A randomized controlled trial evaluating the efficacy of systemic ceftiofur administration for metritis therapy in dairy cows and the effect of metritis cure on economically important outcomes

The main objective of this study was to evaluate the effect of ceftiofur on metritis cure, milk yield, reproductive performance, and culling up to 300 DIM. The secondary objective was to evaluate the effect of metritis cure at 5 (ECURE) and 14 (LCURE) d after diagnosis on milk production, reproduction, and culling. A total of 422 Holstein cows diagnosed with metritis from 4 herds located in TX, CA, and FL were enrolled in a randomized clinical trial. Cows diagnosed with metritis (fetid, watery, reddish/brownish uterine discharge) were blocked by herd and parity and were randomly allocated to receive systemic administration of ceftiofur (CEF) or to remain untreated (CON). In addition, 399 non-metritic cows (NMET) were included for comparison purposes. Metritis cure was evaluated at 5 and 14 d after diagnosis and was defined as the absence of metritis clinical signs. Logistic regression models were fitted to the data to assess the effect of treatment on metritis cure. Milk yield was analyzed using a mixed linear model, while logistic regression, Cox proportional hazard and Kaplan-Meier survival analysis models were fitted to culling and reproduction data. Cows treated with CEF had 1.86 (95% CI: 1.22 – 2.81) and 1.68 (95% CI: 1.02 – 2.75) greater odds of being cured than CON cows at 5 and 14 d after diagnosis, respectively. No effect of CEF was observed for milk yield; however, NMET cows had greater milk yield compared with metritic cows (CEF = 36.0, 95% CI = 33.8 – 38.1; CON = 36.1, 95% CI = 33.9 – 38.2; NMET = 36.9 kg/d, 95% CI = 34.8 – 39.4). Likewise, no effect of CEF was observed on reproductive performance and culling. Nonetheless, the likelihood of conceiving for NMET cows was 1.72 (95% CI = 1.41 – 2.12) and


INTRODUCTION
Metritis is a prevalent polymicrobial postpartum uterine disease that affects 15 -40% of dairy cows (Sheldon et al., 2008;LeBlanc et al., 2011;Menta et al., 2022).Metritis is characterized by an abnormally enlarged uterus and a fetid, watery, red-brown uterine discharge within 21 d after parturition with its incidence peaking within the first 10 d postpartum (Sheldon et al., 2006;Galvão, 2012;Pinedo et al., 2020).Metritis is associated with impaired reproductive performance, increased pregnancy losses, anovulation, and increased calving-to-conception interval (Dubuc et al., 2010; A randomized controlled trial evaluating the efficacy of systemic ceftiofur administration for metritis therapy in dairy cows and the effect of metritis cure on economically important outcomes P. R. Menta, 1 L. Fernandes, 1 J. Prim, 2 E. De Oliveira, 4 F. Lima, 4 K. N. Galvão, 2,3 N. Noyes, 5 M. A. Ballou, 1 and V. S. Machado 1 * Giuliodori et al., 2013;Ribeiro et al., 2016).Cows affected by metritis produce less milk early in lactation (Wittrock et al., 2011;de Oliveira et al., 2020).Moreover, due to the losses in production and reproductive performance, metritis is associated with an increased risk of culling (Wittrock et al., 2011).Because of these losses and additional costs related to veterinary care and antibiotic therapy, the metritis economic burden is calculated to range between US$ 267 and 410 per case (Lima et al., 2019).Recently, it was estimated that leaving metritis cases untreated can lead to US$ 250 in profit losses (Silva et al., 2021).
