Persistent effects of neonatal diarrhea on Holstein dairy cow lactation performance through multiple parities

Abstract

Calf diarrhea, one of the most common stresses causing by heifer rearing, affects most Chinese dairy herds, yet its long-term consequences on lactation performance remain poorly quantified. This study explored whether calf diarrhea has a long-term effect on the lactation performance and Dairy Herd Improvement-related indices of dairy cows and its influence during different lactation periods. This retrospective study included a total of 1,907 Holstein dairy cows from a large-scale farm in northern China (700 in the diarrhea group and 1,207 in the healthy group). A mixed effects model and mediation analysis were used to evaluate the effects of calf diarrhea on growth performance, milk yield, milk composition, and other lactation parameters at different parities and lactation periods. The results revealed that the negative effect of calf diarrhea on lactation performance was persistent and varied by parity. The 305-d milk yield and peak milk yield of cows at each parity in the diarrhea group were significantly lower than those in the healthy group (P < 0.05). Growth performance, such as withers height (median: 134.00 vs. 135.00 cm), was more significantly affected in the first parity (P < 0.001). With increasing parity, the difference in body condition scores narrowed, but the loss of milk production continued to accumulate. For the second and third parities, cows with calf diarrhea presented significant reductions in 305-d milk yield, with decreases of 287.36 kg and 1,086.76 kg in the median comparison, respectively (P < 0.05). The corrected milk yield and milk fat percentage during the early postpartum and early lactation periods decreased more significantly in multiparous cows, and the daily milk production in the middle and late lactation periods remained consistently lower than that in the healthy group (P < 0.05). Mediation analysis confirmed that milk production played a key mediating role in the pathway by which diarrhea affected the milk fat percentage. Our research revealed that calf diarrhea is a critical early-life health challenge, with consequences extending beyond the calf period to significantly affect lactation potential. Our findings underscore the need for early-life interventions to mitigate lactation losses in high-parity herds.

With calf diarrhea, the growth performance of the first parity is more significantly affected, but milk production loss accumulates with parity. The corrected milk production and milk fat percentage of multiple parities decreased more significantly during the early postpartum and early lactation periods, and milk production can serve as a key mediator for the impact of diarrhea on the milk fat percentage.

Introduction

Calves constitute both the biological foundation and economic catalyst for sustainable pasture development, as their developmental trajectory directly determines a herd’s genetic potential (Diao et al. 2017). Healthy prepubertal growth has lifelong impacts on production performance (Overton and Dhuyvetter 2020; Zhang et al. 2022). Notably, the replacement heifers’ period, particularly the calf period, represents a critical window for gastrointestinal, immune system, and mammary parenchyma development, during which health conditions may have long-term effects on the lactation performance of cows (Vailati-Riboni et al. 2018; Du et al. 2023). The concurrent immaturity of the digestive and immune systems during this phase significantly elevates susceptibility to enteric disorders, especially diarrhea (Aghakeshmiri et al. 2017; Abuelo et al. 2021). Hence, the calfhood period of dairy cows has the highest morbidity and mortality rates, inducing wide welfare concerns and significant economic losses for the farm (Urie et al. 2018).

Epidemiologic studies have demonstrated that diarrhea is one of the most common causes of morbidity and mortality in calves (Urie et al. 2018). Calfhood diarrhea, a prevalent neonatal disorder and common stress in heifer raising affecting 29.9% to 39.0% of dairy calves in China, has been extensively studied for its acute impacts on growth performance and mortality rates (Kim et al. 2021; Schinwald et al. 2022). Existing evidence suggests that poor health during calfhood may limit subsequent lactation performance, potentially through altered metabolic processes and immune activation (Amin and Seifert 2021). Recent studies have shown that calves with diarrhea may have reduced milk production during their first lactation (Yue et al. 2021; Crannell and Abuelo 2023); calves that experienced preweaning diarrhea between 15 and 22 d of life produce 422 kg less milk in 305-d milk yield (Goh et al. 2024). Emerging evidence implicates gastrointestinal pathologies—including mucosal damage and microbial imbalance—as key drivers of malabsorption in diarrheic calves. Concurrent stress-induced immune dysregulation may disrupt mammary tissue homeostasis through persistent inflammatory states, compromising both immunological maturation and lactation performance differentiation (Abuelo et al. 2021; Goh et al. 2024). However, whether diarrhea during the calf stage has a long-term parity-dependent effect on the lactation performance of dairy cows remains underexplored. Moreover, it is necessary to further investigate whether diarrhea during the calf period influences lactation performance and dairy herd improvement (DHI)-related indices across different lactation stages.

This study investigated the longitudinal associations between calfhood diarrhea, parity, lactation period, and lactation performance metrics, including milk yield and lactation days, and their modulation within DHI evaluation frameworks. The aim of this study was to quantify the long-term effects of diarrhea during the calf period on lactation performance, including the effects on different lactation stages. We hypothesized that calfhood diarrhea has a long-term negative impact on lactation performance of Holstein cows, including milk yield and milk composition.

Materials and methods

This study was conducted in accordance with the recommendations of the Administration of Affairs Concerning Experimental Animals (Ministry of Science and Technology, China, revised in 2004). The protocol was approved by the Institutional Animal Care and Use Committee of Northwest A&F University.

Animals and data collection

A retrospective cohort study was conducted using herd records from a large dairy farm located in Hebei Province, China (115.40°E, 37.23°N). This farm intensively farmed (unified management and the application of a standardized feed formulation) and strictly recorded Holstein dairy cows’ management data, including health state (e.g. diarrhea), parity, birth season, withers height, weight, body condition score, milk yield, and milk components. The farm owners were willing to share their data. The data used for enrollment in this study ranged from October 2, 2018, to November 17, 2024, and included a total of 1,907 Holstein calves, with birth dates ranging from October 2018 to November 2021, and 26,860 records, with a collection frequency of approximately once a month. Briefly, the reason for selecting these 1,907 dairy cows was that they possess complete, officially recognized DHI data, parity, condition score, birth weight, weaning weight, withers height, etc. and individuals with incomplete records were excluded from the analysis.

The classification criteria for lactation stages were as follows: early postpartum (<15 d), early lactation (16–100 d), middle lactation (101–200 d), and late lactation (>200 d) (Dhom et al. 2018; Nigussie 2018; Arora et al. 2019).

Coding of disease status

The symptom of interest in this study was diarrhea in calves. All the recorded diarrheal calves were cured, excluding samples and records of dead dairy cows. The age of onset of diarrhea ranged from 0.03 to 5.53 mo, with a median (IQR) = 0.30 (0.16–0.39), and the duration of treatment ranged from 2 to 32 d, with a median (IQR) = 8 (6–9). A total of 700 cows exhibiting diarrhea during the calf period were included in this study. Among these, 273 dairy cows had one complete lactation, 340 dairy cows had two complete lactations, and 87 dairy cows had three complete lactations. A total of 1,207 cows without calf diarrhea and with healthy status during the calf period were included in this study. A total of 866 dairy cows had one complete lactation, 269 dairy cows had two complete lactations, and 72 dairy cows had three complete lactations.

Difference analysis

Normal distribution tests were conducted on all the obtained data, and none of them presented a normal distribution (Table S1). All difference analyses were conducted via the Mann–Whitney U-test for nonparametric tests. Repeated measures analysis was performed via a general linear model to investigate the changes in milk yield and lactation duration. The above analyses were all conducted via IBM SPSS Statistics 27 software and GraphPad Prism 9.5 for figure generation.

