Automatic milking systems (AMS) generate mastitis alert lists that report cows likely to have clinical mastitis (CM). Sensor systems involved in mastitis detection in AMSs include: deviation in electrical conductivity (EC), milk color, individual somatic cell count (SCC), milk temperature, milk yield (MY) and its deviation, milking interval, milk flow (MF), etc. The ISO 20966 describes a minimum standard of 80% sensitivity and 99% specificity for detection of abnormal milk. Khatun et al. (2018), developed and evaluated a multivariable quarter level CM prediction index in AMS. The study found that better CM prediction is possible by using multiple automatically recorded inline sensor data records than a single sensor data record. The best mastitis prediction was possible by incorporating 6 measurements: EC, EC per hour, MY, MY per hour, MF and incomplete milking. Incorporation of cow and quarter previous mastitis history improved the performance of the test procedures. Therefore, several indicators should be used to optimize clinical mastitis detection.
Generally, it is recommended to check all cows that appear on the udder health reports generated by the AMS software twice a day. Often udder health reports present cows that are likely to have mastitis. It is up to farmers to check these alerts and cows and to initiate appropriate treatment if mastitis is confirmed. However, it is very important to realize that not all cows with mastitis will end-up on these udder health reports, and at the same time, that these reports will list cows without mastitis. Not all warning messages mean trouble. The ideal procedure would be to visually check all four quarters of all alerts cows visually and with a California Mastitis Test (CMT) but this would be disruptive, time consuming and labour intensive.
Alternatively, the farmer can check the computer for milk production, number of visits to the AMS, activity, finding clots on the filter sock, etc. This will help decide if there is a problem and if the udder of a reported cow needs to be checked. The alerts on a CM alert list can be ranked-ordered, thereby providing the dairy farmer information about which cows have the highest priority for visual inspection. Checking alerts based on a single alert information variable (e.g., EC), would result in many false-positive cases and missing too many true-positive cases. Using a combination of alert information variables, is the best way to select cows that need further attention. Armed with this information, a farmer can decide which steps to take: monitor or treat a cow.
A decision tree was developed to create a structured approach to mastitis alerts management (See figure 1). When a mastitis alert based on multiple parameters appears, the cow should be checked and abnormalities in the milk should be assessed visually by forestripping the quarters to detect CM. If the cow has CM, the affected quarter must be evaluated for treatment (locally and/or systemically; combination with NSAIDs) or not according to the age of the cow, previous history, pathogen, severity of the case, days in milk, etc. If the milk of a cow with a new alert is visually normal and the alert remains on the list for the subsequent days, a CMT should be performed to detect subclinical mastitis. If this cow has a positive CMT, the next step will be to take a milk sample for bacteriological culturing or PCR. If the culture result is a contagious pathogen (e.g., Staphylococcus aureus), the decision tree suggests to treat subclinical mastitis in heifers and second lactation cows in the first 100 days of lactation (if not origin from the dry period). It is also important to identify acute or chronical cases and act accordingly. Chronic cows, usually with high SCC, are going to have constant alarms. Reducing mastitis treatment and culling would be the solution for these cases.
Take home messages:
- Using a combination of variables to generate alerts is the best way to select cows that need further (visual) investigation.
- Alerts for cows with high values of SCC and/or previous CM cases have high clinical or subclinical mastitis detection success rates.
- Based on a combination of parameters cows with mastitis alerts should be checked visually inspecting the milk and examining the udder to detect CM.
- If the milk of a cow with a new alert is visually normal and the alert remains on the list for the subsequent days, a CMT should be performed. If this cow has a positive CMT, the next step will be to collect a milk sample for bacteriological culturing or PCR.