Antimicrobial drugs are commonly used for metritis therapy.Ceftiofur, a third-generation cephalosporin, is the most commonly used drug for metritis therapy in North America (Haimerl and Heuwieser, 2014;Haimerl et al., 2017).Findings from a meta-analysis concluded that there was limited evidence to support the use of systemic ceftiofur treatment for metritis, at least when considering its impact on lactational performance and fertility (Haimerl et al., 2017).At the time that this 2017 meta-analysis was conducted, there were a small number of randomized controlled trials that included a non-treatment group in the study design.Few studies have evaluated clinical outcomes comparing systemic ceftiofur therapy with a non-treated control group (Chenault et al., 2004;McLaughlin et al., 2012;Giuliodori et al., 2013;Piccardi et al., 2016;de Oliveira et al., 2020).Studies reported that metritis therapy with ceftiofur increased cure (absence of metritis 2 weeks after treatment) from 55 to 62% in untreated cows to 75-78% in treated ones (Chenault et al., 2004;McLaughlin et al., 2012).A recent study including nontreated metritic cows reported that systemic ceftiofur therapy led to a modest increase in milk production within the first 45 DIM, and increased the hazard of pregnancy (de Oliveira et al., 2020).Studies conducted in Argentina reported that ceftiofur treatment of metritis did not impact milk production, but did improve fertility (Giuliodori et al., 2013;Piccardi et al., 2016).Although these results seem to favor the use of ceftiofur to treat metritis in dairy cows, it has also been demonstrated that publication bias is prevalent in animal research (Conradi and Joffe, 2017).Recent findings demonstrated that the 2-dose regimen of ceftiofur administration led to increased quantity of antimicrobial resistance genes in feces of metritic cows (Ossa-Trujillo et al., 2023).Given that antimicrobial resistance has become a major public and animal health concern, and judicious use of antimicrobial drugs in food-producing animals is critical for agriculture sustainability (O 'neill, 2015), further evidence on the effectiveness of ceftiofur for metritis therapy is warranted.Thus, additional investigations using different populations and slightly different methodologies are needed to generate robust, more definitive guidelines regarding systemic ceftiofur therapy for the treatment of metritis.
A recent study reported that cows that did not cure of metritis after antimicrobial therapy experienced reduced milk yield, impaired reproductive performance, and greater risk of culling when compared with cows that recovered from this condition (Figueiredo et al., 2021).In that study, metritis cure was evaluated 10 d after disease diagnosis.However, clinical cure is assessed as early as 5 d after onset of treatment in multiple studies, (Lima et al., 2014;de Oliveira et al., 2020).For instance, de Oliveira et al. ( 2020) demonstrated that 44.7% and 51.8% of untreated and ceftiofur-treated cows experience clinical cure at 6 and 12 d after metritis diagnosis, respectively.However, it is unclear if cows that recover from metritis earlier have better performance than cows that recover later.Our hypothesis is that cows with an early cure (within the first 5 d after diagnosis) will have improved milk production, reproductive performance, and decreased culling rates compared with late-cure cows (between 6 and 14 d after diagnosis).We expect that both early and late metritiscure cows perform better than cows that do not cure by 14 d after metritis diagnosis.
Therefore, the major objective of this multi-location randomized clinical trial was to evaluate the effect of 2 doses of ceftiofur on clinical cure, milk production, reproductive performance, and culling in cows diagnosed with metritis.The secondary objective was to evaluate the effect of timing of metritis clinical cure (evaluated at 5 and 14 d after diagnosis) on milk production, reproductive performance, and culling in dairy cows, regardless of therapy.

MATERIALS AND METHODS
All procedures were reviewed and approved by the Texas Tech University Institutional Animal Care and Use Committee (protocol 19025-02).

Herds Description
A total of 422 cows were enrolled in a randomized clinical trial from January 2020 to November 2021 in 4 commercial dairy herds located in Texas (Herd A), California (Herd B), and Florida (Herds C and D).Herds A, B, C, and D had 2,900, 4,300, 1,800, and 450 lactating Holstein cows, respectively.Cows in herds A, B, and C were milked twice a day, and cows in herd D were milked 3 times daily.The herds' rolling average milk yield ranged from 10,000 to 12,500 kg per cow across all herds.Cows were housed in dry-lot pens in herd A and in free-stalls in herds B, C, and D. Post- partum cows were fed a TMR twice daily with diets formulated to meet or exceed the suggested nutrient concentrations in rations for postpartum cows (NRC, 2001).

