- Email: [email protected]
Relationships between bulk-tank antibodies to Ostertagia ostertagi and herd-management practices and measures of milk production in Nova Scotia dairy herds
Preventive Veterinary Medicine 47 (2000) 79±89
Relationships between bulk-tank antibodies to Ostertagia ostertagi and herd-management practices and measures of milk production in Nova Scotia dairy herds F.J. GuitiaÂna,*, I.R. Dohooa, R.J.F. Markhamb, G. Conboyb, G.P. Keefea a Department of Health Management, Atlantic Veterinary College, University of Prince Edward Island, Charlottetown, PEI, Canada C1A 4P3 b Department of Pathology and Microbiology, Atlantic Veterinary College, University of Prince Edward Island, Charlottetown, PEI, Canada C1A 4P3
Received 24 June 1999; accepted 24 June 2000
Abstract Monthly bulk-tank milk samples were obtained from 415 Nova Scotia dairy herds in each of the months of July±September 1998 and tested using an indirect microtitre ELISA against a crude saline-extract, whole-worm Ostertagia ostertagi antigen. ELISA results (optical densities (ODs)) were consistent across months (r 0:85) but there was considerable variation among herds. A questionnaire was sent by mail to all producers; information on management factors that would potentially influence parasite burdens in the herds was obtained from 239 farms. Data on annual milk production, summer milk production (July±September) and seasonal decline in milk production were obtained from the Animal Productivity and Health Information Network (APHIN) database. Associations between management practices and ODs, and between ODs and milk-production parameters were studied. Some management practices known to be associated with parasite burdens had expected directions of association with the ODs, giving supporting evidence that the ELISA is a reasonable measure of parasite burden. Most notably, ODs were increased with greater exposure of heifers or milking cows to pasture. ODs were not associated with either annual milk production or seasonal decline in milk production. However, there was a substantial relationship between the herd OD value and the level of milk production during the summer. An increase in the OD from 0.58 to 0.83 (the interquartile range of ODs)
* Corresponding author. Present address: Laboratorio Lema & BandõÂn, C./Lepanto 5, E-36201 Vigo, Spain. Tel.: 34-986437400; fax: 34-986431145 E-mail address: [email protected] (F.J. GuitiaÂn).
0167-5877/00/$ ± see front matter # 2000 Elsevier Science B.V. All rights reserved. PII: S 0 1 6 7 - 5 8 7 7 ( 0 0 ) 0 0 1 5 6 - 2
80
F.J. GuitiaÂn et al. / Preventive Veterinary Medicine 47 (2000) 79±89
was associated with a reduction in production of 1.25 kg/cow/day. # 2000 Elsevier Science B.V. All rights reserved. Keywords: Cattle-parasitological disease; Ostertagia ostertagi; ELISA; Dairy herd management; Milk production
1. Introduction Internal parasites (particularly Ostertagia ostertagi) are economically detrimental for dairy herds. Negative effects include a reduction in heifer growth rates and decreased levels of milk production. A recent review by Gross et al. (1999) reported a median increase of 0.63 kg/day in milk production following treatment with anthelmintics in a wide range of studies that were based on various study designs, treatment protocols and products. Unfortunately, monitoring of parasite burdens in dairy herds currently requires the use of faecal egg counts (FEC). FEC are highly variable both within and among animals are time consuming and expensive. In general, egg output from internal parasites in cows is low and FEC do not reliably reflect the parasite burden (Gross et al., 1999). An ELISA developed in the Netherlands (Keus et al., 1981) and evaluated under Dutch conditions (Kloosterman et al., 1993) was recently evaluated in dairy herds in Quebec, Canada (Dohoo et al., 1997). In this last study, ELISA optical densities (ODs; computed as an average of the five most-recently calved heifers) had an acceptable correlation with the average FEC taken from the same five animals (r 0:49 to 0.55). In that study, the within-herd variability (as measured by the intra-class correlation coefficient) was lower for the ELISA ODs than it was for FEC Ð suggesting that antibody titres might be less variable cow-to-cow than FEC. However, comparing ELISA results to FEC assumes that FEC are a good indicator of parasite burdens in adult cows Ð which is certainly not the case at the cow level. Consequently, other means of evaluating the ELISA (to determine if it is a useful measure of parasite burden) are needed. One approach is to determine if factors which would be expected to influence parasite burdens have a predictable effect on ELISA ODs. A second approach is to evaluate relationships between ELISA-test results and measures of productivity. A recently completed study in Prince Edward Island determined a strong relationship between ELISA results from bulk-tank milk samples and the amount of seasonal (summer-fall) decline in milk production that the herd experienced (Hovingh, 1998). A third method would be to monitor milk-yield response to anthelmintic therapy to determine if the ELISA ODs were predictive of treatment response. The objectives of this study were twofold. First, the study tested whether the ELISA measures parasite burden by examining the associations between test results and farmmanagement practices known to be associated with parasite burdens. Second (assuming that the ELISA is reflective of parasite burdens), the study evaluated the associations between the ELISA results and measures of milk production.
