Ethical approval for this study was obtained from the Ethical Committee of Heilongjiang Bayi Agricultural University (Approval No. DWKJXY2023057; approved on 1 August 2023). All procedures involving animals were conducted in accordance with the institutional guidelines for the care and use of animals in research, as well as relevant national regulations governing animal welfare.

The study was carried out on commercial dairy farms under standard management and feeding practices, and no invasive or harmful procedures were performed specifically for research purposes. Data collection was limited to diet composition analysis and nutrient evaluation, along with routine production-related observations. Animal handling, where required, was performed by trained personnel following standard husbandry practices to minimize stress and ensure animal welfare.

Permission was obtained from the owners/managers of all participating farms prior to the commencement of the study. The research adhered to the principles of animal welfare, including reduction of unnecessary intervention and avoidance of distress, ensuring that all activities were conducted under normal farm conditions without disrupting routine management practices.

This animal study was conducted between November 2024 and February 2025 on four commercial Holstein dairy farms located in Heilongjiang Province, Northeast China. The farms operated under intensive housed systems with similar herd sizes (more than 3,000 milking cows) and used freestall barns with concrete flooring and cubicles.

The four farms were purposively selected because they are large-scale dairy farms in Heilongjiang Province, with a large number of dairy cows, and represent a significant sample. Additionally, they already implemented different PG-based postpartum health programs for ketosis prevention and treatment as part of their routine veterinary management. No farm used exactly the same protocol, allowing direct comparison of real-world implementation variations. Farms were not randomized and no control (no-PG) group was included, as this was an observational comparison of existing on-farm practices.

A total of 480 multiparous Holstein cows were initially enrolled, with 120 cows selected from each of the four farms as a fixed cohort. At each of six sampling points (–7 d prepartum, 0 d [calving day], +7 d, +21 d, +50–60 d, +80–100 d postpartum), 20 cows were randomly selected from this cohort (n = 20 × 4 farms × 6 time points = 480 observations). Cows were treated as a random effect nested within farm.

Cows enrolled in all farms were highly similar in parity, age, and BCS at the initiation of the study, with strict inclusion and exclusion criteria applied uniformly across farms. Inclusion criteria: Holstein cows of similar age, BCS, and parity [41.26 ± 0.74 months of age, 3.11 ± 0.02 of BCS] multiparous cows (≥2nd lactation, 2.33 ± 0.06 of parity), clinically healthy at enrollment (no obvious systemic disease), and due to calve within ±7 days of the sampling window. Exclusion criteria: cows with severe clinical mastitis, metritis, displaced abomasum, or lameness score >3 at enrollment. Furthermore, the 120 cows selected from each farm were under continuous long-term observation, with no mortality or culling during the entire experimental period.

All the farms supply total mixed ration (TMR) twice a day, at 7:00 am and 1:00 pm after milking. The composition and nutrient levels of the basal diet are presented in Table 1. Rations were formulated to meet or exceed NRC (2001) nutrient requirements for lactating Holstein cows producing 35–45 kg milk/d [22]. Feed samples were collected weekly and analyzed for dry matter, crude protein, neutral detergent fiber (NDF), acid detergent fiber (ADF), starch, and net energy by a commercial laboratory. The quantification of DM (at 105°C for 4 h) and crude protein (CP) in the feed was performed following AOAC (2006) guidelines, utilizing methods 934.01 for DM and 990.03 for CP [23]. The starch content in the feed was analyzed according to the respective China National Standard (GB/T 20194-2018) [24]. NDF contents were measured using the F57 fiber bag (25 mm porosity) technique (ANKOM Technology, Fairport, NY, USA) in accordance with China National Standard (GB/T 20806-2022) [25]. During the NDF analysis, thermostable α-amylase and sodium sulfite were employed to remove the interference of starch and protein, respectively.

