Effect of two dosages of prepartum cholecalciferol injection on blood minerals, vitamin D metabolites, and milk production in multiparous dairy cows. A randomized clinical trial.

The objective of the present study was to evaluate the effect of 2 dosages of prepartum cholecalciferol injection on blood minerals, vitamin D metabolites, and milk production. Cows entering their 2nd or greater lactation (n = 158) were randomly assigned to a control group ( CON ) or one of 2 treatment groups receiving either 6 × 10 6 IU ( 6VitD ) or 12 × 10 6 IU ( 12VitD ) cholecalciferol intramuscularly on d 275 ± 1.2 of gestation. Concentrations of serum total Ca ( tCa ), phosphate, and Mg were determined on 1, 2, 3, 5, 7, and 10 DIM. For a subsample of 30 cows entering the 3rd lactation (n = 10 per group), these samples were analyzed for cholecalciferol, 25-hydroxycholecal-ciferol ( 25-OHD 3 ), and 24,25-dihydroxycholecalciferol ( 24,25-(OH) 2 D 3 ). In these cows, we also determined 1,25-dihydroxycholecalciferol ( 1,25-(OH) 2 D 3 ) , the biologically most active metabolite, on 1, 2, 3, and 5 DIM. Repeated measures ANOVA was performed to evaluate the effect of different dosages of cholecalciferol on blood minerals, vitamin D metabolites, and milk yield over the first 5 test days after calving. Binary outcomes such as retained placenta ( RP) and metri-tis were analyzed using a Chi 2 test. While the 12VitD treatment increased tCa concentrations on 1, 2, and 3 DIM compared with CON, administration of 6VitD increased tCa concentrations only on 1 DIM. Compared with CON cows and 6VitD cows, 12VitD cows had greater serum phosphate concentration during the first 10 DIM. Furthermore, 6VitD cows had a greater serum phosphate concentrations compared with CON cows. On the contrary, 12VitD cows had lower serum Mg concentrations during the first 10 DIM compared with


INTRODUCTION
The intramuscular administration of cholecalciferol 5 to 7 DIM before calving is a common strategy to prevent clinical hypocalcemia in dairy cows used in European and Asian countries (Venjakob et al., 2017;Yamagishi et al., 2000).The hydroxylation of cholecalciferol to 25-hydroxycholecalciferol (25-OHD 3 ) in the Effect of two dosages of prepartum cholecalciferol injection on blood minerals, vitamin D metabolites, and milk production in multiparous dairy cows.A randomized clinical trial.
liver is not regulated tightly (Jones, 2008).Therefore, a high dose of cholecalciferol results in increased serum concentrations of 25-OHD 3 and subsequently increased Ca (tCa) concentrations, but informal reports on the efficiency differ (Hodnik et al., 2020;Venjakob et al., 2022).However, randomized controlled studies evaluating the efficacy are scarce.
In an early study on 182 multiparous cows entering ≥3rd lactation, 10 × 10 6 IU cholecalciferol injected intramuscularly approximately 7 d before expected parturition, reduced the incidence of milk fever, but only in cows that had developed clinical hypocalcemia in previous lactation (Julien et al., 1977).Sadri et al. (2021) administered 8 × 10 6 IU cholecalciferol intramuscularly to 12 Holstein Friesian cows 2 to 8 d prepartum and reported decreased serum concentrations of ionized Ca, parathyroid hormone (PTH), and bone markers around calving in comparison to the placebo treatment.They concluded that a single cholecalciferol injection did not improve Ca homeostasis around calving (Sadri et al., 2021).
Our group previously evaluated the effect of a greater dose (12 × 10 6 IU cholecalciferol as an intramuscular injection, 5 d before expected calving) on calcium homeostasis, uterine health, and milk production (Venjakob et al., 2022).Cows that did not calve within 7 d after treatment were reinjected with 10 × 10 6 IU cholecalciferol.Serum tCa concentrations were greater in treated cows.Gestation length (GL) was affected by the treatment (cows treated once: 278.4 d; cows treated twice: 284.7 d; control 281.3 d).Furthermore, we observed a greater risk for retained placenta (RP; cows treated once: 7.7%; cows treated twice: 4.0%; control: 2.0%) and metritis (cows treated once: 39.3%; cows treated twice: 33.3% control: 21.6%) as well as a reduction in milk yield in treated cows, compared with control cows (control 42.5 kg; cows treated once: 38.8 kg; cows treated twice: 38.7 kg).However, due to the study design (reinjection after 7 d), group allocation (treated once or twice) was not completely independent from GL and our results had to be interpreted carefully.
The present study was conducted (1) to confirm our previous findings on tCa homeostasis, GL, uterine health, and milk yield without injecting the cows for a second time and (2) to gain further insight into the potential mode of action by expanding our analyses of vitamin D metabolites as we had speculated that elevated serum concentrations of cholecalciferol itself or its metabolites might be related to the detrimental effects observed.This knowledge is important in respect to future recommendations on the use of cholecalciferol.Our hypotheses were that a treatment with 6 × 10 6 IU cholecalciferol would result in improved postpartum Ca homeostasis without exerting negative side-effects on health status and milk production.Based on our former results, we assumed that cows treated with 12 × 10 6 IU would have a shortened GL and a reduction in milk production in comparison to untreated animals.To verify or reject this hypothesis, the study was conducted without reinjecting cows that had not calved within 7 d in order not to bias the results on GL.

