RESUMO
BACKGROUND: The accuracy of body temperature measurement is very important in children. The accuracy and reliability of the forehead skin thermometer (FST) and infrared tympanic thermometer (TT) are inconclusive. OBJECTIVES: To determine: 1) the mean difference, agreement, and accuracy of body temperature detected in children by FST and ITT, as compared with the gold standard rectal mercury-in-glass thermometer (RMT); and, 2) the cut-off level for FST and ITT in the detection of fever MATERIAL AND METHOD: Children ≤ 2 years old with the chief complaint of "fever" were recruited for this study. Body temperature (BT) was measured by 3 different methods, including FST 5 times; ITT 3 times; and, RMT, 1 time, for each patient. Each measurement method was performed by 3 well-trained staff members, all of whom were blinded to the recorded data of the other methods. The mean difference in body temperature, agreement by Bland-Altman method, and the verified performance of FST and ITT by diagnostic test were assessed. RESULTS: A total 312 children were recruited. Body temperatures measured by FST and ITT were lower than those measured by RMT, with a mean difference of 1.04°C (p < 0.001) and 1.03°C (p < 0.001), respectively. In subgroup analysis by levels of fever the mean differences between rectal temperature and both forehead and ear temperature were statistically significantly different (p < 0.001) for all levels of fever. Regarding the diagnostic test to verify performance, the data revealed the most practical cut-off point to be 37.0°C for both FST and ITT as compared to the gold standard RMT. CONCLUSION: Both FST and ITT were found to be accurate temperature screening methods for daily clinical use. However, the cut-off points to detect fever should be lowered to 37.0°C to be consistent with gold standard measurement.