Enrollment Criteria and Treatment allocation
Metritis diagnosis followed a systematic monitoring protocol.After morning milking, vaginal discharge (VD) was evaluated on 4, 7, and 10 d in milk (DIM) by members of the research team.Each geographical location (TX, CA, and FL) had a different research team.Additionally, cows with signs of systemic illness (lack of appetite, dullness, decreased milk production) identified by farm employees had their vaginal contents examined by the research team; hence, cows could be diagnosed with metritis on days other than 4, 7, and 10 DIM.Vaginal discharge was assessed using the Metricheck device (Simcro Limited, Hamilton, New Zealand), and was visually inspected and scored using a modified 0 to 5 scale: 0 = no secretion material retrieved, 1 = clear mucus, 2 = flecks of pus in the vaginal discharge, 3 ≤ 50% pus in the vaginal discharge, 4 > 50% pus in the vaginal discharge, 5 = watery, fetid, red-brownish vaginal discharge (Chenault et al., 2004).Cows with fetid, watery, red-brownish VD (score = 5) were diagnosed with clinical metritis.Additionally, at enrollment, rectal temperature (RT) was assessed using a digital thermometer (GLA Agriculture Electronics, CA), and fever was defined as RT >39.5°C.Cows that received antimicrobial treatments within 14 d before parturition, received antimicrobials or antiinflammatory drugs after calving, were diagnosed with other diseases (e.g., mastitis, pneumonia) at time of metritis diagnosis, and that had calving resulting in abortion (gestation length less than 260 d) fetotomy, C-section, or uterine prolapse were not eligible to be enrolled in the study.
Cows were blocked by parity (primiparous and multiparous) and herd, and then randomly enrolled in the study on day of metritis diagnosis.Cows were allocated into 2 different groups: CON (n = 211; primiparous = 110, multiparous = 101) and CEF (n = 211; primiparous = 110, multiparous = 101).Randomization was completed in Excel 2019 (Microsoft Corp.), and pre-randomized sheets were used by the research crew to enroll cows in the study.Cows allocated in the CEF group received 6.6 mg/kg of ceftiofur crystalline-free acid (Excede, Zoetis, Kalamazoo, MI) on the day of diagnosis and 3 d later.Cows allocated in the CON remained untreated at time of metritis diagnosis.Both groups could receive supportive therapy, such as propylene glycol, calcium boluses, vitamins, and intravenous fluids.In addition, a subset of 399 non-metritic cows (NMET) were randomly selected and included in this data set for reproductive performance, culling and milk production comparison purposes.Those cows matched metritic cows based on parity, herd and calving period of metritic cows enrolled in the study.Of note, one cow was removed from NMET group because she was a duplicate.

Metritis cure definition, escape therapy, and data collection.
Clinical cure was evaluated by each research team at d 5 (early cure, ECURE) and 14 (late cure, LCURE) after metritis diagnosis and was characterized as VD score <5.To ensure proper welfare for study cows, escape therapy was considered for all treatment groups based on the following signs: persistent dehydration, anorexia, depression, systemic shock, or any other clinical signs that could be attributable to metritis with or without elevated rectal temperature.Escape therapy need was assessed by farm employees.They were blinded to the treatment groups, and administered the escape therapy following standard farm protocols, which included alternative antibiotics (e.g., oxytetracycline, penicillin, ampicillin).In total, 24 cows in the CEF and 30 in the CON group received escape therapy after enrollment and therefore were considered as non-cured.Additionally, 31, 6, 13, and 4 cows received escape therapy in herds A, B, C, and D, respectively.Cows that received escape therapy, died, or were sold because of metritis were considered as non-cured.Information regarding monthly milk production, parity, dystocia, twins, stillbirth, male offspring, retained placenta (RP), and sold/ death was extracted from the herds' software databases.The herd management software used by Herd A, Herd B, Herd C, and Herd D was BoviSync Herd Management Software (Bovisync), DairyComp 305 (Valley Ag Software), PCDart (Dairy Records Management Systems), and AfiFarm 5.4 (Afimilk), respectively.Daily milk weights were determined by milking parlors and extracted from the herd's software database as monthly measurements.Retained placenta was defined as membranes not detached by 24 h after parturition, and it was diagnosed by farm personnel.Body condition score (BCS) was evaluated by each research crew at enrollment in all metritic cows based on a 5-point scoring system (Ferguson et al., 1994).

