Measurement from axillary site with digital thermometer has been accepted as the most accurate method. But this method is time consuming. Tympanic and forehead measurements are often used but don't always seem to be more appropriate. Another site, umbilical region, could be an alternative site. This study aims to compare the measurements with axillary digital thermometer and non-contact infrared thermometers at sites from umbilicus and forehead to determine whether umbilical site could be used accurately in children. For each method, 2,048 measurements in total were performed. Using axillary method as gold standard, with a cut-off temperature of 38oC, the sensitivities and specificities, positive and negative predictive values of umbilical and forehead temperatures and area under the ROC curve were determined in non obese children. There was a significant positive correlation between axillary and umbilical temperatures with a correlation coefficient of 0.78. The average difference between the mean of both axillary and umbilical temperatures was -0.47 ± 0.65°C. The Bland-Altman plot showed good accuracy with only 2.5 % of the readings falling outside the 95% level of confidence. Umbilical measurements showed sensitivity of 71.7% and specificity of 95.8%. The area under the ROC curve was 0.93. The easy application may lead noncontact measurements from umbilicus site to be the preferable method for health care providers, but agreement limits mentioned in this study should be considered.

Keywords: children, non-contact infrared thermometer, umbilical region