Randomized clinical trial evaluating the effects of administering acidogenic boluses at dry-off on udder health, milk yield, and herd removal.

Acidogenic boluses can mitigate potential negative effects of high milk yield at dry-off on udder health. This randomized controlled trial aimed to investigate the effect of administering acidogenic boluses at dry-off on dry period IMI dynamics and on milk production parameters, somatic cell count linear score ( LSCC ), clinical mastitis ( CM ), and herd removal in the next lactation. A total of 901 cows from 3 dairy farms were randomly allocated to a control ( CON , n = 458; no administration of acidogenic boluses at dry-off) or treatment group ( TRT , n = 443; administration of 2 acidogenic boluses at dry-off). Quarter milk samples were collected at dry-off and after calving and submitted for bacteriological milk culture. The effects of treat-ment on the presence of quarter-level postpartum IMI, cure of existing IMI, and acquisition of new IMI, and on the prevalence of cow-level high LSCC (LSCC ≥4) in the first 30 d in milk ( DIM ) were analyzed using mixed effects logistic regression. Mixed linear regression was used to analyze cow-level milk production parameters (i.e., milk yield, fat corrected milk, fat and protein yield, and LSCC) in the first 90 DIM and until 300 DIM. For CM and herd removal, Cox proportional hazard regression models were used. In addition to treatment group, lactation group at dry-off, presence of high LSCC in the last test-day, average milk yield in the week before dry-off, presence of CM in the lactation of enrollment, and biologically relevant interactions were offered in all models. There was no evidence of a difference in IMI dynamics or in milk, fat corrected milk, protein or fat yields in the subsequent lactation between groups. The TRT group had a lower LSCC in the first 2 mo postpartum compared with the CON group (2.58 ± 0.3 vs. 2.92 ± 0.3 and 2.42 ± 0.3 vs. 2.81 ± 0.3, for first and second month postpartum). The prevalence of high LSCC in the first 30 DIM was 9.1% lower in the TRT compared with the CON group (16.3% vs. 25.5%; risk difference: −9.2; 95% CI: −15.8, −2.5). Cows in the TRT group exhibited reduced hazards of CM in the subsequent lactation compared with cows in the CON group (HR: 0.75; 95% CI: 0.63, 0.89) as well as a reduced hazard of herd removal (HR: 0.82, 95% CI: 0.77, 0.88). The administration of acidogenic boluses as a component of dry-off management is a promising approach to maintain good udder health and reduce the hazard of CM and herd removal during the subsequent lactation.


INTRODUCTION
The dry period represents a vital period for the maintenance of optimal milk yield and udder health in the next lactation.This nonlactating period between 2 lactations is important for the renewal of the mammary gland epithelial tissues for the next lactation (Capuco et al., 1997;Kuhn and Hutchison, 2005;van Knegsel et al., 2013).Thus, managing high-producing cows around dry-off is recognized as a crucial moment in the dairy cow's production cycle and has received attention in recent decades (Capuco and Akers, 1999;Bachman and Schairer, 2003;Choudhary, 2014).Although the dry period is crucial for the cure of intramammary infections (IMI) and the prevention of new IMI (Bradley and Green, 2004;Halasa et al., 2009), the abrupt cessation of milking of high producing dairy cows can substantially increase the risk of new IMI (Pantoja et al., 2009;Oliver and Murinda, 2012;Leelahapongsathon et al., 2015).The incidence of clinical mastitis (CM) is increased in early lactation, with most cases being associated with IMI acquired during the nonlactating period (Green et al., 2002).Likewise, higher milk yield at dry-off is also associated with increased risk of high SCC in the next lactation (Gott et al., 2017).
Previous studies suggest that reducing milk yield at dry-off is essential to mitigate the negative effects of high milk yield during this period on udder health (Dingwell et al., 2004;De Prado-Taranilla et al., 2020;Vilar and Rajala-Schultz, 2020).This recommendation stems from the abrupt cessation of milking after dryoff in high-producing dairy cows, resulting in increased intramammary pressure (O'Driscoll et al., 2011).This, in turn, has been linked to pain, discomfort, changes in activity behavior, and a delay in the keratin plug -a teat barrier protection -formation (Dingwell et al., 2004;Zobel et al., 2015;Maynou et al., 2018).To maintain udder health throughout the dry period, it has been suggested that employing different strategies to reduce milk yield at dry-off could be beneficial.
A gradual reduction in milk yield before dry-off, achieved through measures such as reducing the number of daily milkings, and decreasing the quantity and quality of the feed offered to cows before dry-off, can alleviate some of the repercussions associated with an abrupt cessation of milking in high-yielding dairy cows (Tucker et al., 2009;Ollier et al., 2014;Rajala-Schultz et al., 2018).The feasibility of these strategies in commercial herds is limited.Thus, alternative options are required to reduce milk yield to facilitate the management of high producing dairy cows around dry-off.Recently, the use of acidogenic boluses at dry-off has also been shown to reduce milk yield (Maynou et al., 2018) and is suggested as an alternative strategy to decrease milk yield before dry-off without requiring drastic changes in feeding or milking management.However, this strategy was only investigated in one small study (n = 84; Maynou et al., 2018) whose primary objective was to measure milk production and feed intake but not to examine udder health outcomes.Therefore, conducting a larger-scale randomized controlled trial is warranted to determine the impact of this dry-off management strategy on postpartum udder health and performance.
The objectives of this randomized controlled trial were to evaluate the effects of administering acidogenic boluses at dry-off on postpartum udder health, milk yield, and herd removal in the first 90 DIM and up to 300 DIM in the next lactation.We hypothesized that the use of acidogenic boluses at dry-off would result in a reduction in the risk of new IMI, decreased postpartum IMI prevalence and a greater IMI cure risk during the dry period.In addition, we hypothesized that this strategy will decrease somatic cell count linear score (LSCC), the incidence of CM and the risk of herd removal, as well as increase milk yield.

