Simple infrared thermometry in fever detection: consideration in mass fever screening

The need to rapidly screen patients during outbreaks has prompted Cutaneous Infrared Thermometry (CIT) use. Little is known of CIT performance in this context. What are the performance characteristics of simple CIT in detecting fever?

Prospective cohort, sequential convenience sample.

All patients presenting to the study Emergency Department for care.

CIT and oral temperature measurements.

Fever defined as oral temperature≥38°C. CIT is measured simultaneously with oral temperatures. Comparisons of temperatures are expressed as means and 95% confidence intervals. Means are compared using Student's t test. Limits of agreement are measured using Bland-Altman. Receiver operating characteristics are determined.

There are 548 cases comprising 224 males, 324 females, with mean age 26 years. The mean temperature difference is 12.95°C, (13.18-9.08°C) p≤0.0001. Bland-Altman demonstrates bias at 8.680 (-9.084 to -8.275) p≤0.0001 with upper and lower level bias values of 18.124 (18.819-17.435) and 0.768 (0.076-1.459), respectively. Based on Receiver Operator Characteristics analysis, detection of hyperpyrexia at a CIT of 35.3°C provided sensitivity of 0.236 (0.143-0.359), specificity 0.977 (0.959-0.989), positive predictive value 0.589 (0.325-0.810), negative predictive value 0.904 (0.891-0.919), and accuracy of 0.888 (0.861-0.913).

The use of a readily available CIT measurement device predicted hyperpyrexia about 59 percent of the time and the absence of hyperpyrexia about 90 percent of the time. This is consistent with previous reports of more complex infrared measurement devices. Although commonly used in mass fever screening, the current performance characteristics of CIT are limited and may add little to detection of target diseases in a mass screening context.