The effect of repeated, four-weekly eprinomectin treatment on milk production in pasture-based, seasonally-calving dairy cattle

Veterinary Parasitology 189 (2012) 250–259

Contents lists available at SciVerse ScienceDirect

Veterinary Parasitology journal homepage: www.elsevier.com/locate/vetpar

The effect of repeated, four-weekly eprinomectin treatment on milk production in pasture-based, seasonally-calving dairy cattle W.A. Mason ∗ , W.E. Pomroy, K.E. Lawrence, I. Scott Institute for Veterinary, Animal and Biomedical Science, Massey University, Private Bag 11222, Palmerston North 4442, New Zealand

a r t i c l e

i n f o

Article history: Received 12 March 2012 Received in revised form 3 May 2012 Accepted 7 May 2012 Keywords: Cattle nematodes Eprinomectin Milk-production Ostertagia milk ELISA Selective treatments Dairy

a b s t r a c t A randomised clinical trial from the North Island of New Zealand was conducted to assess the effect of repeated anthelmintic treatment on milk production, and to assess factors that affect treatment response. Nine hundred and twenty three multiparous, lactating dairy cattle from three pasture-based, spring-calving dairy herds were enrolled in this trial. Within each herd, cattle were stratified on age and calving date, and were randomly allocated to treatment (n = 319) or control (n = 604) groups. The treatment group received ≥0.05 mg/kg of topical eprinomectin every 28 days for eight treatments during lactation. Pooled-milk from treated cows and bulk-milk samples were obtained at each treatment and analysed with an Ostertagia antibody ELISA, expressed as optical density ratios (ODR). Bi-monthly milk data were collected and expressed as energy-corrected milk (kg/day; ECM). A linear mixed model was used to analyse ECM, with cow as the random effect. The effect of anthelmintic treatment on days from calving, and start-of-mating, to conception were analysed with Cox-proportional hazard models. ODR values ranged from 0.6 to 1.3; there were no differences in ODR between herds (p = 0.12), or between pooled-milk from treated cows and bulk-milk (p = 0.26). Repeated treatments had no effect on daily ECM yields (p = 0.74). However, there was a significant treatment × herd interaction (p = 0.03); treatment increased ECM in one herd by 0.781 kg/cow/day (p = 0.015), but resulted in a non-significant decrease in the other two herds. A curvilinear interaction existed between days-in-milk and treatment response (p = 0.039); the greatest treatment effect occurred during mid-lactation. Previous year milk production (p = 0.46) and age (p = 0.11) did not influence the effect of treatment on ECM. Treatment had no effect on any reproductive parameter. In conclusion, under New Zealand pastoral conditions, anthelmintic treatment increased milk production in one herd, but had no effect in two other herds. Further work is needed to identify why this variation in gastro-intestinal parasitism occurs. © 2012 Elsevier B.V. All rights reserved.

1. Introduction Gastro-intestinal parasitism (GIP) is one of the major diseases of grazing cattle worldwide (Vercruysse and Claerebout, 2001). Although clinical disease attributed to GIP is rare in adult dairy cattle, parasitism can still have

∗ Corresponding author. Tel.: +64 027 269 2050. E-mail address: [email protected] (W.A. Mason). 0304-4017/$ – see front matter © 2012 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.vetpar.2012.05.003

profound sub-clinical consequences. In a review of 87 clinical trials assessing the effect of anthelmintic treatment on milk production, treatment increased milk production by a median of 0.63 kg/cow/day (p < 0.001; Gross et al., 1999). A meta-analysis of 75 trials revealed a smaller, but still significant, increase in milk production of 0.35 kg/cow/day (Sanchez et al., 2004). Due to the predominantly pasturebased system of lactating dairy cattle in New Zealand, exposure to nematode larvae is likely to be high. Three previous New Zealand trials have demonstrated an advantage

W.A. Mason et al. / Veterinary Parasitology 189 (2012) 250–259

of anthelmintic treatment for adult dairy cattle (McQueen et al., 1977; Bisset et al., 1987a; McPherson et al., 2001). It can be concluded that GIP has a negative effect on milk production in New Zealand dairy cattle. Despite the overall positive effect of anthelmintic treatment, considerable variation exists between cows (Bisset et al., 1987a; McPherson et al., 2001), herds (Ploeger et al., 1989) and trials (Sanchez et al., 2004). From a total of 75 clinical trials reviewed for the meta-analysis, 59 reported a positive result and 16 a negative result; only 40 (53%) of these detected a significant (p ≤ 0.05) increase in milk production (Sanchez et al., 2004). This contradiction of available evidence makes it difficult to recommend anthelmintic control strategies. It would be of greater importance to be able to identify animals and/or herds that are likely to respond positively to anthelmintic treatment; to ‘define a threshold’ for when treatment would be beneficial (Vercruysse and Claerebout, 2001). Routine diagnostic tests for GIP in young animals, i.e. faecal egg counts (FEC) and pepsinogen, have a low sensitivity in detecting and quantifying GIP in adult dairy cattle (Eysker and Ploeger, 2000). Consequently, a milk production response to anthelmintic treatment has historically been used as an indirect indicator of GIP in adult dairy cattle. An Ostertagia antibody ELISA has shown great promise in predicting whether a herd (Charlier et al., 2007) or cow (Sanchez et al., 2002a) will respond positively to anthelmintic treatment – i.e. as a diagnostic tool for subclinical parasitism. However, the farming systems in the two previously mentioned studies were very different to those in New Zealand, and the ELISA test has not been verified in New Zealand cattle. Previous New Zealand trials have involved anthelmintic treatment during the dry-period (Bisset et al., 1987a), or a single treatment in early-lactation (McPherson et al., 2001). Multiple lactating treatments of a potent anthelmintic should minimise the effects of GIP. The aims of this study were to evaluate the impact of GIP on milk production in pasture-based dairy cattle via repeated anthelmintic treatments during lactation, and to determine if the response to treatment is dependent on age, production, herd, or daysin-milk (DIM). A secondary aim was to assess the effect of GIP on reproductive outcomes.

