Transition cow nutrition and management strategies of dairy herds in the northeastern United States: Part III—Associations of management and dietary factors with analytes, health, milk yield, and reproduction

The objective was to evaluate relationships between putative periparturient management and dietary factors at the pen and herd levels with metabolic-and inflammation-related analytes, health disorders, milk yield, and reproductive performance. Multiparous and primiparous cows from 72 farms in the northeastern United States were enrolled in a prospective cohort study. Farms were visited 3 times during the prepartum and postpartum periods: during the far-off dry, close-up dry, and fresh periods. Pen measurements were taken at each visit for the pens where cows sampled were housed, and particle size was determined for the total mixed ration for the pen. A survey was used to acquire data on herd-level management variables. Blood samples were collected from the same 11 to 24 cows per farm during the close-up and fresh period visits. Whole blood was analyzed for postpartum β-hydroxybutyrate (BHB) concentrations, and plasma was analyzed for prepartum and postpartum nonesterified fatty acids (NEFA) and postpartum haptoglobin (Hp) concentrations. Health event, milk yield, and reproductive records were acquired through the records management software program used on the farm. For the pen-level analysis, primiparous and multiparous cows were analyzed separately. For the pen-and herd-level analysis, a simple linear regression was conducted on all possible explanatory variables. Variables were included in the full multivariable general linear model if P < 0.20, and a manual backward stepwise elimination process ensued until all variables had P < 0.10. Our results indicate that pen-and herd-level management factors are associated with blood biomarkers, health, milk yield, and reproductive performance. For the prepartum period, our results support


INTRODUCTION
The transition cow period is characterized as a time of many changes for the dairy cow.Frequent pen moves, adaptation to lactation, dietary changes, increases in social interactions, and other environmental factors can all contribute to success or lack of success during the early postpartum period.Although limited data exist on feeding strategies to optimize cow health and productivity during the transition period, even fewer research studies have investigated management factors or nutritional factors on a continuous basis during this timeframe.Current recommendations are often driven by field observations originating from concepts evaluated in small, controlled research studies.Bach et al. (2008) reported a 13.2 kg/d range in the mean milk yield across 47 herds that were fed the exact same TMR and determined that 50% of the observed variation was attributable to non-nutritional management factors.Research has demonstrated that non-nutritional management factors, such as stocking density, pen moves, and commingling of primiparous and multiparous cows, can influence health and milk yield (Cook and Nordlund, 2004;von Keyserlingk et al., 2008;Huzzey et al., 2012).Controlled research trials typically attempt to evaluate the change in one management factor while minimizing changes in additional management factors; therefore, the magnitude of the effects of management factors has not been fully elucidated.In addition, most studies evaluating management factors have not been conducted during the transition period.
Our objective was to evaluate relationships between putative periparturient management factors at the pen level, during the far-off, close-up, and fresh periods, and at the herd level with metabolic-and inflammation-related analytes, disorder incidence, milk yield, and reproductive performance.We hypothesized that differences in management of pen movement, stocking density factors, commingling, dietary factors, and feed delivery would be associated with blood biomarkers, disease, milk production, and reproductive performance.

Study Population and Study Design
A more complete description of the study population and study design can be found in Kerwin et al. (2022b).Briefly, a prospective cohort study was conducted from a convenience sample of 72 farms located in New York and Vermont between November 2012 and August 2015.All procedures involving cows in this study were approved by the Cornell University Institutional Animal Care and Use Committee, protocol # 2012-0124.Inclusion criteria for farms were (1) Holstein herds, (2) ≥400 milking cows, (3) freestall housing, (4) TMR-fed herds, and (5) enrolled in monthly DHI testing or have on-farm milk recording with record management by Dairy Comp 305 (Valley Ag Software) or PCDART (Dairy Records Management System).Farms were visited 3 times for data collection focused on the same cohort of cows during the far-off dry (49 to 28 d before expected parturition), close-up dry (21 to 0 d before expected parturition; 4 wk after the first visit), and fresh periods (0 to 21 DIM; 16 to 21 d after the second visit).Pen measurements, including stall dimensions, feed bunk space, water space, and facility design, were taken at each visit for the pens where sampled cows were housed.Cows were observed for health disorders of interest in the first 30 DIM by farm personnel, and case definitions for the health disorders were provided to the farm for recording and consistency purposes, as described previously (Kerwin et al., 2022b).
A more complete description of the blood sampling scheme has been published previously (Kerwin et al., 2022b).A convenience sample of the same 11 to 24 cows was blood sampled twice: once during the closeup visit and again during the fresh period visit.To reflect herd demographics, the cows selected on the close-up visit were approximately one-third primiparous cows and two-thirds multiparous cows.Primiparous cows were defined as cows entering their first lactation, and multiparous cows were defined as entering the second or greater lactation.Whole blood was analyzed for BHB concentrations on the fresh period visit.Plasma from the close-up and fresh period visits were analyzed for nonesterified fatty acids (NEFA) concentrations, and plasma from the fresh period visit was analyzed for haptoglobin (Hp) concentrations on cows 0 to 12 DIM.
A more complete description of the diet analysis has been published previously (Kerwin et al. 2022b).Briefly, the formulated diets fed to the cows observed at the time of the visit were collected from the nutritionist or herd manager.Total mixed ration samples fed to pens housing the cows of interest were collected at each visit and were analyzed by near-infrared spectroscopy and evaluated for particle size with the 3-sieve Penn State Particle Separator (PSPS).The addition of the 4-mm sieve was used to estimate physically effective NDF (peNDF) in the diet (Mertens, 1997;Heinrichs and Jones, 2013).Physically effective undigested NDF after 240 h of in vitro fermentation (peuNDF240) was calculated by multiplying the proportion of TMR above the 4-mm sieve by the average analyzed diet undigested NDF after 240 h of in vitro fermentation (uNDF240) on a DM basis (Miller et al., 2020).The formulated diet and corresponding analyzed forage sample compositions were input into the Cornell Net Carbohydrate and Protein System (version 6.1, Cornell University), and diet Cornell Net Carbohydrate and Protein System files were imported into the Nutritional Dynamic System Professional (NDS Professional version 3.8.10.06,RUM&N Sas) to extract estimated total fermentable carbohydrates (% of DM), forage NDF (% of DM), MP (g/kg of DM), and ME (Mcal/kg of DM).
Health event, milk yield, and reproductive records were acquired through the record management software Kerwin et al.: TRANSITION MANAGEMENT AND NUTRITION ASSOCIATIONS program used on the farm when all the observed cows were at least 150 DIM.