MATERIALS AND METHODS
A multi-site randomized controlled trial to investigate the effect of acidogenic boluses was performed on 3 commercial dairy herds in Minnesota.Cows were enrolled between April and September 2021 and followed up until September 2022.All interventions and procedures used to conduct this experimental study were approved by the University of Minnesota Institutional Animal Care and Use Committee (2012-38706A).

Herd and cow inclusion criteria
Herds included in this study had unique individual cow identification tags, electronic herd management systems (DairyComp 305, Valley Agriculture Software), and participated in regular herd DHIA testing (minimum of 8 herd tests per year).Herds were required to comply with research protocols and allow the research team to access cow and herd software records on a weekly basis during the study period.All enrolled herds were located in Minnesota and used a free-stall housing system for dry and lactating cows.All 3 herds fed anionic salts to close-up dry cows and pre-and postpartum diets were formulated to meet the nutrient requirements for dry and lactating dairy cows according to breed and conditions of each farm.Descriptive characteristics of the 3 herds enrolled in this randomized clinical trial are presented in Table 1.
In total, 901 cows were enrolled in this randomized controlled clinical trial across 3 dairy herds in Minnesota.Cows presenting signs of clinical disease, severely lameness (lameness score >4 using a 5-point lameness scoring system (Sprecher et al., 1997), with BCS <2 (using 5-point scale with 0.25-point increments (Ferguson et al., 1994) at the time of dry-off, and with records of CM in the last 14 d were not eligible for enrollment in the study.Enrolled cows that aborted, had a dry period length shorter than 30 or greater than 90 d, or were mistakenly enrolled (n = 3) were excluded.Body condition score and lameness at enrollment were assessed visually by a trained scorer in the research team.

Experimental design
Each herd was visited by the research team once per week on the herd's regular day of dry off.All lactating cows that met inclusion criteria were enrolled in the study.At enrollment, eligible cows were blocked by herd and dry-off date and randomly assigned to a treatment (TRT) or control (CON) group.For each herd and dry-off date, a list of random numbers was created in blocks of 6 using a random number generator in Microsoft Excel (version 16;Microsoft Corp.).Before the final milking, 2 aseptic milk samples were collected from each quarter by the research team for use in the later microbiological cultures for the identification of the microorganism causing the IMI at dry-off.Briefly, after milking staff performed their usual pre-milking teat disinfection routine, research team members, using disposable gloves, discarded the first 3 streams of milk, scrubbed the teat-end with cotton gauze soaked in 70% ethanol, and collected duplicate quarter milk samples (approximately 10 mL) using aseptic procedures following National Mastitis Council guidelines (Adkins and Middleton, 2017).Samples were stored on ice immediately and frozen at −20°C within 4 h of collection.Following unit detachment after the final milking, all quarters from all cows were disinfected in the same manner previously described and subjected to treatment with intramammary antibiotics (Dry-Clox; Boehringer Ingelheim Animal Health USA Inc.) and internal teat sealant (Lockout; Boehringer Ingelheim Animal Health USA Inc.), followed by dipping all teats with a post-milking teat disinfectant.
Cows allocated to the TRT group received 2 acidogenic boluses containing 111.2 g of calcium chloride, 44.5 g of calcium sulfate, and 20.4 g of ammonium chloride per bolus (Bovikalc Dry; Boehringer Ingelheim Animal Health USA Inc.).Acidogenic boluses were administered by members of the research team after the final milking (within 20 min) and dry-off procedures were finalized.This approach allowed the research team to enroll and administer the study intervention during the same visit.In herd A and on one site of Herd B, boluses were given in the parlor before the cow was released.At the second site in Herd B, cows were restrained in headlocks to receive the boluses, while in Herd C, cows were restrained in a chute.Cows allocated to the CON group did not receive acidogenic boluses.Farm staff were masked to treatment assignments.Following the dry-off and administration of study treatments, cows from both groups were commingled and managed in accordance with the specific routines of the herd throughout the dry period.
Aseptic duplicate quarter milk samples were collected again within the first 13 d after calving by research staff.These samples were also stored on ice immediately upon collection and frozen at −20°C until further analysis.Post-calving, cows in both groups continued to be managed as a commingled group, following the herd's standard operating procedures, and were monitored up to 300 DIM.After the designated follow-up period, information about milk yield, SCC, the dry period length, CM, and other relevant events (e.g., herd removal), for both the enrollment and subsequent lactation were gathered using electronic herd records.The standard CM definition, as described elsewhere (Oliveira et al., 2013), was discussed with the herd personnel at the start of the study; however, CM diagnosis and treatment were implemented by each herd following their own standardized care plan.Descriptive characteristics of cows enrolled in the study by treatment group are presented in Table 2.