2. Materials and methods 2.1. Herds Dairy cattle from three spring-calving herds from the lower North Island of New Zealand were enrolled in a randomised clinical trial. The herds were conveniently chosen from dairy farms serviced by Massey University Teaching Hospital. Key selection criteria for the enrolled herds included herd sizes greater than 450 cows, farm managers with expected high compliance with the treatment protocols, and access to accurate milk production and reproductive data. The enrolled farms had not used any anthelmintics on adult dairy cattle in the 12 months prior to initiation of the study and all cattle had access to pasture for greater than 18 h per day during lactation.

251

2.2. Treatment protocol This was a randomised clinical trial, with cattle allocated to the treatment group receiving ≥0.05 mg/kg of topical eprinomectin (Ivomec Eprinex, Merial NZ Ltd.) along the back from the withers to the tail-head. Treatments were repeated at four-weekly intervals for eight treatments and were performed during afternoon milking. The study population consisted of all multiparous lactating cows that had calved by 49 days after the start of calving. These enrolment dates were chosen so that cattle would be first treated as early in lactation as possible, whilst ensuring sufficient numbers were enrolled. The number of treated animals per herd was designed to be less than 20% of the total herd size. Power analysis was conducted to determine appropriate group size to detect a daily milk-solid (fat + protein; kg) increase of 0.06 kg (SD 0.16 kg), which was double the increase noted by McPherson et al. (2001) in response to a single anthelmintic treatment. On a individual herd basis, a range of 90–119 cows were randomly allocated to the treatment group, which were matched to at least one control cow; power for detecting a difference in each herd ranged from 82.5% to 90.8%. Overall across the three farms, 319 cows were allocated to the treatment group and 604 cows to the untreated control group, resulting in an overall power of 99.9%. Within each herd, eligible cattle were stratified first on age, then on calving date. Every third cow was allocated to the treatment group, starting with a computer-generated random number between one and three for each age strata. Cattle remained within the allocated group for the entirety of the study. Treated and control animals were mixed immediately after they had been milked and no placebo was used. An animal was classified as belonging to a certain breed if the animal’s parentage was registered as ≥75% of a certain breed. If an animal did not belong to one breed, it was recorded as a cross-bred animal. As there were only 16 cows classified as Jersey cattle, these were combined with cross-bred cattle to form a new breed variable, ‘other’. 2.3. Milk production and activity data Milk production data for Herds 1 and 2 were available from bi-monthly herd-testing, as part of New Zealand Livestock Improvement Corporation’s breeding-worth scheme (LIC, Hamilton, NZ). Twenty four hour test-day milk data were obtained from each of these occasions, consisting of milk volume, fat and protein percentage and DIM at each herd-test. Information was also obtained on the milk-yield for each cow from the previous season (Milk2010). Daily information on milk production and composition was available for Herd 3 from in-shed technology supplied by Afimilk® , which measures a range of parameters at each milking, including milk volume, milk fat and milk protein. Dates midway between the other two herds’ herd-tests were used to pick appropriate test-days. Milk production data from these four dates were collated together with information from the other two herds. In Herd 3, pedometers were attached to the legs of all cows; information was collected on the average number of

252

W.A. Mason et al. / Veterinary Parasitology 189 (2012) 250–259

steps taken per hour, updated at every milking. This information was collated and expressed as the average number of steps taken per hour over the entire lactation. Multivariable linear regression was carried out, assessing the significance of treatment, breed and age on activity. Criteria for including and excluding variables were as stated for reproductive analysis below. 2.4. Reproductive data Based on pregnancy ultrasound diagnoses, data were gathered on the proportion of animals that conceived within the first six weeks of the mating period (6WICR), proportion of pregnant animals during the entire mating period (PR), days from calving to conception and days from the start of mating to conception. 2.5. Faecal nematode egg counts Faecal samples from 20 to 25 randomly selected control cows were collected rectally at every treatment visit to assess nematode faecal egg counts and were subsequently cultured to assess nematode species present. In addition, 25 faecal samples were collected for FEC from a random sample of the treated cohort prior to the first treatment. A modified McMaster egg counting technique was used, where each egg counted represented 50 eggs per gram of faeces (Stafford et al., 1994). 2.6. Ostertagia ostertagi antibody ELISA At the time of each treatment, pooled samples were randomly collected from approximately 30 treated cows, and a bulk-tank sample from the milk vat, containing milk form treated and untreated cows, were collected and analysed for Ostertagia antibody ELISA optical density ratios (ODR; Svanovir® , Boehringer Ingelheim Svanova, Uppsala, Sweden). 2.7. Statistical analysis The primary outcome measure of interest was daily energy-corrected milk (ECM; kg), with 1 kg of ECM standardised to 3.096 MJ ME/kg milk (Santschi et al., 2011). This was calculated using the following formula (Santschi et al., 2011): 0.2595 × milk weight (kg) + 12.55 × fat (kg) + 7.39 × protein (kg) Secondary milk production outcomes of interest were test-day fat percentage, protein percentage, milk solids (kg fat + protein; MS) and milk volume. 2.7.1. Descriptive statistics and univariate associations The number of animals treated in each herd, breed, age group and herd-test were tabulated. Continuous variables were assessed for normality using histograms and quantile–quantile plots. ODRs were compared using univariate linear regression; treatment or herd were considered significant if p ≤ 0.05.

