Sepsis presents a global challenge to emergency departments (ED) due to its varied presentation and life-threatening outcomes. The quick Sequential Organ Failure Assessment score's introduction expanded sepsis diagnostics beyond the traditional Systemic Inflammatory Response Syndrome, which considered body temperature. However, body temperature remains a vital clinical marker. This study analyzes body temperature patterns in ED sepsis patients and their impact on treatment and outcomes.

This retrospective cohort study analyzed the data from the MIMIC-IV database, focusing on sepsis patients diagnosed within 12 h of ED triage. Patients were split into derivation (n = 3872) and replication (n = 1660) cohorts in a 7:3 ratio and categorized into body temperature trajectory subgroups using group-based trajectory modeling. The model's effectiveness was assessed using the Bayesian Information Criterion, with validation in a replication cohort. Analysis of variance and chi-squared tests were used to evaluate patient characteristics and outcomes.

Four distinct body temperature patterns were identified: Hypothermic (0.9 %), Normothermic (74.1 %), Progressive Fever (10.7 %), and Fever Resolver (14.4 %). The Hypothermic group had the highest mortality rate (37.5 %), while the Normothermic, Progressive Fever, and Fever Resolver groups had mortality rates of 11.7 %, 9.14 %, and 9.78 %, respectively (p < 0.001).

Fever is not always a reliable sepsis indicator; many patients present normothermic or hypothermic, notably the highest mortality cohort. Including diverse temperature patterns in sepsis assessments could enhance patient outcomes.