MATERIALS AND METHODS
The study was conducted on a commercial dairy farm in northern Germany between July 2020 and December 2020.The farm is located in the federal state of Brandenburg and has approximately 2,600 milking cows, with an average 305-d milk yield of 9,600 kg.All procedures reported herein were approved by the federal authorities (protocol 2347-48-2019).Sample size calculation was performed based on the results of our previous study in which cows entering their ≥2 lactation were treated with one dose of 12 × 10 6 IU cholecalciferol 5 d before calving had a 3.8 kg reduction in milk yield at first test day compared with untreated cows of the control group (Venjakob et al., 2022).Assuming 80% power, a 95% confidence level, and a SD of 8.0 kg 51 animals per group were needed to detect a similar effect on milk production.

Transition Cow Management
Transition cows were managed as recently described (Venjakob et al., 2022).Briefly, cows were dried-off at d 223 ± 10.7 of gestation and moved to the close-up pen at approximately d 255 of gestation.During the close-up period, cows received a negative DCAD diet (DCAD: −31 mEq/kg of dry matter) containing 2,000 IU of cholecalciferol per kg of DM.The TMR was formulated to meet or exceed minimum nutritional requirements (NEL, nutrients, minerals, and vitamins) for dairy cows (NRC, 2001).Ingredients and chemical composition of the close-up diet were described previously (Venjakob et al., 2022).In a subsample of approximately 10 closeup cows per wk urinary pH was assessed.The average urinary pH was 6.7 (±0.99; n = 25), 6.7 (±0.98; n = 25) and 6.8 (±0.99; n = 27) for CON, 6VitD, and 12VitD cows, respectively.At calving, calving ease (eutocia = unassisted calving; dystocia = calving assisted by 1 or more person) were recorded.The calves were separated from the dams immediately after calving and cows received an oral Ca bolus (Bovikalc, Boehringer Ingelheim, Ingelheim am Rhein, Germany) before being moved to the fresh cow pen.Milk records from the federal DHIA equivalent testing system were obtained from the on-farm computer system (HerdeW,version 5.8,Ketzin,Germany)

Treatment Allocation
Before the start of the study, 158 multiparous cows were randomly assigned to one of 3 treatment groups, based on a random list created in Excel (Office 2019, Microsoft Deutschland Ltd., Munich, Germany).Cows were enrolled every Mondays and Thursdays during the study period, based on their gestation length (GL 275 ± 2 d).At enrollment, the body condition score (BCS) was evaluated based on a 5-point scale with 0.25 increments (Ferguson et al., 1994).While cows of the control group were left untreated (CON, n = 52), cows of the 2 treatment groups received a single intramuscular injection of 6 × 10 6 IU cholecalciferol (6VitD, n = 52) or 12 × 10 6 IU of cholecalciferol (12VitD, n = 54; Ursovit D3, Serumwerk Bernburg, Bernburg, Germany), respectively.Cows were injected into one of the 2 hind limbs (i.e., M. semimembranosus, M. semitendinosus).All injections as well as BCS assessment were conducted by the research team.Researchers were not blinded to the treatment.