During the experiment, dairy cows in each farm were densely raised in cattle sheds with the same strict hygiene conditions and free movement, receiving the same standardized care and being able to drink water freely. Throughout the experimental period, dairy cows from four farms were initially administered fluid nutrient supplements to ameliorate their nutritional status; subsequent steps involved clinical screening for ketosis and implementation of a PG health program. The detailed formulation of the nutritional supplement is presented in Table 2.

Each farm applied its established PG-based health programs without modification for the study. Programs differed in three key aspects: (1) the timing of first PG administration, (2) whether PG was given prophylactically or only to diagnosed cases, and (3) the number of consecutive treatment days. Details are presented in Table 3. For dairy cows in Farm A, 500 mL of the nutritional supplement was administered via oral drenching on the day of calving; ketosis screening was subsequently performed at 7 and 14 days postpartum, and cows diagnosed with ketosis were given 500 mL of PG daily for 2 consecutive days. In Farm B, cows received co-administration of 500 mL nutritional supplement and 500 mL PG via oral drenching on the day of calving; ketosis detection was carried out at 7 and 14 days postpartum, with a single 500 mL dose of PG administered to all cows diagnosed with ketosis. For Farm C, 500 mL of the nutritional supplement was orally drenched to cows on the day of calving, followed by ketosis screening at 7 and 14 days postpartum. Cows identified with ketosis at 7 days postpartum were given 500 mL of PG daily for 2 consecutive days, while those diagnosed at 14 days postpartum received a single 500 mL dose of PG. In Farm D, cows were administered 500 mL of the nutritional supplement via oral drenching on the day of calving; ketosis detection was conducted at 7 and 14 days postpartum, and a single 500 mL dose of PG was given to any cow diagnosed with ketosis. During the oral administration process, all cows demonstrated good tolerance to PG administration, with no adverse clinical signs observed. Treatment acceptance was high, and cows completed the drenching procedure without resistance.

On the day of calving and 14 days postpartum, the specific enzyme-linked immunosorbent assay kit for cattle (Shanghai Enzyme-Linked Biotechnology Co., Ltd., Shanghai, China) was used to measure BHBA in the experimental cow’s serum, thereby screening for ketosis. Subclinical ketosis was defined as BHBA ≥1.20 mmol/L, and clinical ketosis was diagnosed when BHBA ≥3.0 mmol/L [26] and at least two of the following signs were present: reduced feed intake, decreased milk yield, dullness/lethargy, or detectable abnormal odor on breath. All measurements were performed by trained farm veterinarians blinded to farm protocol allocation during sample collection.

The definitions of other diseases in the experiment are as follows. A retained placenta is defined as the failure of the placenta to be expelled within 24 h after calving [27]. When a cow experiences a sharp decline in milk yield one week before onset, accompanied by a significant decrease in rumination and activity levels, and percussion and auscultation reveal a high-pitched “pinging”, it is diagnosed with true displaced abomasum [28]. When a cow shows signs of miscarriage after confirmation of pregnancy, comes into heat again after confirmation of pregnancy, or is found to have failed to conceive upon re-confirmation of pregnancy, this is defined as a miscarriage [29]. When the somatic cell count in cow’s milk is ≥ 200,000 cells/mL, it is diagnosed as mastitis [30]. The characteristics of metritis in dairy cows include foul-smelling, watery, reddish-brown discharge, with or without fever [31]. Dystocia is usually defined as any delivery process that requires assistance [32]. Clinical milk fever in dairy cows is defined as a serum calcium concentration below 2.0 mmol/L, accompanied by clinical symptoms such as recumbency [33]. Hoof disease is defined as a condition often accompanied by increased hoof temperature, pain, and subsequent lameness [34].