Correlations between predictors were identified using Pearson’s correlations to identify any collinearity between predictors. Unconditional associations between predictors and ECM were assessed using student’s t-test; predictors were included in a multivariable model if p < 0.25. The relationships between continuous predictors and ECM were assessed with locally-weighted scatter-plot smoothers. Due to small numbers of cattle in each age group from eight to twelve, these cattle were recategorised as being eight years of age or older. Age was modelled as a linear continuous variable and as a categorical variable; the model with the lowest Akaike information criterion (AIC) was chosen. The effect of DIM on milk production varies throughout lactation, and polynomial functions including quadratic and the Wilmink’s function (a linear DIM term and DIM raised to the power of 0.05; Nodtvedt et al., 2002) were included separately in a multivariable model to assess the relationship. The DIM function with the lowest AIC was included in the final multivariable model. DIM was centred around its mean (154.6 days) to reduce collinearity when polynomial terms were used. 2.7.2. Multivariable analysis In order to account for the hierarchical and longitudinal data structure, a mixed model with a random effect for cow and an error correlation structure was fitted following the criteria suggested by Zuur et al. (2009) and Dohoo et al. (2009). The clustering of animals within a herd was accounted for by including herd as a fixed effect. Other fixed effects initially included were breed, age, DIM, DIM2 , Milk2010 and treatment. Interactions were only considered between the predictor of interest (treatment) and other significant predictor variables. The fixed effect structure was determined by backward elimination using maximum likelihood; starting with interaction terms, fixed terms were dropped from the model if the log-likelihood ratio (LLR) test between nested models had a significance of p > 0.05, until all remaining fixed effects were significant at p ≤ 0.05. Treatment was retained in the model, regardless of significance. A variable was considered a confounder if coefficients or standard errors of remaining fixed effects altered by ≥20%; these were retained in the model even if not significant. Final results were reported using restricted maximal likelihood. Least-squares means for significant, biologically plausible interactions were inspected, and interaction error-bar plots were utilised to illustrate interactions. Various correlation (variance–covariance) matrices (i.e. compound symmetry, first level auto-regressive, unstructured, auto-regressive with moving averages and no correlation) were modelled for herd-test within cow to account for the longitudinal nature of the data. A first order, auto-regressive (AR1) correlation structure was chosen as the best fit based on AIC and biological relevance (Zuur et al., 2009). Finally, model assumptions of homogeneity of variance and normality of residuals for the fixed and random effects were assessed through graphical techniques, and influential observations were identified and investigated. Apart from least-squares means performed using the PROC MIXED procedure in SAS, all graphics and data analysis

W.A. Mason et al. / Veterinary Parasitology 189 (2012) 250–259

253

Table 1 The number of cows enrolled in the treated (topical eprinomectin treatment, ≥0.5 mg/kg, every 28 days for eight treatments) group and control group, by herd, breed and age. Variable Herd 1 2 3 Breed Holstein-Friesian Other Age 3 4 5 6 7 8+

Treated

Control

Total

119 91 109

241 183 179

360 274 288

151 166

282 320

433 486

39 81 66 54 38 41

79 153 122 101 69 79

118 234 188 155 107 120

were undertaken using R Version 2.10 (R Development Core Team, 2009). The above process was repeated with fat percentage, protein percentage, MS and milk volume as outcome variables, respectively. 2.8. Reproductive parameters Multivariable logistic regression was used to analyse the effect of treatment on PR and 6WICR. Fixed effects included treatment, breed and herd. Fixed effects and interactions were dropped from the model if p-value from a LLR test between nested models was ≥0.05; treatment was retained in the final model regardless of significance. The unconditional relationships between time to conception data and the predictors (treatment, breed or herd) were assessed with Kaplan–Meier graphs and cumulative survival curves. Cox proportional hazard (CPH) models were used to analyse the time to conception survival data. Fixed effects and model selection techniques were the same as were used in the logistic models above. The proportional hazard assumption for the CPH model was assessed by plotting smoothed scaled Schoenfeld residuals against time, and calculating covariate specific score tests (Hosmer et al., 2008). The fit of the model was assessed using Pearson residuals, and the model was checked for outliers and influential observations. 3. Results 3.1. Descriptive statistics The number of animals in each treatment group within herd, breed and age can be seen in Table 1. A total of 923 cows were enrolled; 319 cows were allocated to the treatment group and 604 cows to the control group. Two controls were matched to every treated cow from Herd 1 and 2. Approximately three control cows were matched to two treated cows in Herd 3. Ten cattle from Herd 2 had calving dates recorded incorrectly, and their DIM at first treatment were greater than 360 days. These animals were removed from the trial and were not replaced. Twenty cows died, or were removed

Fig. 1. Distribution of daily herd-test energy-corrected milk (kg/cow/day) for 3516 observations from 923 cows.

from a herd, after enrolment, but prior to the first test date; four were treated cows, and 16 were control cows. The disease rates that occurred on each farm were consistent with the normal incidence on a commercial dairy farm, and did not appear to be related to treatment. Treated animals were allowed to miss one treatment event throughout the course of the trial; if greater than one treatment event was missed, then milk measurements collected after this were excluded from the analysis. However, every animal that met this criterion had died or had been removed from the farm, so did not contribute any further milk production data. The mean DIM for when the first treatment was applied was 31 days (SD = 11.6 days, range 3–60 days). ECM across all four herd-tests was normally distributed with a mean of 21.9 kg/day (SD = 5.26 kg/day; Fig. 1). 3.2. Diagnostic tests The overall mean FEC was 5.1 eggs/gram (median = 0, range 0–150). Ostertagia spp. were the most commonly cultured nematode (70.1% of all cultured larvae), followed by Trichostrongylus spp. (14.7%), Cooperia spp. (12.2%) and all other nematodes (3.0%). Herd 1 had a lower proportion of Ostertagia spp. (41% vs. 79%) and a greater proportion of Trichostrongylus spp. (30% vs. 10%) cultured, compared to the average of the other two herds over the entire lactation. The mean ODR value across all measurements was 0.9 (range 0.6–1.3). There was no difference between herds (p = 0.12; Fig. 2), or between pooled-milk from treated cows and bulk-milk samples (p = 0.26). No seasonal trends in ODR values from the three herds were evident. 3.3. Univariate associations There was a polynomial relationship between DIM and ECM; milk production peaked at 100 DIM. Once centred around its mean, DIM and a quadratic term for DIM provided a better fit for the data than the Wilmink function. ECM increased from three- to six-year-old-cattle, then

