Skip to main content

Evaluation of 6 commercial ELISA kits for the detection of BVDV-specific antibodies in serum and milk using a large collection of field samples

Miet De Baere1,* Jean-Baptiste Hanon 1,2 Camille De la Ferté1 Sophie Roelandt2 Yves Van der Stede2 Brigitte Cay1

  • 1Unit enzootic and (re)emerging viral diseases, 2 Unit coordination of veterinary diagnosis, epidemiology and risk assessment, Veterinary and Agrochemical Research Centre (CODA-CERVA), Brussels, Belgium



Serological surveillance using antibody ELISAs is a commonly used tool to ascertain freedom of disease. Based on a previous study, 6 ELISA kits were selected for evaluation on a large number of serum and milk samples obtained in the field. The study goal was to evaluate performance characteristics of these kits and to verify the observed lack of sensitivity of blocking ELISAs in serum and of milk testing, to evaluate whether pools of sera/individual milk samples can be used, and to assess whether it is possible to certify/monitor “BVD free herds” by means of serology on serum and/or (bulk) milk samples in the Belgian BVDV control programme.

Materials and Methods:

51 herds representative for the Belgian cattle sector (location, herd type, herd size, BVD status and vaccination status) were visited once. Serum samples were taken from all young stock (6-18 months) and from 20 randomly selected cows (mixed herds: 20 dairy and 20 beef cows). In dairy herds, one bulk milk sample and individual milk samples from the same cows which had been sampled for serum were taken. During the visit a questionnaire was filled out with the farmer to collect epidemiological information regarding BVD in their herd and vaccination, to be able to assign a BVD status to each herd: BVD free; BVD eradicated; BVD controlled; BVD infected less than 1 year; and BVD infected.

6 ELISAs were selected: 2/1 blocking and 2/1 indirect ELISA for serum/milk. Depending on the outcome of all ELISAs and using the virus neutralisation test (VNT) as a gold standard for serum, each serum and milk sample was assigned a positive, negative or discordant serological status. Next, all sera from unvaccinated herds were tested in VNT and VNT results were used as a reference for both serum and individual milk samples from the same animal. Using 2 by 2 tables we calculated true and false positive/negative results for each ELISA. These results were imported in STATA to calculate sensitivity (Se), specificity (Sp), Youden index, positive and negative predictive values, Kappa of Cohen and area under ROC curve.

Finally, weak/strong positive sera and individual milk samples were diluted in negative serum/milk to mimic pooled sera or bulk milk. Diluted samples were tested and analyzed.


3159 sera, 557 individual and 28 bulk milk samples were tested using 6 BVDV ELISAs and were assigned a positive, negative or discordant status. For serum samples, the percentage of positive samples was slightly higher for the indirect ELISAs than for the blocking ELISAs. For milk, the blocking ELISA performed better, though not significant, than the indirect ELISA, since the percentage of positively identified samples (73,2%) was similar to the percentage of the corresponding sera (75,9%), whereas the indirect ELISA seemed to be less sensitive (64,6%).

To assess the performance of each ELISA, only field samples (1308 sera and 293 individual milk samples) from unvaccinating herds (19) were used to exclude interference of vaccination. 3 out of 4 evaluated ELISAs on serum scored well for all parameters, whereas the anti-E0 blocking ELISA showed a lack of Se (93%), compensated by an increased Sp (99,1%). For milk, the blocking ELISA gain performed slightly better than the indirect ELISA both in terms of Se and Sp.

Pooling of serum and milk samples was investigated in each of the selected ELISAs since pooling of sera could reduce the costs of testing and to mimic bulk milk samples which are convenient and cheap samples to obtain. Our study showed that pooling should not be performed, since weak positive sera and individual milk samples were missed by all evaluated kits at small pool sizes. We also found that at this moment, a bulk milk sample does not reflect the BVDV status of a herd, since even in free or cleared herds a large proportion of cows are seropositive.


All kits performed well on individual serum samples, observing no difference between blocking or indirect ELISAs for samples from non-vaccinating herds. This led to the decision that both types of kits may be used for serological monitoring of BVDV free herds using individual serum samples in the Belgian BVDV eradication programme in the future. The decreased Se of the ELISAs using individual and pooled milk samples as a matrix and the finding that a bulk tank milk sample does not reflect the BVD status of a herd at the moment, indicates that it may be impossible to guarantee a “BVDV free” status of a herd based only on bulk milk testing.

  • Proceedings of the 29th World Buiatrics Congress, Dublin, Ireland, 3-8 July 2016 - Oral Communication and Poster Abstracts