The following variables were recorded: (1) The BCS of dairy cows is assessed on a 5-point scale (1 = emaciated, 5 = obese) [35], assessed by two trained veterinarians (average used). (2) The daily milk yield (kg) of dairy cows is recorded by an electronic milk meter and is recorded using specific software (Afifarm, Afimilk, Kibbutz Afikim, 1514800, Israel) to record age, parity, milk yield, and disease incidence of the cows. (3) The milk components (including fat, protein, urea nitrogen) are measured monthly DHI laboratory. (4) The disease events in dairy cows (including ketosis, retained placenta, displaced abomasum, abortion, mastitis, metritis, dystocia, milk fever, and hoof disease) are diagnosed and recorded by veterinarians. (5) The observation of the reproductive status of dairy cows, including the number of services, calving to first service interval, calving to first estrus interval, calving interval, and conception rate (calculated as total conception rate × conception rate at estrus × [1 - empty rate]).

Blood (10 mL) was collected from the coccygeal vein before morning feeding into plain serum tubes. Samples were centrifuged at 4,000 × g for 10 min at 4°C within 2 h of collection. Serum was stored at -80°C until analysis. Glucose, calcium, phosphorus, magnesium, and potassium were measured using an automated biochemistry analyzer (Mindray BS-830S) using commercial kits (Nanjing Jiancheng Bioengineering Institute Co., Ltd, Nanjing, China). BHBA and insulin were detected using bovine-specific ELISA kits (Shanghai Enzyme-linked Biotechnology Co., Ltd., Shanghai, China). Intra- and inter-assay CV were <10% for all assays.

Data were analyzed using IBM SPSS 26.0 software (SPSS Inc., Chicago, IL, USA). A linear mixed model was used to analyze continuous variables (BCS, milk yield, blood metabolites), with farm, time point, and their interaction as fixed effects, and cow nested within farm as a random effect. Tukey’s Honest Significant Difference post hoc test was applied to identify specific inter-group differences with adjustment for multiple comparisons. Categorical variables (disease incidence) were compared using Chi-square tests; odds ratios (OR) and 95% confidence intervals (95% CI) were computed. Significance was declared at p < 0.05; extremely significant was set at p < 0.01, and trends at 0.05 ≤ p < 0.10. Results were presented as means ± SEM.

The calculation method for the economic efficiency of cattle farms is as follows:

Economic benefit of ketotic cows = total × morbidity rate × (milk yield × milk price - cost).

Economic benefit of healthy cows = total × healthy rate × (milk yield × milk price - cost).

Total economic benefit = economic benefit of sick cows + economic benefit of healthy cows - cost of PG feeding.

The price of milk was 3.21 ¥/kg.

At 7 days postpartum, the BCS of cows in Farm C (3.65 ± 0.35) and Farm D (3.62 ± 0.31) was significantly higher than that of cows in the other two farms (p < 0.05). At 21 days postpartum and from 50 to 60 days postpartum, the BCS of cows in Farm A and Farm C was significantly higher than that of cows in the other two farms (p < 0.05). From 80 to 100 days postpartum, the BCS of cows in Farm A (3.43 ± 0.24) was significantly higher than that of cows in the other three farms (p < 0.05) (Table 4, Figure 1a). At 21 days postpartum and from 50 to 60 days postpartum, cows in Farm A and Farm B had significantly higher milk yield than those in the other two farms (p < 0.05) (Table 5, Figure 1b).

From 50 to 60 days postpartum and from 80 to 100 days postpartum, no significant differences were observed in protein content among cows. From 50 to 60 days postpartum and from 80 to 100 days postpartum, the fat-to-protein ratio of cows in Farm D was significantly higher than that of cows in the other three farms (p < 0.05). From 50 to 60 days postpartum, the somatic cell count of cows in Farm D (26.2 ± 66.0) was significantly higher than that in the other three farms (p < 0.05). From 80 to 100 days postpartum, the somatic cell count of cows in Farm D (36.5 ± 74.0) was significantly higher than that in the other three farms (p < 0.01). No significant differences were observed in urea nitrogen in cows from 50 to 60 days postpartum. From 80 to 100 days postpartum, the urea nitrogen of cows in Farm D (15.6 ± 2.6) was significantly higher than that in the other three farms (p < 0.01) (Table 6).