254

W.A. Mason et al. / Veterinary Parasitology 189 (2012) 250–259

Fig. 2. Monthly Ostertagia ELISA optical density ratios (ODR) for bulk-milk (solid line) and pooled-milk from eprinomectin-treated cows (dashed line) for the three herds.

decreased in seven-years and older cattle; age was included as a categorical variable in the multivariable model. All predictor variables were significant at p < 0.20, and were included in the initial multivariable model. 3.4. Multivariable analysis All predictors, apart from breed, were included in the final multivariable linear mixed model. Table 2 summarises the significant variables, and their interactions. Treatment had no overall effect on daily ECM (0.107 kg/cow/day; Table 2 Regression coefficients, standard errors and p-values from a multivariable linear mixed-effect model of the effect of repeated topical eprinomectin treatment (every 28 days for eight treatments) on daily energy-corrected milk (kg/cow/day) production over four test days (3157 observations from 812 cows). DIM = days-in-milk; Milk2010 = milk production from the previous season (litres). Variable

Coefficient

Std. error

p-Value

Intercept Treatment Herd Herd 1 Herd 2 Herd 3 Age 3 4 5 6 7 8+ DIM DIM2 Milk 2010 Treatment × DIM Treatment × DIM2 Treatment × herd interaction Treatment × herd 1 Treatment × herd 2 Treatment × herd 3

13.49 1.207

0.606 0.382

<0.001 0.0016 0.038

Reference 1.042 0.713

0.325 0.315 <0.001

Reference 0.414 0.837 0.518 0.133 −1.162 −0.024 −0.00015 0.00169 −0.00021 −6.4E−05

0.376 0.395 0.421 0.448 0.422 0.001 1.62E−05 0.00013 0.00195 2.79E−05

Reference −0.926 −1.226

0.556 0.487

<0.001 <0.001 <0.001 0.91 0.023 0.033

Fig. 3. Least-squares means from a multivariable linear mixed model of daily energy-corrected milk (kg/cow) for eprinomectin-treated (dashed line) and control cattle (solid line) in three herds. Bars represent the standard error of the means. NS = not significant, *p < 0.05.

p = 0.62). However, there was a significant interaction between treatment and herd (Fig. 3; p = 0.033). Treatment increased ECM in Herd 1 by 0.78 kg/cow/day (p = 0.015); there was a non-significant decrease in ECM between treated and control cattle in the other two herds (−0.094 kg/cow/day, p = 0.82 and −0.37 kg/cow/day, p = 0.27 for Herd 2 and Herd 3, respectively). Due to the significant treatment interactions, the coefficient and significance of treatment in Table 2 cannot be interpreted in isolation, without considering herd and DIM. A quadratic interaction existed between DIM and treatment (p = 0.023). There was no difference between treated or control cattle during early-lactation. Cattle responded maximally to treatment during mid-lactation (approximately 150 days post-partum). The difference became smaller after this time, until treated and control cattle were producing equal ECM weights by late-lactation (approximately 250 days post-partum). The shape of this curve was similar for all three farms. There was no evidence of treatment interacting with age (p = 0.11). Despite a different spread of ages amongst the three farms, age did not confound the relationship between treatment and ECM, or affect the herd × treatment interaction. There were large amounts of missing data for Milk2010 (94 cows and 359 milk recordings), predominantly due to 80 cows from Herd 2 that were not present on the farm during the previous lactation. Data were analysed with, and without Milk2010; coefficients and standard errors from fixed variables of interest did not alter greatly. As the relationship between Milk2010 and ECM, and the interaction with treatment, were of interest, this variable was included in the final model. Milk2010 was associated with ECM (p < 0.001). However, there was no interaction between treatment and Milk2010, modelled as a continuous variable (p = 0.46), or, as a categorical variable divided into quartiles (p = 0.51). An error plot of effect of treatment on Milk2010 divided into quartiles can be found in Fig. 4. There was

W.A. Mason et al. / Veterinary Parasitology 189 (2012) 250–259

255

−0.032 kg/cow/day; p = 0.23 for Herd 2 and Herd 3, respectively). 3.6. Effect of treatment on reproduction There was no effect of treatment on PR or the 6WICR; odds ratios for treated compared to control cattle were 1.25 (95% CI 0.72, 2.24) and 1.13 (95% CI 0.83, 1.54) for PR and 6WICR, respectively. Treatment had no effect on either of the days-toconception outcomes; the hazard ratios for treated vs. control cattle from the CPH model were 1.00 (95% CI 0.87, 1.16) for calving-to-conception, and 0.99 (95% CI 0.86, 1.14) for start-of-mating-to-conception. There were no treatment by herd interactions, and there was no effect of time of first treatment on either days-to-conception outcome. 3.7. Effect of treatment on activity Fig. 4. Least-squares means from a multivariable linear mixed model for daily energy-corrected milk (kg/cow) from eprinomectin-treated (dashed line) and control (solid line) cattle within quartiles of total milk-yield (L/cow/year) from the previous lactation for each cow. Bars represent the standard error of the means.

no evidence that production level affected the treatment response. The validity of the model was tested using a variety of measures. Residuals for the fixed and random effects were approximately normally distributed. There were two animals that had greater influence on the model than the remainder of the animals; these were older cattle from Herd 2. The removal of these two animals did not alter the coefficients or standard errors greatly of any of the fixed effects, and no reasons were found to exclude them. 3.5. Secondary milk parameter outcomes Treatment did not affect overall milk volume, MS, fat or protein milk percentage (Table 3). A significant herd × treatment interaction for MS existed (p = 0.03), and there was a trend for a significant interaction for milk volume (p = 0.09). This followed what occurred with ECM, with MS increasing in Herd 1 (0.063 kg/cow/day; p = 0.01), and no significant difference in MS noted in the other two herds (−0.013 kg/cow/day; p = 0.73, and

Table 3 Regression coefficients, standard errors and p-values, along with p-value for herd × treatment interaction from multivariable linear mixed effect models of the effect of repeated topical eprinomectin treatment (every 28 days for eight treatments) on daily milk volume (L/cow/day), daily milksolids (fat + protein; kg/cow/day), daily fat % and daily protein %, with, or without a herd × treatment interaction. Outcome variable

