This study addresses the challenge of predicting mortality in sepsis among burn patients. Given the heterogeneity of sepsis, especially in the context of burn injuries, this study aims to identify reliable biomarkers for mortality prediction. The study is a retrospective review, focusing on the evaluation of various biomarkers and their changes over time in a burn patient cohort. Conducted in the Burn Intensive Care Unit of Hangang Sacred Heart Hospital, the study involved a retrospective review of 1,659 adult burn patients from January 2010 to December 2022. Key biomarkers analyzed include lactate levels, pH, platelets, procalcitonin, and others. Advanced clustering methodologies, such as dynamic time warping and hierarchical clustering, were utilized to classify patients into distinct groups based on their biomarker profiles and clinical outcomes. The study identified four patient clusters with unique lactate level trajectories. Significant findings include the identification of procalcitonin, pH, and platelets as key predictors of mortality, with varying degrees of efficacy across different clusters. For instance, in the "Persistent Rise" cluster, pH and platelet count showed Area Under the Curve (AUC) values of 0.756 and 0.753, respectively, indicating their strong predictive power. The study concludes that a combination of biomarkers, especially lactate dynamics, can effectively predict mortality in burn-induced sepsis. The results advocate for a more personalized approach in managing sepsis in burn patients, considering the specific biomarker trajectories. These findings are crucial for enhancing treatment strategies and improving patient outcomes in burn care.
Keywords: Burn patients; Clustering analysis; Lactate Dynamics; Mortality prediction; Sepsis.
© 2024. The Author(s).