Among all monitored indicators, including ketosis, retained placenta, displaced abomasum, abortion, mastitis, metritis, dystocia, milk fever, and hoof disease, the incidence rate in cows in Farm D was significantly higher than that in the other three farms (Table 7, Figures 2a-i). Notably, the incidence rates of various diseases in cows in Farm A and Farm B were relatively low, and the incidence in cows in Farm B was slightly lower than that in cows in Farm A. Cows in Farm C had relatively high incidence rates of diseases such as retained placenta, milk fever, and hoof disease, but its overall incidence was lower than that in cows in Farm D and higher than that in cows in Farm A and Farm B (Table 7, Figures 2b, h–i).

On the day of calving, no significant differences were observed in glucose and insulin concentrations in cows across the four farms, and BHBA concentrations in cows from Farm B (0.6 ± 0.4) and Farm D (0.8 ± 0.4) were significantly higher than those in cows from Farm A (0.4 ± 0.2) and Farm C (0.5 ± 0.3) (p < 0.05) (Table 8, Figures 3a-c). On 21 days postpartum, there were no significant differences in glucose concentrations in cows among the four farms, and BHBA concentrations in cows from Farm B (0.8 ± 0.2) and Farm C (0.7 ± 0.4) were significantly lower than those in the other two farms (p < 0.05) (Figures 3a and b). The insulin concentration in cows from Farm D (19.31 ± 3.42) was significantly lower than that in the other three farms (p < 0.05) (Table 8, Figure 3c).

On the day of calving, no significant differences were observed in calcium, magnesium, and potassium concentrations in cows among the four farms (Table 9, Figures 4a, c, and d). However, the phosphorus concentrations in cows in Farm A (1.4 ± 0.2) and Farm C (1.5 ± 0.3) were significantly lower than those in the other two farms (p < 0.05) (Table 9, Figure 4b). At 21 days postpartum, there was no significant difference in magnesium and potassium concentrations in cows among the four farms (Table 9, Figures 4c and 4d). The calcium concentration in cows in Farm D (2.50 ± 0.64) was significantly lower than that in the other three farms (p < 0.05), whereas at 21 days postpartum no significant difference in calcium concentration was observed among cows in Farm A, Farm B, and Farm C (Table 9, Figure 4a). The phosphorus concentrations in cows in Farm B (1.8 ± 0.2) and Farm C (1.7 ± 0.4) were significantly lower than those in the other two farms (p < 0.05) (Table 9, Figure 4b).

The number of matings in cows in Farm B was lower than that in the other three farms, but the difference was not significant. The calving to first service interval in cows in Farm A was shorter than that in the other three farms, but there was no significant difference. The calving to first estrus interval in cows in Farm D was significantly higher than that in the other three farms (p < 0.01), whereas that in cows in Farm B was the lowest but did not differ significantly from that in the other two farms. The calving interval in cows in Farm D (406.10 ± 45.79) was significantly longer than that in the other three farms (p < 0.05), whereas that in cows in Farm B (381.83 ± 33.83) was the lowest but did not differ significantly from that in the other two farms. The conception rate in cows in Farm A (82.1) and Farm B (80.0) was significantly higher than that in cows in Farm D (73.2) (p < 0.05) and was also higher than that in cows in Farm C (78.3), although the latter difference was not significant (Table 10).

Cows were fed PG at ¥15 per dose, and the average lactation output of diseased and healthy cows during the first 100 days postpartum was obtained from each farm and is shown in Table 11. The daily rearing costs included expenses for feed, treatment, and breeding. The average daily cost per cow among the four farms was ¥68, with a milk price of 3.21 ¥/kg. Based on the incidence of ketosis in the four farms, the economic benefits of ketotic and healthy cows were calculated for each farm.

