Pre-culture Hypothermia and Short-term Mortality Among ICU Patients with Positive Blood Cultures: A Retrospective Cohort Study Using MIMIC-IV

Abstract

Temperature is often treated as a simple trigger for infection work-up in the intensive care unit (ICU), yet the absence of fever can be clinically deceptive. We conducted a retrospective cohort study using MIMIC-IV v3.1 to examine whether temperature patterns before the first qualifying positive blood culture were associated with short-term mortality in adults with operationally defined ICU-acquired bloodstream infection (BSI). The index event was the first positive blood culture collected at least 48 hours after ICU admission; patients with an earlier positive blood culture in the same hospitalization before ICU hour 48 were excluded. Temperature patterns during the 24 hours before culture were classified as normothermic, febrile, hypothermic, or mixed dysregulated. Among 525 patients, pre-culture temperature data were available for 511. Hypothermia occurred in 51 patients and was associated with higher in-hospital mortality than normothermia after adjustment for age, sex, time from ICU admission to culture, vasoactive infusion, and invasive ventilation (odds ratio [OR] 2.20, 95% confidence interval [CI] 1.13-4.28). The association was similar for ICU mortality and observed 28-day mortality within available hospital/ICU follow-up. A complete-case laboratory-adjusted model showed a similar direction but wider uncertainty. These findings support pre-culture hypothermia as a nurse-sensitive surveillance signal rather than a causal claim or a stand-alone prediction rule.