Statistical Analyses
Considering previous findings (Chenault et al., 2004;McLaughlin et al., 2012), we calculated our sample size to detect a statistical difference in cure risk between CON and CEF cows (62% and 75%, respectively).As-Menta et al.: Effectiveness of ceftiofur for metritis therapy suming α = 0.05 and a power of 80%, a total sample size of 400 cows was necessary to enable detection of this difference.Considering approximately 5% attrition, 422 animals were enrolled in the study.With this number of animals enrolled, we would be able to detect a difference in milk yield of 1.4 kg/d considering a standard deviation of 5 kg/d (α = 0.05, β = 0.20).This enrollment number (i.e., 422) would also enable us to detect a difference in survival rate for pregnancy from 0.73 to 0.60, assuming α = 0.05 and β = 0.20, which was previously reported by Oliveira et al., 2020.Sample size calculation was undertaken using MedCalc.version 18.11.6software (MedCalc Software, Mariakerke, Belgium), and it was not accounted for farm effect.Descriptive statistics for the number of animals enrolled per group, parity, dystocia, vaginal laceration, stillborn calves, RP, BCS and RT at enrollment was performed using the chi-squared and ANOVA functions of JMP 14 (SAS Institute Inc., Cary, NC).
Logistic regression models were fitted in SAS (SAS Institute Inc.) using the GLIMMIX procedure to examine the effect of the treatment (CEF vs CON) on cure at 5 and 14 d after diagnosis.The LSMEAN option was used to calculate the adjusted risk of cure.The independent categorical variables (and biologically plausible interactions) offered to the models were treatment (CEF or CON), fever (yes or no), parity (primiparous or multiparous), dystocia (yes or no), RP (yes or no), vulvovaginal laceration (yes or no) and stillbirth (yes or no).The continuous variables offered to these models were DIM at diagnosis, RT (only offered to models if fever and fever*treatment were not retained in the models), and BCS.
To assess the effect of CEF and cure at 5 and 14 d after metritis diagnosis on time to pregnancy and time to culling, multivariable Cox proportional hazard models were fitted using the PHREG procedure in SAS.For the reproduction outcome, cows were right-censored if not diagnosed as pregnant before culling, or until 300 DIM.For time to culling, cows were right-censored if they were alive at the end of the data collection period, also fixed as 300 DIM.To illustrate and estimate the median calving to pregnancy interval, and the median time to culling (with their respective 95% CI) for the independent variables of interests, Kaplan-Meier survival analysis was performed using MedCalc.Additionally, logistic regression models were fitted to the data to assess the odds of pregnancy and odds of culling by treatment and metritis cure status.
Repeated measures linear models were fitted to the data using the MIXED procedure of SAS to assess the effect of treatment, cure by 5 (ECURE) and 14 d (LCURE) after metritis diagnosis on milk production.The data comprised of repeated measures of monthly milk yield (kg/d).To account appropriately for withincow correlation, the error term was modeled by imposing a first-order autoregressive covariance structure for all models; the selection of this covariance structure was based on the lowest Akaike's information criterion.The interaction terms factor × time were forced into all models.The SLICE option was used to assess the mean differences by month of lactation using the Tukey-Kramer multicomparison adjustment method.Visual evaluation of the distribution plot of the studentized residuals was used to confirm that the residuals were normally distributed.
For all multivariable models described above evaluating the effect of treatment or metritis cure status on milk yield, reproductive performance, and culling, initial models including only metritic cows (CON vs CEF or ECURE vs LCURE vs NCURE) were built, to assess the potential importance of to the model of variables that were not available for cows in NMET group and interaction terms between treatment and cure status with those variables.Hence, interaction terms between treatment and variables such as fever, DIM at diagnosis, BCS, and vulvovaginal laceration were assessed.If none of those variables were confounders (changed the estimate of treatment or cure status in more than 20%) or interaction terms were not retained in the models, final models including NMET cows were fitted to the data.Therefore, independent categorical variables (and biologically plausible interactions) offered to the final multivariable models fitted to milk yield, reproduction, and culling data were treatment (CEF, CON, and NMET) or metritis cure (ECURE, LCURE, NCURE, or NMET), parity (primiparous or multiparous), dystocia (yes or no), RP (yes or no), and stillbirth (yes or no).
For all multivariable models described above, the manual backward stepwise selection procedure was applied, and variables and interaction terms were retained in the model if P < 0.10.The variables treatment and parity, or metritis cure and parity were forced into all models evaluating the effect of treatment or timing of metritis cure on outcome reported herein.Herd was included as a random variable in all multivariable models.

