Disrupted circadian temperature rhythm is commonly observed in elderly patients in the intensive care unit (ICU), but the association between circadian temperature rhythm and mortality in elderly patients is unclear. Adult patients with a relatively complete record of body temperature (BT) during the first 24 hours of ICU stay in the Multi-parameter Intelligent Monitoring in Intensive Care IV (MIMIC-IV) database were included in this retrospective cohort study. The circadian rhythm of body temperature was blunted as a ratio of the maximum BT between 12:00 and 24:00 divided by the minimum BT between 0:00 and 12:00, and we defined it as BT fluctuation ratio. The associations of BT fluctuation ratio with 28-day mortality were assessed separately using Cox proportional hazards model in elderly patients and non-elderly patients. The overall cohort comprised 12 767 patients. After adjusting for covariates, the analysis showed that the BT fluctuation ratio (%) was significantly associated with mortality at 28 days in total patients (hazard ratio: 1.044; 95% CI 1.001-1.088; P = 0.042), and still significantly in elderly patients (hazard ratio 1.055, 95% CI as 1.004-1.109, p = 0.035), but not significantly in non-elderly patients. The implementation of restricted cubic splines demonstrated a nonlinear correlation between the ratio of BT fluctuation and the hazard ratio of 28-day mortality, indicating that increased diurnal temperature fluctuations are linked to elevated risk of mortality. This study revealed that the augmented amplitude of the circadian rhythm of body temperature in the elderly patients constitutes a risk factor for the rise of 28-day mortality. Additionally, the circadian body temperature rhythm may facilitate the early detection of critically ill elderly patients.
Keywords: Body temperature; Multi-parameter Intelligent Monitoring in Intensive Care IV; circadian rhythm; intensive care; mortality.