A predictive model to analyze the factors affecting the presence of serious chest injury in the occupants on motor vehicle crashes: Logistic regression approach.

Objective: Chest injuries that occur in motor vehicle crashes (MVCs) include rib fractures, pneumothorax, hemothorax, and hemothorax depending on the injury mechanism. Many risk factors are associated with serious chest injuries from MVCs. The Korean In-Depth Accident Study database was analyzed to identify risk factors associated with motor vehicle occupants' serious chest injury.

Methods: Among 3,697 patients who visited the emergency room in regional emergency medical centers after MVCs between 2011 and 2018, we analyzed data from 1,226 patients with chest injuries. Vehicle damage was assessed using the Collision Deformation Classification (CDC) code and images of the damaged vehicle, and trauma scores were used to determine injury severity. Serious chest injury was defined as an Abbreviated Injury Scale (AIS) score for the chest code was more than 3. The patients were divided into two groups: serious chest injury patients with MAIS ≥ 3 and those with non-serious chest injury with MAIS < 3. A predictive model to analyze the factors affecting the presence of serious chest injury in the occupants on MVCs was constructed by a logistic regression analysis.

Results: Among the 1,226 patients with chest injuries, 484 (39.5%) had serious chest injuries. Patients in the serious group were older than those in the non-serious group (p=.001). In analyses based on vehicle type, the proportion of light truck occupants was higher in the serious group than in the non-serious group (p=.026). The rate of seatbelt use was lower in the serious group than in the non-serious group (p=.008). The median crush extent (seventh column of the CDC code) was higher in the serious group than in the non-serious group (p<.001). Emergency room data showed that the rates of intensive care unit (ICU) admission and death were higher among patients with serious injuries (p<.001). Similarly, the general ward/ICU admission data showed that the transfer and death rates were higher in patients with serious injuries (p<.001). The median ISS was higher in the serious group than in the non-serious group (p<.001). A predictive model was derived based on sex, age, vehicle type, seating row, belt status, collision type, and crush extent. This predictive model had an explanatory power of 67.2% for serious chest injuries. The model was estimated for external validation using the confusion matrix by applying the predictive model to the 2019 and 2020 data of the same structure as the data at the time of model development in the KIDAS database.

Conclusions: Although this study had a major limitation in that the explanatory power of the predictive model was weak due to the small number of samples and many exclusion conditions, it was meaningful in that it suggested a model that could predict serious chest injuries in motor vehicle occupants (MVOs) based on actual accident investigation data in Korea. Future studies should yield more meaningful results, for example, if the chest compression depth value is derived through the reconstruction of MVCs using accurate collision speed values, and better models can be developed to predict the relationship between these values and the occurrence of serious chest injury.