The economic benefit of cows with ketosis was highest in Farm D (¥59262.72) and second highest in Farm A (¥35233.15). The economic benefit of healthy cows was highest in Farm B (¥133970.40) and second highest in Farm A (¥132190.54). The total economic benefit of cows on the farm was highest in Farm A (¥166823.69) (Table 12).

During the transition period, dairy cows undergo significant physiological changes. Due to insufficient energy intake, fat is excessively mobilized in the body, and triglycerides in the liver are further converted into ketone bodies, triggering ketosis, which has adverse effects on the health, milk yield, and reproductive performance of dairy cows. This study compared health management programs at different stages to identify the most effective program, thereby providing a scientific basis for feeding management and health care in dairy cows. The aim was to improve BCS, reduce disease incidence, and thereby increase economic benefits.

The optimal BCS for lactating cows ranges between 3.25 and 3.50, at which productive performance is in a favorable state. If BCS exceeds or falls below this range, the risk of postpartum diseases increases [36, 37]. When BCS is below the optimal range, it is more likely to cause NEB, which reduces production performance [38, 39]. At the same time, when BCS exceeds the optimal range, it can lead to obesity, and cows become prone to diseases such as fatty liver, ovarian cysts, and dystocia, while milk yield is reduced [40, 41]. Supplementing cows with PG during the transition period can reduce fat mobilization, lower NEFA concentrations, maintain glucose supply to the body, sustain milk yield while stabilizing BCS, and prevent excessive postpartum BCS loss [42, 43]. In this study, there were differences in BCS among farms. After implementing different PG health programs, the BCS of cows in Farm A was more in line with the standard range (3.25–3.50) and was associated with better lactation and productivity. Cows in Farm B had low BCS, whereas cows in Farm C had high BCS, both of which increased the incidence of diseases such as ketosis and reduced lactation. The BCS of cows in Farm D was similar to that of cows in Farm A, approaching the standard level. Although the supplemental PG content in Farm D was higher than that in Farm A, excessive body fat led to stronger fat mobilization, resulting in reduced milk yield. The milk yield levels of cows in the four farms, from highest to lowest, were Farm A > Farm B > Farm C > Farm D. This indicates that both excessive and insufficient body condition can have adverse effects on milk yield, and that reasonable supplementation with PG can mitigate these effects.

Excessive fat mobilization and excessive postpartum body condition decline in dairy cows can both trigger ketosis and lead to performance decline [44, 45]. NEB is also an important cause of the high incidence of postpartum diseases [46, 47], and when it occurs, it causes abnormal liver function, weakens antioxidant capacity, and triggers fatty liver, which further exacerbates the NEB state [48, 49]. PG, as a precursor of gluconeogenesis, can alleviate NEB, inhibit ketone synthesis, reduce oxidative stress, and thereby lower disease incidence [50]. Among the four dairy farms in this trial, cows in Farm D had the highest disease incidence compared with the other three farms, with the highest incidence of ketosis exceeding 35%, whereas cows in Farm B had the lowest incidence of ketosis, approximately 20%. Furthermore, the ketosis incidence rate in cows in Farm A decreased to 28.9%. Although this value was higher than the ~7% reported in a randomized controlled trial (RCT) using PG administration on days 0, 7, and 14 (500 mL of PG), the present PG health program still effectively reduced ketosis risk [51]. Similarly, our treatment effect aligns with the trend reported by McArt et al. [20, 52].

However, the primary reason for the higher ketosis incidence in Farm D was the greater BCS loss among its cows during the 7–21 days postpartum period, coupled with NEB, both of which collectively contributed to the elevated disease incidence. Consequently, even when cows in Farm D were fed high levels of PG, their ketosis incidence did not decrease, indicating that the PG supply method was inappropriate and failed to achieve the desired mitigating effect. On the other hand, prolonged or high-dose use of PG may potentially suppress the appetite of dairy cows, which might be one of the factors contributing to the failure of the PG health program in Farm D [53].