Linear mixed effects model construction

The associations between early-life diarrhea and subsequent lactation performance were evaluated via linear mixed-effects models (LMMs). To build the multivariable model, we considered both statistical significance and biological plausibility. Initially, univariable linear regression was performed to screen candidate predictors (including healthy state, birth weight, weaning weight, weaning age, parity, and birth season as fixed effects) to be carried forward to multivariable modelling. Variables with a statistical association (P < 0.20) in the univariable analysis were considered for the multivariable mixed-effects model. The primary models included cow-level random intercepts and random slopes for measurement time to allow individual-specific baseline and temporal trends; models were compared with a random-intercept-only model using likelihood-ratio tests (REML refitted to ML for comparison) and AIC, and unstructured covariance provided the best fit (Table S2). Two outcomes were analyzed: 305-d milk yield and peak milk yield. Prior to model fitting, biologically implausible outliers were removed based on the interquartile range (IQR) method. Continuous variables were visually checked for linearity using LOESS smoothing, and multicollinearity was assessed via generalized variance inflation factors (GVIFs) to ensure that the adjusted GVIF of all predictors in the model is less than 2. No variables were excluded due to collinearity, allowing us to retain all biologically relevant predictors in the final model without compromising the stability of the estimates. The normality of the residual errors was evaluated visually using histograms. While minor deviations from a perfect normal distribution were observed, the large sample size (n = 15,840) ensured the robustness of the LMM estimates against valid normality violations (Figure S1A and B) (Schielzeth et al. 2020). The homoskedasticity of the residual errors were evaluated visually using scatterplots of residuals versus fitted values. Visual inspection confirmed that the model assumptions were reasonably met (Figure S1C and D). Marginal and conditional R² values and intraclass correlation coefficients (ICCs) are reported. Analyses were performed in R via lme4, lmerTest, MuMIn, r2glmm, and performance. The following model was fitted:

$$\mathit{Yij}\mathrm{ }=\mathrm{ }{\beta }_0+{\beta }_1{d}_{ij}+{\beta }_2{bw}_{ij}+{\beta }_3{ww}_{ij}+{\beta }_4{wa}_{ij}+f\left(p_{ij}\right)+b_{0j}+{\mathrm{b}}_{ij}{t}_{ij}+{\epsilon }_{ij}$$

here Yij represents the 305-d milk yield or peak milk yield for cow j at time i; bwij, waij, and wwij represent birth weight, weaning age, and weaning weight, respectively; f(p) denotes a natural spline function with 3 degrees of freedom applied to parity; and b0j and b1j represent the cow-specific random intercept and slope for time, respectively, which are assumed to follow a multivariate normal distribution with mean zero and unstructured covariance. Model estimation was performed via REML with the lmer() function from the lme4 package in R.

Power analysis

To examine the statistical power to detect the effect of early-life diarrhea, a Monte Carlo simulation-based power analysis was performed under the fitted LMM. The final model was refitted via maximum likelihood (ML), and 1,000 datasets were generated via the simulate() function in lme4. For each simulated dataset, the model was refitted, and the P-value for the diarrhea effect was extracted. Power was calculated as the proportion of simulations in which the diarrhea effect reached statistical significance (α = 0.05). Power estimates stabilized at approximately 0.80–0.85 when the number of cows approached the full dataset (about 1,800 cows). With the current sample size (n = 691 for the diarrhea group, n = 1,120 for the healthy group), the study achieved a statistical power of 0.83 and 0.84 to detect the observed difference in 305-d milk yield and peak milk yield, respectively. The results indicated that the sample provided adequate power to detect the observed effect of diarrhea on milk yield.

Lactation curve construction

The daily milk production of each parity of diarrheal and healthy dairy cows was recorded for 305 d to construct the lactation curve. The data were fitted via the locally weighted regression (loess) method to show the variation trend of lactation volume with the number of lactation days via the R package “dplyr” (Cavani et al. 2024).

Mediation analysis

The mediation analysis was performed with the “lavaan” package (Ballen and Salehi 2021), including regression, variances, and defined parameters, and the standardized coefficient was evaluated. The analytical procedure involved estimating three parameters: the effect of health state on milk yield, the effect of milk production on the milk fat percentage, and the direct effect of health state on the milk fat percentage. This was followed by the quantification of milk production’s mediating effect on the health state–milk fat percentage association. To ensure robust statistical inference and account for the non-normal distribution of the indirect effect, bias-corrected bootstrapping with 1,000 replications was employed. Results were reported as unstandardized regression coefficients (Estimates) with standard errors (SE) and 95% confidence intervals (95% CI). Mediation effects were considered statistically significant if the 95% CI excluded zero.

Results

Descriptive results

A total of 1,159 cows in the first parity were included; 24% of them experienced diarrhea in calves, and the other 76% of them remained healthy during the calf period (Figure 1A). These two groups presented similar distributions of withers height and condition scores, but diarrheal cows presented lower peak milk production, slightly delayed peak milk days and a lower 305-d milk yield (Figure 1B–F). A total of 609 cows in the second parity were included; 56% of them experienced diarrhea in calves, and the other 44% of them remained healthy during the calf period (Figure 1G). Diarrhea cows assumed more distributed withers height and condition score, and the impact on lactation performance decreased (Figure 1H and I). A total of 159 cows in the third parity were included, 55% of whom experienced diarrhea in calves and 45% of whom were healthy individuals (Figure 1M). The withers height and condition score were similar in these two groups, but diarrheal cows still presented lower peak milk contents, 305-d milk yields and shorter peak milk days (Figure 1N–R).

Figure 1.

Basal information of cows of different parities in the diarrhea and healthy groups. (A) Pie chart depicting the distribution of the first parity among individuals with diarrhea (dark purple) and healthy individuals (light blue), along with their respective counts and percentages. (B–F) The distributions of withers height (B), condition score (C), peak milk yield (D), peak milk yield (E) and 305-d milk yield (F) of cows in the first parity between diarrhea and health. (G) Pie chart depicting the distribution of the second parity among individuals with diarrhea and healthy individuals, along with their respective counts and percentages. (H–L) Distributions of withers height (H), condition score (I), peak milk yield (J), peak milk duration (K) and 305-d milk yield (L) of cows in the second parity interval between diarrhea and health. (M) Pie chart depicting the distribution of third parity among individuals with diarrhea and healthy individuals, along with their respective counts and percentages. (N–R) The distributions of withers height (N), condition score (O), peak milk yield (P), peak milk days (Q) and 305-d milk yield (R) of cows in the third parity interval between diarrhea and health.

In further exploration of lactation performance, regardless of the number of parities, the distribution of lactation time between the diarrhea group and the healthy group was more consistent (Figure 2A, D, and G). Milk production was greater in the healthy group and tended to be more concentrated, whereas milk production in the diarrhea group was relatively discrete (Figure 2B and C), and lactation in the second parity group was similar to that in the first parity group (Figure 2E and F). In the third parity, the sample size was small, but the trend still showed that healthy cows produced more milk than did cows with diarrhea (Figure 2H and I). Diarrhea had a negative effect on milk performance and milk yield in cows and was consistent across multiple parities. The normality test found that all the variables involved did not conform to the normal distribution (Figures S2–S4).

Figure 2.

Distribution plots for lactation days and milk yield across different parities, comparing individuals with diarrhea and healthy individuals. (A) Lation days of the first parity of cows with diarrhea and healthy health. (B and C) Milk yield of the first parity of cows with diarrhea (B) and healthy cows (C). (D) The lactation days of the second parity of cows with diarrhea and healthy conditions. (E and F) Milk yield of the second parity of cows with diarrhea (E) and healthy cows (F). (G) The lactation days of the third parity of cows with diarrhea and healthy health. (H and I) The milk yield of the third parity of cows with diarrhea (H) and healthy cows (I).

