In the present study severe thoracic trauma showed a strong impact on patient mortality, but lower than severe head trauma. The odds ratios for death in the multivariate analysis were quite similar for thoracic, facial, and abdominal/pelvic traumas. In addition, three-quarters of those with a severe thoracic trauma and who died did so within the first 24 hours. Surprisingly, Grubmüller et al. found that the presence of a severe thoracic trauma had no influence on mortality 15, contrary to that found herein results and other previous investigations 9. They hypothesized that this finding could be led by the high case load of their level I trauma centre (>100/centre) whereas our study included level I to III trauma centres. Indeed, some authors suggested that the hospital volume of severely injured patients may be an independent predictor of survival 16,17. Furthermore, their results are based on a single-centre experience where penetrating trauma and patients suffering unsuccessful emergency resuscitation were excluded. In the present study, road network (highway and rural road compared to city street), road user (car occupant) and being a male were significantly associated with a higher risk of death while trauma severity (AIS) was included in the multivariate analysis. Because we dichotomised level of severity in two groups (AIS=2 vs AIS≥3), the distribution of AIS≥3 could not be homogenous between road network categories, user categories and sex and could resulted in higher frequency of severe (AIS 4), critical (AIS 5) and beyond treatment (AIS 6) injuries in some groups among previous cited categories which could explain our results.
In the present study, mortality was similar to what is reported elsewhere, Grubmüller et al. reported mortality was 13.2% 15, Huber et al. reported a 17.5% mortality 9 and Veysi et al. mortality was 18.7% in their cohort 5. In contrast, in the German multicentre database TraumaRegister DGU®, Horst et al.. 18 reported an overall mortality rate around 5% in their 10-year-period study. So did Chrysou et al. with a reported mortality of 5.5%.12 However, only patients reaching the hospital with vital signs were included whereas in the present study we did not exclude on-scene dead patients. If we had excluded these patients, we would have found a mortality rate of 6.2%. One strength of the present study was the inclusion of on-scene deceased patients without inducing any misclassification bias because an autopsy was systematically undertaken in those patients to ensure the cause of death and provide a complete injury assessment based on AIS scoring.
The present study found a lower incidence of thoracic trauma compared to that reported elsewhere 9,19,20. This was probably underestimated herein as those suffering minor thoracic injuries (AISThorax=1), including single rib fractures, were not considered. Peek et al. recently reported in their nationwide studies that single rib fracture were frequently encountered injuries (20% of rib fractures) 21,22. We chose not to include these patients because previous studies have demonstrated that these single rib fractures were not associated with mortality 9. We also noted that Chrysou et al. reported lower incidence of sternal fracture (15.5%) but a higher incidence of concomitant spine injuries (53.6%); which may be explained by the inclusion of patients suffering fall from height and snow sport accidents in their study 12 which have been previously reported as a significant source of traumatic spinal cord injuries 23,24.
Another interesting finding of the present study is the evolution of AISThorax 2 and 3 with age. AIS 3 injuries incidence was the highest in children (0-9 years) and decreased with age. This is in line with Samarasekera et al. findings reported a very high incidence of AIS>2 injuries (88%) among children (<15years) thoracic injuries with 65% of them being lung contusions 25 and Ostermann et al. recently reported half of thoracic injuries being lung contusions (AIS 3) 11. One explanation is the difference in anatomy; the thoracic skeleton of children being incompletely calcified and more compressible allowing for the transmission of large forces to the structures of the thoracic cavity, making rib fractures uncommon. Thus, high-energy impact trauma may cause major internal injuries with little evidence of external injuries or fractures of the bony thorax as reported by many authors 11,26,27. Furthermore, as children are more frequently involved in RTA as pedestrians they may be at higher risk of projection and severe trauma caused by direct impact with the front bumper because of their smaller height. In contrast, the incidence of AIS=2 thoracic injuries incidence increased with age. These injuries are mainly rib fractures. which, in the elderly, have been reported as the most frequent after a RTA 28. They required less force compared to young patients because of osteoporosis and loss of muscle mass.
This study has a number of limitations. One of them is the update of AIS during the study period 29. Nevertheless, we chose to base severity assessment on AIS because it has been extensively used in the literature and is an accurate, objective and validated method to independently evaluate each body region impact on mortality 30 contrary to the ISS that evaluates the overall severity 31. Besides, Hsu et al.32 found that the AIS update had no impact on mean ISS when considering the thoracic body region. Although it is a register study with the known weaknesses, the present multicentre study is based on data prospectively collected over a 20-year-period from level I to III trauma centres and included patients of all ages whereas many studies focused on thoracic trauma are single centre studies and only included patients ≥16 years 5,8,12,15,19,33. Thus, our findings are generalisable to other high-income countries. Besides, data were collected by emergency physicians, blinded to which of the risk factors and outcomes were being used in the present study. Another limitation is that the Rhône registry only includes victims of RTA. Therefore, the registry represents a trauma population with a majority of blunt trauma and other reported thoracic mechanisms such as falls, or stab/fire gun injuries, have not been studied here. Also, the objective relies on an anatomic system of injury classification (AIS and ISS) since the database does not report physiological assessment nor medical management. There is therefore no data regarding prehospital management, time to surgery when needed, intensive care unit management such as airway and ventilation management, use of vasopressors or massive blood transfusion, and no detail on organ and respiratory failure or pre-existing cardiopulmonary disease which have been reported to influence outcome in thoracic trauma patients 6,8,9,15. The value of anatomic scoring systems in outcome prediction of trauma patients has been debated and compared to physiologic scores. Some authors found ISS to be a best severity predictor compared to the revised Trauma Scale (RTS) and the simplified acute physiology scale II (SAPS II) 34. Hence, we believe that using anatomic scoring in the present study is as efficient as physiologic scoring in predicting outcomes.