Abstract
BACKGROUND: We sought to evaluate agreement between a new and widely implemented method of temperature measurement in critical care, temporal artery thermometry and an established method of core temperature measurement, bladder thermometry as performed in clinical practice. METHODS: Temperatures were simultaneously recorded hourly (n = 736 observations) using both devices as part of routine clinical monitoring in 14 critically ill adult patients with temperatures ranging >/=1 degrees C prior to consent. RESULTS: The mean difference between temporal artery and bladder temperatures measured was -0.44 degrees C (95% confidence interval, -0.47 degrees C to -0.41 degrees C), with temporal artery readings lower than bladder temperatures. Agreement between the two devices was greatest for normothermia (36.0 degrees C to < 38.3 degrees C) (mean difference -0.35 degrees C [95% confidence interval, -0.37 degrees C to -0.33 degrees C]). The temporal artery thermometer recorded higher temperatures during hypothermia (< 36 degrees C) (mean difference 0.66 degrees C [95% confidence interval, 0.53 degrees C to 0.79 degrees C]) and lower temperatures during hyperthermia (>/=38.3 degrees C) (mean difference -0.90 degrees C [95% confidence interval, -0.99 degrees C to -0.81 degrees C]). The sensitivity for detecting fever (core temperature >/=38.3 degrees C) using the temporal artery thermometer was 0.26 (95% confidence interval, 0.20 to 0.33), and the specificity was 0.99 (95% confidence interval, 0.98 to 0.99). The positive likelihood ratio for fever was 24.6 (95% confidence interval, 10.7 to 56.8); the negative likelihood ratio was 0.75 (95% confidence interval, 0.68 to 0.82). CONCLUSIONS: Temporal artery thermometry produces somewhat surprising disagreement with an established method of core temperature measurement and should not to be used in situations where body temperature needs to be measured with accuracy.