Management Explanatory Variables of Interest
Pen Level.For the pen-level analysis, we assessed pen-related management and nutritional explanatory variables during the far-off, close-up, and fresh period visits.Stall stocking density was calculated as the number of cows in the pen at the time of the visit divided by the number of usable stalls in the pen.If the pen was a bedded pack, a stall was considered 11 m 2 of pack space (Nordlund, 2009).Bunk space density was calculated as the number of cows in the pen at the time of the visit divided by the number of headlocks.If a pen did not have headlocks, a headlock was considered 61 cm of rail space (NFACC, 2009).The feed pushup frequency within each day for a pen was dichotomized as <5×/d or ≥5×/d (Miller-Cushon and DeVries, 2017).The feeding frequency within each day was not evaluated during the prepartum period due to too few observations (<5) feeding the pen more than once per day.Commingling of primiparous and multiparous cows was also assessed at the pen level.Only the 19-mm sieve of the PSPS was assessed, due to the industry's focus on diet sortability.Other dietary factors assessed were peuNDF240, total fermentable carbohydrates (% of DM), forage NDF (% of DM), the interaction between total fermentable carbohydrates and forage NDF, MP (g/kg of DM), and ME (Mcal/kg of DM).Field observations suggest that there may be an interaction between total fermentable carbohydrates and forage NDF.
Herd Level.For the herd-level analysis, explanatory variables assessed included whether cows were routinely vaccinated in the calving pen and fresh pen, whether the herd used a maternity pen or a calving pen, the number of pen moves during the prepartum and postpartum periods, time spent in the calving pen and fresh pen, and time spent locked up in the fresh pen.A maternity pen was classified as a pen cows moved into at least 3 to 0 d before expected calving, whereas a calving pen was classified as a pen cows moved into when exhibiting signs of labor.The prepartum pen moves were the number of pen moves from dry-off or 60 d before expected calving to parturition for multiparous and primiparous cows, respectively (>2 vs. ≤2).The move from the lactating pen to the far-off dry cow pen was included in this measure for multiparous cows.The postpartum pen moves were the number of pen moves from parturition to 90 DIM (>2 vs. ≤2).The time spent in the calving pen before moving to the fresh pen after parturition was dichotomized as ≤8 h or >8 h.Eight hours was chosen as the cut point to reflect the time between milkings, assuming a herd is milked 3×/d.Approximately 75% of fresh period pens were milked 3×/d in this study (Kerwin et al., 2022b).The time spent in the first pen moved into after parturition was dichotomized as ≤10 d or >10 d (Hoseyni et al., 2020).Time spent locked up in the fresh pen for health checks was categorized as follows: (1) locked up <1×/d for <1 h, (2) locked up daily for <1 h, or (3) locked up daily for ≥1 h.

Statistical Analysis
The continuous outcomes of interest were calculated by parity within herd and were as follows: (1) prevalence of elevated prepartum NEFA concentrations in multiparous cows (≥0.17 mmol/L), (2) prevalence of elevated postpartum NEFA concentrations (≥0.59 mmol/L), (3) prevalence of postpartum BHB concentrations (≥1.2 mmol/L), (4) prevalence of elevated postpartum Hp concentrations (≥0.45 g/L), (5) disorder incidence (DI; incidence of one or more displaced abomasum, clinical ketosis, metritis within 30 DIM), (6) average milk yield at 4 wk of lactation (WK4MY), (7) average 305-d mature equivalent milk yield at the fourth test day (ME305; mean ± SD: 114 ± 13 DIM), (8) 21-d herd pregnancy rate (PR), ( 9) herd probability of pregnancy (PP), and (10) the pregnancy risk to first service (PRFS).All outcomes were calculated for the cohort of cows observed, except PR and PP, as described subsequently.Biomarker concentration thresholds were chosen as they were the thresholds chosen for the herd-alarm levels associated with an increase in DI for primiparous and multiparous cows (Kerwin et al., 2022a).The prevalence of elevated prepartum NEFA was only evaluated for multiparous cows because a herd-alarm level was not identified for primiparous cows.The herd PP was determined by averaging the PP for the first 2 estrus cycles after the herd voluntary waiting period for the group of cows that calved within the same calving date range as the cows observed using Dairy Comp software calculations (the percent of services with confirmed pregnancy; Eicker et al., 2000).The 21-d PR was determined by averaging the two 21-d periods after the herd voluntary waiting period for the group of cows that calved within the same time frame as the cows observed using Dairy Comp software calculations (calculated as the heat detection rate × PP; Eicker et al., 2000).Cows that were never bred were not included in the PRFS analysis (n = 155).Due to the 21-d PR and PP being calculated at the herd level, it was not possible to assess these outcomes at the pen level.Data from culled cows were included in all outcomes.
A sample size calculation was conducted to estimate the prevalence of cows with hyperketonemia within a farm, as described previously (Kerwin et al., 2022a).Raw data were entered into Microsoft Excel (Microsoft Corp.).Data cleaning was conducted to correct human recording errors before statistical analysis.
All statistical analyses were calculated using SAS software (SAS 9.4,SAS Institute Inc.).Descriptive statistics and the prevalence of elevated biomarkers were calculated using PROC MEANS and PROC FREQ.Analyses were conducted at the pen and herd levels.
Pen Level.For the pen-level analysis, primiparous and multiparous cows were analyzed separately.One pen was represented for each farm within each period for the primiparous cows.Respectively, 4, 1, and 3 farms had 2 pens represented for the far-off, close-up, and fresh periods for the multiparous cows, and the remaining farms only had 1 pen.The faroff period was not assessed for primiparous cows due to too many missing observations.Pens representing <4 cows were removed from the analysis; this would remove primiparous pens with less than 50% of our maximum sampling population for primiparous cows.The fresh period visit explanatory variables were not assessed for the prevalence of elevated prepartum NEFA concentrations outcome because the outcome occurred before the explanatory variables were measured.A simple linear regression was conducted on all possible continuous explanatory variables, interactions, and categorical explanatory variables, which occurred before the outcome of interest, using PROC REG and PROC GLM, respectively to determine the univariable association.Plots were assessed visually to determine whether a linear or quadratic relationship existed between the outcome of interest and the continuous management variables.Leverage was assessed by observing Cook's distance, and outliers were removed one at a time if Cook's D >0.13 (Cody, 2011).Explanatory variables with P < 0.20 were offered to a multivariable general linear model (PROC GLM) for each outcome.Calving season was included as a covariate if P < 0.10 in a univariate analysis, and a manual backward stepwise elimination process ensued until all variables had P < 0.10.
Herd Level.For the herd-level analysis, outcomes were calculated by parity within a herd.Number of prepartum pen moves was the only explanatory variable assessed for the prevalence of elevated prepartum NEFA concentrations outcome since all the other explanatory variables occurred after the outcome was measured.A multivariable linear regression analysis (PROC MIXED) was conducted on all possible explanatory variables and included parity, the interaction with parity, and the random effect of farm.Calving season was included in the multivariable linear regression analysis as a covariate if season was associated with the outcome (P < 0.10; BHB, DI, PR, PP, and PRFS).If the interaction with parity resulted in P ≥ 0.20, then it was removed and the multivariable linear regression analysis was repeated and the P-value reported.Because ME305 is a calculation that accounts for parity, the interaction between parity and the explanatory variable was initially assessed, but the interaction and parity were removed if the interaction had P ≥ 0.20.Leverage was assessed by observing Cook's distance, and outliers were removed if Cook's D >0.13 (Cody, 2011).Explanatory variables with P < 0.20 were offered to the full multivariable general linear model (PROC MIXED) for each outcome with herd as a random effect, and a manual backward stepwise elimination process ensued until all variables had P < 0.10.The P-values for variables with more than 2 categories (time locked in fresh pen) were corrected for multiple comparisons with a Tukey's honest significant difference test.If an interaction was detected between parity and a categorical variable, an F-test was conducted to determine at which parity level the variable of interest differed, using the SLICE option in the LSMEANS statement.If the F-test resulted in P < 0.10 for at least one of the parity groups and the explanatory variable had more than 2 levels, Bonferroni tests were conducted for each comparison within the parity in the LSMESTIMATE statement to correct P-values for multiple comparisons.
Pen and Herd Level.Before the initial analysis of the full model for the pen-and herd-level models, collinearity was assessed.For continuous variables, collinearity was assessed by using PROC REG with the TOL and COLLINOINT options.If variables had a tolerance <0.40 or if variables in an eigenvector with a condition number >10 were contributing more than 50% to the variance decomposition of the eigenvector, then the variable that had the lowest risk of committing a type I error with the outcome was kept in the model and the other was removed (Allison, 2012); however, we did not observe collinearity in any of our models.For categorical variables, collinearity was assessed via Fisher's exact chi-squared test calculated using PROC FREQ, and if P < 0.10, the variable that had the lowest risk of committing a type I error with the outcome was kept in the model; however, we did not observe collinearity in any of our models.Plots of studentized residuals were visually assessed for homogeneity and normality of variance.Leverage was assessed by observing Cook's distance.Extreme outliers based on studentized residual plots and observations with Cook's D >0.13 were removed.The restricted maximum likelihood parameter estimates and least squares means with standard error are reported throughout for models that yielded at least one explanatory variable with P < 0.10.

