RELIABILITY OF INFRARED TYMPANIC THERMOMETRY IN THE DETECTION OF RECTAL FEVER IN CHILDREN

Study objective: Recently published clinical guidelines for the manage ment of febrile children are based on studies that used rectal tempera ture data to stratify the risk of bacteremia and septic complications. Appropriate management decisions rely on accurate detection and categ orization of fever. Accordingly, this study compared the newer infrare d tympanic thermometry (ITT) to rectal thermometry in this regard. Des ign: Prospective observational study. Setting: Urban teaching hospital ED with annual census of 60,000. Participants: Consecutive children 6 months to 6 years old who had rectal temperatures measured. Intervent ions: Triage nurses recorded rectal temperatures and bilateral ITT tem peratures. Temperatures were correlated by Pearson correlation coeffic ients and compared using paired t tests with significance set at P<.01 . Children were categorized by degree of fever using rectal temperatur e (afebrile, less than 100.4-degrees-F; low fever, 100.4 to 102.9-degr ees-F; and high fever, more than 102.9-degrees-F), and the accuracy of ITT in detecting fever and high fever was determined. Results: Three hundred seventy patients were enrolled in the study. The mean age was 18.4+/-11.3 months; boys comprised 56% of patients. The mean temperatu res were rectal, 101.0+/-2.0-degrees-F; right tympanic membrane, 100.4 +/-1.9-degrees-F; and left tympanic membrane, 100.3+/-1.9-degrees-F. T he tympanic membrane temperatures were significantly lower than rectal readings (P<<.001 for both right and left versus rectal). Rectal temp eratures showed good correlation with both right and left tympanic mem brane temperatures (r=.83 and .85, respectively). ITT was 76% sensitiv e and 92% specific in detecting fever of 100.4-degrees-F or more (posi tive predictive value, 0.92; negative predictive value, 0.76). In the detection of high fever, ITT was only 57% sensitive but 98% specific ( positive predictive value, 0.90; negative predictive value, 0.90). Rec tal and TM temperatures differed by at least 0.5-degrees-F in 70% of t he patients, 1.0-degrees-F in 41%, 2.0-degrees-F in 12%, and 3.0-degre es-F in 3%.Conclusion: Despite the statistical correlation between ITT and rectal temperatures, the modalities may yield significantly diffe rent temperatures. The poor sensitivity of ITT in detecting fever and high fever may result in clinically important miscategorizations of in dividual patients. Current clinical management that is based on the pr esence and height of fever may be adversely affected if ITT is used.