Differences in growth and lactation performance between cows with diarrhea and healthy cows

In the first parity, compared with those of healthy individuals, the weaning weight (Mann–Whitney U = 154,670, Z = 7.698, P < 0.001), withers height (Mann–Whitney U = 106,846.5, Z = 3.295, P < 0.001), peak milk yield (Mann–Whitney U = 114,868, Z = 2.084, P = 0.37), and 305-d milk yield (Mann–Whitney U = 105,161, Z = 3.911, P < 0.001) significantly decreased in dairy cows with diarrhea (Table 1), whereas the birth weight (Mann–Whitney U = 126,693, Z = 1.794, P = 0.073), condition score (Mann–Whitney U = 98,122.5, Z = 1.336, P = 0.182), and peak milk duration (Mann–Whitney U = 97,029, Z = −1.950, P = 0.051) did not significantly differ between these 2 groups (P > 0.05; Table 1). In the second parity, the cows with diarrhea also presented a significantly lower withers height (Mann–Whitney U = 33,993.5, Z = 5.963, P < 0.001), peak milk yield (Mann–Whitney U = 51,171, Z = 3.042, P = 0.002), and 305-d milk yield (Mann–Whitney U = 34,616, Z = 2.390, P = 0.017) than healthy individuals did (P < 0.05, Table 1). The peak milk duration (Mann–Whitney U = 39,789.5, Z = −2.327, P = 0.02) was significantly later in cows with diarrhea (Table 1), and there were no significant differences in birth weight (Mann–Whitney U = 42,302.5, Z = −1.594, P = 0.111), weaning weight (Mann–Whitney U = 43,230, Z = −1.160, P = 0.246) or condition score (Mann–Whitney U = 42,394, Z = −1.648, P = 0.099) (Table 1). In the third parity, there were significant differences in peak milk yield (Mann–Whitney U = 3,972.5, Z = 2.908, P = 0.004) and 305-d milk yield (Mann–Whitney U = 2,584, Z = 3.32, P < 0.001), which were lower in dairy cows with diarrhea (Table 1). There were no significant differences in the number of lactation days between diarrhea and healthy dairy cows in the first parity (Mann–Whitney U = 690,297.5, Z = 0.126, P = 0.900) and third parity (Mann–Whitney U = 3,514.5, Z = 1.325, P = 0.185) nor were there significant differences in the milk yield between diarrhea and healthy dairy cows in the first parity (Mann–Whitney U = 690,847.5, Z = 1.954, P = 0.510) and second parity (Mann–Whitney U = 75,118, Z = 0.815, P = 0.415) (Table 1). However, after diarrhea occurred in calves, dairy cows had longer lactation days in the second parity (Mann–Whitney U = 62,644, Z = −3.329, P < 0.001) and a lower milk yield in the third parity (Mann–Whitney U = 4,113, Z = 3.394, P < 0.001) (Table 1). A comparison of the changes in lactation performance of dairy cows at different parities revealed that there was no significant difference in the number of lactation days among the parities (Figure 3A). From a trend point of view, the milk production of healthy cows increased with increasing parity, while the milk production of diarrheal cows increased in the second parity and basically remained stable in the third parity (P < 0.001; Figure 3B).

Table 1.

Comparison of growth and lactation performance metrics between dairy cows with a history of neonatal diarrhea and healthy controls (total number of 1,907) in different parities from a commercial herd in Hebei, China, analyzed using the Mann–Whitney U-test.

	Group, median (IQR)		P-value
	Diarrhea	Health
1 parity	n = 273	n = 866
Birth weight, kg	38.00 (35.00–40.00)	38.45 (36.00–41.00)	0.073
Weaning weight, kg	97.50 (94.00–101.50)	101.00 (97.00–105.00)	0.001
Condition score	3.50 (3.25–3.50)	3.50 (3.25–3.50)	0.182
Withers height, cm	134.00 (132.00–137.00)	135.00 (133.00–137.00)	0.001
Peak milk, kg	58.98 (49.13–66.87)	60.12 (54.45–66.16)	0.037
Peak milk days, d	55.00 (35.25–81.75)	50.00 (36.00–69.00)	0.051
305-d milk yield, kg	10,830.96 (9,080.68–12,518.10)	11,492.13 (10,214.55–12,759.04)	0.001
2 parities	n = 340	n = 269
Birth weight, kg	38.00 (35.00–42.00)	38.00 (35.00–40.00)	0.111
Weaning weight, kg	98.00 (94.00–101.50)	97.00 (93.75–101.00)	0.246
Condition score	3.50 (3.25–3.50)	3.50 (3.25–3.50)	0.099
Withers height, cm	132.00 (130.00–135.00)	134.00 (132.00–137.00)	0.001
Peak milk, kg	60.56 (47.17–68.86)	62.77 (55.95–68.77)	0.002
Peak milk days, d	51.00 (31.00–78.00)	46.00 (28.00–66.00)	0.02
305-d milk yield, kg	10,669.12 (7,273.99–12,391.96)	10,956.48 (10,214.55–12,595.48)	0.017
3 parities	n = 87	n = 72
Birth weight, kg	38.00 (35.00–41.00)	38.00 (34.00–40.25)	0.201
Weaning weight, kg	98.00 (93.25–102.00)	97.00 (92.75–101.25)	0.32
Condition score	3.50 (3.25–3.50)	3.50 (3.25–3.50)	0.99
Withers height, cm	130.00 (128.50–132.00)	131.00 (126.50–132.00)	0.963
Peak milk, kg	55.61 (47.08–62.36)	61.26 (54.38–67.68)	0.004
Peak milk days, d	42.50 (25.00–64.75)	48.50 (32.00–69.25)	0.134
305-d milk yield, kg	10,342.99 (8,993.81–11,738.02)	11,429.75 (10,185.93–12,892.88)	<0.001
	n = 700	n = 1,207
Lactation days of the first parity, d	302.00 (284.00–342.00)	294.00 (285.00–334.00)	0.9
Milk yield of the first parity, kg	11,251.98 (9,907.91–12,666.42)	11,245.35 (10,103.12–12,566.00)	0.51
	n = 427	n = 341
Lactation days of the second parity, d	291.00 (286.00–343.00)	288.00 (285.00–318.00)	<0.001
Milk yield of the second parity, kg	12,889.33 (10,824.28–14,533.49)	12,821.48 (11,584.26–14,237.13)	0.415
	n = 87	n = 72
Lactation days of the third parity, d	288.00 (285.00–325.00)	292.00 (286.25–334.25)	0.185
Milk yield of the third parity, kg	12,061.26 (10,271.19–14,144.36)	13,539.20 (12,210.69–14,944.24)	<0.001

Figure 3.

Trends in lactation performance in diarrheal and healthy dairy cows with parity. (A) Trends in lactation days of cows with diarrhea and health conditions under different parities. (B) Trends in the milk yield of cows with diarrhea and healthy conditions under different parities. Diarrhea (n = 87), health (n = 72).

Effects of health state, lactation period, and parity on dairy herd improvement

During the early postpartum period, there were no significant differences between cows with diarrhea and healthy condition of the first parity. However, at the second and third parities, the daily milk yield (second parity: Mann–Whitney U = 43,692.5, Z = 2.428, P = 0.015; third parity: Mann–Whitney U = 12,825.5, Z = 4.991, P < 0.001) and corrected milk yield (second parity: Mann–Whitney U = 36,877, Z = 2.997, P = 0.003; third parity: Mann–Whitney U = 10,331, Z = 4.010, P < 0.001) of cows with diarrhea were lower than those of healthy cows (Table 2). Briefly, the corrected milk yield was obtained from the DHI report, which referred to the standardization of actual daily milk production to a theoretical value under the following reference conditions: “150 d in milk (DIM)” and “3.5% milk fat percentage.” Moreover, the fat-to-protein ratio (Mann–Whitney U = 35,699.5, Z = 2.292, P = 0.022) of cows with diarrhea was also lower in the second parity (Table 2).

