[Validity of the tympanic infrared thermometer in geriatric inpatients]

The purpose of this study was to assess the agreement between rectal temperature and infrared emission detection (IRED). The specificity and sensitivity, the positive predictive value and the negative predictive value were calculated. It was a prospective, observational and unblinded study. Thermoscan-pro-1 was used to collect pairs of data in 41 patients, all hospitalised on an acute geriatric ward during a 4-month period. 41% of the patients were febrile. The 'bias' between rectal and tympanic measurement (left ear) was -0.64 degree C (d) and the standard deviation 0.58 degree C (s). The '95% Confidence Interval' for the bias was -0.82 degree C and -0.46 degree C. The '95% limits of agreement' between the rectal measurement and the tympanic measurement (left ear) was -1.78 degrees C and 0.50 degree C. Regression analysis taught us that bias was increasing with increasing temperature. If we take into account one measurement with the IRED ear thermometry, the sensitivity was 41%, the specificity 83%. The positive predictive value was 64% en the negative predictive value 67%. We can conclude that the IRED ear thermometry is not a valuable alternative for the rectal, digital thermometry in the elderly, frail and febrile hospital inpatients.