Disease Diagnosis
Until 10 DIM, daily health checks were performed by the research team including assessment of the general behavior, vaginal discharge, and manure consistency.Detection of clinical mastitis was performed 3 times daily by the farm personnel during regular milking.Clinical mastitis was defined according to Vasquez et al. (2017) as visible signs of inflammation such as redness, swelling, pain, or heat, and alterations such as clots, flakes, discoloration, or abnormal consistency of secretions.When fetal membranes were not expelled within the first 24 h after calving, cows were diagnosed with RP.On 7 DIM, vaginal discharge was assessed using the Metricheck device (Simcro, Hamilton, New Zealand).According to Urton et al. (2005), clear or no mucus was categorized as score 0, mucus with flecks of pus was categorized as score 1, foul smelling and mucopurulent (≤50% pus present) as score 2, foul smelling and mucopurulent (>50% pus present) as score 3, and putrid (foul smelling, watery, red or brown color) as score 4. Cows with score 4 were diagnosed with metritis (Sheldon et al., 2006).
For a subsample of 30 cows entering their 3rd lactation, we quantified the concentrations of 4 vitamin D metabolites.The concentration of 1,25-dihydroxycholecalciferol (1,25-(OH) 2 D 3 ) was determined by a commercial laboratory (Immundiagnostik AG, Bensheim, Germany) using the ELISA method.Intra-and interassay coefficients were 6.69% and 9.00%, respectively.The lower detection level was 4.80 pg/mL.

Statistical Analyses
Test day data of the first 5 DHIA equivalent test days of each cow and the results from blood analyses were combined using Access (Office 2010, Microsoft Deutschland Ltd., Munich, Germany), exported to Excel spreadsheets, and analyzed using SPSS for Windows (version 25.0, IBM Corp., Ehningen, Germany).Univariable models were calculated to test whether GL at enrollment, GL, interval between enrollment and calving, previous 305-d milk yield, parity, and calving ease was evenly distributed among CON, 6VitD, and 12VitD cows.To analyze the effect of treatment on postpartum blood minerals, vitamin D metabolites, and milk yield, repeated measures ANOVA with firstorder autoregressive covariance was performed using GENLINMIXED procedure of SPSS.Eight separate models were calculated to evaluate the effect of the cholecalciferol treatment on serum tCa, P i , Mg, cholecalciferol, 25-OHD 3 , 1,25-(OH) 2 D 3 and 24,25-(OH) 2 D 3 concentrations and milk yield.For serum concentrations of tCa, P i , Mg, cholecalciferol, 25-OHD 3 , and 24,25-(OH) 2 D 3, repeated measures were conducted on 1, 2, 3, 5, 7, and 10 DIM.For 1,25-(OH) 2 D 3 , repeated measures were conducted on 1, 2, 3, and 5 DIM.Milk yield analysis was conducted with repeated measures based on data obtained on test d 1 to 5. According to Dohoo et al. (2009) each explanatory variable was separately analyzed in a univariable model.Explanatory variables tested in the univariable models were treatment (CON vs. 6VitD vs. 12VitD), time (1, 2, 3, 5, 7, 10 DIM for blood minerals and cholecalciferol, 25-OHD 3 , and 24,25-(OH) 2 D 3 concentration; 1, 2, 3, 5 DIM for 1,25-(OH) 2 D 3 concentration; test d 1, 2, 3, 4, 5 for milk production), parity (parity 2 vs. parity 3 vs.parity ≥ 4), 305-d milk yield in previous lactation (continuous), and BCS at enrollment (normal; BCS of 2.75 to 3.25 vs. fat; BCS ≥3.5).If the univariable models resulted in a P-value <0.1, parameters were included in the final mixed model.In the model of tCa, Pi, Mg, and milk yield, all variables tested were included into the mixed models.Analyzing the effect of treatment on vitamin D metabolites, parity was excluded from the models, as all cows were in 3rd lactation.Furthermore, 305-d milk production of previous lactation (P = 0.89) and BCS at enrollment (P = 0.36) were not included in the mixed model of cholecalciferol, and BCS at enrollment (P = 0.76) was not included in the mixed model for 1,25-(OH) 2 D 3 .Selection of the model that best fit the data was performed using a backward stepwise elimination procedure that removed all variables with P > 0.1 from the model.All biologically plausible interactions such as time by treatment, treatment by parity and time by treatment by parity were tested.Whenever the inclusion of time by treatment led to a lower Akaike information criterion, the interaction was forced to remain in the model.In all models, the P-value was adjusted using a Bonferroni correction, to account for multiple comparisons.Variables were declared statistically significant when P < 0.05.As the sample size is not sufficient to analyze binary outcomes such as RP and metritis, disease incidences are described using a Chi 2 test.