Microbiological culture of milk samples
Milk samples were submitted to the Laboratory of Udder Health at the University of Minnesota for routine aerobic culture.Samples were thawed at room temperature and homogenized by gentle inversion.Using a calibrated loop, 10 μL of milk were plated onto Columbia CNA agar with 5% sheep blood (CNA) and MacConkey agar.The plates were subsequently incubated aerobically at 37°C and assessed for the growth of distinct isolates by a trained technician at both 18 to 24 h and again at 42 to 48 h.Samples were classified as contaminated if 3 or more morphologically distinct isolates were recovered and, in such cases, the duplicate sample was submitted for milk culture (Adkins and Middleton, 2017).The identity of representative colonies was investigated using matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF MS; Bruker Daltonics Inc.) as previously described (Jahan et al., 2021).Briefly, the peak profiles of each isolate were compared with a reference spectra in the Biotyper reference library (Microflex).Following manufacturer's recommendations, confidence scores were used in the following way: > 2.0: species-level diagnosis; 1.8 to 2: genus-level diagnosis; and < 1.8: MALDI-TOF diagnosis not recorded and traditional identification methods used.Intramammary infection, independently of LSCC, was defined as the presence of growth of at least 1 colony forming unit (cfu) of any culturable isolate (i.e., 100 cfu/mL), except for nonaureus Staphylococcus (NAS) and Bacillus spp.when a minimum of 2 (≥200 cfu/mL) and 5 (≥500 cfu/mL) colonies, respectively, were necessary to define the presence of an IMI (Dohoo et al., 2011;Rowe et al., 2020).

Sample size calculation
The presence of postpartum quarter-level IMI was considered the primary outcome of interest.Sample size calculations were performed to estimate the minimum sample size needed for an α (type I error) of 0.05, power of 0.80, risk of postpartum IMI of 20% in the CON group (Arruda et al., 2013;Johnson et al., 2016), and a detectable minimum difference of 4% in postpartum quarter-level IMI.The resulting sample size was then arbitrarily multiplied by 1.2 to account for non-independence of observations within each cow and herd.Based on these assumptions, we estimated that we would need to enroll a total of 3,593 quarters (898 cows), with 1,796 quarters (449 cows) in each study group.Sample size was calculated using online tools (https: / / epitools .ausvet.com.au/samplesize).Post-hoc power calculations for clinical mastitis (power = 0.62),  subclinical mastitis in the first 30 DIM (power = 0.97), and herd removal (power = 0.40) were conducted using JMP Pro 17 (SAS Institute Inc.)