F.J. GuitiaÂn et al. / Preventive Veterinary Medicine 47 (2000) 79±89
81
2. Materials and methods 2.1. Study design and study population This was a cross-sectional study in which measures of antibody level, herdmanagement practices and measures of milk production were determined between December 1997 and November 1998. The study population consisted of all dairy herds in Nova Scotia (a province on the east coast of Canada). Dairy herds in this province range in size from approximately 15 to 500 milking and dry cows. Management practices range from total-confinement housing with a total mixed ration to herds being pastured in the summer and being fed a component-based ration. However, most herds have some access to pasture during the summer. 2.2. Sample collection and laboratory methods Bulk-tank milk samples were obtained from all Nova Scotia dairy herds for each of the months of July±September 1998 (N 415). Samples were stored frozen and subsequently tested using an indirect microtitre ELISA against a crude saline-extract, whole-worm O. ostertagi antigen as previously described (Keus et al., 1981). ODs from blank wells in each plate were subtracted from the sample values and the average of the results from 3 months was computed. Negative control sera derived from 3-month-old helminth-naive calves and positive control sera derived from hyperimmunized calves were included on each plate to ensure consistency of results between plates but sample ODs were not adjusted by the values of the positive or negative controls. For the same test (using the whole-parasite antigen from Cooperia), likelihood ratios for herd average optical densities (HOD, obtained by averaging individual milk samples' ODs) for predicting high/low herd average FEC had been computed by Dohoo et al. (1997). After reducing HOD to two categories, both the sensitivity and specificity of the HOD had been estimated to be 79%. 2.3. Farm-management practices At the end of August, all producers received a brief closed-response questionnaire by mail asking for basic information on factors that would potentially influence parasite burdens in their herds. Questions referred to extent of exposure to pasture, common use of pastures by heifers and dry or milking cows, use of parasite-control procedures in heifers and milking cows and pasture-management practices. (In this paper, the term ``heifer'' refer to nulliparous cows; definitions of all management-practice variables are presented in Table 1 and follow the wordings used on the questionnaire.) 2.4. Milk-production data Twelve months (from December 1997 to November 1998) of individual-cow milkproduction data for all herds on production recording (Atlantic Dairy Livestock Improvement Corporation Ð ADLIC) were extracted from the Animal Productivity and
F.J. GuitiaÂn et al. / Preventive Veterinary Medicine 47 (2000) 79±89
82
Table 1 Description and descriptive statistics of qualitative variables used in the analysis of associations between ELISA ODs, herd-management practices and milk production in Nova Scotia dairy herds (1998) Variable Heifer housing Heifer_confinement Heifer_yard Heifer_paddock Heifer_pasture
Description
Heifers kept in total-confinement housing Heifers given access to a concrete or gravel-surface exercise yard Heifers given access to a small grassed paddock for exercise Heifers given access to pasture for grazing
N 239
Heifer_pasture2 No pasture Pasture
Heifers given access to pasture (pasture or paddock)
239
Heifer_tx_fall No Yes
Heifers dewormed in the fall of 1997
239
Heifer_tx_spring No Yes
Heifers dewormed in the spring of 1998
239
Heifer_tx_bolus No Yes
Heifers given a sustained-release bolus in summer 1998
239
Heifer_dry_cows No Yes
Heifers graze on pastures also grazed by dry cows
213
Heifer_milk_cows No Yes
Heifers graze on pastures also used by milking cows
183
Milking_cow_housing Cow_confinement Cow_yard Cow_paddock Cow_pasture
Cows kept in total-confinement housing Cows given access to a concrete or gravel-surface exercise yard Cows given access to a small grassed paddock for exercise Cows given access to pasture for grazing
239
Cow_pasture2 No Yes
Cows given access to pasture (pasture or paddock)
239
Cow_tx_oral
Milking cows dewormed with oral product in the last 12 months
232
Milking cows treated with pour-on or injectable treatment at drying off
232
No Yes Cow_tx_dry_of No Yes
% of factor
Mean OD
8 1
0.43 0.52
5
0.66
86
0.70
9 91
0.44 0.69
50 50
0.67 0.67
72 28
0.68 0.65
93 7
0.67 0.71
51 49
0.66 0.73
90 10
0.73 0.69
11 3
0.41 0.46
9
0.57
77
0.73
15 85
0.42 0.71
95 5
0.67 0.70
88 12
0.67 0.67
F.J. GuitiaÂn et al. / Preventive Veterinary Medicine 47 (2000) 79±89
83
Table 1 (Continued ) Variable
Description
N
Cow_tx_calving No Yes
Cows treated with pour-on or injectable treatment at calving
232
Control_access No Yes
Pastures are managed using some form of controlled access grazing (rotation or strip) vs continuous access
209
Manure No Yes
Manure spread on pastures used for grazing
213
Dragged No Yes
Pastures dragged or harrowed
211
Clipped No Yes
Pastures clipped
214
% of factor
Mean OD
93 7
0.67 0.65
31 69
0.73 0.72
64 36
0.71 0.75
64 36
0.72 0.72
32 68
0.72 0.72
Health Information Network (APHIN) database (Dohoo, 1992). From these data, herd average values of individual-cow milk production (kg/cow/day) were computed for annual milk production, summer milk production (July±September) and relative fall production (average of September±November expressed as a proportion of average of May±July). Herd averages for annual and summer (July±September) days in milk, lactation number and log somatic-cell count were also computed. 2.5. Data analyses Descriptive statistics were computed. Linear-regression models evaluated multivariable relationships between management factors and ODs and between ODs and milkproduction parameters. All analyses were carried out using Version 5.0 of the statistical package Stata (Stata Corporation, College Station, TX) and associations were deemed significant if P 0:05 (all two-sided). Two linear-regression models were fit to evaluate the relationships between management practices and ELISA-test results. The first was based on data from all herds and included level of exposure to pasture as a predictor along with other management factors that applied to all herds. The second model was based only on herds which allowed the cows to have access to pasture and was fit to evaluate the effects of pasture-management variables on ODs. Both linear models were developed using a backwards stepwise elimination of non-significant effects (P > 0:05 by the Wald's test). Once significant main effects had been identified, all two-way interactions were assessed for significance. Multiple linear regression (backwards elimination) was also used to evaluate the associations between the ELISA ODs and each of these three measures of milk production:
84
F.J. GuitiaÂn et al. / Preventive Veterinary Medicine 47 (2000) 79±89
herd average annual milk production, herd average summer milk production, and seasonal decline in milk production (fall production divided by spring production). Before fitting these models, the variables representing heifer and milking cow housing were dichotomized into ``pasture'' vs ``no pasture'' (pasture/paddock vs confinement/ yard) to represent any direct exposure to grassed areas or not. Also, in each of these models, ODs were forced into the model to evaluate the direct relationship between OD and milk production. To avoid confounding, a number of factors which are known to have strong relationships with production parameters (e.g. pasture access, herd average days in milk, herd average lactation number, herd average log somatic-cell count), were made available to the linear-regression models. 3. Results and discussion 3.1. Descriptive statistics Milk samples were tested from 415 herds and usable questionnaires were returned by 239 producers (58% of those contacted) (Table 1). Production data were obtained for 243 herds on ADLIC of which 147 had returned usable questionnaires. The wide range of values for the ODs suggests considerable variation between herds in levels of antibodies to internal parasites (Table 2). The interquartile range for ODs was Table 2 Description and descriptive statistics of quantitative variables used in the analysis of associations between ELISA ODs, herd-management practices and milk production in Nova Scotia dairy herds. Data from 402 Nova Scotia dairy herds for which average ODs were available and 243 Nova Scotia dairy herds for which milkproduction data were available (1998) Variable
Description
Optical_density Milk_annual
Average of three monthly values of OD from ELISA Average milk production (kg/cow/day) from December 1997 to November 1998 Average milk production (kg/cow/day) in July±September 1998 Fall milk production (September±November 1998) divided by spring milk production (May±July 1998) expressed as proportion Herd average natural log somatic-cell counts from December 1997 to November 1998 Herd average natural log somatic-cell counts in July±September 1998 Herd average days-in-milk from December 1997 to November 1998 Herd average days-in-milk in July±September 1998 Herd average lactation number from December 1997 to November 1998 Herd average lactation number in July±September 1998
Milk_summer Milk_fall LogSCC_annual LogSCC_summer DIM_annual DIM_summer Lactation_annual Lactation_summer
Mean
S.D.