RESULTS
Of the 158 multiparous Holstein Friesian cows, 52, 52, and 54 cows were allocated to the CON, 6VitD and 12VitD group, respectively.Parity (P = 0.47), energy corrected 305 milk yield of previous lactation (P = 0.26), GL at the day of enrollment (P = 0.88), and calving ease (P = 0.48) did not differ between the groups.Treatment with 12 × 10 6 IU of cholecalciferol resulted in a difference in GL (P < 0.01) and interval between treatment and calving (P < 0.01; Table 1).
Serum P i concentration was affected by time relative to calving (P < 0.01) and increased by prepartum treatment with cholecalciferol (P < 0.01).Compared with control cows, and 6VitD cows, 12VitD cows had greater serum P i concentrations during the first 10 DIM.Furthermore, 6VitD cows had greater serum P i concentrations compared with CON cows (1.52, 1.68, and 1.86 mmol/l for CON, 6VitD, and 12VitD, respectively; P < 0.01).We also observed a negative association with parity (P < 0.01) and 305-d milk yield in previous lactation (P < 0.01).
Serum Mg concentration was affected by time relative to calving (P < 0.01), negatively associated with parity (P < 0.01) and BCS at enrollment (P < 0.01) and decreased by prepartum treatment with cholecalciferol (P < 0.01).12VitD cows had lower serum Mg concentrations during the first 10 DIM compared with CON and 6VitD cows (0.85, 0.81, and 0.77 mmol/l for CON, 6VitD, and 12VitD, respectively; P < 0.01.Furthermore, the interaction of time by treatment (P = 0.08) remained in the model.

Vitamin D Metabolites
Treatment increased the serum concentrations of cholecalciferol (P < 0.05), while concentrations were below the detection limit in the untreated CON group (Table 2).After calving, serum concentrations decreased rather rapidly (Figure 2A).Treatment affected likewise the concentrations of 25-OHD 3 (P < 0.01) and 24,25-(OH) 2 D 3 (P < 0.01) throughout the entire observation period (Figure 2B and C; Table 2).Serum concentrations of 1,25-(OH) 2 D 3 showed a more dynamic pattern with the greater increase in the CON group compared with the treatment groups (Figure 2D).

DISCUSSION
As expected, treatment with cholecalciferol increased serum concentrations of cholecalciferol, 25-OHD 3 and 24,25-(OH) 2 D 3. 12VitD treatment and to lesser extent also 6VitD treatment increased tCa and Pi concentrations and decreased Mg concentration.In 12VitD cows, GL and milk yield were reduced, while the incidence was RP and metritis were higher in comparison to CON animals.