Statistical analysis
Statistical analysis was conducted using R version 4.2.3 (R Core Team) and Stata version 17 (StataCorp LLC.).Code and output are available in an online repository (https: / / fepenamosca .github.io/Bovikalc .github.io).Data management was performed using functions from the "tidyverse" package in R (Wickham and Bryan, 2023).The prevalence of IMI by each microorganism was explored using the "table1" package in R (Rich, 2021).
Treatment was assigned at the cow level, while outcomes were assessed at the quarter (nested within each cow) or cow level.Quarter-level outcomes included new and postpartum IMI risk and dry period cure likelihood.All quarters were considered at risk for new and postpartum IMI, but only quarters with an IMI present at enrollment were at risk of cure.Postpartum IMI was defined as the presence of IMI in the postpartum samples.New IMI was defined as the presence of postpartum IMI by a microorganism not present in the same quarter at dry-off.Cured IMI was defined as the absence of postpartum IMI by a microorganism present at dry-off.Intramammary infections prevalence at dry-off and calving were reported per microorganism and treatment group.A secondary analysis categorized infections into major groups: Staphylococcus spp., Streptococcus spp.and Streptococcus-like organisms (SSLO), gram-negative microorganisms, and other microorganisms (Rowe et al., 2020;Peña-Mosca et al., 2023).The impact of the administration of acidogenic boluses at dry-off on postpartum IMI, new IMI, and IMI cure was assessed via generalized linear mixed models, as implemented in Stata's "meglm" command.These models employed a binomial family with a logit link (logistic regression), incorporating herd-ID and cow-ID as random effects to account for non-independence of observations.Adjusted proportions were derived with Stata's "margins" command.Risk differences (RD) were estimated using marginal standardization (Rowe et al., 2020).
The cow level outcomes investigated included milk yield, fat-corrected milk (FCM; calculated as 3.5% FCM = [(kg of milk x 0.4324) + (kg of milk fat x 16.216)]), fat and protein yield, first case of CM and herd removal up to 90 and 300 DIM, and the prevalence of high LSCC (LSCC ≥4) in the first postpartum test.Mixed linear regression models assessed the effect of the administration of acidogenic boluses at dry-off on next lactation milk yield, FCM production, fat and protein yield, and LSCC, as implemented in the "nlme" package in R (Pinheiro et al., 2017).An interaction between treatment and postpartum month (i.e., 30 d intervals starting at calving) was forced into these models.Herd-ID and cow-ID (nested within each herd) were included in the models as random effects, using an autoregressive correlation structure -selected for its lower Akaike Information Criterion.Normality assumption was checked by visual inspection of model's residuals.For each of the models, the "emmeans" package in R (Lenth et al., 2022) was utilized to estimate the treatment's effect on all outcomes for each postpartum month and to calculate marginal means in the multivariable models.
The effect of administration of acidogenic boluses at dry-off on the presence of high SCC in the first postpartum herd test was analyzed using a generalized linear mixed model with a binomial family and logit link (logistic regression) using Stata's "meglm" command.For this specific outcome, herd-ID was incorporated as a random effect.Cox proportional hazards regression examined the effect of administration of acidogenic boluses at dry-off on incidence of CM (i.e., first case of CM) and herd removal utilizing the "survival" package in R (Therneau, 2022).Herd-ID was included using a robust sandwich estimator to account for clustering of observations within each farm.The proportional hazards assumption was tested using the Schoenfeld residuals.Kaplan-Meier curves were constructed to describe the time to CM and herd removal for the cows in the different treatment groups.
The covariates selected using directed acyclic graphs and offered into the models included lactation group (1st, 2nd, and 3rd and greater), presence of high SCC in the last herd test before dry-off, presence or absence of CM in the previous lactation, average milk yield in the last week before dry-off, and any biologically relevant interaction.In addition, presence of quarter-level IMI at dry-off was considered as a potential covariate for the investigation of new IMI or postpartum IMI.Subsequently, covariates were added into the multivariable models and backward stepwise.Variables offered to the models that changed the estimates by 10% or more were considered to be an intervening variable and kept in the models.Effect modification by lactation group and milk yield at dry-off were evaluated, and excluded from the models when P-value >0.05.Results were interpreted based on model estimates and confidence intervals instead of P-values (Poole, 2001).

Descriptive statistics
Information about predominant breed, herd size, lactation group, average milk yield, bulk tank SCC, incidence of CM, and bedding material used in each herd is presented in Table 1.In addition, Table 1 contains information describing the lactation group distribution, incidence of CM in the enrollment lactation, high SCC at enrollment, milk yield at dry-off, DIM at enrollment, and dry period length for the cows enrolled in the study.The baseline characteristics for the cows enrolled in the 2 treatment groups are presented in Table 2.
Three cows were not included in the data analysis due to non-compliance with assigned treatment and another 3 cows due to CM treatment in the last 14 d before dry-off (Figure 1).Thirty 5 cows were retrospectively excluded from analysis due to short (<30 d) or long (>90 d) dry periods or because of an abortion (CON, n = 13; TRT, n = 22).At dry-off, 26.2% (926/3,535) of the quarter level milk samples were defined as contaminated and excluded from the IMI dynamics analysis (CON: 27.5% [493/1,794]; TRT: 24.9% [433/1,741]).During the dry-period or first 2 weeks after calving, 33 cows died or were sold and were not sampled postpartum (dry period: CON: n = 9 cows; TRT: n = 8 cows; postpartum: CON: n = 10 cows; TRT: n = 6 cows).During postpartum, another 20 cows (CON: n = 12; TRT: n = 8) were not sampled due to an inability to locate them during the visits; thus, these cows were excluded from the IMI dynamics analysis.Among postpartum quarter samples, 21.4% were contaminated (CON: 21.3% [359/1,689]; TRT: 21.6% [358/1,660]).These exclusions resulted in 41.5% (1,391/3,349) of the quarters being excluded from the new and cured IMI analysis due to missing or contaminated samples at dry-off or postpartum.Overall, the number of samples excluded from the new and cured IMI analysis was similar in both treatment groups (CON: 43.3% [732/1,689]; TRT: 39.7% [659/1,660]; P = 0.18).
During the dry period, new IMI were observed in 33.4% (320/957) of quarters in the CON group and 30.2% (303/1,001) in the TRT group.The crude proportions of IMI cure during the dry period were 83.8% (331/395) for the CON group and 85.3% (355/416) for the TRT group.Table 3 and Table 4 provide comprehensive lists of microorganisms isolated from dry-off and postpartum samples by treatment group, respectively.

