Ketosis is one of the most common metabolic diseases of dairy cows and occurs at the beginning of lactation. The incidence of clinical ketosis is low and is only 1.3% in Austria (Fürst-Waltl and Schwarzenbacher, 2021). Subclinical ketosis occurs much more frequently but is not routinely recorded. Egger-Danner et al. (2016) and Liebminger (2021) reported that 10 to 20% of cows in Austria are affected with subclinical ketosis. The range of ketosis incidence differs for each country, as for example in the Netherlands, much higher values were reported by Vanholder et al. (2015) in 23 dairy farms, 11.6% of cows had clinical ketosis and 47.2% had subclinical ketosis. Lean et al. (2023) conducted a retrospective meta-analysis based on individual cow data from Australia, Canada and the United States. They reported the incidence of clinical ketosis was 3.3%, whereas the incidence of subclinical ketosis was 26.8%.

Ketosis can be diagnosed by measuring ketone bodies present in blood, urine, or milk. Various handheld blood ketone meters are available to farmers to detect cows with subclinical ketosis (e.g. Sailer et al., 2018; Khol et al., 2019). However, due to practical limitations associated with individual blood sampling and testing, routine screening of all at-risk animals with blood tests is not feasible. Analysis of individual milk samples by mid-infrared (MIR) spectrometry remains the most promising method for detecting ketosis because it allows complete screening of the health status of the animal and the herd (Benedet et al., 2019). In recent years, many novel MIR-predicted traits related to metabolic health have been derived (de Roos et al., 2007; Grelet et al., 2016; Franceschini et al., 2022). A new ketosis risk indicator trait, KetoMIR, based on clinical ketosis diagnoses and MIR-predicted traits, was developed at the Regional State Association for Performance and Quality Inspection in Animal Breeding of Baden Württemberg as part of the OptiMIR project (Hamann et al., 2017; Drössler et al., 2018). The new ketosis risk indicator is intended as a herd management tool that allows farmers to monitor the metabolic situation of their herd during the first 120 days of lactation. In Baden-Württemberg, KetoMIR has been routinely used since 2015, and in Austria since 2017. The aim of the present study was to investigate the relationship of ketosis risk based on KetoMIR with milk yield, health and fertility at herd level in Austrian dairy farms. For this purpose, the phenotypic associations between these traits were analyzed.

In our study, the mean incidence of cows with a moderate or high ketosis risk based on KetoMIR was 14% at the herd level ( Table 1 ). Highly elevated incidences of >30% were found in 8.6% of farms ( Figure 1 ).

The Kruskal-Wallis test determined that there are significant differences in all investigated variables between farms at different risk of ketosis. The subsequently-applied Dwass, Steel, Critchlow-Fligner multiple comparison procedure showed that the farms with the lowest ketosis risk (≤10% of cows in the herd with a moderate or high ketosis risk) differ in all variables from the farms with the highest ketosis risk (>50% of cows in the herd with a moderate or high ketosis risk) ( Table 2 ). The increased ketosis risk based on KetoMIR was associated with lower average herd milk yield (-1,975 kg milk). Mean herd somatic cell count in first and higher lactations was increased by 60,500 and 134,400 cells/ml, respectively. The interval from calving to first service was prolonged by +36.5 days, as was the calving interval with +58.2 days.

The reported incidence of ketosis based on KetoMIR in Austria is at the lower end of the range published in the literature for dairy cattle. A study in 10 European countries indicated that the overall prevalence of ketosis, defined as serum β-hydroxybutyrate concentration ≥ 1,200 to 1,400 µmol/L in cows within 2 to 15 days in milk was 21.8%, ranging from 11.2 to 36.6% (Suthar et al., 2013). Berge and Vertenten (2014) reported that 39% of the cows in western European dairy herds were classified as having ketosis, defined as a milk based keto test result ≥ 100 μmol/L. The herd average of ketosis was 43% in Germany, 53% in France, 31% in Italy, 46% in the Netherlands, and 31% in the United Kingdom. Of the 131 investigated farms, 112 (85%) had 25% or more of their fresh cows resulting as positive for ketosis. Brunner et al. (2019) reported prevalence of subclinical ketosis (β-hydroxybutyrate concentration ≥1.2 mmol/L) in dairy cows in Central and South America, Africa, Asia, Australia, New Zealand, and Eastern Europe. Across all investigated countries, the prevalence of subclinical ketosis was 24.1%, ranging from 8.3% up to 40.1%.

In our study, a higher incidence of cows at moderate or high risk of ketosis on the farm was associated with lower milk yield, higher somatic cell count, longer time from calving to first service and prolonged calving interval, which is in agreement with previous studies. A negative association between the herd prevalence of ketosis and milk yield was also recently reported by Couto Serrenho et al. (2023). As mentioned by Daros et al. (2022) it is likely that herds that produce more milk also have better general management and thus may have implemented more effective ketosis prevention protocols.

In agreement with our results, van Straten et al. (2009) reported that cows with ketosis were 44% more likely to have an event with a somatic cell count >250,000 cells/ml than cows without ketosis. In addition, it was found that cows with extreme relative body weight loss in early lactation were more likely to have an increased somatic cell count throughout lactation.

An undesirable relationship of ketosis and fertility was reported by numerous authors (Reist et al., 2000; Walsh et al., 2007; Rutherford et al., 2016). Compton et al. (2015) found that concentration of β-hydroxybutyrate in blood ≥ 1.2 mmol/L at any stage within 5 weeks post-calving was associated with decreased 6-week pregnancy rate (78 vs. 85%). Recently, Alemu et al. (2023) reported a negative association between elevated milk β-hydroxybutyrate within 42 d and reproductive performance after the voluntary waiting period.

Milk is the most promising matrix to identify cows with ketosis, because it is easy to sample and allows a complete screening of the herd. Based on routinely available clinical ketosis diagnoses and MIR-predicted traits, a new ketosis risk indicator (KetoMIR) was developed to provide a new screening tool of the herd at milk recording. The ketosis risk was modeled for the first 120 days of lactation. Overall, an increased incidence of cows with a moderate or high ketosis risk on the farm indicates a problem and was associated with lower milk production, increased somatic cell count, longer time from calving to first service and prolonged calving interval. The use of milk MIR-predicted traits for herd-level ketosis screening could be an important factor in preventing ketosis to limit as much as possible the economic losses caused by this disease.

The data used in the current study are not publicly available due to privacy restrictions of the data provider and owner (LKV Austria). Requests to access these datasets should be directed to F-JA Franz-Josef.Auer@lkv-austria.at.

This study was an observational study of farms in their normal routines, and no interventions were conducted on any animals or humans for the purpose of this article. It thus did not require Institutional Animal Care and Use Committee or Institutional Review Board approval.

AK: Conceptualization, Formal analysis, Methodology, Visualization, Writing – original draft, Writing – review & editing. LD: Writing – original draft, Writing – review & editing. AW: Writing – original draft, Writing – review & editing. MM: Writing – original draft, Writing – review & editing, Data curation. F-JA: Writing – original draft, Writing – review & editing. CE-D: Conceptualization, Funding acquisition, Project administration, Supervision, Writing – original draft, Writing – review & editing.