Descriptive Results and Univariable Associations
A more complete description of the study population was reported previously (Kerwin et al., 2022a,b).The simple linear regression results for the association between the pen-and herd-level explanatory variables and the outcome of interest for primiparous and multiparous cows can be found in Supplemental Tables S1 to S5 (https: / / doi .org/ 10 .17632/n8mhp2n7rh .1;Kerwin et al., 2022c).Exclusions for each analysis can be found in Supplemental Tables S6 to S8 (https: / / doi .org/ 10 .17632/n8mhp2n7rh .1;Kerwin et al., 2022c).

Prevalence of Elevated Biomarker Concentrations
Prepartum NEFA.The pen-level results for the prevalence of elevated prepartum NEFA concentrations analysis are reported in Table 1.No explanatory variables remained in the far-off model for multiparous cows.We found no evidence for the herd-level explana-tory variable to be associated with the prevalence of elevated prepartum NEFA concentrations.
Total fermentable carbohydrates was the only variable remaining in the close-up period pen model for multiparous cows.A 1-percentage unit increase in total fermentable carbohydrates was associated with a 1.7-percentage unit decrease in the proportion of multiparous cows with elevated prepartum NEFA concentrations (R 2 = 0.08; P = 0.02).
Postpartum NEFA.For the prevalence of elevated postpartum NEFA concentrations, the pen-and herdlevel results are reported in Table 2.No explanatory variables remained in the fresh period model for primiparous cows nor the far-off period model for multiparous cows.
For the pen-level analysis, total fermentable carbohydrates during the close-up period was associated with an increase (R 2 = 0.06; P = 0.05) in the proportion of primiparous cows with elevated postpartum NEFA concentrations.Bunk space density (P = 0.04) and water space (P = 0.04) for multiparous cows during the close-up period were associated with an increase in the Bunk space density = number of cows/number of headlocks × 100.If a pen had a rail at the feed bunk, a headlock was considered 61 cm. 4 Water space = cm of linear water space/number of cows in the pen.
proportion of multiparous cows with elevated postpartum NEFA concentrations (model R 2 = 0.10).Forage NDF in the fresh period was negatively associated with the proportion of multiparous cows with elevated postpartum NEFA concentrations (R 2 = 0.07; P = 0.02).
For the herd-level analysis, keeping cows in the calving pen for >8 h after parturition was associated with a greater proportion of cows with elevated postpartum NEFA concentrations compared with herds that kept cows in the calving pen for ≤8 h (43.6 vs. 21.0 ± 6.0%; P < 0.001).
BHB.The pen-and herd-level results for the postpartum BHB analysis are reported in Table 3.No explanatory variables remained in the BHB models for the close-up period for primiparous cows.
For the pen-level analysis, the proportion of particles on the 19-mm sieve during the fresh period was positively associated with the proportion of primiparous cows with elevated postpartum BHB concentrations (P = 0.008).Commingled fresh period pens were associated with a higher proportion of primiparous cows with elevated BHB concentrations compared with noncommingled fresh period pens (15.9 vs. 4.0 ± 4.1%; P = 0.02).Fresh period pens that were fed >1×/d were associated with a lower proportion of primiparous cows (3.0 vs. 16.9 ± 4.2%; P = 0.009) and multiparous cows (21.7 vs. 40.1 ± 4.9%; P = 0.003) with elevated BHB concentrations compared with pens that were fed ≤1×/d.Water space during the far-off period was negatively associated with the proportion of multipa- Means within parity group with different superscripts differ (P < 0.05) as determined by an F-test. x,y Means within parity group with different superscripts differ (0.05 ≤ P < 0.10) as determined by an F-test.
1 Range = the range (continuous) or number of observations (categorical) for variables remaining in the model.rous cows with elevated postpartum BHB concentrations (P = 0.04).Predicted ME during the close-up period was positively associated with the proportion of multiparous cows with elevated postpartum BHB concentrations (P = 0.05).
For the herd-level analysis, we observed an interaction between parity group and the time spent in the calving pen after parturition such that keeping multiparous cows in the calving pen for more than 8 h was associated with a lower proportion of multiparous cows with elevated postpartum BHB concentrations (4.7 vs. 17.7 ± 7.0%; P = 0.08) and keeping primiparous cows in the calving pen for more than 8 h was associated with a higher proportion of primiparous cows with elevated postpartum BHB concentrations (26.6 vs. 7.2 ± 5.4%; P = 0.001) compared with keeping primiparous or multiparous cows in the calving pen for ≤8 h.
Hp.For the Hp analysis, the pen-and herd-level results are reported in Table 4.No explanatory variables remained in the Hp models for the close-up period for multiparous cows.
For the pen-level analysis, pushing up feed ≥5×/d during the close-up and fresh periods was associated with a 22.9% (R 2 = 0.08; P = 0.07) and a 22.7% (R 2 = 0.08; P = 0.06) increase in the proportion of primiparous cows with elevated Hp concentrations, respectively.The proportion of particles on the 19-mm PSPS sieve during the far-off period was negatively associated with the proportion of multiparous cows with elevated Hp concentrations (R 2 = 0.04; P = 0.10).Commingling in fresh period pens was associated with a lower proportion of multiparous cows with elevated Hp concentrations compared with non-commingled fresh period pens (36.5 vs. 47.7 ± 4.4%; P = 0.03), and fresh period total fermentable carbohydrates was positively associated with the proportion of multiparous cows with elevated Hp concentrations (P = 0.07).
For the herd-level analysis, we observed an interaction between the number of prepartum pen moves and parity; moving primiparous cows ≤2× from 60 d before expected calving to parturition was associated with a lower proportion of primiparous cows with elevated Hp concentrations compared with moving primiparous cows >2× (57.1 vs. 69.0± 5.6%; P = 0.04).We observed no evidence that the number of pen moves from dry-off to parturition was associated with the prevalence of elevated Hp concentrations for multiparous cows (≤2× vs. >2×: 42.0 vs. 37.2 ± 7.2%; P = 0.50).An interaction occurred between parity group and the time spent in the calving pen after parturition; herds with primiparous cows kept in the calving pen for >8 h were associated with a greater proportion of primiparous cows with elevated Hp concentrations compared with herds that kept primiparous cows in the calving pen for ≤8 h (79.4 vs. 46.7 ± 8.3%; P < 0.001).We found no evidence that time spent in the calving pen was associated with the prevalence of elevated Hp concentrations for multiparous cows (>8 h vs. ≤8 h: 37.2 vs. 42.0 ± 7.2%; P = 0.57).We observed an interaction between parity group and time locked in the fresh pen.There was no evidence that time locked in the fresh pen was associated with the prevalence of elevated Hp concentrations for multiparous cows (P = 0.56).Herds that had primiparous cows locked up daily for ≥1 h were associated with a lower proportion of primiparous cows with elevated Hp concentrations (53.3 ± 7.9%) than herds that had primiparous cows locked up daily for <1 h (72.3 ± 5.7%; P = 0.11) but did not differ from herds that had primiparous cows locked up for <1×/d for <1 h (63.6 ± 5.6%; P = 0.72).

