Hypothermia in Adult ICUs: Changing Incidence But Persistent Risk Factor for Mortality.

Abstract

This study examined whether hypothermia (< 36.0°C) incidence among critically ill patients varied over time, the determinants of change, and the associated risk for ICU mortality. Interrupted time series analysis among adults admitted to ICUs in Calgary, Canada over 8.5 years. Changes in the incidence of hypothermia within the first 24 hours of ICU admission were modelled using segmented regression. Among 15,291 first admissions to ICU, hypothermia incidence decreased from 29% to 21% during the study period. Implementation of a new temporal artery thermometer (TAT) was associated with the majority of the decrease in incidence (10%; 95% CI 7.1-13%; P < .0001). However, subgroup analysis revealed important differences between medical and surgical patients. Hypothermia incidence decreased among surgical patients before TAT implementation (0.4% per quarter, 95% CI 0.1-0.7%, P = .009), but not after, whereas in medical patients, the incidence increased after (1.0% per quarter, 95% CI 0.6-1.4%, P < .0001) but not before TAT implementation. Segmented logistic regression suggested that increases in the proportion of patients with non-traumatic neurologic admission diagnoses were associated with hypothermia incidence among medical patients, whereas there was no measurable clinical factor associated with the observed time trends among surgical patients. Hypothermia at ICU admission was independently associated with ICU mortality in medical and surgical patients throughout the entire study. The incidence of hypothermia at ICU admission was dependent on medical versus surgical status, and the method of non-invasive temperature measurement, but was persistently associated with ICU mortality.