Table 2.

Comparison of dairy herd improvement in the early postpartum period between dairy cows with a history of neonatal diarrhea and healthy controls (total number of 1,390 observations) in different parities from a commercial herd in Hebei, China, analyzed using the Mann–Whitney U-test.

	Group, median (IQR)		P-value
	Diarrhea	Health
First parity	n = 215	n = 320
Daily milk yield, kg	27.50 (21.6–31.8)	27.95 (22.90–31.70)	0.383
Milk fat percentage, %	3.79 (3.21–4.68)	3.91 (3.31–4.64)	0.669
Protein rate, %	3.44 (3.16–3.79)	3.46 (3.16–3.86)	0.484
Fat to protein ratio	1.10 (0.97–1.32)	1.13 (0.97–1.29)	0.633
Somatic cell count, ×104 cells/mL	7.00 (4.00–18.00)	9.00 (4.00–21.00)	0.206
Somatic cell score	3.00 (2.00–4.00)	3.00 (2.00–4.00)	0.256
Milk loss, kg	0.00 (0.00–0.30)	0.00 (0.00–0.50)	0.326
Milk shortage	0.00 (0.00–1.00)	0.00 (0.00–1.60)	0.317
Economic loss	0.00 (0.00–1.60)	0.00 (0.00–2.50)	0.318
Corrected milk yield, kg	18.40 (14.23–22.88)	18.90 (15.10–24.30)	0.155
Second parity	n = 213	n = 366
Daily milk yield, kg	35.50 (28.20–42.75)	37.75 (30.50–44.78)	0.015
Milk fat percentage, %	4.12 (3.39–4.89)	4.21 (3.65–4.78)	0.21
Protein rate, %	3.61 (3.26–3.97)	3.50 (3.13–3.92)	0.088
Fat to protein ratio	1.15 (0.98–1.34)	1.20 (1.04–1.35)	0.022
Somatic cell count, ×104 cells/mL	8.00 (3.00–20.75)	5.00 (3.00–15.00)	0.069
Somatic cell score	3.00 (1.00–4.00)	2.00 (1.00–4.00)	0.063
Milk loss, kg	0.00 (0.00–0.60)	0.00 (0.00–0.00)	0.12
Milk shortage	0.00 (0.00–2.08)	0.00 (0.00–0.80)	0.112
Economic loss	0.00 (0.00–3.15)	0.00 (0.00–0.15)	0.112
Corrected milk yield, kg	23.60 (16.73–30.73)	26.60 (20.33–32.00)	0.003
Third parity	n = 141	n = 135
Daily milk yield, kg	34.70 (25.85–42.55)	42.10 (35.00–47.40)	0.001
Milk fat percentage, %	4.11 (3.39–4.85)	4.18 (3.57–4.78)	0.676
Protein rate, %	3.53 (3.09–4.00)	3.43 (3.15–3.79)	0.263
Fat to protein ratio	1.15 (0.97–1.34)	1.21 (1.01–1.36)	0.19
Somatic cell count, ×104 cells/mL	9.00 (4.00–42.00)	8.00 (3.00–24.50)	0.444
Somatic cell score	3.00 (2.00–5.00)	3.00 (1.00–4.00)	0.437
Milk loss, kg	0.00 (0.00–1.35)	0.00 (0.00–0.78)	0.815
Milk shortage	0.00 (0.00–4.75)	0.00 (0.00–2.58)	0.818
Economic loss	0.00 (0.00–7.40)	0.00 (0.00–4.05)	0.817
Corrected milk yield, kg	21.90 (16.80–28.00)	27.05 (21.08–33.35)	0.001

In the first parity, the milk fat percentage (Mann–Whitney U = 1,471,612, Z = 5.194, P < 0.001), fat-to-protein ratio (Mann–Whitney U = 1,461,841, Z = 4.834, P < 0.001) and amount of corrected milk (Mann–Whitney U = 1,438,303.5, Z = 3.966, P < 0.001) were lower in cows with diarrhea than healthy cows (Table 3). In the second parity, the daily milk yield (Mann–Whitney U = 1,224,428, Z = 3.756, P < 0.001), milk fat percentage (Mann–Whitney U = 105,786.5, Z = 3.865, P < 0.001), fat-to-protein ratio (Mann–Whitney U = 1,088,802.5, Z = 5.274, P < 0.001), and corrected milk yield (Mann–Whitney U = 1,095,803, Z = 5.592, P < 0.001) of cows with diarrhea decreased significantly, whereas the protein percentage (Mann–Whitney U = 903,142, Z = −3.179, P < 0.001), somatic cell count (Mann–Whitney U = 887,801.5, Z = −3.891, P < 0.001), somatic cell score (Mann–Whitney U = 885,752, Z = −4.026, P < 0.001), milk loss (Mann–Whitney U = 925,211.5, Z = −2.794, P = 0.005), milk shortage (Mann–Whitney U = 925,278.5, Z = −2.790, P = 0.005), and economic loss (Mann–Whitney U = 925,233, Z = −2.793, P = 0.005) of cows with diarrhea increased significantly compared with those of healthy cows (Table 3). In the third parity, only the daily milk yield (Mann–Whitney U = 316,223.5, Z = 6.330, P < 0.001) and corrected milk yield (Mann–Whitney U = 258,979.5, Z = 4.193, P < 0.001) were markedly lower in cows with diarrhea (Table 3).

Table 3.

Comparison of dairy herd improvement in the early lactation between dairy cows with a history of neonatal diarrhea and healthy controls (total number of 8,204 observations) in different parities from a commercial herd in Hebei, China, analyzed using the Mann–Whitney U-test.

