ObjectivesTo investigate the clinical accuracy of infrared ear thermometer derived and equilibrated rectal temperatures in estimating core body temperature. The clinical bias (i.e., mean difference between body sites), and variability (SD of the differences) of simultaneous temperatures were compared with pulmonary artery temperatures. Clinical repeatability (pooled SD of triplicate reading differences) was also examined for three ear infrared thermometers.DesignProspective clinical study.SettingA multidisciplinary, adult intensive care unit.PatientsTwenty patients with an existing pulmonary artery catheter were studied in a multidisciplinary, adult intensive care unit.InterventionsA single operator using optimum ear infrared technique and masked to ear and rectal temperatures recorded triplicate measurements with each of three infrared ear thermometers, each over a 4-min period with each infrared thermometer, while an assistant recorded temperatures. Infrared and rectal temperatures were compared with a simultaneous pulmonary artery temperature.Measurements and Main ResultsInfrared ear thermometers and rectal thermometers were calibrated daily, and pulmonary artery catheters were calibrated on removal from the patient. Patients were grouped into afebrile and febrile groups, based on initial pulmonary artery temperature. Bias and variability were compared between thermometers using analysis of variance.Clinical bias, but not variability, was significantly different between three ear infrared thermometers (0.16 +/- 0.46 degrees C, 0.07 +/- 0.38 degrees C, and -0.22 +/- 0.47 degrees C). The repeatability was not different between ear infrared thermometers (range 0.13 degrees C to 0.14 degrees C). Rectal temperature had a significantly greater bias (average 0.3 degrees C), but less variability (average 0.2 degrees C). Bias was increased, and variability decreased for both rectal and infrared ear temperatures when pulmonary artery temperature was increased.ConclusionsThe three infrared ear thermometers studied provided a closer estimate of core body temperature than equilibrated rectal temperature. Clinical bias was greatest in febrile vs. afebrile intensive care unit patients.(Crit Care Med 1996; 24:1501-1506)