Postpartum Performance Outcomes
Disorder Incidence.The pen-and herd-level results for the DI analysis are reported in Table 5.No explanatory variables remained in the DI models for the close-up period for multiparous cows.Herds (n = 6) were removed from the DI analysis if they did not record one of the disorders (Supplemental Tables S6 to  S8).
For the pen-level analysis, bunk space density during the close-up period was positively associated (P = 0.03) and ME was negatively associated (P = 0.08) with DI for primiparous cows.The interaction between total fermentable carbohydrates and forage NDF during the fresh period was associated with DI for primiparous cows (P = 0.002).The predicted DI response for the primiparous fresh period model, using the range of values observed for total fermentable carbohydrates and forage NDF, is presented in Figure 1.Fresh period forage NDF >25.8% of DM was positively associated with DI for primiparous cows and fresh period forage NDF <25.8% of DM was negatively associated with DI for primiparous cows, as fresh period total fermentable carbohydrate concentration increased.Fresh period total fermentable carbohydrates >41.6% of DM was positively associated with primiparous DI and fresh period total fermentable carbohydrates <41.6% of DM was negatively associated with primiparous DI, as fresh period forage NDF concentration increased.The proportion of particles on the PSPS 19-mm sieve during the far-off period was negatively associated with DI for multiparous cows (P = 0.08).Feeding fresh period pens >1×/d was associated with a lower DI for multiparous cows than feeding pens ≤1×/d (7.5 vs. 14.8 ± 2.9%; P = 0.04).

WK4MY.
The pen-and herd-level results for the association between management factors and WK4MY are reported in Table 6.No explanatory variables remained in the far-off WK4MY models for multiparous cows.One herd was removed from the pen-and herdlevel analyses due to missing WK4MY (Supplemental Tables S6 to S8).Means within parity group with different superscripts differ (P < 0.05) as determined by Bonferroni tests (time locked in fresh pen × parity group) or an F-test.
x,y Means, within parity group if applicable, with different superscripts differ (P < 0.10) as determined by an F-test (interaction).The prepartum pen moves were the number of pen moves from dry-off or 60 d before expected calving to parturition for multiparous and primiparous cows, respectively (>2 vs. ≤2).The move from the lactating pen to the far-off dry cow pen was included in this measure for multiparous cows.
For the pen-level analysis, the proportion of particles on the PSPS 19-mm sieve (P = 0.06) and total fermentable carbohydrates (P = 0.06) for the close-up pen diets were positively associated with WK4MY for primiparous cows.Metabolizable energy in the fresh period diet for primiparous cows was positively associated with WK4MY (P = 0.003).Commingling in the fresh period pens was associated with lower WK4MY for primiparous cows than non-commingled pens (32.9 vs. 35.0± 0.7 kg/d; P = 0.01).Metabolizable protein in the close-up period diet for multiparous cows was negatively associated with WK4MY (P = 0.03).The interaction between total fermentable carbohydrates and forage NDF during the fresh period was associated with WK4MY for multiparous cows (P = 0.004).The predicted WK4MY response for the multiparous fresh period model, using the range of values observed for total fermentable carbohydrates and forage NDF, is presented in Figure 1.Fresh period forage NDF >23.6% of DM was negatively associated with WK4MY for multiparous cows and fresh period forage NDF <23.6% of DM was positively associated with WK4MY for multiparous cows, as fresh period total fermentable carbohydrates concentration increased.Fresh period total fermentable carbohydrates >40.9% of DM was negatively associated with multiparous WK4MY and fresh period total fermentable carbohydrates <40.9% of DM was positively associated with multiparous WK4MY, as fresh period forage NDF concentration increased.
For the herd-level analysis, we observed a calving pen vaccination by parity group interaction (P = 0.05).Means with different superscripts differ (P < 0.10) as determined by Tukey's honestly significant difference test.Vaccinating multiparous cows in the calving pen was associated with lower WK4MY than not vaccinating in the calving pen (42.7 vs. 46.8± 1.8 kg/d; P = 0.04); however, we found no evidence of a difference in WK4MY for primiparous cows (vaccinated vs. not vaccinated: 33.1 vs. 33.4± 1.2 kg/d; P = 0.80).

ME305.
For the ME305 analysis, the pen-and herdlevel results are reported in Table 7. Two herds were removed from the analysis due to missing ME305 (Supplemental Tables S6 to S8).No explanatory variables remained in the far-off period model for multiparous cows.
For the pen-level analysis, the interaction between total fermentable carbohydrates and forage NDF dur-ing the close-up period was associated with ME305 for primiparous cows (P = 0.004).The predicted ME305 response for the primiparous close-up period model, using the range of values observed for total fermentable carbohydrates and forage NDF, is presented in Figure 1.Close-up period forage NDF >39.2% of DM was positively associated with ME305 for primiparous cows and close-up period forage NDF <39.2% of DM was negatively associated with ME305 for primiparous cows, as close-up period total fermentable carbohydrate concentration increased.Close-up period total fermentable carbohydrates >23.1% of DM was positively associated with primiparous ME305 and close-up period total fermentable carbohydrates <23.1% of DM was negatively associated with primiparous ME305, as close-up period forage NDF concentration increased.Close-up period diet ME was negatively associated with ME305 for multiparous cows (P = 0.01).We observed a negative association between peuNDF240 during the fresh period and ME305 for primiparous (P = 0.002) and multiparous cows (P = 0.02).

Reproductive Performance
Herds were removed from the pen-and herd-level analysis for the reproductive outcomes if the farm used natural service (n = 2) or if the reproductive management was compromised (n = 1) or missing data (n = 1; only for PR and PP outcomes; Supplemental Tables S6 to S8).

