We recently conducted 10-minute axillary temperature measurements on 699 healthy individuals, ranging in age from 5 to 83 years (mean: 36.5±15.3 years).
Axillary temperature readings are sometimes inaccurate because the thermometer is not inserted in the correct place (i.e. the point of the highest temperature) or due to incomplete closure of the axilla during measurement. As the result of analysis of temperature rises during the 10-minute axillary temperature measurement, we found that the results are not always accurate because of incorrect conditions of measurement such as when the thermometer reading does not reflect the surrounding temperature.
In this study, the following temperature readings were regarded to be inaccurate when: (1) the temperature temporarily fell during the rising phase; and (2) the rise in temperature was accelerated during measurement and the temperature had not become stable after 10 minutes. As a result of analyzing 10-minute axillary temperature measurements, we believe the temperature rise during the measurement must be considered to obtain more accurate readings.
When analyzing the readings of short-term temperature measurements using a predictive algorithm, awareness required of possible errors which may be caused by the measurement method employed. Error factors other than the algorithm used for prediction become large when the thermometer is temporarily withdrawn from the axillary pit or its direction is changed during measurement to check an interim reading. After error factors associated with the measurement method have been eliminated, high accuracy is obtained and the difference between predicted reading and 10-minute becomes 0.01±0.13°C. A difference smaller than±0.2°C has been achieved in 98.0% of all measurements. Such a difference causes no problem from a clinical point of view.