A series of 34 side impact tests conducted at the Medical College of Wisconsin and at Ohio State University using post-mortem human subjects in a Heidelberg type sled, were examined for the purpose of developing and assessing thoracic injury criteria for side impact. The effects of three test conditions were investigated: test speed (24 or 32 kph), impact surface (padded or rigid), and pelvic offset (present or absent). The post-mortem human subjects were instrumented with accelerometers on the ribs and spine and chest bands around the thorax and abdomen to characterize their mechanical response during the impact. Load cells at the walls measured the impact force at the level of the thorax, abdomen, pelvis, and lower extremity. The resulting injuries were determined through radiography and detailed autopsy and their severity was coded according to the AIS 90 Scale. Rib fractures were the most common injury type with injury severity ranging from AIS=O to AIS=5. Chest deflections were derived by using the chest band data to compute the chest contours at every millisecond during the event.
The test data were analyzed using statistical techniques such as ANOVA, linear regression, logistic regression, and categorical analysis. Several existing candidates for side impact injury criteria were evaluated such as Thoracic Trauma Index (TTI), Average Spinal Acceleration (ASA), chest deflection, chest velocity, chest VC, peak and average contact force, stored energy criteria (SEC) and energy storing rate criteria (ESRC) for their injury prediction ability. The age of the subject was found to influence injury severity while gender and mass were found to have little or no influence on injury response. Accelerations filtered with SAE Class 180 filters were better predictors of injury than accelerations filtered with SAE class 600,60 or FIR100 filters. Maximum normalized chest deflection (dmaxn) was a better predictor of rib fractures (R2=0.54, p-value=O.OOOl) and injury severity based on AIS (score p-value=O.OOOl, Gamma=O.71) than any other existing injury criteria with TTI being the next best predictor ofinjury severity based on AIS (score p-value=0.0012, Gamma=0.64). Maximum normalized resultant upper spine acceleration (rspul80n) was the best individual predictors of injury severity based on rib fractures and maximum AIS levels with a pvalue= O.OOOl . A model using a linear combination of age, dmaxn, and rspul80n was a significantly better predictor of rib fractures and injury based on AIS (p-value=O.OOOl, Gamma=0.86). Similarly, a model using a linear combination of age and the product of dmaxn and rspul80n was also a good predictor of injury severity (p-value=O.OOOl, Gamma=O.85).