ABSTRACT
Body temperature is often the sole determinant of whether or not the neutropenic cancer patient is admitted to the hospital for empiric antibiotic therapy. Recently developed infrared tympanic thermometers offer rapid readings, but their accuracy has not been established. We studied two infrared thermometers (FirstTemp and Thermoscan) and a thermistor (IVAC) in children with cancer. Mean infrared measurements did not differ significantly between right and left ear canals, and the mean IVAC temperature did not differ significantly from the left to the right axilla (P greater than 0.05, paired t test). IVAC predictive mode mean temperature was 0.2 degrees C lower than monitor mode mean temperature in the axilla (P less than 0.0001), but 0.1 degree C higher than monitor mode orally (P less than 0.0001). Aiming the infrared instrument at the tympanic membrane using an ear tug resulted in a 0.2 degree C higher mean temperature than casual placement in the ear canal (P less than 0.0001). After compensation for the mean difference in reference oral glass-mercury versus test instrument temperatures, the FirstTemp, Thermoscan, and oral and axillary predictive mode IVAC measurements yielded sensitivities for the detection of fever of 84%, 84%, 82%, and 86%; specificities of 100%, 99%, 100%, and 100%; positive predictive values of 100%, 93%, 100%, and 100%; and negative predictive values of 95%, 98%, 98%, and 98%, respectively. We conclude that each of these instruments detects fever with comparable reliability. Infrared instruments are especially attractive alternatives due to their time efficiency.
