Comparative evaluation of digital rectal and non-invasive infrared thermometers for measuring equine body temperature

Abstract

This study evaluated the accuracy of infrared digital thermometry (IDT) compared to rectal temperature (RT) in 476 clinically healthy Thoroughbred racehorses prior to turf competitions. Temperatures were measured in the neck and shoulder regions using IDT and compared with RT, considered the gold standard. Median and interquartile ranges were: neck 36.2 °C (36.0-36.4), shoulder 36.3 °C (36.2-36.5), and rectal 37.6 °C (37.4-37.9). Bland-Altman analyses revealed clinically relevant mean biases: neck vs. rectal=-1.59 °C (limits -2.85 to -0.33 °C), shoulder vs. rectal=-1.37 °C (limits -2.26 to -0.47 °C), and neck vs. shoulder=-0.22 °C (limits -1.20 to 0.76 °C). Intraclass correlation coefficients (ICC) between IDT and RT were low (ICC=0.00823, 95 % CI: -0.013 to 0.034; ICC=0.00446, 95 % CI: -0.009 to 0.02; ICC=0.371, 95 % CI: 0.23 to 0.487, for neck-rectal, shoulder-rectal, and neck-shoulder, respectively), indicating measurements are not interchangeable. A linear mixed model showed that wet conditions decreased surface temperature by 0.59 °C (p<0.001), while coat presence was not significant (p=0.32); ambient temperature increased surface temperature by 0.013 °C per 1 °C increase (p<0.001). Marginal R² was 0.734 and conditional R² 0.77, indicating that most variability is explained by fixed effects (region, environmental conditions) and that individual horse effect is minor. Based on these findings, IDT measurements at the neck and shoulder are neither equivalent nor interchangeable with RT for hyperthermia diagnosis in field conditions, although it may be useful as a complementary tool under controlled conditions.