Background and Objectives
To investigate sensitivity and specificity of infrared ear thermometry compared to rectal thermometry to detect fever in children.

Methods
Systematic review of studies comparing rectal and infrared ear temperatures in children.

Results
Sensitivity and specificity estimates were highly heterogeneous, and displayed an inverse relationship suggestive of a threshold effect, due in part to the different offsets used to obtain adjusted tympanic temperatures depending on the ear thermometer mode. To account for this threshold effect, results from each study were summarized as a diagnostic odds ratio (DOR). These varied extensively across studies, suggesting that heterogeneity between study estimates is not fully explained by the threshold effect. Pooled estimates of sensitivity and specificity from random effects models were 63.7% (95% CI 55.6%, 71.8%) and 95.2% (95% CI 93.5%, 96.9%), respectively.

Conclusion
Pooled estimates of measures of diagnostic accuracy from these studies suggest that infrared ear thermometry would fail to diagnose fever in three or four out of every 10 febrile children (with fever defined by a rectal temperature of 38°C or above). These findings support our previous concerns about the use of infrared ear thermometers in situations where a failure to detect fever has serious implications.