Utility of infrared thermography for screening febrile subjects

To assess the utility of remote-sensing infrared thermography as a screening tool for fever.

Cross-sectional study comparing body temperatures measured by remote-sensing infrared thermography (maximum for frontal, forehead, or lateral views) with core temperatures measured by aural or oral methods.

Accident and Emergency Department, Queen Mary Hospital, Hong Kong.

A total of 1517 patients (747 men, 770 women) with or without fever; 34 of whom entered a substudy to measure the effects of distance on recorded temperature.

The proportions of subjects with fever (core temperature of 38°C or above) detected by remote-sensing infrared thermography compared with the proportion detected by conventional thermometry.

The correlations between infrared thermography temperatures and core temperature were only moderate (r=0.36-0.44), albeit statistically significant. The temperature recorded by infrared thermography was inversely proportional to the distance from the camera. There were 113 (7.4%) subjects with a core temperature of 38°C or above. The areas under the receiver operating characteristic curves for the three infrared thermography measurements were around 0.8. However, the maximum sensitivity achieved at a low cut-off temperature of 35°C was only 0.87 (for frontal and lateral infrared thermography views), resulting in 13% of febrile subjects being missed. The maximum forehead temperature in general had the poorest performance among the three infrared thermography views.

Forehead infrared thermography readings from a distance should be abandoned for fever screening. Although maximum lateral or frontal infrared thermography temperatures have reasonable correlations with core temperatures and areas under the receiver operating characteristic curves, the sensitivity-specificity combination might still not be high enough for screening febrile conditions, especially at border crossings with huge numbers of passengers.