	Group, median (IQR)		P-value
	Diarrhea	Health
First parity	n = 1,422	n = 2,100
Daily milk yield, kg	39.80 (34.40–45.10)	40.10 (35.40–44.90)	0.223
Milk fat percentage, %	3.45 (2.99–3.93)	3.60 (3.09–4.13)	0.001
Protein rate, %	3.21 (3.01–3.42)	3.22 (3.01–3.45)	0.849
Fat to protein ratio	1.08 (0.94–1.21)	1.11 (0.97–1.25)	0.001
Somatic cell count, ×104 cells/mL	6.00 (3.00–14.00)	5.00 (3.00–13.00)	0.418
Somatic cell score	2.00 (1.00–3.00)	2.00 (1.00–3.00)	0.271
Milk loss, kg	0.00 (0.00–0.00)	0.00 (0.00–0.00)	0.196
Milk shortage	0.00 (0.00–0.00)	0.00 (0.00–0.00)	0.197
Economic loss	0.00 (0.00–0.00)	0.00 (0.00–0.00)	0.199
Corrected milk yield, kg	30.50 (24.90–37.20)	32.00 (26.00–38.50)	0.001
Second parity	n = 1,036	n = 2,185
Daily milk yield, kg	49.30 (41.70–56.40)	50.90 (44.60–56.70)	0.001
Milk fat percentage, %	3.50 (2.98–4.01)	3.63 (3.09–4.19)	0.001
Protein rate, %	3.18 (2.98–3.40)	3.13 (2.92–3.38)	0.001
Fat to protein ratio	1.10 (0.94–1.24)	1.15 (0.98–1.32)	0.001
Somatic cell count, ×104 cells/mL	6.00 (3.00–18.00)	5.00 (2.00–15.00)	0.001
Somatic cell score	2.00 (1.00–4.00)	2.00 (1.00–4.00)	0.001
Milk loss, kg	0.00 (0.00–0.70)	0.00 (0.00–0.23)	0.005
Milk shortage	0.00 (0.00–2.50)	0.00 (0.00–0.90)	0.005
Economic loss	0.00 (0.00–4.00)	0.00 (0.00–1.35)	0.005
Corrected milk yield, kg	34.70 (28.33–42.30)	37.30 (30.70–44.30)	0.001
Third parity	n = 785	n = 676
Daily milk yield, kg	50.3 (40.9–58.05)	53.55 (47.3–59.78)	0.001
Milk fat percentage, %	3.58 (3.08–4.21)	3.62 (3.06–4.18)	0.881
Protein rate, %	3.09 (2.93–3.33)	3.13 (2.94–3.34)	0.42
Fat to protein ratio	1.15 (0.99–1.33)	1.15 (0.97–1.34)	0.913
Somatic cell count, ×104 cells/mL	6.00 (2.00–16.00)	5.00 (2.00–15.00)	0.402
Somatic cell score	2.00 (1.00–4.00)	2.00 (1.00–4.00)	0.28
Milk loss, kg	0.00 (0.00–0.50)	0.00 (0.00–0.00)	0.525
Milk shortage	0.00 (0.00–1.60)	0.00 (0.00–0.00)	0.53
Economic loss	0.00 (0.00–2.50)	0.00 (0.00–0.00)	0.53
Corrected milk yield, kg	34.90 (27.20–42.60)	37.00 (30.85–44.73)	0.001

During the mid-lactation period, the daily milk yield (first parity: Mann–Whitney U = 1,909,611, Z = 4.522, P < 0.001; second parity: Mann–Whitney U = 922,086, Z = 4.568, P < 0.001; third parity: Mann–Whitney U = 170,169.5, Z = 6.992, P < 0.001) and corrected milk yield (first parity: Mann–Whitney U = 1,917,677, Z = 11.041, P < 0.001; second parity: Mann–Whitney U = 755,774, Z = 4.010, P = 0.030; third parity: Mann–Whitney U = 139,584.5, Z = 4.868, P < 0.001) of cows with diarrhea decreased from one to three parities (Table 4). However, the milk fat percentage (Mann–Whitney U = 1,878,053, Z = 4.193, P = 9.753), protein percentage (Mann–Whitney U = 1,693,554, Z = 3.753, P < 0.001), and fat-to-protein ratio (Mann–Whitney U = 1,859,694.5, Z = 9.157, P < 0.001) of cows with diarrhea significantly decreased only in the first parity, and the milk fat percentage (Mann–Whitney U = 676,010.5, Z = −2.436, P = 0.015) and protein percentage (Mann–Whitney U = 597,077, Z = −6.999, P < 0.001) of cows with diarrhea significantly increased in the second parity, whereas the protein percentage (Mann–Whitney U = 130,259, Z = 2.754, P = 0.015) of cows with diarrhea significantly increased in the third parity (Table 4).

Table 4.

Comparison of dairy herd improvement in the mid-lactation between dairy cows with a history of neonatal diarrhea and healthy controls (total number of 7,277) in different parities from a commercial herd in Hebei, China, analyzed using the Mann–Whitney U-test.

	Group, median (IQR)		P-value
	Diarrhea	Health
First parity	n = 1,422	n = 2,100
Daily milk yield, kg	39.70 (34.30–44.70)	40.50 (36.20–45.00)	0.001
Milk fat percentage, %	3.57 (3.04–4.10)	3.84 (3.28–4.45)	0.001
Protein rate, %	3.34 (3.11–3.56)	3.37 (3.17–3.59)	0.001
Fat to protein ratio	1.07 (0.94–1.20)	1.13 (0.99–1.29)	0.001
Somatic cell count, ×104 cells/mL	6.00 (3.00–13.00)	6.00 (3.00–15.00)	0.308
Somatic cell score	2.00 (1.00–3.00)	2.00 (1.00–4.00)	0.271
Milk loss, kg	0.00 (0.00–0.00)	0.00 (0.00–0.00)	0.076
Milk shortage	0.00 (0.00–0.00)	0.00 (0.00–0.00)	0.079
Economic loss	0.00 (0.00–0.00)	0.00 (0.00–0.00)	0.078
Corrected milk yield, kg	41.90 (35.20–48.30)	45.40 (39.00–52.60)	0.001
Second parity	n = 961	n = 1,735
Daily milk yield, kg	43.90 (37.10–50.45)	45.70 (39.90–51.40)	0.001
Milk fat percentage, %	3.69 (3.23–4.20)	3.63 (3.05–4.20)	0.015
Protein rate, %	3.37 (3.17–3.59)	3.27 (3.08–3.49)	0.001
Fat to protein ratio	1.09 (0.97–1.23)	1.19 (0.94–1.27)	0.626
Somatic cell count, ×104 cells/mL	8.00 (4.00–20.00)	7.00 (3.00–19.00)	0.001
Somatic cell score	3.00 (2.00–4.00)	2.00 (1.00–4.00)	0.003
Milk loss, kg	0.00 (0.00–0.70)	0.00 (0.00–0.70)	0.422
Milk shortage	0.00 (0.00–2.50)	0.00 (0.00–2.40)	0.423
Economic loss	0.00 (0.00–3.90)	0.00 (0.00–3.70)	0.424
Corrected milk yield, kg	44.85 (37.40–52.48)	45.40 (38.50–52.90)	0.001
Third parity	n = 622	n = 437
Daily milk yield, kg	43.70 (37.38–51.00)	47.70 (41.90–54.55)	0.001
Milk fat percentage, %	3.66 (3.13–4.29)	3.665 (3.14–4.24)	0.704
Protein rate, %	3.24 (3.03–3.46)	3.31 (3.08–3.50)	0.006
Fat to protein ratio	1.13 (0.98–1.30)	1.11 (0.97–1.25)	0.061
Somatic cell count, ×104 cells/mL	8.00 (3.00–22.00)	7.00 (3.00–19.00)	0.176
Somatic cell score	3.00 (1.00–4.00)	2.00 (1.00–4.00)	0.214
Milk loss, kg	0.00 (0.00–0.80)	0.00 (0.00–0.68)	0.093
Milk shortage	0.00 (0.00–2.80)	0.00 (0.00–2.35)	0.095
Economic loss	0.00 (0.00–4.30)	0.00 (0.00–3.73)	0.096
Corrected milk yield, kg	42.60 (35.70–50.40)	45.90 (39.65–53.18)	0.001

During the late lactation period, the daily milk yield (Mann–Whitney U = 2,131,495.5, Z = 6.003, P < 0.001), milk fat percentage (Mann–Whitney U = 1,767,355.5, Z = 2.189, P = 0.029), fat-to-protein ratio (Mann–Whitney U = 1,788,520, Z = 2.832, P = 0.005) and corrected milk yield (Mann–Whitney U = 1,936,480.5, Z = 7.323, P < 0.001) of cows with diarrhea significantly decreased, whereas the somatic cell count (Mann–Whitney U = 1,581,301.5, Z = −3.465, P < 0.001), somatic cell score (Mann–Whitney U = 1,575,278, Z = −3.693, P < 0.001), milk loss (Mann–Whitney U = 1,643,403.5, Z = −2.005, P = 0.045), milk shortage (Mann–Whitney U = 1,642,712, Z = −1.993, P = 0.046), and economic loss (Mann–Whitney U = 1,641,764, Z = −2.028, P = 0.043) were significantly lower than those of healthy cows in the first parity (Table 5). In the second parity, diarrhea in calves only negatively affected the daily milk yield (Mann–Whitney U = 757,771.5, Z = 5.591, P < 0.001), milk fat percentage (Mann–Whitney U = 534,060, Z = −3.007, P = 0.003), fat-to-protein ratio (Mann–Whitney U = 540,176.5, Z = −2.585, P = 0.010), and corrected milk yield (Mann–Whitney U = 628,429, Z = 3.512, P < 0.001) (Table 5). In the third parity, only the daily milk yield (Mann–Whitney U = 96,325, Z = 4.538, P < 0.001), and corrected milk yield (Mann–Whitney U = 87,690, Z = 4.349, P < 0.001) of cows with diarrhea decreased significantly (Table 5).