21-d PR.
The 21-d PR herd-level analysis results are reported in Table 8.The F-test conducted for the parity and maternity pen interaction resulted in P = 0.32 for multiparous cows and P = 0.10 for primiparous cows, such that moving primiparous cows into a maternity pen where they were expected to calve in the next 3 to 0 d was associated with a lower 21-d PR than moving primiparous cows into a calving pen when the primiparous cow is showing signs of labor (26.4 vs. 29.7 ± 1.8%).
PP.The herd-level results for the PP analysis are reported in Table 9. Keeping cows in the calving pen for >8 h after parturition was associated with a lower PP than keeping cows in the calving pen for ≤8 h (35.8 vs. 40.3± 2.4; P = 0.09).
PRFS.The pen-and herd-level analysis results from the PRFS are reported in Means within parity group with different superscripts differ (P < 0.05) as determined by an F-test. 1 Range = the range (continuous) or number of observations (categorical) for variables remaining in the model. 2PSPS = Penn State Particle Separator (Nasco). 3Ferm.CHO = total fermentable carbohydrates. 4 Commingling = primiparous and multiparous cows are present in the same pen. 5FNDF = forage NDF. 6Interaction between total fermentable carbohydrates (% of DM) and forage NDF (% of DM).Interaction between total fermentable carbohydrates (% of DM) and forage NDF (% of DM).variables remained in the close-up period model for primiparous cows.
For the pen-level analysis, fresh period diet peu-NDF240 was positively associated with PRFS for primiparous cows (P = 0.10).For the far-off period for multiparous cows, diet MP was negatively associated (P < 0.001) with PRFS, and pushing feed up ≥5×/d was associated with a higher PRFS than pushing feed up <5×/d (35.5 vs. 28.6 ± 3.0%; P = 0.05).The interaction between total fermentable carbohydrates and forage NDF during the far-off period was associated with PRFS for multiparous cows (P = 0.004).The predicted PRFS response for the multiparous far-off period model, using the range of values observed for total fermentable carbohydrates and forage NDF, is presented in Figure 1.Far-off period forage NDF >44.8% of DM was negatively associated with PRFS for multiparous cows and far-off period forage NDF <44.8% of DM was positively associated with PRFS for multiparous cows, as far-off period fermentable total carbohydrate concentration increased.Far-off period total fermentable carbohydrates >20.4% of DM was negatively associated with multiparous PRFS and far-off period total fermentable carbohydrates <20.4% of DM was positively associated with multiparous PRFS, as far-off period forage NDF concentration increased.The proportion of particles on the PSPS 19-mm sieve (P = 0.06) and diet ME (P = 0.07) for the close-up period for multiparous cows were negatively associated with PRFS, whereas diet MP was positively associated with PRFS (P = 0.07).For the fresh period model for multiparous cows, bunk stocking density (P = 0.06) was negatively associated and MP (P = 0.02) was positively associated with PRFS.
For the herd-level analysis, we observed an interaction between the amount of time locked in the fresh pen and parity group (P = 0.08); however, we found no evidence that PRFS differed within parity (multiparous: P = 0.25; primiparous: P = 0.16).Keeping cows in the calving pen for >8 h after parturition was associated with a lower PRFS than keeping cows in the calving pen for ≤8 h (24.5 vs. 38.2± 4.2%; P = 0.003).

DISCUSSION
Our study identified relationships between transition cow management and nutritional factors and metabolic-(NEFA and BHB) and inflammation-related (Hp) biomarkers, health, milk yield, and reproductive performance.Notably, our results support increasing the proportion of particles on the PSPS 19-mm sieve for the prepartum diets, optimizing bunk space during the close-up period, and not overfeeding ME during the close-up period.For the fresh period, our results support optimizing bunk space, avoiding commingling primiparous and multiparous cows, increasing feeding frequency, avoiding high peuNDF240 and total fermentable carbohydrate diets while suppling adequate high forage NDF, and ensuring adequate diet ME and MP.At the herd level, our results support the practice of avoiding vaccination in the calving or maternity pen, using a calving pen compared with a maternity pen, reducing the number of prepartum and postpartum pen moves for primiparous cows, and avoiding long stays (≥8 h) in the calving or maternity pen after parturition.Little to no work has evaluated management factors during the transition cow period on postpartum performance outcomes related to blood biomarkers, health, milk yield, or reproductive performance.Previous studies evaluating management factors have focused on mid-to late-lactation cows and were performed in a controlled research environment or were controlled research trials evaluating the effects of management factors on the behavior of transition cows.In addition, to our knowledge, transition period nutritional parameters have not been evaluated on a continuous basis with our outcomes of interest.Controlled research studies evaluating nutrient parameters during the transition  .  Interaction between total fermentable carbohydrates (% of DM) and forage NDF (% of DM). 6 Calving season was dichotomized into warm (May through September) versus cool (October through April).
8 Bunk space density = number of cows/number of headlocks × 100.If a pen had a rail at the feed bunk, a headlock was considered = 61 cm.
be mindful of the number of observations, particularly for categorical variables, and the range observed for continuous variables.

