ABSTRACT
A cross-sectional study was conducted in an urban emergency department to determine if predictive variables existed that would identify a patient who would be afebrile by oral temperature measurement and febrile by rectal temperature measurement. This study included 366 patients. Five variables studied achieved statistical significance by univariant analysis: mouth breathing (P = .002), respiratory rate (P = .001), supplemental oxygen (P = .009), pulse (P = .0001), and supplemental oxygen via mask (P = .01). Placing these variables in a logistic regression model left two variables that significantly explained the variance of the model: pulse (odds-risk ratio, 1.032/increase in pulse of 1 from 0; 95% confidence interval, 1.020 to 1.039) and mouth breathing (odds-risk ratio, 2.113; 95% confidence interval, 1.41 to 3.43). There was poor linear correlation between oral and rectal temperatures (r = 0.2). If a patient has an unexplained tachycardia and/or is breathing by mouth and is afebrile orally, a rectal temperature measurement should be obtained to determine if fever exists. The results of this study suggest that good linear correlation does not exist between oral and rectal temperature measurements.