Treatment effect on the quarter level postpartum prevalence, new, and cured intramammary infections
Cows in the TRT (36.4%) and CON (37.1%) groups had similar adjusted IMI risk postpartum (RD: −0.7%; 95% CI: −4.7, 3.2; P = 0.71) in quarter level IMI after accounting for lactation group, the presence of high SCC in the last herd test, and the presence of CM in the enrollment lactation.After adjusting for herd-ID and cow-ID as random effects, the risk of new IMI was not different between treatment groups (CON: 32.8% and TRT: 29.6%; RD [95%CI]: −3.2% [-7.5, 1.0]; P = 0.13).After adjusting for presence of high SCC in the last herd test and average milk yield in the last week before dry-off, the estimated proportions of quarters that cured during the dry period was similar between treatment groups, with risk of 84.3% for CON and 85.4% for TRT (RD (%) [95%CI]: −1.1 [-4.0, 6.1]; P = 0.67).Similarly, no statistical or biological difference was observed when comparing the risk of postpartum IMI, new IMI, and cured IMI by the different bacterial groups in our generalized linear mixed models.The adjusted risk of postpartum IMI, new IMI, and cured IMI and the risk differences for all IMI are presented in Table 5 and by the bacterial group in Supplemental Table S1 (https: / / osf .io/2srkb ?view _only = 707eab-da2d044c47b2fb23cd2be2c98c).
The adjusted LSCC was lower in the TRT group compared with the CON group during the first 2 postpartum months in both investigated time domains.The adjusted risk of high SCC in the first postpartum month was 9.1% lower (95% CI: −15.8%, −2.5%, P < 0.01) in cows in the TRT group (Adjusted risk: 16.3%) compared with cows in the CON group (Adjusted risk: 25.5%).Because none of the covariates caused the estimates to change by more than 10%, no covariates were included in the final model.In the first 90 DIM, 12.2% of cows in the TRT group and 14.7% in the CON group were diagnosed with CM (RD (%) [95%CI]: −2.5 [-7.0, 1.9]; P = 0.27).The final Cox proportional hazard model investigating the effect of the administration of acidogenic boluses at dry-off on CM mastitis during this period indicated that the  Fixed effects covariates included postpartum month (P < 0.001), lactation group (P = 0.02), milk yield at the last week before dry-off (P < 0.001), treatment by postpartum month interaction (P = 0.88) for 90 DIM and postpartum month (P < 0.001), lactation group (P < 0.001), milk yield at the last week before dry-off (P < 0.001), treatment by postpartum month interaction (P = 0.54) for 300 DIM. 4 Fixed effects covariates included postpartum month (P = 0.006), lactation group (P = 0.33), milk yield at the last week before dry-off (P < 0.001), treatment by postpartum month interaction (P = 0.97) for 90 DIM and postpartum month (P < 0.0001), lactation group (P < 0.001), milk yield at the last week before dry-off (P < 0.001), treatment by postpartum month interaction (P = 0.39) for 300 DIM. 5 Fixed effects covariates included postpartum month (P < 0.001), lactation group (P = 0.32), milk yield at the last week before dry-off (P < 0.001), treatment by postpartum month interaction (P = 0.81) for 90 DIM and postpartum month (P < 0.001), lactation group (P < 0.001), milk yield at the last week before dry-off (P < 0.001), treatment by postpartum month interaction (P = 0.17) for 300 DIM. 6 Fixed effects covariates included postpartum month (P = 0.52), lactation group (P = 0.68), milk yield at the last week before dry-off (P < 0.01), high SCC (>200,000 cells/mL) at dry off (P = 0.01), treatment by postpartum month interaction (P = 0.98) for 90 DIM and postpartum month (P < 0.001), lactation group (P = 0.05), milk yield at the last week before dry-off (P < 0.001), high SCC (>200,000 cells/mL) at dry off (P = 0.03), treatment by postpartum month interaction (P = 0.32) for 300 DIM.

DISCUSSION
This multi-site randomized controlled trial exploring the impact of administering acidogenic boluses at dryoff revealed no evidence of a difference (either biological or statistical) in dry period IMI dynamics between the TRT and CON groups.In addition, no differences were observed in milk yield, FCM, and protein and fat yield when comparing both treatment groups by 90 and up to 300 DIM in the subsequent lactation.In contrast, the administration of acidogenic boluses at dry-off led to reductions in LSCC before 60 DIM, a lower prevalence of high LSCC in the first month postpartum, and diminished hazards of CM and herd removal in the following lactation.