Range
0.69 25.7
0.19 4.5
0.16, 1.14 12.6, 39.3
26.6
4.7
11.9, 40.6
0.85
0.11
0.44, 1.13
5.39
0.42
4.07, 6.23
5.49
0.48
3.88, 6.97
172
17
132, 270
177 2.81
23 0.42
108, 270 1.87, 4.48
2.83
0.47
1.94, 5.50
F.J. GuitiaÂn et al. / Preventive Veterinary Medicine 47 (2000) 79±89
85
0.58±0.83. Within-herd correlations between monthly samples were high (r 0:85). The consistency within herds, across months suggests that little is to be gained through the testing of multiple samples if the test were to be included in a routine herd-monitoring program. Descriptive statistics of qualitative and quantitative variables used in subsequent analyses are presented in Tables 1 and 2, respectively. 3.2. Associations between management factors and ELISA results Results for the model that included level of exposure to pasture as a predictor along with other management factors that applied to all herds are presented in Table 3. This model was fitted using all available data once the best final model was selected. Increasing exposure of both heifers and cows to pasture resulted in steadily increasing ELISA ODs. Cows that spent part of their time grazing on pasture had ODs that were (on an average) 0.27 units higher than cows kept in confinement housing. Deworming heifers in the spring was associated with significantly reduced ODs (with pasture use analytically controlled). Other parasite-control practices (e.g. fall treatments of heifers, sustainedrelease bolus use and a variety of treatments of milking cows) did not have significant associations with ODs. This model had an R2 of 0.42 indicating that pasture exposure and spring treatment of heifers explain a reasonable amount of the variability in parasite± antibody levels. The model based only on herds which allowed cows to have access to pasture (Table 4) shows that allowing heifers to graze the same pastures as dry cows resulted in elevated ODs but spring treatment of heifers lowered the ODs. Spreading manure on pastures also was associated with increased ODs. However, the R2 of this model was only 0.06. Correlation coefficients were computed to check for the possibility of serious colinearity problems. Coefficients were calculated in two ways: using the subset of records for which all variables were present and using all records for which each individual pair of values was present. Apart from correlations between different
Table 3 Regression of effects of housing and management practices on test results from an Ostertagia ELISA applied to bulk-tank milk samples (239 Nova Scotia dairy herds in 1998; R2 0:42) Variable
b
Confidence interval
P
Intercept Heifer_confinement Heifer_yard Heifer_paddock Heifer_pasture Heifer_tx_spring Cow_confinement Cow_yard Cow_paddock Cow_pasture
0.34 Baseline 0.08 0.14 0.15 ÿ0.06 Baseline 0.00 0.10 0.27
0.27, 0.41 ± ÿ0.14, 0.29 0.03, 0.25 0.07, 0.22 ÿ0.11, ÿ0.02 ± ÿ0.11, 0.12 0.02, 0.20 0.21, 0.34
0.00 ± 0.47 0.01 0.00 0.00 ± 0.96 0.02 0.00
86
F.J. GuitiaÂn et al. / Preventive Veterinary Medicine 47 (2000) 79±89
Table 4 Regression of effects of herd-management practices on test results from an Ostertagia ELISA applied to bulktank milk samples (192 Nova Scotia dairy herds which used pasture for grazing in 1998; R2 0:06) Variable
b
Confidence interval
P
Intercept Heifers_dry_cows Manure Heifer_tx_spring
0.71 0.04 0.04 ÿ0.04
0.67, 0.74 0.00, 0.08 0.00, 0.08 ÿ0.09, ÿ0.00
0.00 0.05 0.05 0.05
categories of a single variable, the highest correlation is 0.31 excluding the possibility of serious colinearity problems. All of the observed results from the linear regressions are consistent with accepted ideas about the associations of management factors with parasite burdens (Urquhart et al., 1987). Lack of significant association with different parasite-control practices could reflect that they eliminate adult parasites without having an important effect at the herd level parasite control or could be due to an insufficient power to detect effects. 3.3. Associations between ELISA results and production parameters Descriptive statistics of production parameters for different levels of milking cow housing are shown in Table 5. The regression model evaluating the associations between ELISA ODs and annual milk production (R2 0:34) shows that increasing days in milk, herd age, somatic-cell count Table 5 Descriptive statistics of three measures of milk production (annual, summer and seasonal decline) for different levels of milking cow housing. Data from 147 Nova Scotia dairy herds for which both herd-management data and milk-production data were available (1998) Variable
N
Mean
S.D.
