[Effects of injury pattern on the predictability of 4 multiple trauma scores. Presentation of a method for identifying artifacts].

When comparing two populations of multiply injured patients differences in patient characteristics must be controlled for. To measure the overall severity of injury, scaling systems are used. If after adjustment for injury severity, the proportion of deaths in the two data sets are still different, the difference is to be considered as due to the quality of care. However, this conclusion is only valid after excluding the possibility that the scale in use fails to adequately reflect certain injuries. The scope of the study is to demonstrate a method for examining this potential interference. As an example, four widely used scales were applied to the data of 418 multiply injured patients. By means of multiple logistic regression analysis, variables were selected which have an influence on prognosis in addition to a scale, thus indicating a subgroup of patients who are underrepresented by the respective scoring system. For the scales examined, these additional variables were: Head and thoracic trauma for the Polytrauma score (Oestern), abdominal and thoracic trauma for the Trauma index (Schreinlechner), thoracic trauma and age for the Trauma score (Champion), head trauma and age for the Injury Severity Score (Baker). We conclude that each score analyzed had its characteristic weak points. Prognostic quality was affected by casemix. Therefore, comparisons between groups of polytraumatized patients may be invalidated. The method outlined here is a useful means for checking a scoring system for these types of interfering variables. Therefore, it is recommended to search routinely for potentially interfering variables before applying a scale. In a given data set of multiply injured patients, appropriate adjustments can then be made for the deficiencies of the scoring system.