Pen-Level Management Variables
Stall Stocking and Bunk Space Density.Stall stocking density was not associated with any outcomes of interest.Feed bunk space density was positively associated with DI (close-up primiparous) as well as with the prevalence of elevated postpartum NEFA concentrations (close-up multiparous), and negatively associated with PRFS (fresh multiparous).Nordlund et al. (2006) noted that bunk space is more likely to be an important risk factor for ketosis than stall stocking density, as stall stocking density typically does not account for the risk of decreased DMI.In a prospective study of 67 farms in Michigan, Cameron et al. (1998) observed that herds with suboptimal prepartum feed bunk management (bunk space <30 cm/cow, or bunk space was 30 to 60 cm/cow and the diet was feed restricted) had increased incidence of displaced abomasum.Contrary to our results, Deming et al. (2013) reported that milk yield was positively associated with an increase in feed bunk space in 13 herds using automatic milking systems.In an observational study, Sova et al. (2013) observed that an increase in feed bunk space was associated with a decrease in milk fat percentage, whereas stall stocking density did not remain in any of the milk yield, milk component, or DMI models.
Stall stocking density and feed bunk density do not always correspond with one another, due to differences in the number of rows within a pen.We did not account for the number of rows within the pen when assessing stall stocking density.Most controlled research trials evaluating stall stocking or bunk space density have only assessed behavioral differences among cows (DeVries et al., 2005;Huzzey et al., 2006;Hill et al., 2009;Proudfoot et al., 2009;Lobeck-Luchterhand et al., 2015).
Water Space.Water space was negatively associated with the prevalence of elevated BHB concentrations (far-off multiparous) and positively associated with the prevalence of elevated postpartum NEFA concentrations (close-up multiparous).To our knowledge, no controlled research trials have investigated the effects of water space on production and performance outcomes.In an observational study evaluating the highest-producing group of cows, Sova et al. (2013) observed a positive relationship between increased water space per cow and test day milk yield, FCM, and ECM; however, a negative relationship was observed with MUN and milk protein.Providing ample access to water, as well as feed and resting space, can be crucial to health and performance (Chebel et al., 2016), although no research has demonstrated the importance of water space during the transition cow period.
Commingling.Commingling primiparous and multiparous cows was associated with increased prevalence of elevated BHB concentrations (fresh primiparous), decreased WK4MY (fresh primiparous), and decreased prevalence of elevated Hp concentrations (fresh multiparous).Boyle et al. (2013) did not observe milk production (yield or component) differences, serum cortisol concentration differences, or differences in BCS or BW loss during the fresh period between primiparous cows that were commingled or not commingled with multiparous cows during the close-up dry cow period.
Feeding and Feed Pushup Frequency.Increased feeding frequency was associated with decreased prevalence of elevated BHB concentrations (fresh primiparous and multiparous), decreased DI (fresh multiparous), and increased prevalence of elevated Hp concentrations (close-up primiparous).Feed is consumed more evenly after each feed delivery and cows tend to increase the amount of time spent feeding when fed >1×/d (DeVries et al., 2005;Mäntysaari, 2006).Feed sorting is decreased with an increase in feeding frequency (DeVries et al., 2005;Sova et al., 2013), which results in a more consistent diet being consumed throughout the day.An increase in feeding frequency can promote increase in DMI (Sova et al., 2013;Hart et al., 2014;Miller-Cushon and DeVries, 2017), reduce the risk of subclinical acidosis (Plaizier et al., 2008), and can result in less sorting against long particles (DeVries et al., 2005;Endres and Espejo, 2010).Increasing DMI can reduce the extent of NEB during early lactation and therefore reduce blood NEFA concentrations.Severe negative energy balance can result in health disorders such as ketosis, displaced abomasum, and metritis (Grummer et al., 2004); however, to our knowledge, studies evaluating the effects of increased feeding frequency on blood metabolites and health disorders have not been evaluated during the transition cow period.
An increase in pushup frequency was associated with increased prevalence of elevated Hp concentrations (fresh primiparous), increased DI (fresh primiparous), and increased PRFS (far-off multiparous).Scientific evidence does not exist to suggest that an increase in pushup frequency results in increased DMI (DeVries, 2017); however, Bach et al. (2008) observed herds that pushed up feed produced more milk (3.9 kg/d) than herds that did not push up feed.In our data set, we did not observe any herds that never pushed up feed; herds pushed up feed at least 3×/d in the close-up and fresh periods (Kerwin et al., 2022a).It is plausible that herds may increase feed pushup frequency to stimulate eating and improve DMI and milk as a way of compensating for poor forages; however, data have shown that feed delivery is the primary stimulus for eating and not feed pushup or milking (DeVries, 2017).
Penn State Particle Separator.The proportion of particles on the PSPS 19-mm sieve was negatively associated with the prevalence of elevated Hp concentrations (far-off multiparous) and DI (far-off multiparous), and positively associated with WK4MY (closeup primiparous) and the prevalence of elevated BHB concentrations (fresh primiparous).Most of the work assessing particle size has evaluated the effects on sorting, rumination, and DMI, particularly comparing longer to shorter particles.There is evidence of decreased milk yield and components when sorting against long particles (Sova et al., 2013;Miller-Cushon and DeVries, 2017;Coon et al., 2018).Coon et al. (2019) observed that an increase in particle size often leads to sorting and a less stable rumen environment, putting the cow at a greater risk of subacute ruminal acidosis.Havekes et al. (2020) observed an increase in prepartum DMI, less sorting against longer particles, and lower postpartum BHB concentration for cows consuming a prepartum diet with a short straw length (chopped with a 2.54-cm screen) compared with a long straw length (chopped with a 10.16-cm screen).We did not account for the size of the particles found on the 19-mm sieve, which could confound our results.It has been recommended to chop hay or straw so that particles are <4 cm (Nydam et al., 2017); therefore, most of the herds enrolled in this study may have chopped the hay or straw to an adequate size to reduce sorting and prevent postpartum health problems and maximize milk yield.
PeuNDF240.Physically effective uNDF240 was negatively associated with ME305 (fresh primiparous and multiparous) and positively associated with PRFS (fresh primiparous).Evaluating peuNDF240 in diets is a relatively new concept; therefore, there is limited research evaluating the effects during the transition cow period.Our results are in agreement with previous studies (Grant et al., 2018;LaCount, 2019) indicating that as peuNDF240 increases, intake may be limited and milk yield may be compromised.We did not observe an association between peuNDF240 and the prevalence of elevated postpartum BHB or NEFA concentrations, unlike LaCount (2019), who observed greater NEFA and BHB concentrations for cows fed a high-fiber fresh diet (12.2% uNDF240; 23.2% peNDF) compared with a low-fiber fresh diet (9.5% uNDF240; 21.6% peNDF).Use of peuNDF240 may be more beneficial than peNDF or uNDF240 alone, as it accounts for uNDF240 and physically effective fiber, and has been demonstrated to be a better predictor for DMI and ECM than uNDF240 for typical diets fed in the northeastern United States (Grant et al., 2018).Contrary to our hypothesis, we did not observe an association between peuNDF240 and the prevalence of elevated Hp concentrations.LaCount (2019) observed an interaction between treatment and parity group for Hp such that second-lactation cows had lower Hp concentrations when fed the high-fiber fresh diet compared with the low-fiber fresh diet; however, treatment differences were minor (0.23 vs. 0.17 g/L; P = 0.15).Feeding a high-fiber diet, with adequate particle size (<4 cm; Nydam et al., 2017), may help reduce sorting and subacute ruminal acidosis (Coon et al., 2019), reducing the risk of negatively affecting the gut barrier integrity and systemic inflammation (Horst et al., 2021).
Total Fermentable Carbohydrates and Forage NDF.Total fermentable carbohydrates were negatively associated with the prevalence of elevated prepartum NEFA concentrations (close-up multiparous) and positively associated with the prevalence of postpartum NEFA concentrations (close-up primiparous) and Hp concentrations (fresh multiparous), and WK4MY (close-up primiparous).Similar to our results, feeding more starch and energy during the prepartum period has been demonstrated to reduce prepartum NEFA concentrations and increase postpartum NEFA concentrations (Janovick et al., 2011;Mann et al., 2015), although milk production effects were not observed.Cows are at risk of subacute ruminal acidosis in the early postpartum period as a result of the sudden increase in fermentable total carbohydrates (Dann, 2017).Gastrointestinal barrier dysfunction can occur as a result of subacute ruminal acidosis (Emmanuel et al., 2008;Khafipour et al., 2009;Gott et al., 2015), stimulating an inflammatory response.
Forage NDF was negatively associated with the prevalence of elevated postpartum NEFA concentrations (fresh multiparous).In contrast, Piantoni et al. (2015) reported lower postpartum NEFA concentrations for multiparous cows fed a lower forage NDF (20%) diet compared with a higher forage NDF (26%) diet during the fresh period.
The interaction between total fermentable carbohydrates and forage NDF was associated with PRFS (far-off multiparous), DI (fresh primiparous), WK4MY (fresh multiparous), and ME305 (close-up primiparous).Limited studies have evaluated or reported total fermentable carbohydrates and forage NDF, and, to our knowledge, no studies have evaluated the interaction between total fermentable carbohydrates and forage NDF on transition cow health and performance outcomes.
MP. Metabolizable protein was negatively (far-off multiparous) and positively (close-up and fresh multiparous) associated with PRFS.Evaluating MP supply in transition cow diets is relatively new; previous studies have focused on CP, RDP, and RUP (Lean et al., 2013).Studies evaluating CP, RDP, or RUP will not be discussed, because only a weak relationship with MP exists (Dann, 2017).To our knowledge, only one study has investigated the relationship between MP and reproductive outcomes.Underwood et al. (2022) observed improved follicular development at 72 DIM with increased MP supply during the close-up dry period, although the sample size was small.Our results correspond with current recommendations; feeding less MP far-off compared with close-up and ensuring that adequate MP is fed in the close-up period (90 to 100 g of MP/kg of DM) to maintain protein reserves during the transition period (Van Saun and Sniffen, 2014).
ME. Metabolizable energy was positively associated with the prevalence of elevated BHB concentrations (close-up multiparous) and WK4MY (fresh primiparous) and negatively associated with DI (close-up primiparous), ME305 (close-up multiparous), and PRFS (close-up multiparous).Several studies have reported reduced BHB concentrations in multiparous cows fed a controlled energy diet during the close-up period compared with a high-energy diet (Janovick et al., 2011;Mann et al., 2015;Richards et al., 2020).Janovick et al. (2011) 2001) model] in a study by Richards et al. (2020) and reported greater BHB concentrations in cows fed the high-energy close-up diet, but differences were not observed in milk yield.