Annual milk production Cow_confinement Cow_yard Cow_paddock Cow_pasture
18 5 13 111
29.34 31.05 29.74 25.54
4.80 2.24 2.47 4.26
Summer milk production Cow_confinement Cow_yard Cow_paddock Cow_pasture
18 5 13 111
29.08 30.30 29.18 25.27
5.25 2.01 3.39 4.54
Seasonal decline in milk production Cow_confinement Cow_yard Cow_paddock Cow_pasture
18 5 13 111
0.89 0.91 0.87 0.84
0.11 0.02 0.07 0.10
F.J. GuitiaÂn et al. / Preventive Veterinary Medicine 47 (2000) 79±89
87
and use of pasture for grazing had the expected directions of association with this parameter. OD was not significantly (b ÿ3:1, P 0:11) associated with annual milk production. The model evaluating the association with the seasonal decline in milk production (R2 0:06) shows that there is also no significant association between OD and the seasonal decline in milk production. Exposure to pasture was the only factor which exhibited a significant association with this parameter. These results are in contrast to those reported from PEI herds, where OD scores were highly correlated to the seasonal decline in milk production (Hovingh, 1998). However, the amount of seasonal decline demonstrated by Nova Scotia herds was considerably less than that shown by PEI herds (relative fall production of 0.85 in Nova Scotia herds vs 0.75 in PEI herds). As a result, the power of our study to detect factors which would affect the decline would be lower. The results from the regression model evaluating the effects of factors on summer (July±September) production is shown in Table 6. The ELISA OD had a significant negative association with daily milk production during these months (b ÿ5:0, P 0:02), as did grazing, increasing days in milk and somatic-cell counts. With an interquartile range 0.58±0.83 and a coefficient of ÿ5.0, a herd at the 25th percentile for OD would be expected to produce 1.25 kg (5:0 0:83 ÿ 0:58) less milk per head per day than a herd at the 75th percentile. This represented the effect of the parasite burden after controlling for exposure to pasture and several other potential confounding variables. However, the R2 was only 0.28. The above effect of parasite burden on summer milk production must be interpreted with caution for a couple of reasons. First, it is based on the assumption that the OD is a valid measure of parasite burden. If it is a measure which only roughly approximates parasite burden, then it is likely that the effects of the parasite burden will be underestimated since non-differential information bias will tend to bias the results toward the null. On the other hand, there may have been unmeasured confounding variables which affected this estimate and biased it towards or away from the null. For example, exposure to pasture was measured in a very crude manner (pasture vs no pasture). If increasing levels of exposure to pasture resulted in both higher ODs and lower milk production, then the estimate of the effect of the ODs may be biased upwards. Confirmation of the relationship between parasite burdens as predicted by OD values and lost milk production will most likely come from trials in which response to anthelmintic treatment is compared to previous estimates of OD. Such trials have the Table 6 Regression of effects of ELISA ODs and other herd parameters on the herd average daily milk production during July±September, i.e. milk_summer (147 Nova Scotia dairy herds on summer pasture for which both herdmanagement data and milk-production data were available in 1998; R2 0:28) Variable
b
Confidence interval
P
Intercept Optical_density Cow_pasture2 DIM_summer LogSCC_summer
57.79 ÿ5.02 ÿ2.53 ÿ0.05 ÿ3.12
47.67, 67.91 ÿ9.20, ÿ0.84 ÿ4.69, ÿ0.37 ÿ0.09, ÿ0.02 ÿ4.62, ÿ1.63
0.00 0.02 0.02 0.00 0.00
88
F.J. GuitiaÂn et al. / Preventive Veterinary Medicine 47 (2000) 79±89
potential to both confirm the validity of the ELISA as a monitoring tool for parasite burdens and better quantify the effect of parasite burden on milk production. Previous studies that have examined the ability of ELISA-test results to predict milk-production response to anthelmintic treatment have been inconclusive. One study (Ploeger et al., 1989) found a positive correlation (r 0:36) between the herd average milk-production response and the herd average Ostertagia antibody titre while a subsequent study (Ploeger et al., 1990) did not find such a relationship. The first study was carried out in 1985 and the second in 1986. The authors suggested that differences in the overall level and chronology of exposure, probably related to the different meteorological conditions in both years, could explain the results. Different anthelmintic treatments (ivermectin and albendazole) were used in both studies and in both cases anthelmintic treatment resulted in a significant increase in milk production. 4. Conclusions In our study, some of the factors that would be expected to affect parasite burdens in dairy herds had their predicted directions of association with the ELISA ODs. Evaluation of a test's ability to measure parasite burdens in dairy herds requires the use of surrogate ways because of the limitations of FEC in lactating dairy cows, and in the absence of other plausible explanations for the observed associations, it seemed reasonable to us to assume that the ELISA is measuring parasite burdens in the herds. This, taken with other previously published evidence of the value of the ELISA as a monitoring tool, suggests that the ELISA may be a useful tool for monitoring parasite burdens in dairy herds. There appeared to be a substantial relationship between the herd OD value and the level of milk production during the summer in herds exposed to grass. This suggests that herds with high OD readings may benefit from more-intensive parasite-control programs in the milking herd, assuming that the test is measuring parasite burden. Acknowledgements The authors thank the Nova Scotia Milk Producers Association and the Nova Scotia Agri-Futures Fund for their support of this project and Ms. Judy Sheppard for her technical assistance. References Dohoo, I.R., 1992. Dairy APHIN Ð an information service for the dairy industry in Prince Edward Island, Canada. Prev. Vet. Med. 12, 259±268. Dohoo, I.R., Caldwell, V., Markham, R.J.F., Conboy, G., Boucahrd, E., Descoteaux, L., 1997. Evaluation of an ELISA for monitoring parasite burdens in dairy herds. In: Proceedings of the International Symposium on Veterinary Epidemiology and Economics, Paris, France.
F.J. GuitiaÂn et al. / Preventive Veterinary Medicine 47 (2000) 79±89
89
Gross, S.J., Ryan, W.G., Ploeger, H.W., 1999. Anthelmintic treatment of adult dairy cows and the effect on milk production. Vet. Rec. 144, 581±587. Hovingh, E., 1998. An investigation into factors affecting summer/fall milk production and profitability in PEI dairy herds. Ph.D. Thesis. University of Prince Edward Island, Canada. Keus, A., Kloosterman, A., Van den Brink, R., 1981. Detection of antibodies to Cooperia spp. and Ostertagia spp. in calves with the enzyme linked immunosorbent assay (ELISA). Vet. Parasitol. 8, 229±236. Kloosterman, A., Verhoeff, J., Ploeger, H.W., Lam, T.J.G.M., 1993. Antibodies against nematodes in serum, milk and bulk tank milk samples as possible estimators of infection in dairy cows. Vet. Parasitol. 47, 267± 278. 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. Ploeger, H.W., Kloosterman, A., Bargeman, G., Wuijckhuise, L.V., Van den Brink, R., 1990. Milk yield increase after anthelmintic treatment of diary cattle related to some parameters estimating helminth infection. Vet Parasitol. 35, 103±116. Urquhart, G.M., Armour, J., Duncan, J.L., Dunn, A.M., Jennings, F.W., 1987. Veterinary Parasitology. Longman Scientific & Technical, Essex, 286 pp.