Descriptive statistics and the effect of treatment on clinical cure
Descriptive statistics regarding the number of animals enrolled per group, number of cows in each parity, number of cows that experienced calving problems (dystocia, vaginal laceration, stillborn calves, RP), Menta et al.: Effectiveness of ceftiofur for metritis therapy and average BCS and RT, and proportion of cows with fever at enrollment are presented in Table 1.In Herd A, 107, 104, and 200 cows were included in groups CEF, CON, and NMET, respectively.In Herd B, 52, 54, and 100 cows were included in the groups CEF, CON, and NMET, respectively.In Herd C, 36, 34, and 69 cows were included in groups CEF, CON, and NMET, respectively.In Herd D, 16, 19, and 30 cows were included in groups CEF, CON, and NMET, respectively.The only difference observed between CEF and CON groups was in the proportion of cows that experienced stillbirth, which was 2 times greater among CEF cows than CON counterparts.Also, the average DIM at enrollment was numerically greater for CEF cows (P = 0.08).As expected, cows in NMET group had lower proportion of cows that experienced calving related issues when compared with CEF and CON cows (P < 0.01).

Menta et al.: Effectiveness of ceftiofur for metritis therapy
Table 1.Descriptive statistics for the cows enrolled in a randomized controlled trial evaluating the effect of ceftiofur therapy for metritis.Cows diagnosed with metritis (MET; watery, fetid, red-brownish vaginal discharge) were randomly allocated to received subcutaneous administration of 6.6 mg of ceftiofur crystalline-free acid/kg of BW administered at metritis diagnosis and 72 h later (CEF) or to remain untreated (CON).Non-metritic cows (NMET) were randomly selected and included in the study for milk yield, reproduction, and culling comparison purposes  (Ferguson et al., 1994).
Ceftiofur therapy did not improve milk yield for metritic cows (P = 0.37).The average milk yield during the first 10 mo of lactation for CEF, CON, and NMET cows was 36.00 (95% CI = 33.8-38.1), 36.1 (95% CI = 33.9-38.2), and 36.9 (95% CI = 34.8-39.4) kg/d, respectively.The monthly milk production by treatment group is illustrated in Figure 2. The variables retained in this model were parity (P < 0.01), month of lactation (P < 0.01), treatment (CEF, CON, NMET; P = 0.06), and the interaction between treatment and month of lactation (P < 0.01).Although no differences in average milk production were observed between CEF and CON cows in the first 10 mo of lactation (P = 0.37), NMET produced more milk in the first and second months of lactation (P < 0.01).In the first month of lactation, NMET cows produced 3.1 and 2.2 kg/d more milk when compared with CEF and CON cows, respectively (P < 0.01).In the second month of lactation, NMET cows produced 3.1 and 2.9kg/d more milk when compared with CEF and CON cows, respectively (P < 0.01).Moreover, CEF cows produced 2.2 kg/d less milk than NMET counterparts in the third month of lactation (P < 0.01).Finally, NMET tended to produce 1.56 and 1.47 kg/d more milk than CEF and CON cows in the fourth month of lactation, respectively (P ≤ 0.08).

DISCUSSION
In North America, ceftiofur is the first-choice treatment for metritis therapy because its labeled use is  not followed by a milk withhold period (Garzon et al., 2022).However, with new evidence demonstrating that ceftiofur therapy for metritis can potentially increase the risk of antimicrobial resistance in dairy farms (Ossa-Trujillo et al., 2023), further research reinforcing the evidence supporting the use of ceftiofur for the treatment of metritis is needed.Our study was designed to evaluate the effect of systemic ceftiofur administration on metritis cure risk, reproductive performance, milk yield, and culling up to 300 DIM in dairy cows.The efficacy of ceftiofur in increasing metritis cure risk is better documented in the literature than its impact on lactational performance (Haimerl and Heuwieser, 2014;Garzon et al., 2022).Our study included herds from 3 geographical locations within the US which might increase the external validity of our findings in comparison to a previous report that also reported on the effect of ceftiofur therapy on performance of metritic cows (de Oliveira et al., 2020).However, it is essential to highlight that this study was still conducted with a limited number of herds.Our data set demonstrated that increased the cure rate by 14 and 10 percentage points compared with CON cows, at 5 and 14 d after diagnosis, respectively.Our results confirm the findings of previous studies that reported on the efficacy of systemic ceftiofur therapy for metritis (Chenault et al., 2004;McLaughlin et al., 2012;de Oliveira et al., 2020).Chenault et al. (2004)  ).The scientific literature has consistently shown that ceftiofur is efficacious against bacteria associated with metritis development, such as Fusobacterium necrophorum and Escherichia coli (Malinowski et al., 2011;Bicalho et al., 2012;Jeon et al., 2015;Jeon et al., 2021).In addition, an in vitro study demonstrated ceftiofur's effectiveness against E. coli collected from intrauterine fluid of cows affected with metritis (Malinowski et al., 2011).The current findings reinforce ceftiofur's use to improve metritis clinical cure outcomes.However, it also demonstrates that more than half of cows recover from metritis 14 d after initial diagnosis without the need for antimicrobial treatment.Thus, reinforcing the need for the development of targeted therapy strategies that can lead to the reduction of antimicrobial drug, especially given the importance of judicious use of antimicrobial drugs in animal agriculture (O 'neill, 2015).