Herd-Level Management Variables
Vaccination.Routinely vaccinating cows in the calving pen was associated with increased DI (primiparous and multiparous) and decreased WK4MY (multiparous).Vaccinating cows stimulates an acute phase protein response (Stokka et al., 1994;Cooke and Arthington, 2013).Transition cows are already in a state of immune dysregulation (Sordillo and Raphael, 2013;Vlasova and Saif, 2021), and further insults may make the transition cow more susceptible to health disorders and disease (Sordillo and Raphael, 2013).Immune activation requires repartitioning glucose from the mammary gland, which could result in reduced milk production (Kvidera et al., 2017).Our results support the recommendation to avoid vaccinating in the calving pen (Guterbock, 2004).
Maternity Pen.The use of a maternity pen was associated with decreased 21-d PR (primiparous).Although limited work has evaluated moving cows around the time of parturition (Cook and Nordlund, 2004;Proudfoot et al., 2013), to our knowledge, no research has evaluated long-term production effects.Field investigations have observed elevated NEFA concentrations and an increased risk of ketosis and displaced abomasum for cows that were on a maternity pack for 3 or more days compared with cows that calved within 2 d of being on the maternity pack.It is plausible that moving cows a few days before parturition could increase stress, due to a new environment and new social interactions (Arthington et al., 2003;Cook and Nordlund, 2004), stimulate an acute phase protein response, increase the risk of retained placenta, metritis, or both (Huzzey et al., 2009), and increase the risk of poor reproductive performance (Huzzey et al., 2015;Kerwin et al., 2022b).Further research needs to evaluate short-and long-term effects of moving a cow a few days before parturition versus when parturition is imminent.
Pen Moves.Increased prepartum pen moves were associated with increased prevalence of elevated Hp concentrations (primiparous), and increased postpartum pen moves were associated with decreased ME305 (primiparous).Pen moves have been associated with increased negative social interactions or avoidance behaviors (Kondo and Hurnik, 1990;Grant and Albright, 2001;Cook and Nordlund, 2004).Social behavior can play a critical role in the health of the transition cow, such that cows diagnosed with severe metritis engaged in fewer aggressive interactions, had depressed prepartum DMI, and produced less milk during the first 3 wk of lactation, compared with healthy cows (Huzzey et al., 2007).In a commingled pen, primiparous cows tend to be the low-rank or subordinate cow and may exhibit more avoidance behaviors than multiparous cows (Cook and Nordlund, 2004).Although Hp concentrations were not observed by Huzzey et al. (2007), metritis has been significantly associated with elevated Hp concentrations (Huzzey et al., 2009;Kerwin et al., 2022b).Increased Hp concentrations have also been observed in cows and calves due to a stressful event, such as transportation (Arthington et al., 2003;Lomborg et al., 2008).Limited research has evaluated transition cow postpartum pen moves on milk yield; however, mid-lactation, multiparous cow studies have generally observed a decrease in short-term milk yield due to a pen move (Cook and Nordlund, 2004;von Keyserlingk et al., 2008).
Time Spent in Calving Pen.Moving cows out of the calving pen within 8 h after parturition was associated with decreased prevalence of elevated postpartum NEFA (primiparous and multiparous), BHB (primiparous), and Hp (primiparous) concentrations, and increased 21-d PR, PP, and PRFS; however, we observed increased prevalence of elevated BHB concentrations for multiparous cows.To our knowledge, previous research has not evaluated the time spent in the calving pen on postpartum health, blood biomarkers, milk yield, or reproductive performance.Our results indicate that it may be beneficial to move the after calving so the cow can resume being a herd animal (Cook and Nordlund, 2004).Previous work has demonstrated that cows prefer to be secluded in the hour before and after calving but will rejoin cows in the pen afterward (Proudfoot et al., 2014).It is important to note that not all herds used a single-cow calving pen; some herds used a maternity pen where cows would calve among others.It is plausible that the longer a cow remains isolated from the herd, the more stressed she will become, which could negatively affect eating, putting the cow at a greater risk of increased negative energy balance and increased mobilization of NEFA.An increase in NEFA concentrations has been associated with poorer reproductive performance (Santos and Staples, 2017;Roche et al., 2018;Cardoso et al., 2020).
Time Locked Up in Fresh Pen.Locking up cows for <1×/d for <1 h was associated with lower DI compared with herds that had cows locked up daily for <1 h.To our knowledge, previous research has not evaluated the amount of time locked up in the fresh pen for health checks on postpartum health, blood biomarkers, milk yield, or reproductive performance.It is plausible that herds that had cows locked up more frequently did a more thorough fresh cow examination and identified more sick cows, potentially resulting in higher DI.Alternatively, herds that did not have cows locked up daily may have a lower DI due to (1) underreporting and missing cases or because (2) the herd truly has a low DI and, therefore, the herd does not do daily fresh cow health checks to minimize disruptions to the cow's schedule.Our results do not advocate against the current recommendation of locking cows up for <1 h/d to minimize daily time-budget disruptions (Nordlund, 2009); however, research should investigate this practice further.

Strengths and Limitations
A strength of this study includes evaluating management and nutritional factors during the transition period on 72 commercial farms.By observing management and nutritional factors on commercial farms, we are able to identify variables that are associated with outcomes without minimizing potential stressors, as is often done in controlled research trials.Although we attempted to evaluate these factors accurately, there are limitations to the study.We believe that our conservative statistical approach with removing influential observations was a strength; however, for categorical variables, this may have led to a small number of observations within a category, making it difficult to interpret the data.Additional data points within these categories would enhance the analysis.We had a reduced number of observations for the pen-level analysis due to missing data (far-off primiparous) or too few cows within the pen.It is difficult to know whether the ranges in the exploratory variables assessed are representative of what is observed in the industry, as limited data exists describing these variables during the same time frame of our data collection.However, we sampled a variety of large farms feeding different nutritional strategies, which gives us confidence that our data set is representative of large commercial farms within the northeastern United States.The measurements acquired for the management and nutritional variables were a snapshot of the current herd practices and pen demographics.There is inherent variation in measuring the variable once, as the measurement may change from day to day.We analyzed each period separately, as periods may be confounded.Farms may have similar management styles across the different periods (e.g., similar or the same diet and particle size between the far-off and close-up period, similar feed pushup frequencies, etc.); therefore, it is difficult to differentiate where the management variable has the most influence and whether it truly does have an influence during the period where we identified an association.As an example, we observed a negative association between feed pushup frequency during the close-up and fresh periods with WK4MY in the univariable analysis; however, the proportion of pens that had feed pushed up ≥5×/d was very similar between periods (Kerwin et al., 2022b), making it difficult to differentiate the management factor between the 2 periods.With the exception of total Kerwin et al.: TRANSITION MANAGEMENT AND NUTRITION ASSOCIATIONS fermentable carbohydrates and forage NDF, we did not assess interactions between variables.The summation of multiple management stressors may have a more pronounced effect on health and postpartum performance than any one management factor alone (Campbell and Grant, 2016); however, our evaluation identifies the management and nutritional factors that have the most pronounced associations with postpartum outcomes without minimizing other potential stressors.We relied on herd personnel for recording health disorders; although we tried to standardize disorder definitions, interpretation may vary within and between farms, which may have contributed to potential reporting issues.Although we evaluated PSPS, we did not evaluate the size of the particles on the top sieve.The particle size may have a greater effect on our observed outcomes compared with just evaluating the proportion of particles on the top screen, although we did not evaluate the TMR refusals to determine the extent of sorting in this study; this should be considered and investigated in the future.Water quality was not assessed, which may be another confounder when evaluating water space.Although plots were visually inspected for quadratic effects, quadratic effects were not evaluated due to few observations at the tail-ends of our range for parameters.Quadratic effects for nutrient parameters should be investigated in the future.In addition, it is plausible that interactive effects between nutrient parameters may occur across time periods (e.g., interactive effect between fermentable total carbohydrates fed during the close-up and fresh periods).Having individual or pen-level DMI would have provided crucial information with regard to energy balance.We also did not account for the amount of time spent in each of the pens, particularly in the fresh period.It has been demonstrated that DMI may be limited depending on the amount of time spent on a diet and the forage NDF concentration (LaCount, 2019).Because this was a pen-and herdlevel study, we were not able to account for cow-level factors, such as BCS, which may be associated with our outcomes.
This study enrolled farms in the northeastern United States.In general, these results can be applied to any dairy farm; however, they may not apply fully to dairy production systems with different diets (particularly dietary variables), housing environments, climate, or management style (e.g., robotic milking systems).Due to limited data or contradicting results, further research should evaluate short-and long-term effects of feed pushup frequency, water space, total fermentable carbohydrates, forage NDF, ME, the use of a calving pen versus a maternity pen, and time spent in the calving or maternity pen, and the amount of time locked up in the fresh pen.