Relationships between bulk-tank antibodies to Ostertagia ostertagi and herd-management practices and measures of milk production in Nova Scotia dairy herds F.J. GuitiaÂna,*, I.R. Dohooa, R.J.F. Markhamb, G. Conboyb, G.P. Keefea a Department of Health Management, Atlantic Veterinary College, University of Prince Edward Island, Charlottetown, PEI, Canada C1A 4P3 b Department of Pathology and Microbiology, Atlantic Veterinary College, University of Prince Edward Island, Charlottetown, PEI, Canada C1A 4P3
Received 24 June 1999; accepted 24 June 2000
Abstract Monthly bulk-tank milk samples were obtained from 415 Nova Scotia dairy herds in each of the months of July±September 1998 and tested using an indirect microtitre ELISA against a crude saline-extract, whole-worm Ostertagia ostertagi antigen. ELISA results (optical densities (ODs)) were consistent across months (r 0:85) but there was considerable variation among herds. A questionnaire was sent by mail to all producers; information on management factors that would potentially influence parasite burdens in the herds was obtained from 239 farms. Data on annual milk production, summer milk production (July±September) and seasonal decline in milk production were obtained from the Animal Productivity and Health Information Network (APHIN) database. Associations between management practices and ODs, and between ODs and milk-production parameters were studied. Some management practices known to be associated with parasite burdens had expected directions of association with the ODs, giving supporting evidence that the ELISA is a reasonable measure of parasite burden. Most notably, ODs were increased with greater exposure of heifers or milking cows to pasture. ODs were not associated with either annual milk production or seasonal decline in milk production. However, there was a substantial relationship between the herd OD value and the level of milk production during the summer. An increase in the OD from 0.58 to 0.83 (the interquartile range of ODs)
* Corresponding author. Present address: Laboratorio Lema & BandõÂn, C./Lepanto 5, E-36201 Vigo, Spain. Tel.: 34-986437400; fax: 34-986431145 E-mail address: [email protected] (F.J. GuitiaÂn).
0167-5877/00/$ ± see front matter # 2000 Elsevier Science B.V. All rights reserved. PII: S 0 1 6 7 - 5 8 7 7 ( 0 0 ) 0 0 1 5 6 - 2
80
F.J. GuitiaÂn et al. / Preventive Veterinary Medicine 47 (2000) 79±89
was associated with a reduction in production of 1.25 kg/cow/day. # 2000 Elsevier Science B.V. All rights reserved. Keywords: Cattle-parasitological disease; Ostertagia ostertagi; ELISA; Dairy herd management; Milk production
1. Introduction Internal parasites (particularly Ostertagia ostertagi) are economically detrimental for dairy herds. Negative effects include a reduction in heifer growth rates and decreased levels of milk production. A recent review by Gross et al. (1999) reported a median increase of 0.63 kg/day in milk production following treatment with anthelmintics in a wide range of studies that were based on various study designs, treatment protocols and products. Unfortunately, monitoring of parasite burdens in dairy herds currently requires the use of faecal egg counts (FEC). FEC are highly variable both within and among animals are time consuming and expensive. In general, egg output from internal parasites in cows is low and FEC do not reliably reflect the parasite burden (Gross et al., 1999). An ELISA developed in the Netherlands (Keus et al., 1981) and evaluated under Dutch conditions (Kloosterman et al., 1993) was recently evaluated in dairy herds in Quebec, Canada (Dohoo et al., 1997). In this last study, ELISA optical densities (ODs; computed as an average of the five most-recently calved heifers) had an acceptable correlation with the average FEC taken from the same five animals (r 0:49 to 0.55). In that study, the within-herd variability (as measured by the intra-class correlation coefficient) was lower for the ELISA ODs than it was for FEC Ð suggesting that antibody titres might be less variable cow-to-cow than FEC. However, comparing ELISA results to FEC assumes that FEC are a good indicator of parasite burdens in adult cows Ð which is certainly not the case at the cow level. Consequently, other means of evaluating the ELISA (to determine if it is a useful measure of parasite burden) are needed. One approach is to determine if factors which would be expected to influence parasite burdens have a predictable effect on ELISA ODs. A second approach is to evaluate relationships between ELISA-test results and measures of productivity. A recently completed study in Prince Edward Island determined a strong relationship between ELISA results from bulk-tank milk samples and the amount of seasonal (summer-fall) decline in milk production that the herd experienced (Hovingh, 1998). A third method would be to monitor milk-yield response to anthelmintic therapy to determine if the ELISA ODs were predictive of treatment response. The objectives of this study were twofold. First, the study tested whether the ELISA measures parasite burden by examining the associations between test results and farmmanagement practices known to be associated with parasite burdens. Second (assuming that the ELISA is reflective of parasite burdens), the study evaluated the associations between the ELISA results and measures of milk production.