Although systemic administration with ceftiofur improved clinical outcomes when evaluating clinical cure, no benefits of systemic ceftiofur on milk production, reproduction, and culling of metritic cows were observed.We expected to observe a positive effect of ceftiofur therapy on the reproductive performance of metritis cows, as the literature suggests that ceftiofur administration increases the hazard of pregnancy in metritic cows (Piccardi et al., 2016;de Oliveira et al., 2020).Oliveira et al. ( 2020) demonstrated that metritic cows treated with ceftiofur had 156 median days open compared with 181 d for untreated metritic cows.Piccardi et al. (2016) demonstrated similar results in grazing cows, observing that ceftiofur-treated cows had an increased hazard of pregnancy compared with untreated animals.The present first study was the first one to evaluate the impact of ceftiofur on the reproductive performance of metritic cows and not identify any benefits.Although our findings are inconsistent with the literature, it is reasonable to expect that ceftiofur-treated cows would have an increased hazard  of pregnancy, since we demonstrated that treatment led to clinical cure, and cure has been associated with an increased hazard of pregnancy (Figueiredo et al., 2021).
In our study, we also did not observe any effect of ceftiofur administration on milk yield.In a recent study, it was observed that ceftiofur-treated cows produced more milk compared with untreated counterparts, but this effect was only observed during the first 60 DIM (de Oliveira et al., 2020).However, similar to our find-ings, a study performed on grazing cows found that ceftiofur therapy for metritis did not lead to increased milk production (Piccardi et al., 2016).The authors hypothesized that different results among papers evaluating milk production in metritic cows may be explained by different metritis diagnostic criteria and/ or ceftiofur protocols used in those experiments.For instance, Piccardi et al. (2016) used a single dose of 6.6 mg/kg of ceftiofur crystalline-free acid at time of metritis diagnosis, while Oliveira et al. ( 2020) used a protocol consisting of 2 doses of 6.6 mg/kg of ceftiofur crystalline-free acid 72 h apart.In addition, we observed that NMET cows had improved milk production only at the first 2 mo of lactation compared with metritic cows.Metritic cows are expected to produce less milk than healthy cows because of their lower feeding and rumination time during the first weeks after lactation (Stangaferro et al., 2016;Merenda et al., 2021).Other studies have demonstrated that losses in milk production mainly occur in early lactation (Wittrock et al., 2011;Stangaferro et al., 2016).
We observed that systemic ceftiofur administration did not impact culling of metritic cows, consistent with a previous report (de Oliveira et a., 2020).The lack of effect of treatment on culling was expected because treatment did not impact milk yield and reproductive performance, which are known factors associated with culling decisions in dairy operations (Bascom and Young, 1998).We observed a higher likelihood of culling for CEF and CON cows compared with NMET cows.Therefore, our results support other studies that demonstrated a higher risk of culling for metritic cows (Wittrock et al., 2011;Mahnani et al., 2015).