Milk volume Milk-solids Fat % Protein %

Treatment × herd interaction

Treatment effect

Coefficient

SE

p-Value

p-Value

0.005 0.011 0.031 0.032

0.182 0.016 0.045 0.019

0.98 0.50 0.49 0.08

0.09 0.03 0.29 0.84

Activity data from 109 treated and 177 control cattle were available from Herd 3; the mean number of steps taken for all cattle was 206.8/h (SD = 24.5). Age and breed were not significantly associated with the activity, so were not included in the model. There was a trend for treatment to increase the activity of cattle by 5.6 steps/h (p = 0.059) compared to control cattle. 4. Discussion This randomised clinical trial assessed the effect of repeated anthelmintic treatment on milk production in pasture-based dairy cattle in New Zealand, and, some of the factors that may affect this milk production response. Whilst repeated treatment had no overall effect on milk production, treatment did result in a significant increase in milk production in one of the three herds. A number of studies have investigated the impact of anthelmintic treatment on milk production. A review of the literature (Gross et al., 1999) and a meta-analysis (Sanchez et al., 2004) indicated that anthelmintic treatment increased overall milk production. New Zealand data have shown similar findings; a single, early-lactation eprinomectin treatment of cattle from three herds in New Zealand produced a 0.6 L/cow/day (p = 0.005) and 0.03 kg MS/cow/day (p = 0.031) increase compared to controls (McPherson et al., 2001). However, considerable variation exists between clinical trials. Thirty five of 75 (46%) clinical anthelmintic trials did not show a significant difference in milk production between treated and control cattle (Sanchez et al., 2004). There were several possible reasons why no overall effect was found in the present study. Grazing paddocks that had previously been grazed by calves has been identified as an important risk factor for a milk production response to anthelmintics (Bisset et al., 1987b). Across all three farms, no young stock were grazed on any of the three farms, and dairy cattle had no contact with either the grazing calves or paddocks that had been grazed by calves. Although not quantified, this likely decreased the amount of nematode larvae consumed by the cows. A second possible reason was that the study was designed such that less than 20% of the total herd was

256

W.A. Mason et al. / Veterinary Parasitology 189 (2012) 250–259

treated, in an effort to maintain the normal parasite population on pasture. In other studies, milk production responses were 0.4 kg/day lower in clinical trials that involved partial herd treatment, compared to whole herd treatment (p < 0.001; Sanchez et al., 2004). Another possible reason for the lack of effect could be related to the management of cattle post-treatment. It was not possible to keep treated and control cattle separate after each treatment, and cograzing occurred between the treated and control cattle. Licking of topically applied macrocyclic lactones (ML) can result in systemic absorption of the compound (Laffont et al., 2003; Bousquet-Melou et al., 2011). Allo-licking is a recognised outcome of topical ML treatment and partial efficacy to ivermectin has been recorded in untreated heifers managed with topically treated heifers. As treated and control cattle were not kept separate, allo-licking could have allowed control cattle to obtain partial treatment. No attempt was made in this present study to measure whether this occurred. There was considerable variation in response to anthelmintic treatment observed between herds in the present study. The significant increase that occurred in Herd 1 equated to 214.6 kg ECM over a 275-day lactation, or, a 3.6% increase in ECM in treated cattle compared to controls. In addition, total lactation MS production in Herd 1 was increased by 17.3 kg/cow. This increase was of considerable economic benefit. In contrast, milk production decreased in the other two herds in response to treatment, although the differences were not significant. Causal relationships for the difference in treatment response between herds are difficult to ascertain with only three herds enrolled. However, possible explanations will be discussed. A major difference in management of Herd 1 was that greater than 80% of the farm was irrigated; the other two farms did not irrigate. Moisture is required to allow parasites to migrate away from the dung pat to surrounding pastures (Stromberg, 1997) and it would be expected that larvae would be more available under irrigation. However, further work needs to be done on the effect of irrigation and parasite control in dairy cattle before recommendations can be made. Another difference between the herds was that grazed pasture for Herd 1 constituted approximately 10% more of the total annual dry matter, compared to the other two herds, which were fed more maize silage, palm kernel extract and soya bean meal. As pasture is the predominant source of ingested larvae, it would be fair to assume that a greater amount of pasture eaten would result in more larvae ingested. This variation between herds is consistent with the literature. In one large, multi-herd clinical trial from New Zealand, only one of 47 herds demonstrated a statistically significant increase in milk production, despite an overall significant effect (Bisset et al., 1987a). There was even a large amount of variation in the number of herds that demonstrated a positive response, significant or not (Gross et al., 1999). In this study, animals were treated every 28 days for eight treatments. The time between treatments and total treatment duration was designed to minimise the effect of Ostertagia spp. for the majority of the lactation. Persistent activity for topical eprinomectin treatment has been demonstrated at 28 days against Ostertagia spp.