Table 5.

Comparison of dairy herd improvement in the late lactation between dairy cows with a history of neonatal diarrhea and healthy controls (total number of 7,249) in different parities from a commercial herd in Hebei, China, analyzed using the Mann–Whitney U-test.

	Group, median (IQR)		P-value
	Diarrhea	Health
First parity	n = 1,446	n = 2,648
Daily milk yield, kg	35.50 (30.08–40.50)	36.90 (32.23–41.50)	0.001
Milk fat percentage, %	3.79 (3.33–4.31)	3.85 (3.31–4.45)	0.029
Protein rate, %	3.42 (3.22–3.66)	3.41 (3.21–3.64)	0.243
Fat to protein ratio	1.10 (0.99–1.24)	1.13 (0.98–1.28)	0.005
Somatic cell count, ×104 cells/mL	8.00 (3.00–18.00)	7.00 (3.00–17.00)	0.001
Somatic cell score	3.00 (1.00–4.00)	2.00 (1.00–4.00)	0.001
Milk loss, kg	0.00 (0.00–0.50)	0.00 (0.00–0.50)	0.045
Milk shortage	0.00 (0.00–1.80)	0.00 (0.00–1.70)	0.046
Economic loss	0.00 (0.00–2.88)	0.00 (0.00–2.60)	0.043
Corrected milk yield, kg	49.85 (41.63–57.80)	52.70 (45.33–60.20)	0.001
Second parity	n = 978	n = 1,365
Daily milk yield, kg	36.35 (29.00–42.60)	38.60 (32.50–44.20)	0.001
Milk fat percentage, %	3.73 (3.19–4.26)	3.62 (3.05–4.20)	0.003
Protein rate, %	3.42 (3.22–3.65)	3.40 (3.21–3.61)	0.167
Fat to protein ratio	1.08 (0.95–1.22)	1.06 (0.91–1.21)	0.01
Somatic cell count, ×104 cells/mL	10.00 (4.00–22.00)	9.00 (4.00–26.00)	0.813
Somatic cell score	3.00 (2.00–4.00)	3.00 (2.00–4.00)	0.685
Milk loss, kg	0.00 (0.00–0.60)	0.00 (0.00–0.80)	0.437
Milk shortage	0.00 (0.00–2.20)	0.00 (0.00–2.90)	0.432
Economic loss	0.00 (0.00–3.40)	0.00 (0.00–4.55)	0.432
Corrected milk yield, kg	47.20 (38.30–55.90)	49.20 (41.45–57.05)	0.001
Third parity	n = 454	n = 358
Daily milk yield, kg	35.00 (29.38–41.90)	38.65 (32.40–44.60)	0.001
Milk fat percentage, %	3.75 (3.30–4.32)	3.63 (3.16–4.26)	0.202
Protein rate, %	3.34 (3.15–3.59)	3.35 (3.12–3.59)	0.643
Fat to protein ratio	1.11 (1.00–1.25)	1.10 (0.95–1.25)	0.295
Somatic cell count, ×104 cells/mL	9.00 (4.00–26.00)	12.00 (4.00–30.00)	0.189
Somatic cell score	3.00 (2.00–4.00)	3.00 (2.00–5.00)	0.185
Milk loss, kg	0.00 (0.00–0.80)	0.00 (0.00–1.20)	0.079
Milk shortage	0.00 (0.00–2.65)	0.00 (0.00–4.30)	0.078
Economic loss	0.00 (0.00–4.15)	0.00 (0.00–6.70)	0.077
Corrected milk yield, kg	43.40 (36.85–51.80)	47.20 (40.15–55.78)	0.001

Predictors of 305-d milk yield and peak milk yield

This model is based on 26,860 observations, and after the missing values and outliers were removed, a total of 15,840 observations were included in the model construction. The results of the univariable analysis demonstrated significant associations between 305-d milk yield/peak milk yield and all investigated predictors, including diarrhea state, birth weight, wean weight, wean age, and parity (all P < 0.001; Table S3 and S4, Figure S5). For all included variables, the adjusted generalized variance inflation factors (GVIFs) were lower than 2, which indicated that there was no severe multicollinearity among the independent variables, ensuring the stability and reliability of the subsequent model parameter estimates (Table S5).

For 305-d milk yield, early-life diarrhea was associated with a significant decrease in 305-d milk yield (Estimate = −132.82, SEM = 49.73, P = 0.008), explaining 0.05% of the variance. Parity had significant effects: ns(parity)1 (Estimate = 2,595.58, SEM = 71.68, P < 0.001) and ns(parity)2 (Estimate = −3,751.75, SEM = 133.73, P < 0.001) collectively explained 4.64% of the variance. Birth weight was positively associated with 305-d milk yield (Estimate = 33.74, SEM = 9.36, P < 0.001), contributing 0.13% of the variance. Wean weight also showed a positive effect (Estimate = 24.48, SEM = 7.53, P = 0.001), explaining 0.10% of the variance. Wean age tended to decrease 305-d milk yield (Estimate = −22.31, SEM = 11.46, P = 0.051, marginally significant), accounting for 0.04% of the variance (Table 6).

Table 6.

Fixed effects results of linear mixed-effects models for 305-d milk yield and peak milk yield (total number of 15,840 observations) from a commercial herd in Hebei, China.

Variable	Description	305-d milk yield				Peak milk
		Estimate	SEM	P-value	Variance (%)	Estimate	SEM	P-value	Variance (%)
Intercept	Intercept	8,253.61	730.97	<0.001	86.75	39.26	3.62	<0.001	64.74
Healthy state	Healthy	0.00 (Ref)	–	–	–	0.00 (Ref)	–	–	–
	Diarrhea	−132.82	49.73	0.008	0.05	−0.71	0.22	0.001	0.17
Parity	Parity 1	0.00 (Ref)	–	–	–	0.00 (Ref)	–	–	–
	Parity 2	2,595.58	71.68	<0.001	4.64	−2.38	0.25	<0.001	27.02
	Parity 3	−3,751.75	133.73	<0.001		25.79	0.45	<0.001
Birth weight	Birth weight	33.74	9.36	<0.001	0.13	0.1	0.05	0.035	0.13
Wean age	Wean age	−22.31	11.46	0.051	0.04	−0.07	0.06	0.191	0.04
Wean weight	Wean weight	24.48	7.53	0.001	0.1	0.06	0.04	0.101	0.07

For peak milk, early-life diarrhea significantly reduced peak milk (Estimate = −0.71, SEM = 0.22, P = 0.001), explaining 0.17% of the variance. Parity had substantial effects: ns(parity)1 (Estimate = −2.38, SEM = 0.25, P < 0.001) and ns(parity)2 (Estimate = 25.79, SEM = 0.45, P < 0.001) collectively accounted for 27.02% of the variance. Birth weight was positively associated with peak milk (Estimate = 0.10, SEM = 0.05, P = 0.035), contributing 0.13% of the variance. Wean age (Estimate = −0.07, SEM = 0.06, P = 0.191) and wean weight (Estimate = 0.06, SEM = 0.04, P = 0.101) showed no significant effects on peak milk, explaining 0.04% and 0.07% of the variance, respectively (Table 6).

