Abstract

Objective:To determine accuracy of infrared thermometer for detection of fever as compared to mercury thermometer.
Study Design: Cross sectional study.
Place andDuration of Study: Department of Medicine, Combined Military Hospital Peshawar, fromMaytoJun 2020.
Methodology: All willing adult patients reporting to the fever desk were selected by consecutivesampling. Exclusion criteria included any dermatological condition affecting forehead and unwillingness. Forehead temperature was first checked twice using Kinlee FT3010 infrared thermometer. Axillary temperature was then recorded using a standard clinical mercury thermometer.
Results: There were 538 patients, including 251 (46.65%) males and 287 (53.35%) females, aged 46.76± 12.44 years. Median temperatures recorded with infrared and mercury thermometers were 97.00°F (interquartile range: 95.10-97.80°F)and 98.30°F (interquartile range: 98.00-98.90°F) respectively (p<0.001). Intra-class correlation was 0.143 (95% CI -0.052, 0.323). There was a weak to moderate correlation (R: 0.366; p<0.001) between temperatures recorded by the two techniques. ROC curveanalysis for temperatures recorded by infrared thermometerrevealed an area under curve of 0.725 at a threshold of 98.6°F and 0.746 at a threshold of 100.4°F defined by mercury thermometer. Infrared thermometer had sensitivity, specificity, positive predictive value and negative predictive value of 13.61% and 9.38%, 97.95% and 99.80%, 71.43% and 75.00%, and 75.10% and 95.57% for thresholds of 98.6°F and 100.4°F respectively.
Conclusion: Infraredthermometer underestimatestemperatures recorded by mercury thermometer. Limits of agreement are too broad, indicating inconsistency in measurements. A significantly lower threshold is required to improve the sensitivity of Infrared thermometerin picking up fever.