Effect of treatment with acidogenic boluses on the quarter level postpartum prevalence, new, and cured intramammary infections
Across all studied bacterial groups, the risk difference used to measure the impact of the treatment on IMI dynamics remained below 1%.Similarly, we observed only a slight variation in the adjusted risk for new IMI when comparing both treatment groups.For example, the postpartum prevalence of IMI showed near equivalence, as represented by the 0.7% risk difference favoring quarters from TRT cows in comparison to those from CON cows.Although this implies a consistent and uniform influence across the groups, these results lack significant biological or statistical relevance.The absence of a significant effect in assessing the influence of administering acidogenic boluses at dry-off on the IMI dynamics could be attributed to the existing implementation of multiple strategies to protect the mammary gland during the dry period (such as intramammary antimicrobial treatment and internal teat sealants).In instances when these strategies are already in place, the inherent risk of new IMI tends to be lower (Dohoo et al., 2007).Consequently, the impact of additional interventions, such as administering acidogenic boluses at dry-off, might not be as pronounced as that of other interventions like antimicrobial therapy or internal teat sealants.For instance, in a meta-analysis examining the use of internal teat sealants (Rabiee and Lean, 2013), the risk of new IMI was 14% lower in cows that received internal teat sealant compared with quarters not receiving intramammary antimicrobials, while the risk difference was only 5% lower when compared with quarters receiving intramammary antimicrobials in addition to internal teat sealant.In our study, all cows and quarters received intramammary long-acting antimicrobial treatment and internal teat sealant at dry-off, limiting our results to demonstrating only an additive effect and decreasing the potential effect of the tested intervention on IMI dynamics around the dry period.Future investigations into the effects of administering acidogenic boluses at dry-off on IMI dynamics in cows not receiving intramammary long-acting antimicrobials (e.g., selective dry cow therapy) are warranted.Such investigations are driven by the need for additional tools to efficiently manage the dry-off process for highproducing dairy cows, especially given the increasing number of farms and animals undergoing dry-off without the use of antimicrobials.
In this manuscript, we have chosen to report our findings using absolute measures of association (risk difference) rather than relative ones (relative risk or odds ratio).This decision was made because we considered that absolute measures of association to be reflective of the actual impact of this intervention on a commercial dairy farm, making interpretation more straightforward.Furthermore, presenting results in relative terms often leads readers to overestimate the impact of an intervention, often disregarding the underlying absolute risks (Noordzij et al., 2017).Considering all the information presented above, it is worth noting that while statistical significance was not achieved within the hypothesis-testing framework, we observed a 3.2% reduction in the risk of new IMI in quarters from cows that received acidogenic boluses at dry-off, with a 95% CI ranging from −7.5% to 1.0%.This translates to a  Asterisks indicate statistically significant differences between both groups within a given time point (P < 0.05).The final fixed effects covariates included treatment (P = 0.04), postpartum month (P = 0.73), lactation group (P = 0.001), milk yield at the last week before dry-off (P = 0.80), high SCC (>200,000 cells/mL) at dry off (P = 0.001), clinical mastitis in lactation of enrollment (P < 0.001), treatment by postpartum month interaction (P = 0.14) for 90 DIM and treatment (P = 0.06) postpartum month (P < 0.001), milk yield at the last week before dry-off (P = 0.85), high SCC (>200,000 cells/mL) at dry off (P < 0.001), clinical mastitis in lactation of enrollment (P < 0.001), treatment by postpartum month interaction (P = 0.02) for 300 DIM.
10% decrease in the baseline risk of the TRT group compared with the CON group (with a relative risk of 0.90 when comparing 29.6% to 32.8%).Considering these variations might offer insights into potential benefits for postpartum udder health and performance, even if these differences are not substantial from both a biological and statistical standpoint.For the sake of comparison, and taking into account practical considerations on the farm, the utilization of prolactin in-hibitors, such as cabergoline, could be considered as an approach that closely compares with the strategy of administering acidogenic boluses at dry-off to mitigate the adverse effects of high milk yield at dry-off.A European study reported that using cabergoline alone, without concurrent dry cow therapy or internal teat sealants, resulted in a 4.5% reduction in the risk of new IMI from major pathogens (Hop et al., 2019).This reduction was observed when compared with a negative control group that lacked both dry cow therapy and internal teat sealant.These observed benefits align with the outcomes of our study, particularly concerning the reduction in the risk of new IMI.This alignment holds true even when considering that the cows in our study received dry cow therapy and internal teat sealant.