F.J. GuitiaÂn et al. / Preventive Veterinary Medicine 47 (2000) 79±89
81
2. Materials and methods 2.1. Study design and study population This was a cross-sectional study in which measures of antibody level, herdmanagement practices and measures of milk production were determined between December 1997 and November 1998. The study population consisted of all dairy herds in Nova Scotia (a province on the east coast of Canada). Dairy herds in this province range in size from approximately 15 to 500 milking and dry cows. Management practices range from total-confinement housing with a total mixed ration to herds being pastured in the summer and being fed a component-based ration. However, most herds have some access to pasture during the summer. 2.2. Sample collection and laboratory methods Bulk-tank milk samples were obtained from all Nova Scotia dairy herds for each of the months of July±September 1998 (N 415). Samples were stored frozen and subsequently tested using an indirect microtitre ELISA against a crude saline-extract, whole-worm O. ostertagi antigen as previously described (Keus et al., 1981). ODs from blank wells in each plate were subtracted from the sample values and the average of the results from 3 months was computed. Negative control sera derived from 3-month-old helminth-naive calves and positive control sera derived from hyperimmunized calves were included on each plate to ensure consistency of results between plates but sample ODs were not adjusted by the values of the positive or negative controls. For the same test (using the whole-parasite antigen from Cooperia), likelihood ratios for herd average optical densities (HOD, obtained by averaging individual milk samples' ODs) for predicting high/low herd average FEC had been computed by Dohoo et al. (1997). After reducing HOD to two categories, both the sensitivity and specificity of the HOD had been estimated to be 79%. 2.3. Farm-management practices At the end of August, all producers received a brief closed-response questionnaire by mail asking for basic information on factors that would potentially influence parasite burdens in their herds. Questions referred to extent of exposure to pasture, common use of pastures by heifers and dry or milking cows, use of parasite-control procedures in heifers and milking cows and pasture-management practices. (In this paper, the term ``heifer'' refer to nulliparous cows; definitions of all management-practice variables are presented in Table 1 and follow the wordings used on the questionnaire.) 2.4. Milk-production data Twelve months (from December 1997 to November 1998) of individual-cow milkproduction data for all herds on production recording (Atlantic Dairy Livestock Improvement Corporation Ð ADLIC) were extracted from the Animal Productivity and
F.J. GuitiaÂn et al. / Preventive Veterinary Medicine 47 (2000) 79±89
82
Table 1 Description and descriptive statistics of qualitative variables used in the analysis of associations between ELISA ODs, herd-management practices and milk production in Nova Scotia dairy herds (1998) Variable Heifer housing Heifer_confinement Heifer_yard Heifer_paddock Heifer_pasture
Description
Heifers kept in total-confinement housing Heifers given access to a concrete or gravel-surface exercise yard Heifers given access to a small grassed paddock for exercise Heifers given access to pasture for grazing
N 239
Heifer_pasture2 No pasture Pasture
Heifers given access to pasture (pasture or paddock)
239
Heifer_tx_fall No Yes
Heifers dewormed in the fall of 1997
239
Heifer_tx_spring No Yes
Heifers dewormed in the spring of 1998
239
Heifer_tx_bolus No Yes
Heifers given a sustained-release bolus in summer 1998
239
Heifer_dry_cows No Yes
Heifers graze on pastures also grazed by dry cows
213
Heifer_milk_cows No Yes
Heifers graze on pastures also used by milking cows
183
Milking_cow_housing Cow_confinement Cow_yard Cow_paddock Cow_pasture
Cows kept in total-confinement housing Cows given access to a concrete or gravel-surface exercise yard Cows given access to a small grassed paddock for exercise Cows given access to pasture for grazing
239
Cow_pasture2 No Yes
Cows given access to pasture (pasture or paddock)
239
Cow_tx_oral
Milking cows dewormed with oral product in the last 12 months
232
Milking cows treated with pour-on or injectable treatment at drying off
232
No Yes Cow_tx_dry_of No Yes
% of factor
Mean OD
8 1
0.43 0.52
5
0.66
86
0.70
9 91
0.44 0.69
50 50
0.67 0.67
72 28
0.68 0.65
93 7
0.67 0.71
51 49
0.66 0.73
90 10
0.73 0.69
11 3
0.41 0.46
9
0.57
77
0.73
15 85
0.42 0.71
95 5
0.67 0.70
88 12
0.67 0.67
F.J. GuitiaÂn et al. / Preventive Veterinary Medicine 47 (2000) 79±89
83
Table 1 (Continued ) Variable
Description
N
Cow_tx_calving No Yes
Cows treated with pour-on or injectable treatment at calving
232
Control_access No Yes
Pastures are managed using some form of controlled access grazing (rotation or strip) vs continuous access
209
Manure No Yes
Manure spread on pastures used for grazing
213
Dragged No Yes
Pastures dragged or harrowed
211
Clipped No Yes
Pastures clipped
214
% of factor
Mean OD
93 7
0.67 0.65
31 69
0.73 0.72
64 36
0.71 0.75
64 36
0.72 0.72
32 68
0.72 0.72
Health Information Network (APHIN) database (Dohoo, 1992). From these data, herd average values of individual-cow milk production (kg/cow/day) were computed for annual milk production, summer milk production (July±September) and relative fall production (average of September±November expressed as a proportion of average of May±July). Herd averages for annual and summer (July±September) days in milk, lactation number and log somatic-cell count were also computed. 2.5. Data analyses Descriptive statistics were computed. Linear-regression models evaluated multivariable relationships between management factors and ODs and between ODs and milkproduction parameters. All analyses were carried out using Version 5.0 of the statistical package Stata (Stata Corporation, College Station, TX) and associations were deemed significant if P 0:05 (all two-sided). Two linear-regression models were fit to evaluate the relationships between management practices and ELISA-test results. The first was based on data from all herds and included level of exposure to pasture as a predictor along with other management factors that applied to all herds. The second model was based only on herds which allowed the cows to have access to pasture and was fit to evaluate the effects of pasture-management variables on ODs. Both linear models were developed using a backwards stepwise elimination of non-significant effects (P > 0:05 by the Wald's test). Once significant main effects had been identified, all two-way interactions were assessed for significance. Multiple linear regression (backwards elimination) was also used to evaluate the associations between the ELISA ODs and each of these three measures of milk production:
84
F.J. GuitiaÂn et al. / Preventive Veterinary Medicine 47 (2000) 79±89
herd average annual milk production, herd average summer milk production, and seasonal decline in milk production (fall production divided by spring production). Before fitting these models, the variables representing heifer and milking cow housing were dichotomized into ``pasture'' vs ``no pasture'' (pasture/paddock vs confinement/ yard) to represent any direct exposure to grassed areas or not. Also, in each of these models, ODs were forced into the model to evaluate the direct relationship between OD and milk production. To avoid confounding, a number of factors which are known to have strong relationships with production parameters (e.g. pasture access, herd average days in milk, herd average lactation number, herd average log somatic-cell count), were made available to the linear-regression models. 3. Results and discussion 3.1. Descriptive statistics Milk samples were tested from 415 herds and usable questionnaires were returned by 239 producers (58% of those contacted) (Table 1). Production data were obtained for 243 herds on ADLIC of which 147 had returned usable questionnaires. The wide range of values for the ODs suggests considerable variation between herds in levels of antibodies to internal parasites (Table 2). The interquartile range for ODs was Table 2 Description and descriptive statistics of quantitative variables used in the analysis of associations between ELISA ODs, herd-management practices and milk production in Nova Scotia dairy herds. Data from 402 Nova Scotia dairy herds for which average ODs were available and 243 Nova Scotia dairy herds for which milkproduction data were available (1998) Variable
Description
Optical_density Milk_annual
Average of three monthly values of OD from ELISA Average milk production (kg/cow/day) from December 1997 to November 1998 Average milk production (kg/cow/day) in July±September 1998 Fall milk production (September±November 1998) divided by spring milk production (May±July 1998) expressed as proportion Herd average natural log somatic-cell counts from December 1997 to November 1998 Herd average natural log somatic-cell counts in July±September 1998 Herd average days-in-milk from December 1997 to November 1998 Herd average days-in-milk in July±September 1998 Herd average lactation number from December 1997 to November 1998 Herd average lactation number in July±September 1998
Milk_summer Milk_fall LogSCC_annual LogSCC_summer DIM_annual DIM_summer Lactation_annual Lactation_summer
Mean
S.D.