In our study, the criterion for metritis diagnosis and therapy was solely the presence of a watery, red-brownish, and foul-smelling vaginal discharge, regardless of presence of fever.It is important to highlight that in other studies, the presence of fever was part of therapy criteria (Pohl et al., 2016;Lora et al., 2021), including in the initial clinical trials evaluating the efficacy of ceftiofur for metritis therapy (Chenault et al., 2004;McLaughlin et al., 2012).Nevertheless, multiple others do not use fever to define which cows should be treated (Giuliodori et al., 2013;Piccardi et al., 2016;Lomb et al., 2018), indicating that the criterion for therapy is inconsistent within the current literature.This inconsistency in using fever in the decision of treating metritis with antibiotics is also observed in the field (Espadalama et al., 2018).Although treating metritis only in the presence of fever may represent judicious use of antimicrobials, Guiliodori et al. ( 2013 strated that cure rates and economic losses associated with metritis were not linked with RT at diagnosis.Leaving cases of metritis untreated is associated with US$ 250 in profit losses (Silva et al., 2021), which suggests that metritis may warrant antimicrobial therapy even in the absence of fever.However, more research is needed to test this hypothesis.Herein, although RT was associated with metritis cure, the effect of ceftiofur in all outcomes evaluated was not dependent on fever.These findings at least partially support the idea that basing metritis treatment decisions solely on fever may not represent the best metritis therapy targeted approach.However, observations from recent studies suggest that RT, combined with other variables such as DIM at diagnosis, circulating concentration of biomarkers of inflammation and metabolism, and behavior data, can be used to predict metritis cure (Machado et al., 2020;Prim et al., 2024;Menta et al., In press).More research is needed to evaluate whether selective strategies for metritis therapy based on different predictive models can lead to significant decreases in antimicrobial use while maintaining adequate welfare and similar productivity.Given the societal pressure, concerns related to antimicrobial resistance development, and high self-cure rates, blanket treatment of metritis no longer represent an adequate therapeutic strategy within the modern dairy industry.
Another objective of our study was to evaluate the effect of time to metritis cure (ECURE and LCURE) on lactation performance up to 300 DIM.We observed that approximately 29% of metritic cows failed to cure by 14 d after diagnosis.Figueiredo et al. (2021) recently reported that cure was associated with improved performance, and that cows that do not experience clinical cure by 10 d after antibiotic therapy have impaired reproductive performance, decreased milk production, and are more likely to leave the herd than cows that recover from metritis.Because the literature suggests that performance is associated with cure, we hypothesized that cows experiencing early clinical cure from metritis would have better performance than cows experiencing cure later.Understanding if performance differences exist between ECURE and LCURE cows could improve how the efficacy of new strategies for metritis therapy is evaluated.However, contrary to our hypothesis, we observed that lactational performance of cows with early metritis cure was not different than cows recovering from metritis later.However, cows that did not experience any clinical cure had decreased milk yield up to 90 DIM, decreased hazard of pregnancy, and increased hazard of culling compared with cows that underwent cure at 5 and 14 d after diagnosis.
Metritis cure failure might be linked to the load of pathogenic bacteria in the uterus at time of treatment.For instance, Jeon (2018) reported that cows that failed to cure had increased concentrations of Fusobacterium, Bacteroides, and Porphyormonas compared with cured counterparts.The increased uterine presence of these pathogens may indicate persistent uterine infection, and thus inflammation, which can partially explain the detrimental impacts of failed clinical cure on reproduction, milk yield, and survivability.For instance, Figueiredo et al. (2021) demonstrated that cows that do not undergo clinical cure by 10 d have persistent endometrial inflammation manifested as cytological endometritis, which could partially explain the reproductive losses observed both by us and by Figueiredo et al. (2021).Additionally, Figueiredo et al. (2021) observed that failure of metritis clinical cure only resulted in milk production losses among multiparous cows.In contrast, we found that the effect of clinical cure on milk yield was not dependent on parity, as cows that failed to cure from metritis had lower milk yield in comparison to cows that underwent cure, regardless of parity.In a literature review, Bradford (2015) explained how inflammation can be associated with impaired milk production throughout lactation.Indeed, cows that did not recover from metritis had greater circulating concentration of haptoglobin, an acute phase protein that serves as a biomarker of inflammation in cows (Machado et al., 2020).It is possible that the sustained inflammation experienced by non-cured cows may have led to the reduced productivity observed in our study, as Machado et al. (2020) demonstrated that non-treated metritic cows with high haptoglobin concentration (>0.54 mg/mL) at metritis diagnosis had decreased milk yield in comparison to non-treated metritic cows with low blood haptoglobin level (≤0.54 mg/mL).Because cure is associated with improved reproductive performance and increased milk yield, it is not surprising that uncured metritic cows were more likely to leave the herd than cows that recovered from metritis.This outcome corroborates Figueiredo et al. (2021) finding that cows that failed to cure up to 10 d after metritis diagnosis had an increased culling hazard.However, it is important to highlight that although clinical cure was associated with improved lactational performance, recovering from metritis earlier did not lead to improved outcomes.