Effect of treatment with acidogenic boluses on cow level postpartum milk and components yield, udder health, and herd removal
Our results indicated no significant differences in milk yield, FCM, and protein and fat yield between treatment groups, consistent with the findings of a prior study that also reported no significant changes in milk yield following acidogenic bolus treatment (Maynou et al., 2018).
Cows receiving acidogenic boluses at dry-off had a lower LSCC during the first 60 DIM and a reduced risk of high LSCC in the first postpartum test compared with the CON cows, in both models (i.e., 90 and 300 DIM follow-up period).One potential explanation for the reduced LSCC and lowered risk of high LSCC could be the fact that administration of acidogenic boluses at dry-off has been related to a reduction in milk yield and udder pressure within the first 48 h after treatment (Maynou et al., 2018).This, as a result, could have a positive impact on postpartum udder health (Vilar and Rajala-Schultz, 2020).In addition, cows in TRT showed reduced hazards of CM compared with the CON group, which concurs with the lower SCC in these cows during early lactation (Astorga-Jorquera et al., 2022).High LSCC usually indicates the presence of an IMI that elicited an inflammatory response (Pantoja et al., 2009).Indeed, prior studies have shown that new IMI during the dry period are associated with the presence of high SCC (Gott et al., 2017) and CM in early lactation (Green et al., 2002).Interestingly, there was no discernible evidence of divergence in the IMI dynamics among the treatment groups.
Despite the predominantly positive impact of administering acidogenic boluses at dry-off on quarter-level udder health parameters in the IMI dynamics results, it is important to highlight that these findings were not as biologically significant as the cow-level udder health results.One plausible reason for this inconsistency could be the high proportion of contaminated milk samples encountered in this study, potentially leading to skewed estimations regarding the impact of administering acidogenic boluses on IMI dynamics, as discussed below.Additional research is warranted to comprehensively elucidate the biological rationale of this finding.
In our study, reduced rates of herd removal were encountered in response to the administration of acidogenic boluses at dry-off.These results exhibit a consistent pattern, mirroring the observed reduction in CM incidence among cows in the TRT group when compared with those in the CON group.Taken together, these results align with prior studies demonstrating that CM contributes to an elevated risk of culling in dairy cows (Gröhn et al., 1998;Hertl et al., 2018).However, it is important to mention that herd removal is complex and influenced by numerous factors, with injuries, reproductive problems, low milk yield, and mastitis reported as some of the most frequent reasons for her removal (Shahid et al., 2015).Despite these complexities, these findings suggest an enduring impact of this singular management event at dry-off on herd removal, as it appears to improve some of the risk factors, such as reducing CM incidence and increasing FCM, associated with herd removal.

Internal validity
One potential threat to causal inference in epidemiological studies is the lack of exchangeability between treatment groups (Greenland and Robins, 2009).In our study, the randomization seemed to be effective -despite not blocking by parity -given the similar distribution of covariates across groups at enrollment.
Other important sources of systematic error are selection bias and misclassification bias.Selection bias could potentially represent an issue for making causal inferences in randomized control trials, when the likelihood of loss during follow-up is affected by both the exposure and outcome (Hernán et al., 2004); in our study, these correspond to the treatment group and IMI dynamics, respectively.Despite efforts to prevent milk sample contamination, such as sampler training before the start of the study and the use of National Mastitis Council guidelines (Adkins and Middleton, 2017), a significant proportion of the milk samples in our study resulted in quarter-level milk contamination.This shows a failure in procedures used during milk sampling, which could possibly be related to the large number of samplers involved in the research team across different visits (a total of 15 samplers were involved in 1 or more herd visits).Fortunately, the proportion of milk samples excluded from the analysis because of contamination were similar across TRT and CON groups, suggesting that likelihood of quarter level contamination was not driven by the treatment allocation.However, the high proportion of milk contamination in our study is a concern about the possibility of selection bias, and decreased precision and power of estimates.
Misclassification bias is a concern when only one sample is used to estimate the IMI prevalence, given the imperfect sensitivity and specificity of milk culture (Dohoo et al., 2011).In the present study, further steps were taken to prevent misclassification by using a cutoff point of 200 cfu/mL for NAS and 500 cfu/mL for Bacillus spp.(Dohoo et al., 2011;Rowe et al., 2020).In addition, the use of standardized protocols that follow National Mastitis Council guidelines, in combination with MALDI-TOF MS provides a higher taxonomic resolution, minimizes the risk of misclassification across different types of microorganisms and has been validated for the identification of mastitis pathogens (Jahan et al., 2021).

External validity
This multi-site randomized controlled trial was performed in 3 Midwest conventional dairy farms where average milk yield at dry-off was 27.6 kg/d.This milk yield represents the performance of dairy cows in the region, aligning with findings from a prior study (Rowe et al., 2020).Within the United States, 93% of cows receive intramammary antibiotics at dry-off, while 36% are administered internal teat sealant -a combination that constitutes effective strategies for safeguarding the mammary gland throughout the dry period (Halasa et al., 2009;Rabiee and Lean, 2013;USDA, 2016).Consequently, our results can be appropriately generalized to other farms employing similar practices of intramammary antimicrobials and internal teat sealant during dry-off.As the practice of selective dry cow therapy continues to evolve, further research is essential to examine the impact of this intervention on farms that opt for selective treatment protocols at dry-off.
The dry-off IMI prevalence in our study was much higher than in previous reports (Newman et al., 2010;Rowe et al., 2019).Additionally, the average bulk tank SCC in our herds exceeded the national average in the United States (USDA, 2021).Exploring the possibility of administering the intervention 24 to 48 h before the planned dry-off time becomes more important.This is especially significant when considering earlier research that showed the lowest milk yield in lactating cows occurred 48 h after administering acidogenic boluses (Maynou et al., 2018).

