Abstract
BACKGROUND: Due to continuous improvements in treatment, more and more severely and seriously injured patients are surviving. The complexity of the injury patterns of these patients means that they are difficult to map in routine data. AIM OF THE WORK: The aim of the data exploration was to identify ICD 10 diagnoses that show an association with an injury severity score (ISS) ≥ 16 and could therefore be used to operationalize severely injured patients in routine data. MATERIAL AND METHODS: The coded four-digit ICD 10 S diagnoses and the calculated ISS of trauma patients from the Armed Forces Central Hospital Koblenz (BwZKrhs) and the University Hospital Düsseldorf (UKD) were analyzed using statistical association measures (phi and Cramer's V), linear regressions and machine learning methods (e.g., random forest). RESULTS: The S diagnoses of facial, head, thoracic and pelvic injuries, associated with an ISS ≥ 16 were identified. Some S diagnoses showed an association with an ISS ≥ 16 in only 1 of the 2 datasets. Likewise, facial, head, thoracic and pelvic injuries were found in the subgroup of 18-55-year-old patients. DISCUSSION: The current evaluations show that it is possible to identify ICD 10 S diagnoses that have a significant association with an ISS ≥ 16. According to the annual report of the trauma register of the German Society for Trauma Surgery (TR-DGU®), injuries with an abbreviated injury scale (AIS) ≥ 3 are particularly common in the head and thoracic regions.