The BUrn Mortality Prediction (BUMP) Score – An improved mortality prediction score based on data of the German burn registry

Abstract

BackgroundBurn injuries constitute the fourth most common injuries globally. Patient outcomes must be currently assessed to provide appropriate patient care with high quality standards. However, existing mortality prediction scoring methods have been shown to lack accuracy in current burn patient populations. Therefore, this study aimed to validate existing scores using current patient data and assess whether new prediction parameters can provide better accuracy. MethodsA retrospective analysis of the patient data from the German Burn Registry between 2016 and 2019 was performed to evaluate all Abbreviated Burn Severity Index (ABSI) score parameters. All patients over 16 years of age who received intensive care were included. Descriptive statistics and logistic regression analysis were used to identify novel prediction parameters based on the parameters documented at admission and establish a new prediction score, the BUrn Mortality Prediction (BUMP) score. The quality of the new score was subsequently compared to that of the original ABSI, modified ABSI, Galeiras, Revised Baux score and TIMM. The new prediction score was then validated using patient data collected in the German Burn Registry in 2020. ResultsIn total, 7276 patients were included. Age; the presence of at least two comorbidities; burn injuries caused by work-related accidents, traffic accidents and suicide attempts; total burn surface area; inhalation trauma and full-thickness burns were identified as independent significant predictors of mortality (p < 0.001). Additionally, we evaluated new age groups to improve prediction accuracy. The number of comorbidities (p < 0.001) and the aetiology (burns occurring at work [p = 0.028], burns caused by traffic accidents [p < 0.001] or burns due to attempted suicide [p < 0.001]) had a significant influence on mortality. The BUMP score, which was developed based on these parameters, showed the best fitness and showed more accurate mortality prediction than all the above-mentioned scores (area under the receiver operating characteristic curve: 0.947 [0.939–0.954] compared to 0.926 [0.915–0.936], 0.928 [0.918–0.939], 0.937 [0.928–0.947], 0.939 [0.930–0.948], 0.940 [0.932–0.949] respectively). ConclusionsA novel score (BUMP score) was developed for the purpose of external quality assessment of burn centres participating in the German burn registry, where observed and expected outcomes are compared on a hospital level, and for scientifically applications. The clinical impact of this score and its generalisability to other patient populations needs to be evaluated.