas the latter was the only significant determinant of outcome in patients who died at the scene. Additionally, in the univariate analysis, associated injuries of chest, liver and spine were significant risk factors of death at the scene compared with haemodynamic instability and open aortic repair, which seem to predict in-hospital mortality. Therefore, risk factors of prehospital and in-hospital death in patients with TAR differ greatly. The in-hospital mortality was not affected by the number or localisation of associated injuries.
Acknowledgments
We thank Professor W Baer, the former director of the Institute of Forensic Medicine from the University of Zurich for helping us with patient recruitment and data entry.
References
Footnotes
-
Contributors Both authors planned the study and conceived the study design. DF performed the literature search, study design, data collection, data analysis, and writing of the manuscript. MG contributed interpreting the data and critically revised the manuscript.
-
Competing interests None.
-
Ethics approval Committee of Ethics from Canton for Zurich.
-
Provenance and peer review Not commissioned; externally peer reviewed.