Abstract
BACKGROUND: Evidence regarding the relationship between temperature and in-hospital mortality is limited. Hence, we aimed to examine the potential association between temperature and in-hospital mortality in patients with congestive heart failure (CHF) and acute myocardial infarction (AMI), while taking into account other variables. METHODS: This was a retrospective cohort study utilizing data from the Medical Information Mart for Intensive Care (MIMIC-IV) database. The patient data of 5057 CHF with AMI were sourced from the MIMIC-IV database. The target independent variable was the first temperature measured at baseline (upon ICU admission), and the dependent variable was in-hospital mortality. Covariates included age, race, sex, severity scores, comorbidities, vital signs, drug information, mechanical ventilation, intubation, and laboratory test indexes. Multivariate regression models, a Generalized Additive Model, smooth curve fitting, and subgroup analysis were applied to evaluate the relationship. RESULTS: The study comprised 5057 participants (average age 73.5 ± 12.4 years; 61.1% male). In-hospital mortality occurred in 713 cases (14.1%). Fully adjusted binary logistic regression analysis demonstrated that temperature was independently correlated with the risk of in-hospital mortality (Odds Ratio [OR] = 0.77, [95% CI 0.64 ~ 0.92]). A U-shaped relationship was identified between body temperature and in-hospital mortality, with a critical point at 37 °C. The effect sizes on both sides of the inflection point were: OR = 0.52 (95% CI 0.31 ~ 0.85) for temperature < 37 °C, and OR = 5.16 (95% CI 1.19 ~ 22.36) for temperature ≥ 37 °C. Subgroup analysis suggested a potentially stronger correlation in patients with higher APSIII scores. CONCLUSION: An independent, U-shaped association exists between body temperature at ICU admission and in-hospital mortality in patients with CHF and AMI. Both low and high body temperatures are linked to an increased risk of mortality. These findings highlight the prognostic value of temperature and suggest that monitoring it may aid in risk stratification, though causal inference is limited by the observational design.