One of the limitations of our study is the limited sample size to make comparisons between ECURE and LCURE, but our findings suggest that evaluating metritis at 14 d post initial diagnosis is a sufficient method for assessing the efficacy of metritis therapy strategies.Another limitation of our study is linked to its multi-herd nature.As described earlier, although improving external validity, conducting a study in multiple herds from different geographical regions added challenges to data collection and standardization.We would like to highlight that the research crew conducting the study in each geographical location received training before onset of the study to ensure that data collected across states were standardized.However, we acknowledge that we did not conduct any prior assessments on intra-observer agreement on data collection related to vaginal discharge score, vulvovaginal laceration score, and BCS.Perhaps this is a challenge that is difficult to overcome when vaginal discharge score and metritis incidence is evaluated in research using the current methodology.For instance, Sannmann and Heuwieser (2015) reported that the inter-and intraobserver reliability of the vaginal discharge characteristics (odor, color viscosity) is only moderate.Additionally, challenges related to a lack of consistency on general management and data collected among the commercial herds included in the study also represent study limitations.However, we would like to highlight that the participating farms have common management practices regarding fresh cow monitoring and reproductive programs, even though differences between farms exist.To address this, we included farm as a random variable in all our statistical multivariable models.

CONCLUSIONS
Collectively, our findings suggest that although ceftiofur treatment for metritis leads to improved clinical cure outcomes, it does not necessarily lead to increased productivity, improved reproductive performance, and lower culling.However, our results should be interpreted in context with findings from previous studies that reported economic losses associated with leaving metritic cows untreated in an untargeted manner (Piccardi et al., 2016;de Oliveira et al., 2020;Silva et al., 2021).Additionally, the impact of leaving cows untreated on animal welfare is not fully understood, although recent findings suggests that behavior of metritic cows is not influenced by ceftiofur therapy (Prim et al., 2022).Considering the rising concern regarding the judicious use of antimicrobials within the dairy industry, we believe that our findings provide further evidence that a reduction in antimicrobial use for metritis therapy can be achieved without impairing important economic outcomes.Further research is needed to develop alternative therapeutic strategies that target only cows that will benefit from antimicrobial therapy, which will be pivotal to optimizing the use of antibiotics in dairy farms.
Figure 2. Least squares means of milk production by month of lactation for cows diagnosed with metritis and treated with ceftiofur (CEF; n = 211), cows diagnosed with metritis and not treated (CON; n = 211) and cows not diagnosed with metritis (NMET; n = 399).The variables retained in this model were parity (P < 0.01), month of lactation (P < 0.01), treatment (CEF, CON, NMET; P = 0.06), and the interaction between treatment and month of lactation (P < 0.01).Herd was included as a random variable.Error bars represent 95% CI.A indicates P < 0.05 for CEF vs. NMET; B indicates P < 0.05 for CON vs. NMET.
Menta et al.: Effectiveness of ceftiofur for metritis therapy Menta et al.: Effectiveness of ceftiofur for metritis therapy

Table 3 .
Cox's proportional hazard analyses evaluating the effect ceftiofur therapy for metritis on the hazard of pregnancy and culling up to 300 after parturition 1 CEF = subcutaneous administration of 6.6 mg of ceftiofur crystalline-free acid/kg of BW administered at metritis diagnosis and 72 h later. 2 CON = untreated control cows.3 NMET = cows not diagnosed with metritis.

Table 4 .
Cox's proportional hazard analyses evaluating the effect of metritis cure 1 at 5 and 14 d after diagnosis on the hazard of pregnancy and culling up to 300 after parturition 1Cure was defined as absence of fetid watery vaginal discharge.2 ECURE = cows cured by 5 d after metritis diagnosis.3 LCURE = cows cured by 14 d after metritis diagnosis.3 NCURE = cows not cured by 14 d after metritis diagnosis.4 NMET = cows not diagnosed with metritis.