Clinical Investigation and Statistical Analysis Improve the Metrological Reliability of Forehead Infrared Radiation Thermometers

Abstract Forehead Infrared Radiation Thermometers (FIRTs) are widely employed in healthcare settings and access control due to their practicality. However, concerns persist regarding their metrological reliability, particularly in clinical applications. These devices operate in two modes: object and adjusted (body temperature). However, current evaluation methodologies lack standardised procedures and statistical criteria for assessing clinical performance. This study aimed to develop a clinical investigation protocol and establish robust statistical criteria for evaluating the metrological performance of FIRTs in adjusted mode. Eighteen FIRTs from ten different models were assessed through clinical measurements conducted on 106 emergency room patients. Simultaneously, a digital contact thermometer was used to obtain reference body temperatures. The analysis included four statistical approaches: clinical tendency analysis (mean error and standard deviation), outcome agreement tests (paired t-test and Wilcoxon signed-rank test), Bland–Altman analysis, and effect size assessment. Only six devices (33%) met the performance criteria across all volunteers. Among febrile patients (≥ 37.5 °C), seven thermometers performed satisfactorily, while only one device met the requirements for non-febrile patients (< 37.5 °C). The proposed clinical methodology proved effective for performance assessment, and the combination of statistical tools facilitated the establishment of evaluation criteria. Among these, the effect size analysis provided the most comprehensive insights. The significant disparity in results between febrile and non-febrile groups suggests a potential influence of correction algorithms embedded in the devices.