The high conditional R2c (>0.9 for both outcomes) indicated that the mixed-effects models fit the data very well, with most of the variance explained by the combined fixed and random effects (Table S6). Both power values exceeded 0.8, suggesting that the mixed-effects models had enough statistical power to identify the real effects on 305-d milk yield and peak milk (Table S7).

Lactation curves for different health states and different parities

The first parity included a total of 11,952 observations, with 4,676 diarrhea cases and 7,276 healthy controls. In the second parity, a total of 8,839 observations were included, with 3,188 diarrhea cases and 5,651 healthy controls. The third parity included a total of 3,608 observations, with 2,002 diarrhea cases and 1,606 healthy controls. Regardless of parity, the same trend in the lactation curve was observed. Milk production in both the diarrhea group and the healthy group increased with increasing lactation time and gradually decreased after approximately 100 d, and milk production in the healthy group was greater than that in the diarrhea group from the beginning to the end (first parity: Mann–Whitney U = 690,847.5, Z = 1.954, P = 0.510; second parity: Mann–Whitney U = 75,118, Z = 0.815, P = 0.415; third parity: Mann–Whitney U = 4,113, Z = 3.394, P < 0.001) (Table 1; Figure 4A–C).

Figure 4.

Lactation curves for milk yield on different lactation days. (A–C) The lactation curves of milk yield in first parity (A), second parity (B) and third parity (C).

Mediating effect of milk yield on the effect of diarrhea on the milk fat percentage

The effect of diarrhea on milk fat percentage by reducing milk yield at each parity was analyzed. In the first parity, diarrhea had a significant negative impact on milk yield (estimate = −0.904, SE = 0.157, 95% CI = −1.205 to −0.597, P < 0.001), and milk yield had a significant negative impact on milk fat percentage (estimate = −0.012, SE = 0.001, 95% CI = −0.014 to −0.010, P < 0.001), and diarrhea still had a significant negative direct impact on the milk fat percentage (estimate = −0.201, SE = 0.018, 95% CI = −0.238 to −0.164, P < 0.001). However, the indirect effect of diarrhea on milk fat percentage through milk yield was significantly positive (estimate = 0.011, SE = 0.002, 95% CI = 0.007–0.015, P < 0.001) (Figure 5A). In the second parity, diarrhea had a significant negative impact on the milk yield (estimate = −2.807, SE = 0.270, 95% CI = −3.347 to −2.305, P < 0.001), and it was significantly enhanced compared to first parity dairy cows. At the same time, milk yield also had a significant negative impact on milk fat percentage (estimate = −0.011, SE = 0.001, 95% CI = −0.013 to −0.009, P < 0.001). However, the direct effect of diarrhea on the milk fat percentage was not significant (estimate = −0.012, SE = 0.023, 95% CI = −0.055 to 0.033, P = 0.611), while the indirect effect remained significant (estimate = 0.031, SE = 0.004, 95% CI = 0.024–0.040, P < 0.001) and the positive effect was significantly enhanced compared to the first parity dairy cows (Figure 5B). In the third parity, the negative impact of diarrhea on milk yield remained significant and further intensified (estimate = −4.338, SE = 0.465, 95% CI = −5.302 to −3.428, P < 0.001), and milk yield still had a significant negative impact on milk fat percentage (estimate = −0.010, SE = 0.001, 95% CI = −0.013 to −0.008, P < 0.001). The direct effect of diarrhea on milk fat percentage remained insignificant (estimate = −0.053, SE = 0.035, 95% CI = −0.126 to 0.015, P = 0.130), while the indirect effect had a significant positive effect and continues to ­intensify (estimate = 0.046, SE = 0.008, 95% CI = 0.031–0.061, P < 0.001) (Figure 5C).

Figure 5.

The mediating analysis between health status and milk fat percentage that affected by milk yield. (A–C) Milk yield played a mediating role in the influence of health state on milk fat percentage in the first parity (A), second parity (B) and third parity (C).

Discussion

Healthy growth during the calf-rearing period is critical to future production performance. Previous studies have indicated that diarrhea accounts for 53% of calf mortality in the United States and is also the most prevalent and severe disease affecting calves in China, Canada, and Switzerland (Cho and Yoon 2014; Pipoz and Meylan 2016; Urie et al. 2018; Zhao et al. 2021; Schinwald et al. 2022). The economic loss per diarrheic calf is estimated at 106–140 euros in Europe, a significant burden considering that Europe is the world’s largest producer of cow milk (Roblin et al. 2023). However, the economic losses caused by diarrhea during the calf stage include not only direct expenditures associated with treatment and calf mortality but also the subsequent reduction in production performance, which may lead to long-term and sustained economic losses (Meganck et al. 2014; Carter et al. 2021). However, the current research on the follow-up performance of calves with diarrhea is still not comprehensive, and our study fills this gap.

Our findings demonstrated that diarrhea during the calf period exerted a persistent and negative influence on the lactation performance of dairy cows. Although the absolute decrease in 305-d milk yield and peak milk yield was modest and health status explained only a small proportion of the total variance, the consistency of negative impact across parities suggested that early-life diarrhea contributed to long-term impair on herd productivity. These findings were consistent with other observational studies, which indicated that diarrhea significantly impairs the weaning weight and withers height of calves, thereby reducing the overall growth performance and subsequent milk production of dairy cows (Wenge et al. 2014; Nicola et al. 2023; Yin et al. 2025). Specifically, previous studies have identified that calf diarrhea clearly and markedly diminished the growth performance, peak milk yield, and 305-d milk yield of first-parity cows (Toftaker et al. 2017; Boccardo et al. 2019; Arnaiz et al. 2021). Similarly, our study revealed that the weaning weight, withers height, and milk production performance of first-parity cows were most severely affected by diarrhea, supporting the concept that early-life health challenges may exert their greatest influence during the initial lactation. But as parity increased, the influence of diarrhea on the body condition score and withers height decreased, yet its detrimental effect on milk production persisted. Through longitudinal cohort analysis, this study extended existing evidence by demonstrating that the consequences of calf diarrhea were not limited to early productive life but remained detectable across multiple lactations, offering a critical foundation for comprehensive health management of dairy cows throughout their full life cycle.

