The accuracy of oral predictive and infrared emission detection tympanic thermometers in an emergency department setting

Abstract

Objective: To assess the accuracy of an oral predictive thermometer and an infrared emission detection (IRED) tympanic thermometer in detecting fever in an adult emergency department (ED) population, using an oral glass mercury thermometer as the criterion standard. Methods: This was a single-center, nonrandomized trial performed in the ED of a metropolitan tertiary referral hospital with a convenience sample of 500 subjects. The temperature of each subject was taken by an oral predictive thermometer, an IRED tympanic thermometer set to 'oral' equivalent, and an oral glass mercury thermometer (used as the criterion standard). A fever was defined as a temperature of 37.8°C or higher. The subject's age, sex, triage category, and diagnostic group were also recorded. Sensitivity, specificity, positive and negative likelihood ratios, positive and negative predictive values, and corresponding 95% confidence intervals were calculated. Logistic regression was used to identify predictors of fever. Results: The sensitivities and specificities for detection of fever of the predictive and the IRED tympanic thermometers were similar (sensitivity 85.7%/88.1% and specificity 98.7%/95.8%, respectively). The predictive thermometer had a better positive predictive value (85.7%) compared with the IRED tympanic thermometer (66.1%). The positive and negative likelihood ratios for the predictive oral thermometer were 65 and 0.14, respectively, and for the IRED tympanic thermometer 21 and 0.12, respectively, indicating that the predictive thermometer will 'miss' 1 in about 7 fevers and the IRED tympanic thermometer will 'miss' 1 in about 8 fevers. Conclusions: Although quick and convenient, oral predictive and IRED tympanic thermometers give readings that cannot always be relied on in the detection of fever. If we are to continue using electronic thermometers in the ED setting, we need to recognize their limitations and maintain the importance of our clinical judgment.