Determinants of temperature abnormalities and influence on outcome of critical illness.

Objective: To determine the occurrence and determinants of temperature abnormalities among patients presenting (<24 hrs) to an intensive care unit and assess their effect on mortality outcome.

Design: Inception cohort.

Setting: French intensive care units participating in the Outcomerea group.

Patients: Adults (≥ 18 yrs) admitted to an intensive care unit between April 2000 and November 2010. Patients undergoing therapeutic hypothermia were excluded.

Interventions: None.

Measurements And Main Results: A total of 10,962 patients were included. The median age was 63 yrs (interquartile range, 49-76), 6639 (61%) admissions were in males, and the median admission Simplified Acute Physiology Score II and Sequential Organ Failure Assessment scores were 39 (interquartile range, 27-54) and 5 (interquartile range, 3-8), respectively. Patients were classified as medical in 8,237 (75%), nonscheduled surgical in 1,507 (14%), and scheduled surgical in 1,218 (11%). At presentation, 1,046 (10%) patients had mild hypothermia (35.0-35.9 °C), 541 (5%) had moderate hypothermia (32-35.9 °C), 72 (1%) had severe hypothermia (<32 °C), 2,264 (21%) patients had mild fever (38.3-39.4 °C), and 559 (5%) had high fever (>39.5 °C). Normothermia was present in 6,133 (55%) and mixed fever/hypothermia abnormalities occurred in 347 (3%) patients overall. Medical patients had the highest occurrence of any fever, whereas hypothermia was more common in surgical patients. The overall intensive care unit case-fatality was 1,944 of 10,962 (18%) and 828 of 6,133 (14%) for normothermia, 235 of 1,046 (22%) for mild hypothermia, 205 of 541 (38%) for moderate hypothermia, 43 of 72 (60%) for severe hypothermia, 412 of 2,264 (18%) for mild fever, 117 of 559 (21%) for high fever, and 104 of 347 (30%) for those with mixed temperature abnormalities. After controlling for confounding variables in logistic regression analyses, fever at presentation was not associated with any significantly increased risk for death. However, hypothermia was a significant independent predictor for death in medical patients.

Conclusions: Temperature abnormalities are common among patients presenting to the intensive care unit. Hypothermia is a major, potentially modifiable factor associated with increased risk for death.