Parity has been associated with lactation performance, such as milk production, milk fat, and milk protein (Kelsey et al. 2003; Marumo et al. 2022; Valldecabres and Silva-del-Río 2022). Our analysis showed that parity accounted for a substantial ­proportion of the variance in 305-d milk yield (4.64%) and peak milk yield (27.02%). Research indicated that multiparous cows exhibited increased milk production than primiparous cows due to the progressive accumulation of secretory tissue and enhanced metabolic capacity (Valldecabres and Silva-del-Río 2022). Our study monitored the milk production of dairy cows across three parities. Healthy cows showed a consistent upward trajectory in milk production across parities, whereas cows that experienced calfhood diarrhea exhibited an increase in milk yield from the first to the second parity, followed by no further significant increase from the second to the third parity in the repeated-measures analysis via general linear model, showing a flattening of the parity-related trajectory, indicating that early-life inflammation might impose a permanent damage on lactation potential. The sensitivity of first-parity cows to calf diarrhea may be directly attributed to their underdeveloped physiological functions (Zhang et al. 2015). Research has demonstrated that the quantity and functionality of mammary gland cells in first parity dairy cows were incompletely developed, resulting in relatively weaker immune responses and metabolic regulatory capacities (Miller et al. 2006; Morales Piñeyrúa et al. 2018). In contrast, second and third-parity cows may partially compensate for early growth deficiencies through subsequent nutritional management or physiological adaptation mechanisms (Morales Piñeyrúa et al. 2018; Ferreira et al. 2021). Our study revealed that after the second and third parities, the disparity in withers height and condition between cows in the diarrhea group and those in the healthy group diminished. Nevertheless, milk production performance did not improve as a result of compensatory growth. This dissociation implied that compensatory growth alone might be insufficient to offset early-life constraints under the increasing metabolic load of successive lactations. Calfhood diarrhea negatively impacted dairy cow performance across all lactation periods (early postpartum, early, middle, and late lactation), with the pattern of effects varying by parity. The early postpartum period, a critical phase initiating the lactation cycle (Pillay and Davis 2023), showed that while first-parity cows with a history of diarrhea exhibited no significant reduction in daily milk yield or corrected milk yield, second- and third-parity cows experienced a decrease in corrected milk yield. This disparity was likely related to the differences in metabolic demand among parities. Primiparous cows had a relatively lower absolute milk yield and a less pronounced negative energy balance (NEB), which may allow them to cope with the subclinical impairments associated with early-life diarrhea without a visible loss in production (Wathes et al. 2007). In contrast, multiparous cows (second and third parity) faced a much steeper lactation curve and more severe NEB immediately postpartum, potentially amplifying the expression of early-life constraints on production (Civiero et al. 2021). The early lactation period, characterized by peak milk yield (Bangar and Verma 2017), presented the most pronounced negative effects associated with diarrhea. First-parity cows with a history of diarrhea presented a significantly reduced milk fat percentage and corrected milk yield. In second-parity cows, the daily milk yield decreased, with further declines in the milk fat percentage and corrected milk yield, alongside elevated somatic cell counts and economic loss indicators. Therefore, these findings suggested that early-life diarrhea may compromise lactation performance through multiple, interacting mechanisms, including altered mammary function, metabolic adaptability, and immune competence, particularly under the physiological stress of early lactation (Zhang et al. 2015). During mid- and late lactation, cows with a history of diarrhea consistently presented lower daily milk yields and corrected milk yields across all parities. However, divergent trends in milk composition metrics indicated that the long-term effects of diarrhea were not mediated through a single pathway but involved complex interactions among nutrient metabolism, the immune response, and mammary cell function (Delosière et al. 2023; Hayes et al. 2023).

Our mediation analysis revealed a dynamic relationship between early-life diarrhea, milk yield, and milk fat percentage that evolved across parities, characterized by competitive mediation mechanisms. In the first parity, the impact of diarrhea was driven by two opposing pathways. We observed a significant negative direct effect on milk fat percentage, suggesting that neonatal diarrhea in calves might cause long-term impacts on mammary metabolic capacity and organ nutrient allocation to affect synthesize milk fat (Miller et al. 2006; Osorio 2020; Anger et al. 2024). However, this was partially counteracted by a significant positive indirect effect. This indirect pathway was explained by the physiological dilution effect (Miglior et al. 2017). Diarrhea significantly reduced milk yield, and due to the inherent negative correlation between milk yield and milk fat percentage, this reduction in milk yield was associated with a higher milk fat percentage through a dilution effect, rather than through enhanced lipid synthesis. These opposing effects highlighted that changes in milk fat percentage could arise from distinct mechanisms, including both functional alterations in lipid synthesis and yield-driven ­concentration effects, which must be interpreted separately. The presence of both significant negative direct and positive indirect effects indicated that in primiparous cows, the metabolic defect in fat synthesis outweighed the concentration effect. In the second and third parities, although cows with a history of calfhood diarrhea exhibited lower milk yield and lower milk fat percentage than healthy cows, mediation analysis indicated that the reduction in milk yield exerted a positive indirect effect on milk fat percentage via a dilution mechanism. This indirect effect reflected a relative dilution effect within diarrheic cows and did not contradict the overall lower milk fat percentage observed at the group level. This aligned with our observation that the reduction in milk yield worsened with age, likely due to the irreversible damage imposed on mammary development. As the milk yield gap widened, the concentration effect became more pronounced. These findings clarified that the long-term legacy of diarrhea was complex, while the direct functional impairment of fat synthesis appeared to wane after the first lactation, the structural limitation on milk yield persisted and intensified, driving changes in milk composition primarily through the concentration of solids.

This study had several limitations. This study was a retrospective cohort analysis that relied on farm record data; the samples were concentrated in the northern region of China, and the applicability of the conclusions to the peripheral environment needed to be verified. The use of mixed-effects models partially addressed the inherent imbalance in our dataset, particularly the unequal number of observations per cow and the differing sample sizes between the diarrhea and healthy groups. Although the efficacy analysis confirmed the feasibility of the model, it could not fully eliminate bias from extreme imbalance, and our results should also be interpreted with caution. Furthermore, the study did not explore specific molecular mechanisms or the effects of intervention measures. Future studies should include multi-herd and omics techniques, and intervention trials can be combined to analyze the key pathways through which calf diarrhea affects lactation performance in detail and develop targeted early prevention and control strategies.

In conclusion, our research indicated that diarrhea during the calf-rearing period had a significant long-term adverse effect on lactation performance. From an economic standpoint, early excluding affected calves from the core breeding herd might serve as a prudent risk mitigation strategy. Furthermore, investing high costs in preventive measures, such as timely vaccination and consistent environmental management during the neonatal stage proved highly cost-effective over time. Such interventions not only enhanced immediate survival rates but also secured future milk production profitability over the subsequent 3–4 years.

Conclusion

This study not only confirmed the negative effects of calf diarrhea but also quantified the persistent losses across three parities. We identified that although physical condition appeared to partially recover in multiparous cows, differences in lactation performance remained evident across parities. This persistent yield gap drove changes in milk components, primarily through the concentration effect caused by reduced milk yield. We conclude that calfhood diarrhea is a critical early-life health challenge, and it is recommended that ranches should re-evaluate the early elimination standards or differentiated management strategies for cattle that have recovered from diarrhea.

Abbreviations: 

DHI

dairy herd improvement

DIM

days in milk

GVIF

generalized variance inflation factor

ICC

intraclass correlation coefficient

IQR

interquartile range

LMM

linear mixed-effects model

ML

maximum likelihood

Author contributions

Xinfeng Hou: Writing–original draft, Investigation, Conceptualization, Methodology. Jingyi Xu: Writing–original draft, writing–review & editing, Investigation, Conceptualization, Methodology. Jianrong Ren: Formal analysis, Conceptualization. Guanglei Liu: Conceptualization. Xiaoxu Wang: Investigation. Qian Du: Investigation. Zheng Niu: Investigation. Junhu Yao: Investigation, Conceptualization, Methodology, Writing–review & editing, Supervision. Yangchun Cao: Investigation, Methodology. Dewen Tong: Supervision, Methodology. Shengru Wu: Investigation, conceptualization, methodology, funding acquisition, writing–original draft, writing–review & editing, project administration.

Supplementary data

Supplementary data is available at Journal of Animal Science online.

Funding

This work was supported by the China Dairy Industry Technology Innovation Center Open Research Topics (2024‐KFKT‐031), the National Natural Science Foundation of China (32573272), and the General Project of the Natural Science Foundation of Xi’an, Shaanxi Province (2025JH-ZRKX-0639).

Conflict of interest statement:

The authors declare no real or perceived conflicts of interest.

Availability of data and material

The data generated or analyzed for this study are included in this paper.