Blood Minerals
In the untreated CON, cholecalciferol concentrations were below our detection limit (5 ng/mL) which is in line with results from Poindexter et al. (2023a).In both treatment groups, administration of cholecalciferol 5 d before expected calving led to an increase not only in cholecalciferol itself but also in serum 25-OHD 3 in comparison to the CON group during the entire observation period (Table 2, Figure 2A, Figure 2B).This result was expected as the hydroxylation of cholecalciferol to 25-OHD 3 is not regulated tightly (Jones, 2008).
Serum concentrations of tCa and P i were greater in both treatment groups.These results confirm the findings of our former study (Venjakob et al., 2022) and results of an experiment by Poindexter et al. (2023a).This group of authors investigated the effect of increasing prepartum serum concentrations of 25-OHD 3 by feeding either 1 or 3 mg 25-OHD 3 instead of feeding 1 or 3 mg cholecalciferol.Serum concentrations of 25-  OHD 3 increased to 93.8 ng/mL (43 animals), 173.6 ng/ mL (46 animals), 58.3 ng/mL (39 animals), and 63.5 ng/mL (45 animals) in cows fed with 1 mg 25-OHD 3, 3 mg 25-OHD 3 , 1 mg cholecalciferol, and 3 mg cholecalciferol, respectively.The authors reported increased postpartum tCa with 1 mg or 3 mg oral 25-OHD 3 (2.15mmol/L and 2.17 mmol/L, respectively) in comparison with cows fed with the same amount of cholecalciferol (2.13 mmol/L and 2.11 mmol/L, respectively).Serum P i was affected accordingly (1.75 mmol/L and 1.80 mmol/L vs. 1.72 mmol/L and 1.68 mmol/L).In the present study, higher serum concentrations of 25-OHD 3 in the treated groups increased the tCa and P i likewise.
The reduction in serum Mg could have been caused by a decrease in renal reabsorption due to lower secretion of parathyroid hormone in the treatment groups associated with greater serum concentrations of tCa (Goff, 2008).From our observations and former results reported in the literature it can be concluded that the effect on serum tCa and P i was probably caused by 25-OHD 3 .
The biologically most active vitamin D metabolite regulating mineral homeostasis is 1,25(OH) 2 D 3 .After binding to the vitamin D receptor (VDR) and formation of a heterodimer with the retinoid X receptor, it affects so called vitamin D responsive elements in the promoter regions of target genes which leads to an activation or repression of the transcription of vitamin D dependent genes (Christakos et al., 2016).However, studies using mice lacking the enzyme that converts 25-OHD 3 to 1,25-(OH) 2 D 3 showed that 25-OHD 3 can bind to the VDR and induce transcription similar to 1,25-(OH) 2 D 3 (DeLuca et al., 2011).But due to the lower affinity of 25-OHD 3 to the VDR it is probably only able to activate the VDR when serum concentrations are above 150 ng/mL (Quesada-Gomez and Bouillon, 2018).
Although a typical increase until 3 DIM was observed in all cows, 6VitD and 12VitD cows had a lower 1,25-(OH) 2 D 3 concentration on 3 and 5 DIM compared with CON cows.From these results we conclude that the high serum concentrations of 25-OHD 3 had repressed the formation of 1,25-(OH) 2 D 3 , either directly as described in the next section, or indirectly via the increase in serum tCa and P i .

Vitamin D Metabolites
The formation of 1,25-(OH) 2 D 3 out of 25-OHD 3 is subject to a strict regulation by several factors like serum Ca, directly and indirectly via parathyroid hormone (PTH), P i , and 1,25-(OH) 2 D 3 itself as a negative feedback (Christakos et al., 2019).The greater concentrations of tCa, especially in the 12VitD group, probably inhibited the activation of 25-OHD 3 to 1,25-(OH) 2 D 3 as indicated by the lower concentrations of this metabolite (Table 2, Figure 2D) and induced the inactivation of 1,25-(OH) 2 D 3 , and 25-OHD 3 .

Health Status and Milk Production
Although the power of this study is limited (CON: 52 cows, 6VitD: 52 cows, 12VitD: 54 cows), the results on GL, RP and metritis are in line with the observations reported in our previous study (untreated cows: 187, treated once: 135, treated twice: 54) and thus support the hypothesis that 12 × 10 6 IU cholecalciferol 12 can have adverse effects on uterine health.This could be directly linked to the shortened GL, as Vieira-Neto et al. ( 2017) demonstrated that short GL is a risk factor for RP and metritis.
Compromised uterine health might have decreased DMI and thus milk yield of 12VitD cows.It was demonstrated that postpartum feed intake is compromised in cows with early lactation diseases such as metritis (Huzzey et al., 2007;Pérez-Báez et al., 2019).A direct impact of the treatment on milk production seems unlikely as Poindexter et al. (2023b) showed a positive correlation between postpartum 25-OHD 3 and milk yield in cows treated with cholecalciferol or 25-OHD 3 orally.In contrast to the present study, health status was not affected in their experiment although serum concentrations of 25-OHD 3 were comparable.However, there was also a trend for decreased prepartum DMI.