and Cooperia spp., and at least 21 days for Trichostrongylus spp. (Cramer et al., 2000). No formal assessment of eprinomectin efficacy was possible on the three farms, as no young stock were grazed on the properties. It is acknowledged that there is a prevalence of up to 100% of ML-resistant Cooperia on cattle properties in the North Island of New Zealand (Waghorn et al., 2006). However, Cooperia is not considered to be a pathogenic nematode in adult dairy cattle, as immunity is well developed after the first grazing season (reviewed by Armour, 1989). The low proportion of Cooperia spp. cultured from control cattle in the current study is consistent with this statement. To date, there have been no confirmed reports of ML-resistant Ostertagia spp. in cattle in New Zealand. It was, therefore, presumed that the efficacy of eprinomectin was high on all three farms. Treatment did not alter the fat and protein percentage of milk, in agreement with Nodtvedt et al. (2002), McPherson et al. (2001), and Ploeger et al. (1989), and there was no overall difference in MS or milk volume between treated and control cattle. However, a significant treatment × herd interaction existed for MS production. In Herd 1, MS increased proportional to milk volume. In can be concluded that the increase in ECM in Herd 1 was driven by an increase in both milk volume and MS, and not due to changes in the composition of the milk. ODR values were high throughout the study period for bulk-milk and pooled-milk from treated cattle. Extrapolating from overseas ODR data, it would have been expected that milk production should have increased in treated cattle in all three herds (Forbes et al., 2008). The number of herds was small in the current trial, and the ODR sampling was limited. Nevertheless, ODR had no ability to predict a milk response at any stage of the year, and no differences were noted in ODR values between Herd 1 and the other two herds. Early evidence demonstrated that antibody levels to Ostertagia ostertagi were positively associated with a milk production response (Ploeger et al., 1989). More recently, ODR values have been used to predict milk responses to anthelmintic treatment (Sanchez et al., 2002a; Charlier et al., 2007, 2010). High bulk-milk ODR levels were associated with lower average annual herd milk-yield; an increase from an ODR of 0.83 to 1.11 (25th and 75th percentile, respectively) was associated with a 1.1 kg/cow/day decrease in average annual milk-yield (Charlier et al., 2005b). Pastured dairy herds in Belgium with pre-treatment bulk-milk ODR values greater than 0.84 had a 4 kg/cow/day milk increase after treatment (Charlier et al., 2007); there were no differences between treated and control herds for the other ODR categories. Based on ODR values throughout Europe, Forbes et al. (2008) suggested that ODR values greater than 0.5 were likely to cause a reduction in milk-yield, with a linear decrease in milkyield as ODR increased beyond 0.5. However, these authors did not recommend relying solely on ODR cut-offs when investigating GIP in adult dairy cattle. ODR values from the current study were greater than those from dairy farms in Belgium (mean = 0.70, SD = 0.14; Charlier et al., 2007) and Canada (mean = 0.54, SD = 0.26; Sanchez and Dohoo, 2002). They were also greater than the mean ODRs (range 0.3–0.6) from a large European survey across six countries (Forbes

W.A. Mason et al. / Veterinary Parasitology 189 (2012) 250–259

et al., 2008). The higher ODR values found in New Zealand are presumably due to grazed pasture being the predominant feed source on the majority of New Zealand dairy farms; the amount of exposure to pasture was positively associated with ODR levels at both the cow and herd level (Sanchez et al., 2002a; Charlier et al., 2007; Forbes et al., 2008). Further validation is needed in New Zealand to infer meaningful results from ODR values, and to assess if a cutpoint exists that can predict a milk production response. Eprinomectin did not alter ODR in the current study. This is in contrast to the findings of Charlier et al. (2007), where mean ODR values were significantly lower in treated herds compared to placebo herds for eight months following a single treatment. Although in the current study, treated cattle were contributing milk to the bulk-tank, it was still expected that there would be a difference in ODR values between pooled-treated and bulk-tank milk. However, another study did not demonstrate an effect of treatment on ODR (Charlier et al., 2005a), and only whole herd treatment decreased ODR values in Canadian dairy herds (Sanchez and Dohoo, 2002). Serum antibody concentrations are more indicative of the uptake of L3 larvae from pasture, rather than the extent of the current infection (Charlier et al., 2009). Therefore, a reduction in ODR values post treatment may not be a useful tool to determine whether treatment was successful. The milk production response to treatment was maximal during mid-lactation (approximately 150 days post-partum), consistent with the literature (Gross et al., 1999; Sanchez et al., 2004). Mid-lactation or strategic (repeated) treatments during lactation resulted in a 0.4 kg/day increase in milk compared to single dryperiod or calving treatments (Sanchez et al., 2004). In an early New Zealand spring-calving study, repeated monthly anthelmintic treatments starting seven weeks after calving resulted in a significant increase in milk-fat and milk-yield over the second half of lactation (p < 0.01), but had no effect during the first half of lactation (McQueen et al., 1977). In the same study, summer treatments (mid-lactation) were as effective as spring treatments (early-lactation), with respect to lactation milk-yield. An argument has been made for treating early in lactation, as this will result in greater days for a treatment response to occur, potentially increasing total milk production/cow/lactation (Ploeger et al., 1989; Charlier et al., 2010). However, the increased DIM needs to be weighed against the expected milk response and the variation between trials is too great to support, or refute, this suggestion. Some of the observed differences may be the result of different trial designs and farming systems, and care must be taken in extrapolating results from non-seasonal to seasonal production systems. The effect of DIM was confounded by season, and it was not possible to separate the two in this current study. Ingestion of parasite larvae and ODR often follow a seasonal pattern; cow and herd ODR values were significantly greater in the autumn, compared to winter and spring (Charlier et al., 2005b; Sanchez et al., 2005; Charlier et al., 2007). However, although there was low power, there was no evidence of seasonal trends for ODR values across the three farms in the current trial, and the mid-lactation increase in milk production was seen in the absence of a change in ODR

