Changes in Body Temperature Patterns Are Predictive of Mortality in Septic Shock: An Observational Study

SIMPLE SUMMARY: We applied, here, mathematical modeling to the body temperature time series (sinusoidal regression and cosinor analysis) in a cohort of septic shock to describe an accurate temperature course and its relation to mortality. Lower mesor (mean temperature) and higher amplitude were associated with mortality and could be considered an interesting prognostic marker in septic shock. In the age of artificial intelligence, the incorporation of such data in an automated scoring alert could compete with physicians to identify high-risk patients during septic shock. Besides, we demonstrated that many factors, such as specific therapeutic strategies, may influence the normal body temperature rhythm. ABSTRACT: Biological rhythms are important regulators of immune functions. In intensive care unit (ICU), sepsis is known to be associated with rhythm disruption. Our objectives were to determine factors associated with rhythm disruption of the body temperature and to assess the relationship between temperature and mortality in septic shock patients; In a cohort of septic shock, we recorded body temperature over a 24-h period on day 2 after ICU admission. For each patient, the temperature rhythmicity was assessed by defining period and amplitude, and the adjusted average (mesor) of the temperature by sinusoidal regression and cosinor analysis. Analyses were performed to assess factors associated with the three temperature parameters (period, amplitude, and mesor) and mortality. 162 septic shocks were enrolled. The multivariate analysis demonstrates that the period of temperature was associated with gender (women, coefficient −2.2 h, p = 0.031) and acetaminophen use (coefficient −4.3 h, p = 0.002). The mesor was associated with SOFA score (coefficient −0.05 °C per SOFA point, p = 0.046), procalcitonin (coefficient 0.001 °C per ng/mL, p = 0.005), and hydrocortisone use (coefficient −0.5 °C, p = 0.002). The amplitude was associated with the dialysis (coefficient −0.5 °C, p = 0.002). Mortality at day 28 was associated with lower mesor (adjusted hazard ratio 0.50, 95% CI 0.28 to 0.90; p = 0.02), and higher amplitude (adjusted hazard ratio 5.48, 95% CI 1.66 to 18.12; p = 0.005) of temperature. Many factors, such as therapeutics, influence the body temperature during septic shock. Lower mesor and higher amplitude were associated with mortality and could be considered prognostic markers in ICU. In the age of artificial intelligence, the incorporation of such data in an automated scoring alert could compete with physicians to identify high-risk patients during septic shock.