Future Research Related to Potential Interactions between Steroid and Vitamin D Metabolism
One of the crucial parts in the physiological cascade to induce calving is the induction of placental CYP17A1 that results in an increased placental transformation of progesterone to estrogen (Braun et al., 2012;Kindahl et al., 2002;Schuler et al., 2018).Interestingly, Novola-Mertínez et al. (2017) could show an upregulation of the RNA expression of CYP17A1 and a downregulation of CYP11A1 in cultured human placental cells treated with 1,25-(OH) 2 D 3 .CYP11A1 mediates the production of pregnenolone, the precursor of progesterone (Schuler et al., 2018).As described above, 25-OHD 3 at high concentrations can bind to the VDR and alter gene expression.In addition, cholecalciferol can be a substrate for CYP11A1, too, which could lead to a reduced formation of progesterone (Slominski et al., 2005).Future studies should investigate whether vitamin D treatment could interfere with steroid metabolism, either by altering the RNA expression of CYP450 enzymes or by competing with cholesterol as substrate for CYP11A1.The difference between our study and the experiment done by Poindexter et al. (2023aPoindexter et al. ( , 2023b) ) who supplemented either 1 mg or 3 mg of cholecalciferol or 25-OHD 3 per day orally and did not observe any effect on reproductive traits is the rapid increase in both cholecalciferol and 25-OHD 3 after the injection of cholecalciferol.Furthermore, the treatment administered by Poindexter et al. (2023aPoindexter et al. ( , 2023b) ) did not increase serum concentrations of cholecalciferol as pronounced as the single injection of 12 × 10 6 IU cholecalciferol which amounts to 300 mg.

Future Research Related to Potential Interactions between Ca Status around Calving and Immune Response
McArt and Neves (2020) differentiated between 3 different types of hypocalcemia evaluating tCa concentrations on 1 and 4 DIM and concluded that a certain degree of transient hypocalcemia around calving is a physiological condition.Other authors showed that there is a relation between endotoxemia and Ca homeostasis.Kvidera et al. (2017) infused cows with LPS, a cell wall component of gram-negative bacteria that elicits a robust and well-characterized immune response.Consequently, ionized Ca concentrations decreased.Horst et al. (2020) quantified the amount of Ca lost during LPS challenge in combination with the eucalcemic clamp technique.After LPS infusion, in half of the cows enrolled ionized Ca concentrations were leveled by infusing.The results indicated that intravenous administration of LPS causes a loss of 13.8 g of Ca within 12 h.Compared with cows that were treated with LPS only and showed a transient hypocalcemia, milk yield in cows that were treated and then leveled with Ca infusion was decreased by 15%.Horst et al. (2021) speculated that inflammation-induced hypocalcemia is a protective strategy to remove endotoxins.In the present study, cows of the 12VitD group that had higher serum concentrations of tCa around calving presented with a higher incidence of metritis.Future studies should further investigate whether treatments to prevent hypocalcemia might also have the potential to interfere with endogenous mechanisms regulating serum tCa as a response to the pro-inflammatory effects of calving.

Study Limitation
The present study was based on a former project where cows treated with 12 × 10 6 IU of cholecalciferol had a 3.8 kg lower milk production at first test day, compared with untreated control cows.Analyzing the association between treatment and milk production, in the present study, the interaction of test day relative to calving by treatment was not significant.We observed however an association with treatment independent from test day between 12VitD and CON and 12VitD and 6VitD cows.As the differences in test day milk yield might be lower between 6VitD and CON cows than between 12VitD and CON cows a more elaborate sample size would be needed to confirm that milk production between 6VitD cows and CON cows is not different.As the study was conducted on a commercial dairy farm and as there had been animal welfare issues after detecting negative downstream outcomes in the treatment group of the previous study, the collaborating institutions decided not to include more cows.

CONCLUSION
The present study confirmed our previous finding that an injection of 12 × 10 6 IU cholecalciferol led to increased tCa concentrations during the first 3 d after calving but shortened GL and negatively affected milk production, compared with control cows.In contrast, no detrimental effects were observed in cows treated with 6 × 10 6 IU cholecalciferol.The application of 6 × 10 6 IU cholecalciferol, however, had only marginal effects at 1 DIM.Therefore, it cannot be recommended for general use to prevent hypocalcemia.
Venjakob et al.: HYPOCALCEMIA IN DAIRY COWS

Figure 3 .
Figure 3.Effect of cholecalciferol treatment 5 d before expected parturition (on d 275 ± 2 of gestation) on energy-corrected milk yield over the first 5 monthly test days after calving of multiparous cows of a single dairy farm (n = 158).Energy corrected milk yield was associated with treatment (P < 0.05), parity (P < 0.01), time (P < 0.01), body condition score at enrollment (P < 0.01), 305-d milk production in previous lactation (P < 0.01).The interaction of time*treatment (P = 0.12) was forced to remain in the model.
Venjakob et al.: HYPOCALCEMIA IN DAIRY COWS