257

values. The period of maximal response coincided with mid-summer, which is often the period of lowest larval pasture contamination due to warm, dry conditions commonly experienced over this period (Bisset and Marshall, 1987). However, this is a generalisation and might not be representative of what occurred on the three farms during that particular year. It is unsure why this occurred, but it may have been related to diet change, or, due to alterations in metabolism and energy partitioning. During the period of highest expected parasitic challenge, autumn, there was no milk response. A possible reason for the lack of difference between treated and control cattle in late lactation may be due to decreasing milk-yields, as the absolute response may be proportional to milk-yield. Absolute increases in milk-yield in response to anthelmintic treatment were greater in Canada than in New Zealand (0.94 kg/cow/day vs. 0.60 kg/cow/day for Canada and New Zealand dairy cattle, respectively; McPherson et al., 2001; Nodtvedt et al., 2002). In the same studies, Canadian dairy cattle had greater daily milk-yields per cow compared to New Zealand dairy cattle (27.75 kg/cow/day vs. approximately 20 kg/cow/day for Canadian and New Zealand dairy cattle, respectively). When reported as a percentage of milk increase compared to controls, there was little difference between the two studies (3.4% vs. 3% for Nodtvedt et al., 2002 and McPherson et al., 2001, respectively). This may explain the lack of effect during late-lactation, as milkyields decreased exponentially in late-lactation. However, if this was true, then following the same logic, it would be expected that higher producing cattle should have greater absolute milk responses. There was no evidence from the current study to conclude that higher producing cattle have a greater response to anthelmintic treatment than lower producing cattle. It has been postulated that genetic selection for increased milk production has resulted in dairy cattle more susceptible to disease processes (Lucy, 2001), including GIP. A similar conclusion has been reached in dairy goats (Hoste et al., 2002). Experimental infection resulted in a greater decrease in milk-yield in high-producing compared to lowproducing dairy goats, and greater milk-yield increases in response to anthelmintic treatment were seen in highproducing dairy goats (Hoste et al., 2002). Less is known on the epidemiological relationship between production level and GIP in dairy cattle. Ploeger et al. (1989) reported a significant positive association between milk production response to anthelmintic treatment over the dry-period and cow production from the previous season. A subsequent study identified that every extra kg milk produced in the previous season was associated with a 1.9%, nonsignificant increase in treatment response the following season (Ploeger et al., 1990). Similarly, data from New Zealand has shown that greater responses occurred in ‘high-quality’ cows compared to ‘low-quality’ cows, as defined by breeding indices (Bisset et al., 1987a). However, the findings of the present study disagree with these early studies. There was no evidence that production level affects milk production response to anthelmintic treatment in the current trial and further work is needed before strategic treatment of high, or low, producing cattle can be recommended.

258

W.A. Mason et al. / Veterinary Parasitology 189 (2012) 250–259

There was no improvement in reproductive parameters following anthelmintic treatment. This is in agreement with other similar studies (Sanchez et al., 2002b; Sithole et al., 2006); a large recent study involving 2381 cows failed to demonstrate an improvement in calving to conception, calving to first insemination and the number of services per conception (Sithole et al., 2006). Another trial showed that whilst there was a trend for a decrease in calving to conception interval in treated cattle, treatment had no effect on the calving to first-service interval (Sanchez et al., 2002b). Therefore, treatment with eprinomectin is unlikely to improve reproduction outcomes. Substantial advances in our understanding of risk factors for GIP in adult dairy cattle have been made over the past decade. However, further work is needed to identify risk factors that can account for the variation in the effect of GIP on milk production. In particular, further epidemiological studies are needed to identify why some herds respond positively to anthelmintics and others do not, and the relevance of ODR values from New Zealand dairy farms. Until then, routine anthelmintic treatment cannot, and should not, be recommended to all adult dairy cattle. Instead, emphasis should be placed on identifying particular animals or herds that are affected by GIP, and on parasite control measures other than anthelmintic treatment. Acknowledgements Thanks to the farm owners and farm managers on the three farms for their help and co-operation, and to all the veterinary students that helped with treating and sample collecting. The authors would also like to thank Anne Tunnicliff and Barbara Adlington for their technical assistance. The Eprinex was funded by Merial NZ Ltd. References Armour, J., 1989. The influence of host immunity on the epidemiology of Trichostrongyle infections in cattle. Vet. Parasitol. 32, 5–19. Bisset, S.A., Marshall, E.D., 1987. Dynamics of Ostertagia spp. and Cooperia oncophora in field-grazed cattle from weaning to 2-year old in New Zealand, with particular reference to arrested development. Vet. Parasitol. 24, 103–116. Bisset, S.A., Marshall, E.D., Morrison, L., 1987a. Economics of a dry-cow anthelmintic drenching program for dairy-cows in New Zealand. Part 1. Overall response in 47 herds. Vet. Parasitol. 26, 107–118. Bisset, S.A., Marshall, E.D., Morrison, L., 1987b. Economics of a dry-cow anthelmintic drenching program for dairy-cows in New Zealand. Part 2. Influence of management factors and other herd characteristics on the level of response. Vet. Parasitol. 26, 119–129. Bousquet-Melou, A., Jacquiet, P., Hoste, H., Clement, J., Bergeaud, J.P., Alvinerie, M., Toutain, P.L., 2011. Licking behaviour induces partial anthelmintic efficacy of ivermectin pour-on formulation in untreated cattle. Int. J. Parasitol. 41, 563–569. Charlier, J., Claerebout, E., De Muelenaere, E., Vercruysse, J., 2005a. Associations between dairy herd management factors and bulk tank milk antibody levels against Ostertagia ostertagi. Vet. Parasitol. 133, 91–100. Charlier, J., Claerebout, E., Duchateau, L., Vercruysse, J., 2005b. A survey to determine relationships between bulk tank milk antibodies against Ostertagia ostertagi and milk production parameters. Vet. Parasitol. 129, 67–75. Charlier, J., Duchateau, L., Claerebout, E., Vercruysse, J., 2007. Predicting milk-production responses after an autumn treatment of pastured dairy herds with eprinomectin. Vet. Parasitol. 143, 322–328. Charlier, J., Hoglund, J., von Samson-Himmelstjerna, G., Dorny, P., Vercruysse, J., 2009. Gastrointestinal nematode infections in adult dairy cattle: impact on production, diagnosis and control. Vet. Parasitol. 164, 70–79.

