Identification of sepsis subphenotypes: current methods and clinical implications in critical care practice. A structured narrative review

Sepsis demonstrates high variability in clinical manifestations, and patients with similar manifestations at the moment of diagnosis usually develop in very different ways. Such differences are explained by the difference in the immune response and organ reactions to infection. To counter this diversity, scientists have started to classify the patients into subphenotypes to determine whether the treatment can be planned in an individualized way. The objective of this review is to determine the methods of determining sepsis subphenotypes and their role in clinical practice. To conduct this review, a search in PubMed was conducted on studies published between 2015 and 2025. Papers that provided explicit information about the process of subphenotype identification were included. Articles that lacked the complete text, publications that were not published in English, and those that lacked the necessary methodological information were filtered out. In the chosen studies, there were several methods used, based on routinely available clinical variables; others were based on biomarkers, machine-learning techniques, time-varying trends in the dysfunction of organs, and transcriptomic data. Even though the methods are different, they have found many groups of patients who have significant differences in terms of inflammation, organ failure patterns, and response to treatment. According to the available evidence, there is no one such condition that sepsis represents, but rather a series of biologically different conditions. Subphenotyping can improve initial diagnosis and treatment, but most current methods are complex for clinical use. More efforts should be made to establish less complex tools, which may be implemented consistently at the bedside.