Impact
This is the first study to investigate the effect of the administration of acidogenic boluses at dry-off on dry period IMI dynamics.The use of acidogenic boluses represents a potential alternative for producers who want to take additional steps to improve the transition from the lactating to nonlactating period and its impact on the udder health during the following lactation.The reduction in LSCC in early lactation in the TRT group signifies that administering the acidogenic bolus at dry-off enhances early lactation udder health and reduces the likelihood of elevated SCC during the first postpartum test.Furthermore, the advantages of using the acidogenic bolus at dry-off for udder health extend to a decrease in the incidence of CM among cows in the TRT group, compared with those in the CON group.When considering these improvements collectively, it is likely that they contributed to the lower herd removal hazard for cows in the TRT group in the subsequent lactation.

CONCLUSIONS
Contrary to our hypothesis, the findings of this randomized clinical trial revealed no evidence of differences in dry period IMI dynamics and subsequent lactation milk yield performance between cows that were administered acidogenic boluses at dry-off and those in the CON group.However, the use of this strategy resulted in a reduced prevalence of high LSCC in the first month postpartum, decreased LSCC in early lactation, and a diminished hazard of CM and herd removal in the subsequent lactation as compared with the CON group.

FlorentinoFigure 1 .
Figure 1.Flowchart describing losses during follow-up at each stage of the study.Number of excluded cows or quarter-level samples along with reasons for exclusion at each phase of the study are shown on the right side of the flowchart.TRT: Cows that were administered 2 acidogenic boluses at dry-off.CON: Cows that did not receive any acidogenic boluses.

Figure 2 .
Figure 2. Somatic cell count linear score (LSCC) during the first 90 (A) and up to 300 (B) days in milk in the subsequent lactation by treatment group.The solid maroon line represents the control group (CON; no acidogenic boluses administered at dry-off; n = 458) cows and the gold solid line represents the treatment group (TRT; 2 acidogenic boluses administered at dry-off; n = 443) cows.Asterisks indicate statistically significant differences between both groups within a given time point (P < 0.05).The final fixed effects covariates included treatment (P = 0.04), postpartum month (P = 0.73), lactation group (P = 0.001), milk yield at the last week before dry-off (P = 0.80), high SCC (>200,000 cells/mL) at dry off (P = 0.001), clinical mastitis in lactation of enrollment (P < 0.001), treatment by postpartum month interaction (P = 0.14) for 90 DIM and treatment (P = 0.06) postpartum month (P < 0.001), milk yield at the last week before dry-off (P = 0.85), high SCC (>200,000 cells/mL) at dry off (P < 0.001), clinical mastitis in lactation of enrollment (P < 0.001), treatment by postpartum month interaction (P = 0.02) for 300 DIM.

Table 1 .
Descriptive characteristics of the three herds and of the cows enrolled in each herd enrolled in a study investigating the effect of the administration of acidogenic bolus at dry-off

Table 2 .
Cow level characteristics of cows enrolled in a study investigating the effect of the administration of acidogenic bolus at dry-off stratified by treatment group 1A total of 942 cows were initially included in the study (TRT = 468 and CON = 474).After excluding cows mistakenly enrolled in the wrong group, those with a case of clinical mastitis 14 d before enrollment, and cows with short (<30 d) or long (>90 d) dry periods, or cows that aborted, 901 cows were deemed eligible for statistical analysis.2Clinicalmastitis in previous lactation at the cow level recorded by farm personnel on herd records.3Somatic cell count >200,000 cells/mL at the last DHIA test before enrollment.

Table 3 .
Quarter-level culture results of the cows enrolled at dry-off (n = 901) stratified by treatment group 1

Table 4 .
Quarter-level culture results of enrolled cows after calving (n = 851) stratified by treatment group 1

Table 5 .
Final multivariable models estimating the effect of administration of acidogenic boluses at dry-off 1 on quarter level intramammary infections (IMI) dynamics during the dry-period Risk difference and 95% CI were derived from generalized linear mixed models in Stata.The final model included random intercepts for herd-ID and cow-ID and other fixed-effect covariates.

Table 6 .
Final multivariable models estimating the effect of administration of acidogenic boluses at dry-off 1 on milk yield, fat corrected milk (FCM) 2 , protein and fat yield, and somatic cell linear score (LSCC) during 90 and 300 DIM 1TRT: Cows that were administered two acidogenic boluses at dry-off.CON: Cows that did not receive any acidogenic boluses. 2 3.5% FCM = [(kg of milk x 0.4324) + (kg of milk fat x 16.216)].3