CONCLUSIONS
Our results indicate that pen-and herd-level management factors are associated with metabolic-and inflammation-related biomarkers, health, milk yield, and reproductive performance.During the prepartum period, our results support increasing the proportion of particles on the 19-mm sieve of the PSPS, optimizing bunk space, and not overfeeding ME.For the fresh period, our results generally support optimizing bunk space, avoiding commingling, increasing the feeding frequency, avoiding high peuNDF240 and high total fermentable carbohydrate diets while optimizing the inclusion of forage NDF, and ensuring adequate diet ME and MP.At the herd level, our results support the practice of avoiding vaccination in the calving or maternity pen, utilizing a calving pen compared with a maternity pen, reducing the number of prepartum and postpartum pen moves for primiparous cows, and avoiding long stays (≥8 h) in the calving or maternity pen after parturition.
and multiparous cows are present in the same pen.

1
Range = the range (continuous) or number of observations (categorical) for variables remaining in the model.2 PSPS = Penn State Particle Separator (Nasco).

1Ferm
Range = the range (continuous) or number of observations (categorical) for variables remaining in the model.2Bunkspace density = number of cows/number of headlocks × 100.If a pen had a rail at the feed bunk, a headlock was considered = 61 cm.3 fermentable carbohydrates (% of DM) and forage NDF (% of DM). 6 PSPS = Penn State Particle Separator (Nasco).

Figure 1 .
Figure 1.Visual interpretation of the forage NDF (FNDF, % of DM) and total fermentable carbohydrates (ferm.CHO, % of DM) interaction associated with (A) disorder incidence for the primiparous fresh period model, (B) milk yield at the fourth week of lactation (WK4MY) for the multiparous fresh period model, (C) 305-d mature equivalent milk at the fourth test day (ME305) for the primiparous close-up period model, and (D) pregnancy risk to first service (PRFS) for the multiparous far-off period model, assuming MP = 76 g/kg of DM, feed pushup frequency is <5×/d, and calving season is cool (October through April).
effective undigested NDF measured at 240 h of in vitro fermentation, calculated by multiplying the proportion of TMR above the 4-mm sieve by the diet content of undigested NDF after 240 h of fermentation, on a DM basis.

Table 1 .
Restricted maximum likelihood parameter estimates for the prevalence of elevated prepartum nonesterified fatty acid concentration (≥0.17 mmol/L) close-up period multiparous cow pen-level analysis (n = 73) for a 72-farm prospective cohort study in the northeastern United States 1 Range = the range (continuous) or number of observations (categorical) for variables remaining in the model.2Ferm.CHO = total fermentable carbohydrates.3

Table 3 .
Kerwin et al.:TRANSITION MANAGEMENT AND NUTRITION ASSOCIATIONS Restricted maximum likelihood parameter estimates for the prevalence of elevated postpartum BHB concentration (≥1.2 mmol/L) pen-and herd-level analysis for a 72-farm prospective cohort study in the northeastern United States

Table 4 .
Kerwin et al.:TRANSITION MANAGEMENT AND NUTRITION ASSOCIATIONS Restricted maximum likelihood parameter estimates for the prevalence of elevated postpartum haptoglobin concentration (≥0.45 g/L) pen-and herd-level analysis for a 72-farm prospective cohort study in the northeastern United States

Table 5 .
Kerwin et al.:TRANSITION MANAGEMENT AND NUTRITION ASSOCIATIONS Restricted maximum likelihood parameter estimates for disorder incidence (incidence of one of more displaced abomasum, clinical ketosis, metritis within 30 DIM; calculated by parity within herd) within 30 DIM pen-and herd-level analysis for a 72-farm prospective cohort study in the northeastern United States

Table 10 .
No explanatoryKerwin et al.: TRANSITION MANAGEMENT AND NUTRITION ASSOCIATIONS

Table 6 .
Restricted maximum likelihood parameter estimates for the milk yield at the fourth week of lactation (WK4MY, kg/d) pen-and herdlevel analysis for a 72-farm prospective cohort study in the northeastern United States

Table 7 .
Restricted maximum likelihood parameter estimates for the 305-d mature equivalent milk at the fourth test day (kg) pen-and herdlevel analysis for a 72-farm prospective cohort study in the northeastern United States a,bMeans within parity group with different superscripts differ (P < 0.05) as determined by an F-test. 1 Range = the range (continuous) or number of observations (categorical) for variables remaining in the model.2Ferm.CHO = total fermentable carbohydrates. 3 FNDF = forage NDF.4

Table 9 .
Kerwin et al.:TRANSITION MANAGEMENT AND NUTRITION ASSOCIATIONS Restricted maximum likelihood parameter estimates for the probability of pregnancy (PP) herd-level analysis (n = 134) for a 72-farm prospective cohort study in the northeastern United States 1

Table 10 .
Restricted maximum likelihood parameter estimates for the pregnancy risk to first service pen-and herd-level analysis for a 72-farm prospective cohort study in the northeastern United States peuNDF240 = physically effective undigested NDF measured at 240 h of in vitro fermentation, calculated by multiplying the proportion of TMR above the 4-mm sieve by the diet content of undigested NDF after 240 h of fermentation, on a DM basis.
1 Range = the range (continuous) or number of observations (categorical) for variables remaining in the model. 2 3Ferm.CHO = total fermentable carbohydrates. 4 FNDF = forage NDF Kerwin et al.: TRANSITION MANAGEMENT AND NUTRITION ASSOCIATIONS