Cross validation of cut–off limit estimations for milk indicators of subclinical ketosis

HANUS OTO, ROUBAL PETR, SAMKOVA EVA, FALTA DANIEL, CHLADEK GUSTAV, VYLETĚLOVA KLIMESOVA MARCELA: Cross validation of cut–off limit estimations for milk indicators of subclinical ketosis. Acta Universitatis Agriculturae et Silviculturae Mendelianae Brunensis, 2013, LXI, No. 6, pp. 1661–1667 Real time analyses of main milk components are attended in milking parlours today. Breeders can know milk composition regularly. Energy (ketosis) milk quotients such as fat/crude protein (F/CP) and fat/lactose (F/L) could be essential for ketosis prevention and losses which can be linked with it. Aim was to do a reciprocal validation of reliability of milk indicators for subclinical ketosis (SK) identifi cation. There were: dairy cows on 1st and other lactations free of clinical and superior subclinical mastitis; 3 herds of Czech Fleckvieh (CF) breed, 3 herds of Holstein (H) breed and 1 herd with CF and H; 329 individual milk samples from summer and winter season and 1st third of lactation. Average milk yield varied from 5,500 to 10,000 kg per lactation. Thresholds of F/CP and F/L were estimated (P 10 mg.l−1). These were: for fi rst lactation 1.27 (CF) and 1.32 (H), for other lactations 1.52 (CF) and 1.42 (H) in F/CP; for fi rst lactation 0.84 (CF) and 0.84 (H), for other lactations 0.87 (CF) and 0.85 (H) in F/L. They were used for cross validation (P 10 mg.l−1; according to: Emery et al., 1964; Vojtisek, 1986; Heuer et al., 2001; Hanus et al., 2001, 2011 a; de Roos et al., 2007) in our relevant previous papers (Hanus et al., 2011 b, 2013). The original AC values were log transformed (de Roos et al., 2007) as the data were not normally distributed. In this case, the arithmetic mean is not suitable parameter for representation of these data sets in all cases. Standard statistical Dairy herd during time: tolerable Healthy dairy cows Successive expansion of production disorders, their subclinical signs and diffusion in the herd Clinical signs of production disorders in animals Prevention field Cure field high! 0 Economical losses: 1: Pattern of prevention setting at reduction of milk yield losses of dairy cows cused by production disorders – especially by metabolic dysbalancies and milk secretion disorders Cross validation of cut–off limit estimations for milk indicators of subclinical ketosis 1663 methods were used to evaluate the results (means, standard deviations and coeffi cient of variation). Linear and non linear regression were used for the correlations using relevant regression equations. In subclinical ketosis threshold validation of F/CP and F/L the procedure of reciprocal estimation (cross calculation) according to linear regression results was used. Data were processed using Microso Excel programme. RESULTS AND DISCUSSION Milk acetone threshold and F/CP and F/L threshold estimation In Tab. I, there are shown main statistical characterictics of evaluated milk indicators according to breed and number of lactation. These have quite typical values for breeds and conditions in the country. SK cut–off values for milk quotients F/L and F/CP were calculated in relation to milk acetone (> 10 mg.l−1) as mentioned already. It was carried out using regression equations (P < 0.05) such as these are shown in Fig. 2 and Tab. II. As maximum through regression analysis, there is possible to I: Main statistical characteristics of evaluated milk indicators in fi rst third of lactation Data set n MY x ± sd AC x ± sd log AC x ± sd xg F x ± sd CP x ± sd L x ± sd F/CP x ± sd F/L x ± sd All 329 28.65 ± 7.86 3.31 ± 5.38 0.2894 ± 0.460 3.86 ± 0.97 3.20 ± 0.33 5.01 ± 0.20 1.21 ± 0.308 0.77 ± 0.200