Oral and Tympanic Membrane Temperatures Are Inaccurate to Identify Fever in Emergency Department Adults

INTRODUCTION: Identifying fever can influence management of the emergency department (ED) patient, including diagnostic testing, treatment, and disposition. We set out to determine how well oral and tympanic membrane (TM) temperatures compared with rectal measurements. METHODS: A convenience sample of consecutively adult ED patients had oral, TM, and rectal temperatures performed within several minutes of each other. Descriptive statistics, Bland–Altman agreement matrices with 95% confidence interval (CI), and measures of test performance, including sensitivity, specificity, predictive values, and interval likelihood ratios were performed. RESULTS: A total of 457 patients were enrolled with an average age of 64 years (standard deviation: 19 years). Mean temperatures were: oral (98.3°F), TM (99.6°F), and rectal (99.4°F). The mean difference in rectal and oral temperatures was 1.1°F, although there was considerable lack of agreement between oral and rectal temperatures, with the oral temperature as much as 2.91°F lower or 0.74°F higher than the rectal measurement (95% CI). Although the difference in mean temperature between right TM and rectal temperature was only 0.22°F, the right TM was lower than rectal by up to 1.61°F or greater by up to 2.05°F (95% CI). Test performance varied as the positive predictive value of the oral temperature was 97% and for tympanic temperature was 55% (relative to a rectal temperature of 100.4°F or higher). Comparative findings differed even at temperatures considered in the normal range; among patients with an oral temperature of 98.0 to 98.9, 38% (25/65) were found to have a rectal temperature of 100.4 or higher, while among patients with a TM of 98.0 to 98.9, only 7% (10/134) were found to have a rectal temperature of 100.4 or higher. CONCLUSION: The oral and tympanic temperature readings are not equivalent to rectal thermometry readings. Oral thermometry frequently underestimates the temperature relative to rectal readings, and TM values can either under- or overestimate the rectal temperature. The clinician needs to be aware of the varying relationship between oral, TM, and rectal temperatures when interpreting readings.