Adiponectin is an adipokine that increases insulin sensitivity [54]. During parturition, glucose stimulates the release of insulin from the pancreas, and the levels of the two are correlated [55]. Cows with high BCS have reduced adiponectin secretion, which leads to decreased insulin sensitivity and tissue reactivity, and reduced tissue glucose uptake [56, 57]. In contrast, the use of PG can cause the body to produce more propionic acid, which is an important precursor for glucose production, thereby increasing plasma glucose concentration [58]. At the same time, PG also alleviates insulin resistance by lowering NEB, reducing fat mobilization, and maintaining a certain balance between glucose and insulin levels [59, 60]. The investigation showed that the BHBA levels of cows in Farm D were significantly higher than those in the other three farms, especially during the 7–21 days postpartum period, exceeding 1.0 mmol/L, which is consistent with the high prevalence of ketosis in cows in Farm D. After supplementing Farm D with high levels of PG, its blood glucose levels were not significantly different from those of the other three farms, indicating that PG played a role in raising blood glucose levels. However, high BHBA levels as well as low insulin levels indicated that ketosis was not being effectively treated in cows in Farm D, suggesting adverse effects of high postpartum BCS loss in these cows.

The estrus number, estrus rate, and conception rate of cows with NEB are significantly lower than those of normal cows [61, 62]. Severe NEB can disrupt reproductive hormone secretion, leading to decreases in estradiol (E2) and progesterone (P4) levels in dairy cows, which in turn reduce reproductive performance [63–65]. Impaired reproductive performance in cows can prolong the production cycle and reduce calving efficiency [66, 67], which causes substantial economic losses to the farm [68]. In this study, cows in Farm A had the best reproductive performance, whereas cows in Farm D had poorer reproductive performance and higher disease incidence than those in the other farms, suggesting that overfeeding of PG could not fully reverse the negative effects of high postnatal body condition loss and NEB in Farm D, which resulted in reduced reproductive performance.

The economic benefit of the different farms was mainly composed of morbid and healthy cows, with morbid cows having lower average economic efficiency; therefore, cow morbidity had a major impact on the economic benefit observed in this study. Due to the high incidence of ketosis in cows in Farm D, the economic benefit of morbid cows accounted for approximately 47.2% of the overall economic benefit, which was much higher than that of the other farms during the same period, and was the main reason for the low economic benefit of cows in Farm D. The analysis of economic benefits showed that cows in Farm A had the highest milk production and a relatively low overall disease incidence rate, thus achieving the highest economic benefits. While the economic analysis provides a practical comparison among farms, it was relatively simplified. Labor and veterinary costs associated with PG drenching were not included, and only the benefits from improved production and reduced disease risk were considered. This represents a limitation of the present study. Furthermore, to verify the robustness of the economic results, a sensitivity analysis was performed based on key parameters (milk price, daily cost per cow, and milk yield), with each parameter varying by ±5% and ±10%. The results showed that the economic trends among farms remained stable, confirming the reliability of the findings.

In this study, although Farm B performed well in terms of disease incidence and metabolic parameters, this might be attributed to universal preventive PG administration to all cows on the day of calving. In contrast, Farm A adopted a targeted, diagnosis-based treatment approach: 500 mL of PG was administered for 2 consecutive days at 7 and 14 days postpartum only to cows diagnosed with ketosis. This approach ensured effective intervention during the peak ketosis risk period (3–14 days postpartum), while avoiding excessive supplementation in healthy animals. The overall effect of Farm A’s health program was significantly better than that of Farm C and Farm D, which could be attributed to the targeted and continuous PG supplementation following ketosis diagnosis. Notably, compared with the other three farms, Farm A achieved better body condition, higher milk yield, and greater economic benefits, indicating that this strategy is more practical and efficient under commercial farm conditions. Therefore, the outcomes from Farm B do not alter the conclusion that Farm A’s health program yielded the best overall performance.