Range
0.69 25.7
0.19 4.5
0.16, 1.14 12.6, 39.3
26.6
4.7
11.9, 40.6
0.85
0.11
0.44, 1.13
5.39
0.42
4.07, 6.23
5.49
0.48
3.88, 6.97
172
17
132, 270
177 2.81
23 0.42
108, 270 1.87, 4.48
2.83
0.47
1.94, 5.50
F.J. GuitiaÂn et al. / Preventive Veterinary Medicine 47 (2000) 79±89
85
0.58±0.83. Within-herd correlations between monthly samples were high (r 0:85). The consistency within herds, across months suggests that little is to be gained through the testing of multiple samples if the test were to be included in a routine herd-monitoring program. Descriptive statistics of qualitative and quantitative variables used in subsequent analyses are presented in Tables 1 and 2, respectively. 3.2. Associations between management factors and ELISA results Results for the model that included level of exposure to pasture as a predictor along with other management factors that applied to all herds are presented in Table 3. This model was fitted using all available data once the best final model was selected. Increasing exposure of both heifers and cows to pasture resulted in steadily increasing ELISA ODs. Cows that spent part of their time grazing on pasture had ODs that were (on an average) 0.27 units higher than cows kept in confinement housing. Deworming heifers in the spring was associated with significantly reduced ODs (with pasture use analytically controlled). Other parasite-control practices (e.g. fall treatments of heifers, sustainedrelease bolus use and a variety of treatments of milking cows) did not have significant associations with ODs. This model had an R2 of 0.42 indicating that pasture exposure and spring treatment of heifers explain a reasonable amount of the variability in parasite± antibody levels. The model based only on herds which allowed cows to have access to pasture (Table 4) shows that allowing heifers to graze the same pastures as dry cows resulted in elevated ODs but spring treatment of heifers lowered the ODs. Spreading manure on pastures also was associated with increased ODs. However, the R2 of this model was only 0.06. Correlation coefficients were computed to check for the possibility of serious colinearity problems. Coefficients were calculated in two ways: using the subset of records for which all variables were present and using all records for which each individual pair of values was present. Apart from correlations between different
Table 3 Regression of effects of housing and management practices on test results from an Ostertagia ELISA applied to bulk-tank milk samples (239 Nova Scotia dairy herds in 1998; R2 0:42) Variable
b
Confidence interval
P
Intercept Heifer_confinement Heifer_yard Heifer_paddock Heifer_pasture Heifer_tx_spring Cow_confinement Cow_yard Cow_paddock Cow_pasture
0.34 Baseline 0.08 0.14 0.15 ÿ0.06 Baseline 0.00 0.10 0.27
0.27, 0.41 ± ÿ0.14, 0.29 0.03, 0.25 0.07, 0.22 ÿ0.11, ÿ0.02 ± ÿ0.11, 0.12 0.02, 0.20 0.21, 0.34
0.00 ± 0.47 0.01 0.00 0.00 ± 0.96 0.02 0.00
86
F.J. GuitiaÂn et al. / Preventive Veterinary Medicine 47 (2000) 79±89
Table 4 Regression of effects of herd-management practices on test results from an Ostertagia ELISA applied to bulktank milk samples (192 Nova Scotia dairy herds which used pasture for grazing in 1998; R2 0:06) Variable
b
Confidence interval
P
Intercept Heifers_dry_cows Manure Heifer_tx_spring
0.71 0.04 0.04 ÿ0.04
0.67, 0.74 0.00, 0.08 0.00, 0.08 ÿ0.09, ÿ0.00
0.00 0.05 0.05 0.05
categories of a single variable, the highest correlation is 0.31 excluding the possibility of serious colinearity problems. All of the observed results from the linear regressions are consistent with accepted ideas about the associations of management factors with parasite burdens (Urquhart et al., 1987). Lack of significant association with different parasite-control practices could reflect that they eliminate adult parasites without having an important effect at the herd level parasite control or could be due to an insufficient power to detect effects. 3.3. Associations between ELISA results and production parameters Descriptive statistics of production parameters for different levels of milking cow housing are shown in Table 5. The regression model evaluating the associations between ELISA ODs and annual milk production (R2 0:34) shows that increasing days in milk, herd age, somatic-cell count Table 5 Descriptive statistics of three measures of milk production (annual, summer and seasonal decline) for different levels of milking cow housing. Data from 147 Nova Scotia dairy herds for which both herd-management data and milk-production data were available (1998) Variable
N
Mean
S.D.
