Technical note: Assessment of milk temperature measured by automatic milking systems as an indicator of body temperature and fever in dairy cows

The objective of this study was to evaluate whether milk temperature (MT) measured by automatic milking system (AMS) is a reliable indicator of body temperature of dairy cows and whether cows with fever could be detected. Data loggers (Minilog 8, Vemco Ltd., Halifax, NS, Canada) measuring body temperature were inserted for 7 ± 1 d into the vaginal cavity of 31 dairy cows and programmed to take 1 reading/min. Milk temperature was recorded at each milking event by the AMS, and values from the vaginal loggers were paired with the corresponding MT. The correlation (r) between vaginal temperature (VT) and MT was 0.52. Vaginal temperature was higher (39.1 ± 0.4°C) than MT (38.6 ± 0.7°C) with a mean difference of 0.5 ± 0.6°C. The ability of MT to identify cows with fever was assessed using 2 approaches. In the first approach, VT could indicate fever at any time of the day, whereas MT could display fever only during the milking events of a given day. Different definitions of fever based on thresholds of VT and duration exceeding these thresholds were constructed. Different thresholds of MT were tested to distinguish between cows with and without fever. The combination of 39.0°C as a threshold for MT and 39.5°C for at least 2 h/d as a threshold for VT resulted in the highest combination of sensitivity (0.65) and specificity (0.65). In the second approach, we evaluated whether MT could identify cows with fever at a given milking event. A threshold of MT >38.7°C delivered the best combination of sensitivity (0.77) and specificity (0.66) when fever was defined as VT ≥39.5°C. Therefore, MT measured by AMS can be indicative of fever in dairy cows to a limited extent.