Diseases, reproductive performance, and changes in milk production associated with subclinical ketosis in dairy cows: a meta-analysis and review.

Many studies have shown that subclinical ketosis (SCK) is associated with an increased risk of developing various diseases, reproductive disorders, and changes in milk production. The present work aims at producing an overview between this disorder and theses outcomes. A meta-analysis of the literature, including 131 different models from 23 papers, or a review when the literature was scarce was conducted. For each outcome, the odds ratio (OR), relative risk, or hazard ratio was presented for various moderators to reduce heterogeneity among the studies. The raw change in milk production associated with SCK was estimated and adjusted, taking into consideration the outcomes known to interact with milk production during the peripartum period. The results showed that 2 main categories of moderators had a significant effect on the adjusted risk. First, the adjustment made by defining SCK as (1) β-hydroxybutyrate concentration >1.4 mM, (2) nonesterified fatty acid (NEFA) concentration >0.4 mM prepartum, or (3) NEFA concentration >1.0 mM postpartum corrected the underestimated risk (despite low significance). This is because several trials reported the relevant risks using lower thresholds for the β-hydroxybutyrate or NEFA values. Using a low threshold leads to lower risk of disease compared with using a high threshold. Second, the correction produced using the polyfactorial terms corrected the overestimation of risk because many trials reported only univariable models. The relative risk or OR (95% confidence interval) related to abomasal displacement, clinical ketosis, early culling and death, metritis, placental retention, clinical mastitis, lameness, and a doubling of the SCC in cases with SCK were 3.33 (2.60-4.25), 5.38 (3.27-8.83), 1.92 (1.60-2.30), 1.75 (1.54-2.01), 1.52 (1.20-1.93), 1.61 (1.24-2.09), 2.01(1.64-2.44), and 1.42 (1.26-1.60), respectively. The precision level of the estimate depended on the outcome. The direct mean ± standard deviation of the 305-d milk losses associated with SCK were 251 ± 73 kg after adjusting for abomasal displacement, clinical ketosis, metritis, and placental retention. The OR (95% confidence interval) for first service calving risk in cases of SCK was 0.67 (0.53-0.83). The calving-to-first-service interval was 8 d longer and the calving-to-conception interval was 16 to 22 d longer in cows with SCK. The relationships among the different reproductive indicators were quantified in only 1 or 2 trials. The present work highlights the need to conduct further studies on the associations between SCK and the risks of diseases, changes in milk production, and reproductive parameters.