Review ArticleIn a systematic review, infrared ear thermometry for fever diagnosis in children finds poor sensitivity
Introduction
We previously undertook a systematic review of studies where rectal and infrared ear temperatures were compared in children [1]. Despite a relatively small pooled mean temperature difference (0.29°C), the wide 95% limits of agreement (−0.74 to 1.32°C) suggested that infrared ear thermometry should not be used as an approximation for rectal thermometry. Having investigated agreement, a necessary first step in any method comparison study, an important clinical question is how many cases of true fever would be missed using ear thermometry. The aim of this study was to determine diagnostic accuracy by examining the sensitivities and specificities from the studies used in our previous work, and to investigate the effect of thermometer mode on these measures of accuracy.
Section snippets
Methods
Studies that compared temperatures taken at the ear (using infrared devices) and rectum (using either an electronic, mercury, or indwelling thermocouple device) in children aged between 0 and 18 were included in the original review, excluding those with hypothermia (rectal temperature <35.0°C) and preterm infants (born at <37 weeks' gestation) [1]. We used rectal temperature as a reference measure, as it is a well-established method of measuring temperature in children [2]. Of the 44 original
Results
Twenty-three of 44 eligible studies identified in the original review provided relevant data for inclusion in this review, giving a total of 4,098 children (69%). The majority of studies defined 38°C as the cutoff for fever for both rectal and tympanic temperatures. However in one study, a cutoff of 38.1°C was taken for rectal temperatures and 37.6°C for tympanic temperatures, while in a second, a cutoff of 37.9°C was used for both temperatures.
Sensitivities ranged from 0 to 100% [unweighted
Discussion
We have extended the analysis of our original systematic review [1] by examining sensitivities and specificities of infrared ear thermometers for detecting fever in children, as determined by rectal temperature. We have assumed that the results from the 23 studies that we were able to include in this analysis are representative of the 44 included in the original review.
The sensitivity and specificity estimates varied extensively in these studies. One reason for this heterogeneity was the
Contributors
Susanna Dodd carried out data extraction and meta-analysis, and drafted and revised the report. Rosalind Smyth conceived the idea for the study and Paula Williamson conceived the idea for this further analysis. Jean Craig obtained IPD and assisted with data extraction. Paula Williamson and Gill Lancaster gave advice on the approach to the analysis and Rosalind Smyth advised on the interpretation of the results. All authors commented on the final draft of the report.
Conflict of interest statement
None declared.
References (9)
- et al.
Infrared ear thermometry compared with rectal thermometry in children: a systematic review
Lancet
(2002) - et al.
The diagnostic odds ratio: a single indicator of test performance
J Clin Epidemiol
(2003) - et al.
Meta-analytic methods for diagnostic test accuracy
J Clin Epidemiol
(1995) - Clinical Effectiveness Team for Fever of Uncertain Source 2–36 Months. Evidence based clinical practice guideline of...
Cited by (84)
Medical Device Technologies: A Systems Based Overview Using Engineering Standards
2020, Medical Device Technologies: A Systems Based Overview Using Engineering StandardsCore temperature is not elevated at rest in people with relapsing-remitting multiple sclerosis
2019, Multiple Sclerosis and Related DisordersFever of Unknown Origin in Childhood
2017, Pediatric Clinics of North AmericaPathogenesis of Fever
2017, Infectious Diseases, 2-Volume SetApproach to evaluation of fever in ambulatory cancer patients receiving chemotherapy: A systematic review
2016, Cancer Treatment ReviewsCitation Excerpt :One study comparing the diagnostic accuracy of tympanic, axillary, and oral thermometers in detecting rectal fever in patients with febrile neutropenia showed tympanic thermometry to have the best performance, with mean sensitivity and specificity of 63.6% and 97.4%, respectively [19]. However, examination of systematic reviews of thermometry beyond our target population showed wide variation in temperature readings across thermometer types [21–29]. No guidelines were identified that addressed patients receiving chemotherapy with a fever prior to a diagnosis of febrile neutropenia.
Thermoregulation: Physiology and Perioperative Disturbances
2016, Smith's Anesthesia for Infants and Children, Ninth Edition