Annual milk production Cow_confinement Cow_yard Cow_paddock Cow_pasture
18 5 13 111
29.34 31.05 29.74 25.54
4.80 2.24 2.47 4.26
Summer milk production Cow_confinement Cow_yard Cow_paddock Cow_pasture
18 5 13 111
29.08 30.30 29.18 25.27
5.25 2.01 3.39 4.54
Seasonal decline in milk production Cow_confinement Cow_yard Cow_paddock Cow_pasture
18 5 13 111
0.89 0.91 0.87 0.84
0.11 0.02 0.07 0.10
F.J. GuitiaÂn et al. / Preventive Veterinary Medicine 47 (2000) 79±89
87
and use of pasture for grazing had the expected directions of association with this parameter. OD was not significantly (b ÿ3:1, P 0:11) associated with annual milk production. The model evaluating the association with the seasonal decline in milk production (R2 0:06) shows that there is also no significant association between OD and the seasonal decline in milk production. Exposure to pasture was the only factor which exhibited a significant association with this parameter. These results are in contrast to those reported from PEI herds, where OD scores were highly correlated to the seasonal decline in milk production (Hovingh, 1998). However, the amount of seasonal decline demonstrated by Nova Scotia herds was considerably less than that shown by PEI herds (relative fall production of 0.85 in Nova Scotia herds vs 0.75 in PEI herds). As a result, the power of our study to detect factors which would affect the decline would be lower. The results from the regression model evaluating the effects of factors on summer (July±September) production is shown in Table 6. The ELISA OD had a significant negative association with daily milk production during these months (b ÿ5:0, P 0:02), as did grazing, increasing days in milk and somatic-cell counts. With an interquartile range 0.58±0.83 and a coefficient of ÿ5.0, a herd at the 25th percentile for OD would be expected to produce 1.25 kg (5:0 0:83 ÿ 0:58) less milk per head per day than a herd at the 75th percentile. This represented the effect of the parasite burden after controlling for exposure to pasture and several other potential confounding variables. However, the R2 was only 0.28. The above effect of parasite burden on summer milk production must be interpreted with caution for a couple of reasons. First, it is based on the assumption that the OD is a valid measure of parasite burden. If it is a measure which only roughly approximates parasite burden, then it is likely that the effects of the parasite burden will be underestimated since non-differential information bias will tend to bias the results toward the null. On the other hand, there may have been unmeasured confounding variables which affected this estimate and biased it towards or away from the null. For example, exposure to pasture was measured in a very crude manner (pasture vs no pasture). If increasing levels of exposure to pasture resulted in both higher ODs and lower milk production, then the estimate of the effect of the ODs may be biased upwards. Confirmation of the relationship between parasite burdens as predicted by OD values and lost milk production will most likely come from trials in which response to anthelmintic treatment is compared to previous estimates of OD. Such trials have the Table 6 Regression of effects of ELISA ODs and other herd parameters on the herd average daily milk production during July±September, i.e. milk_summer (147 Nova Scotia dairy herds on summer pasture for which both herdmanagement data and milk-production data were available in 1998; R2 0:28) Variable
b
Confidence interval
P
Intercept Optical_density Cow_pasture2 DIM_summer LogSCC_summer
57.79 ÿ5.02 ÿ2.53 ÿ0.05 ÿ3.12
47.67, 67.91 ÿ9.20, ÿ0.84 ÿ4.69, ÿ0.37 ÿ0.09, ÿ0.02 ÿ4.62, ÿ1.63
0.00 0.02 0.02 0.00 0.00
88
F.J. GuitiaÂn et al. / Preventive Veterinary Medicine 47 (2000) 79±89
potential to both confirm the validity of the ELISA as a monitoring tool for parasite burdens and better quantify the effect of parasite burden on milk production. Previous studies that have examined the ability of ELISA-test results to predict milk-production response to anthelmintic treatment have been inconclusive. One study (Ploeger et al., 1989) found a positive correlation (r 0:36) between the herd average milk-production response and the herd average Ostertagia antibody titre while a subsequent study (Ploeger et al., 1990) did not find such a relationship. The first study was carried out in 1985 and the second in 1986. The authors suggested that differences in the overall level and chronology of exposure, probably related to the different meteorological conditions in both years, could explain the results. Different anthelmintic treatments (ivermectin and albendazole) were used in both studies and in both cases anthelmintic treatment resulted in a significant increase in milk production. 4. Conclusions In our study, some of the factors that would be expected to affect parasite burdens in dairy herds had their predicted directions of association with the ELISA ODs. Evaluation of a test's ability to measure parasite burdens in dairy herds requires the use of surrogate ways because of the limitations of FEC in lactating dairy cows, and in the absence of other plausible explanations for the observed associations, it seemed reasonable to us to assume that the ELISA is measuring parasite burdens in the herds. This, taken with other previously published evidence of the value of the ELISA as a monitoring tool, suggests that the ELISA may be a useful tool for monitoring parasite burdens in dairy herds. There appeared to be a substantial relationship between the herd OD value and the level of milk production during the summer in herds exposed to grass. This suggests that herds with high OD readings may benefit from more-intensive parasite-control programs in the milking herd, assuming that the test is measuring parasite burden. Acknowledgements The authors thank the Nova Scotia Milk Producers Association and the Nova Scotia Agri-Futures Fund for their support of this project and Ms. Judy Sheppard for her technical assistance. References Dohoo, I.R., 1992. Dairy APHIN Ð an information service for the dairy industry in Prince Edward Island, Canada. Prev. Vet. Med. 12, 259±268. Dohoo, I.R., Caldwell, V., Markham, R.J.F., Conboy, G., Boucahrd, E., Descoteaux, L., 1997. Evaluation of an ELISA for monitoring parasite burdens in dairy herds. In: Proceedings of the International Symposium on Veterinary Epidemiology and Economics, Paris, France.
F.J. GuitiaÂn et al. / Preventive Veterinary Medicine 47 (2000) 79±89
89
Gross, S.J., Ryan, W.G., Ploeger, H.W., 1999. Anthelmintic treatment of adult dairy cows and the effect on milk production. Vet. Rec. 144, 581±587. Hovingh, E., 1998. An investigation into factors affecting summer/fall milk production and profitability in PEI dairy herds. Ph.D. Thesis. University of Prince Edward Island, Canada. Keus, A., Kloosterman, A., Van den Brink, R., 1981. Detection of antibodies to Cooperia spp. and Ostertagia spp. in calves with the enzyme linked immunosorbent assay (ELISA). Vet. Parasitol. 8, 229±236. Kloosterman, A., Verhoeff, J., Ploeger, H.W., Lam, T.J.G.M., 1993. Antibodies against nematodes in serum, milk and bulk tank milk samples as possible estimators of infection in dairy cows. Vet. Parasitol. 47, 267± 278. 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. Ploeger, H.W., Kloosterman, A., Bargeman, G., Wuijckhuise, L.V., Van den Brink, R., 1990. Milk yield increase after anthelmintic treatment of diary cattle related to some parameters estimating helminth infection. Vet Parasitol. 35, 103±116. Urquhart, G.M., Armour, J., Duncan, J.L., Dunn, A.M., Jennings, F.W., 1987. Veterinary Parasitology. Longman Scientific & Technical, Essex, 286 pp.