A comparison of axillary and tympanic membrane to rectal temperatures in children

Background Core body temperature measurement is not commonly done in pediatric populations because it is invasive and difficult to perform. Therefore, axillary and tympanic membrane temperature measurements are preferable, but their accuracy is still debatable. Objective To compare the accuracy of axillary and tympanic temperatures to rectal temperature in children with fever, and to measure the cut-off point for fever based on each temperature measurement method. Methods A diagnostic study was conducted among feverish children aged 6 months to 5 years who were consecutively selected from the Pediatric Outpatient Clinic, Pediatric Emergency Unit, and the inpatient ward in the Department of Child Health, Cipto Mangunkusumo Hospital (CMH), from December 2014 to January 2015. Subjects underwent three measurements within a two minute span, namely, the axillary, tympanic membrane, and rectal temperature measurements. The values obtained from the examination were analyzed with appropriate statistical tests. Results The cut-off for fever on axilla was 37.4oC and on tympanic membrane was 37.4oC, with sensitivity 96% (95%CI 0.88 to 0.98) and 93% (95%CI 0.84 to 0.97), respectively; specificity 50% (95%CI 0.47 to 0.84) and 50% (95%CI 0.31 to 0.69), respectively; positive predictive value/PPV 90% (95%CI 0.81 to 0.95) and 85% (95%CI 0.75 to 0.91), respectively; and negative predictive value/NPV 83% (95%CI 0.61 to 0.94) and 69% (95%CI 0.44 to 0.86), respectively. The optimal cut-off of tympanic membrane and axilla temperature was 37.8oC (AUC 0.903 and 0.903, respectively). Conclusion Axillary temperature measurement is as good as tympanic membrane temperature measurement and can be used in daily clinical practice or at home. By increasing the optimum fever cut-off point for axillary and tympanic membrane temperature to 37.8oC, we find sensitivity 81% and 88%, specificity 86% and 73%, PPV 95% and 91%, and NPV 95% and 91%, respectively.