Charlier, J., Vercruysse, J., Smith, J., Vanderstichel, R., Stryhn, H., Claerebout, E., Dohoo, I., 2010. Evaluation of anti-Ostertagia ostertagi antibodies in individual milk samples as decision parameter for selective anthelmintic treatment in dairy cows. Prev. Vet. Med. 93, 147–152. Cramer, L.G., Pitt, S.R., Rehbein, S., Gogolewski, R.P., Kunkle, B.N., Langholff, W.K., Bond, K.G., Maciel, A.E., 2000. Persistent efficacy of topical eprinomectin against nematode parasites in cattle. Parasitol. Res. 86, 944–946. Dohoo, I., Martin, W., Stryhn, H., 2009. Veterinary Epidemiologic Research, 2nd edition. VER Inc., Charlottetown, Prince Edward Island, Canada, p. 865. Eysker, M., Ploeger, H.W., 2000. Value of present diagnostic methods for gastrointestinal nematode infections in ruminants. Parasitology 120, S109–S119. Forbes, A.B., Vercruysse, J., Charlier, J., 2008. A survey of the exposure to Ostertagia ostertagi in dairy cow herds in Europe through the measurement of antibodies in milk samples from the bulk tank. Vet. Parasitol. 157, 100–107. Gross, S.J., Ryan, W.G., Ploeger, H.W., 1999. Anthelmintic treatment of dairy cows and its effect on milk production. Vet. Rec. 144, 581–587. Hosmer, D.W., Lemeshow, S., May, S., 2008. Applied Survival Analysis: Regression Modelling of Time-to-Event Data, 2nd edition. John Wiley and Sons, Inc., Hoboken, New Jersey, p. 408. Hoste, H., Chartier, C., Le Frileux, Y., 2002. Control of gastrointestinal parasitism with nematodes in dairy goats by treating the host category at risk. Vet. Res. 33, 531–545. Laffont, C.M., Bousquet-Melou, A., Bralet, D., Alvinerie, M., Fink-Gremmels, J., Toutain, P.L., 2003. A pharmacokinetic model to document the actual disposition of topical ivermectin in cattle. Vet. Res. 34, 445–460. Lucy, M.C., 2001. ADSA Foundation Scholar Award – reproductive loss in high-producing dairy cattle: where will it end? J. Dairy Sci. 84, 1277–1293. McPherson, W.B., Gogolewski, R.P., Slacek, B., Familton, A.S., Gross, S.J., Maciel, A.E., Ryan, W.G., 2001. Effect of a peri-parturient eprinomectin treatment of dairy cows on milk production. N. Z. Vet. J. 49, 106–110. McQueen, I.P.M., Cottier, K., Hewitt, S.R., Wright, D.F., 1977. Effects of anthelmintics on dairy cow yields. N. Z. J. Exp. Agric. 5, 115–119. Nodtvedt, A., Dohoo, I., Sanchez, J., Conboy, G., DesCoteaux, L., Keefe, G., 2002. Increase in milk yield following eprinomectin treatment at calving in pastured dairy cattle. Vet. Parasitol. 105, 191–206. Ploeger, H.W., Kloosterman, A., Bargeman, G., Vonwuijckhuise, L., Vandenbrink, R., 1990. Milk-yield increase after anthelmintic treatment of dairy-cattle related to some parameters estimating helminth infection. Vet. Parasitol. 35, 103–116. Ploeger, H.W., Schoenmaker, G.J.W., Kloosterman, A., Borgsteede, F.H.M., 1989. Effect of anthelmintic treatment of dairy cattle on milk production related to some parameters estimating nematode infection. Vet. Parasitol. 34, 239–253. R Development Core Team, 2009. R: A Language and Environment for Statistical Computing, 2.10.1 edition. R Foundation for Statistical Computing, Vienna, Austria. Sanchez, J., Dohoo, I., 2002. A bulk tank milk survey of Ostertagia ostertagi antibodies in dairy herds in Prince Edward Island and their relationship with herd management factors and milk yield. Can. Vet. J. 43, 454–459. Sanchez, J., Dohoo, I., Carrier, J., DesCoteaux, L., 2004. A metaanalysis of the milk-production response after anthelmintic treatment in naturally infected adult dairy cows. Prev. Vet. Med. 63, 237–256. Sanchez, J., Dohoo, I., Leslie, K., Keefe, G., Markham, F., Sithole, F., 2005. The use of an indirect Ostertagia ostertagi ELISA to predict milk production response after anthelmintic treatment in confined and semi-confined dairy herds. Vet. Parasitol. 130, 115–124. Sanchez, J., Dohoo, I., Nodtvedt, A., Keefe, G., Markham, F., Leslie, K., DesCoteaux, L., Campbell, J., 2002a. A longitudinal study of gastrointestinal parasites in Canadian dairy farms – the value of an indirect Ostertagia ostertagi ELISA as a monitoring tool. Vet. Parasitol. 107, 209–226. Sanchez, J., Nodtvedt, A., Dohoo, I., DesCoteaux, L., 2002b. The effect of eprinomectin treatment at calving on reproduction parameters in adult dairy cows in Canada. Prev. Vet. Med. 56, 165–177. Santschi, D.E., Lefebvre, D.M., Cue, R.I., Girard, C.L., Pellerin, D., 2011. Complete-lactation milk and component yields following a short (35-d) or a conventional (60-d) dry period management strategy in commercial Holstein herds. J. Dairy Sci. 94, 2302–2311. Sithole, F., Dohoo, I., Leslie, K., DesCoteaux, L., Godden, S., Campbell, J., Keefe, G., Sanchez, J., 2006. Effect of eprinomectin pour-on treatment

W.A. Mason et al. / Veterinary Parasitology 189 (2012) 250–259 around calving on reproduction parameters in adult dairy cows with limited outdoor exposure. Prev. Vet. Med. 75, 267–279. Stafford, K.J., West, D.M., Pomroy, W.E., 1994. Nematode worm egg output by ewes. N. Z. Vet. J. 42, 30–32. Stromberg, B.E., 1997. Environmental factors influencing transmission. Vet. Parasitol. 72, 247–264. Vercruysse, J., Claerebout, E., 2001. Treatment vs non-treatment of helminth infections in cattle: defining the threshold. Vet. Parasitol. 98, 195–214.

259

Waghorn, T.S., Leathwick, D.M., Rhodes, A.P., Jackson, R., Pomroy, W.E., West, D.M., Moffat, J.R., 2006. Prevalence of anthelmintic resistance on 62 beef cattle farms in the North Island of New Zealand. N. Z. Vet. J. 54, 278–282. Zuur, A.F., Ieno, E.N., Walker, N.J., Saveliev, A.A., Smith, G.M., 2009. Mixed Effects Models and Extensions in Ecology with R. Springer, New York, p. 574.