Accuracy of an infrared tympanic thermometer.

BACKGROUND: The use of infrared thermometry to measure temperatures in hospitalized patients is increasing. Although infrared thermometers have been proven to be accurate when they are used by well-trained personnel, no previous studies have examined their accuracy during routine hospital use.
OBJECTIVE: To determine the accuracy and observer variability of temperatures measured with an infrared tympanic thermometer (TT).
DESIGN: Prospective, observational study.
SETTING: ICUs of a 300-bed teaching community hospital. PATIENTS: Fifty-one critically ill patients. MEASUREMENTS: The mean of three tympanic temperatures measured with the infrared TT (tempTTs) was compared to temperatures simultaneously measured with the thermistor of right heart catheters and rectal mercury thermometers for the following three groups of observers who had been certified in the use of the infrared TT: a single critical care nurse (CCN)/educator (Ed); CCNs, and floor nurses (FNs)/clinical care practitioners (CCPs).
RESULTS: Two rounds of measurements were given to 51 patients by 153 observers. Temperatures of the pulmonary artery (PA) measured with the thermistor of right heart catheters (tempPAs) ranged from 96.5 to 102.6 degrees F, with a mean (-/+ SD) of 99.3+/-1.1 degrees F. The intraobserver variabilities (correlation coefficients) of the tempTTs ranged from 0.90 for those measured by FNs/CCPs, to 0.92 for those measured by CCNs, to 0.98 for those measured by the CCN/Ed. Accuracy, arbitrarily defined as within a deviation of -/+0.5 degrees F of the tempPA, was 100% for the rectal mercury thermometer and 98.0% for the infrared TT when used by the CCN/Ed. The accuracy of the infrared TT was 80% when measured by CCNs and 61% when measured by FNs/CCPs. Differences between tempPAs and tempTTs measured by the CCN/Ed ranged from 0 to 0.7 degrees F, with a mean of 0.2 degrees F. Similarly, differences between tempPAs and tempTTs measured by CCNs ranged from 0 to 2.4 degrees F, with a mean difference of 0.3 degrees F. However, differences between tempPAs and tempTTs measured by FNs/CCPs ranged from 0 to 3.0 degrees F, with a mean of 0.6 degrees F (greater differences than those obtained by the CCNs; p < 0.01). The accuracy of rectal mercury thermometry was 100%. If a temperature > or = 101.0 degrees F had been considered as the threshold at which a fever is present, and if the mean of three measurements had been used to designate temperature, workups that were either inappropriately performed or omitted would have resulted from 2% of tempTTs measured by the CCN/Ed, 1% of those measured by CCNs, and 4% of those measured by FNs/CCPs.
CONCLUSION: When used properly, both tympanic and rectal thermometry are very accurate. However, the infrared TT produced measurements that were both less accurate and less reproducible when used by nurses who routinely used it in clinical practice.