First-hour infection recognition and 24-hour sepsis classification in emergency resuscitation-room encounters: a retrospective cohort study

Abstract Background During the first hour of emergency resuscitation, clinicians must decide whether critical illness is infectious and whether an infected patient has, or develops, infection-associated organ dysfunction. The first decision may be difficult in older or severely ill patients with atypical presentations; the second determines the urgency of monitoring and resuscitation. We assessed how first-hour information supported these linked clinical questions. Methods This single-centre retrospective cohort included 1,195 non-consecutive adult resuscitation-room encounters: 240 non-infectious critical illnesses, 218 infections without sepsis and 737 cases of sepsis. We compared age, sex and vital signs; their combination with complete blood-count indices; and the further addition of lactate and procalcitonin. Analyses developed in 2023–2024 were applied without refitting to a post hoc 2025 sample. Question 1 distinguished final physician-adjudicated infection from non-infectious illness; Question 2 classified 24-hour Sepsis-3 status among adjudicated infections. Results In the 2025 sample, 139 of 174 encounters involved infection. Vital signs and complete blood-count indices distinguished infection from non-infectious illness with an area under the receiver operating characteristic curve (AUROC) of 0.808 (95% confidence interval [CI], 0.733–0.877). Adding lactate and procalcitonin increased the AUROC to 0.868 (0.813–0.921; change, 0.060 [0.029–0.097]). At a high-sensitivity threshold, 127 of 139 infections were identified and 23 of 35 non-infectious encounters were also flagged. Among 139 infected encounters, 110 met the 24-hour sepsis definition. Clinical and complete blood-count information identified 103 cases and showed moderate discrimination (AUROC, 0.780 [0.682–0.865]). Adding lactate and procalcitonin did not improve discrimination (AUROC, 0.766 [0.665–0.856]; change, − 0.014 [− 0.090 to 0.060]). Conclusions In this selected emergency-resuscitation cohort, first-hour vital signs and complete blood-count indices discriminated final infection classification from non-infectious critical illness, and lactate plus procalcitonin improved that discrimination. Among infected patients, first-hour clinical and complete blood-count information also discriminated 24-hour sepsis status, but lactate and procalcitonin added no clear value. These findings support structured early assessment and reassessment; they do not establish a stand-alone treatment rule.