Blast injury to the human skeleton: recognition, identification and differentiation using morphological and statistical approaches.

Abstract

Anthropologists are increasingly called upon to assess trauma to the skeleton and contextualising the nature of this trauma. Blast injury is a type of trauma which is increasingly seen in a variety of contexts, such as terrorism, human rights violations, combat and accidents. The purpose of this study was to examine blast injury in the human skeleton and apply robust multivariate statistical methods, alongside morphological methods, to identify blast trauma based on the distribution of injury in the skeleton. The objectives of the study were to identify patterns in a sample of cases from mass graves in Bosnia and to determine differences between the blast injury cases and gunshot wound cases which can identify indicators of blast injury for future use. This was done using Pearson‘s χ2, cluster analysis and multiple correspondence analysis. Secondly, the identified indicators were applied to two methods of binary logistic regression model to test prediction of the presence or absence of blast injury in the sample, as well as assessing the results of the two methods. Lastly, the results of these analyses were subsequently compared with clinical literature to identify similarities and differences which can aid anthropologists in determining presence of blast injury in large assemblages. This also served to address the specific argument that the injuries seen in the Bosnia sample are combat related, as claimed in court proceedings in ongoing cases at the International Criminal Tribunal for the Former Yugoslavia. It was found that cluster analysis was not useful for the sample examined in the research, however multiple correspondence analysis permitted graphical differentiation between the blast injury and gunshot cases, identifying variables which contributed to the variance and could be used as indicators of blast injury. Binary logistic regression was employed to test the significant contribution of these variables to a model predicting the presence of blast injury in a sample. It was found that presence of trauma to the right shoulder girdle, left forearm, vertebrae, right pelvis and left femur could indicate the presence of blast injury in an assemblage, with correct average classification in 74.86% of cases. The prevalence of trauma in the Bosnia sample was compared with examples from terrorist incidents and combat situations to identify similarities and differences between these and found that there significant differences in the prevalence of trauma in the Bosnia sample. This highlights that this sample does not resemble any combat patterns of injury, answering the question posed in ICTY court proceedings. This work contributes new knowledge to anthropology on the identification and differentiation of blast injury in assemblages as well as demonstrating the use of multivariate statistical methods for trauma analysis. These results can be applied to anthropological investigation of historical contexts as well the modern investigations which will require knowledge of blast injury currently and in the future.