Systematic review
The diagnostic accuracy of infrared tympanic, oral, axillary and temporal thermometry, compared with rectal readings when identifying fever in adult hospitalized patients
Systematic review
|Updated
This report summarizes the documentation of diagnostic accuracy of infrared tympanic thermometry compared to rectal thermometry to identify fever among adult patients in hospital or in nursing homes.
Download
Key message
The use of infrared tympanic thermometry has become a common method of measuring body temperature in Norwegian hospitals. This report summarizes the documentation of diagnostic accuracy of infrared tympanic thermometry compared to rectal thermometry to identify fever among adult patients in hospital or in nursing homes. In addition we have searched for studies that compare oral-, axillary and temporal thermometry with rectal thermometry.
The review shows that the diagnostic accuracy of tympanic thermometry compared to rectal thermometry is sparsely documented. We identified eleven small cross-sectional studies (N=1426). Most studies evaluated tympanic thermometry, some evaluated oral- or axillary thermometry. No studies evaluated temporal thermometry.
Correct and observer-independent use of infrared tympanic thermometry can be challenging in a clinical setting. Comparing temperature measurements of different body sites might also be problematic, because the measurements at different sites are all estimates for what we wish to know, the core temperature. Although rectal measurements are considered as reference standard in this review, we acknowledge that this is imperfect in many ways.
The studies showed that in general, infrared tympanic thermometry did not identify an acceptable part of patients with fever detected by rectal thermometry (low sensitivity). Infrared tympanic thermometry resulted in few false positive readings (high specificity). Since the studies included few patients with fever measured rectally and had different cut offs for fever, the sensitivity values are uncertain.
Given the widespread use of infrared tympanic thermometer, further documentation of diagnostic accuracy and repeatability of newer models used in a clinical setting is needed.
Summary
Background and mandate
The use of infrared tympanic thermometry has become a common method of measuring body temperature in Norwegian hospitals. The Norwegian Knowledge Centre for the Health Services has been asked to summarize the documentation of the diagnostic accuracy of the use of infrared tympanic, oral, axillary and temporal thermometry, compared to rectal readings.
Methods
We performed systematic literature searches in several health related databases (per 1.10.2008). We included clinical prospective cross-sectional studies with rectal thermometry (mercury or digital) as reference test. Only studies carried out in emergency wards, in general hospital wards or in nursing homes were included.
Results
We identified eleven small cross-sectional studies that evaluated infrared tympanic, oral and/or axillary thermometry, compared to rectal thermometry (N=1426). Eight of these studies evaluated infrared tympanic thermometry (N=1115), five evaluated oral thermometry (N=601) and two evaluated axillary thermometry (N=273). The definition of rectal fever varied among studies, with cut off points from 37,5 °C to 38,5 °C. The overall quality of the studies varied.
Six studies reported the sensitivity and specificity of infrared tympanic thermometry compared with rectal thermometry. The sensitivity varied in the six studies, from 14 % (95% CI 3 to 35) to 64 % (95% CI 46 to 79) (6 studies, N=813; 136 patients with rectal fever). The specificity varied from 72 % (95% CI 59 to 82) to 100 %. The mean temperature difference between infrared tympanic readings and rectal readings varied between 0,07 °C (SD ±0,54) and 0,54 °C (SD ±0,41) (6 studies, N=799).
The sensitivity of oral thermometry compared with rectal thermometry varied from 25 (95% CI 1 to 50) to 81 % (95% CI 66 to 96) (3 studies, N =315; 67 patients with rectal fever). The specificity varied from 70 % (95 % CI 62 to 78) to 100 %. The mean difference in temperature between rectal thermometry and oral thermometry was 0,53 °C (SD±0,53) and 0,75 °C (SD ± 0,74) (2 studies, N=291).
The sensitivity of axillary thermometry compared with rectal thermometry was reported only in one study (N=73). The sensitivity was 33 % (95 % CI 7 to 60) and specificity was 100 % (95 % CI 100 to 100). Another study (N= 200) reported a mean difference in temperature of 0,62 °C (SD 0,49) between rectal and axillary thermometry.
Conclusion
Correct and observer-independent use of infrared tympanic thermometry can be challenging in a clinical setting. Comparing temperature measurements of different body sites might also be problematic, because the measurements at different sites are all estimates for what we wish to know, the core temperature. Although rectal measurements are considered as reference standard in this review, we acknowledge that this is imperfect in many ways.
This review shows that few studies have assessed the accordance between infrared tympanic and rectal thermometry in detecting and excluding fever. We found eight small studies that compared different types of infrared tympanic thermometers to rectal measurement. These studies generally showed that infrared tympanic thermometry had low sensitivity but high specificity in detecting and excluding rectal fever compared with rectal measurements. Since these results were based on few patients with elevated temperature, the sensitivity values are uncertain, as expressed by wide confidence intervals. Different cut off values for defining fever in these studies also contributes to uncertainty around sensitivity and specificity.
There was no documentation on the diagnostic accuracy of temporal thermometry, and very few studies that compared oral and axillary thermometry with rectal thermometry.
Given the widespread use of infrared tympanic thermometer, further documentation of diagnostic accuracy and repeatability of newer models used in a clinical setting is needed.