Preventive Veterinary Medicine 143 (2017) 68–78 Contents lists available at ScienceDirect Preventive Veterinary Medicine jo ur nal ho me pag e: www.elsevier.com/locate/prevetmed Schmallenberg virus: Predicting within-herd seroprevalence using bulk-tank milk antibody titres and exploring individual animal antibody titres using empirical distribution functions (EDF) Á.B. Collins a,b , J. Grant d , D. Barrett c , M.L. Doherty b , A. Hallinan e , J.F. Mee a,∗ a Animal and Bioscience Research Department, Teagasc, Moorepark, Fermoy, Co., Cork, Ireland b School of Veterinary Medicine, University College Dublin, Dublin 4, Ireland c Department of Agriculture, Food and the Marine, Surveillance, Animal By Products and TSE Division, Backweston, Celbridge, Co. Kildare, Ireland d Applied Physics and Statistics Department, Teagasc, Ashtown, Dublin 15, Ireland e Department of Agriculture, Food and the Marine, Blood Testing Laboratory, Model Farm Road, Cork, Ireland a r t i c l e i n f o Article history: Received 8 September 2016 Received in revised form 20 January 2017 Accepted 28 April 2017 Keywords: Schmallenberg virus Bulk-tank milk ELISA Regression model Antibody Prevalence Empirical distribution functions (EDF) a b s t r a c t Schmallenberg virus (SBV) is transmitted by Culicoides spp. biting midges and can cause abortions and congenital malformations in ruminants and milk drop in dairy cattle. Estimating true within-herd sero- prevalence is an essential component of efficient and cost-effective SBV surveillance programs. The objectives of this study were: (1) determine the correlation between bulk-tank milk (BTM)-ELISA results and within-herd seroprevalence, (2) evaluate the ability of BTM-ELISA results to predict within-herd sero- prevalence and (3) explore the distributions of individual animal serology results using novel statistical methodology. BTM samples (n = 24) and blood samples (n = 4019) collected from all lactating cows contributing to the BTM in 26 Irish dairy herds (58–444 cows/herd) in 2014 located in a region exposed to SBV in 2012/2013, were analysed for SBV-specific antibodies using IDVet ® ELISA kits. The correlation between BTM-ELISA results and within-herd seroprevalence was determined by calculating Pearson’s correlation coefficient. Linear regression models were used to assess the ability of BTM-ELISA results to predict within-herd seroprevalence. The distributions of individual animal serology results were explored by determining the empirical distribution functions (EDF) of the individual animal serum ELISA results in each herd. EDFs were compared pairwise across herds, using the Kolmogorov-Smirnov statistical test. Herds with similar BTM-ELISA results, herds with similar within-herd seroprevalence and herds with similar mean-herd serology ELISA results were stratified in order to explore their respective paired-herd EDF comparisons. Statistical significance was set at p < 0.05. Twenty-two herds were BTM-ELISA-positive (within-herd seroprevalence 30.6–100%) and two herds were BTM-ELISA-negative (within-herd seroprevalence 10.7 and 16.2%) indicating BTM-ELISA-negative herds can have seropositive animals present. BTM-ELISA results were highly correlated (r = 0.807, p < 0.0001) with, and predictive of (R 2 = 0.832, p < 0.0001) of within-herd seroprevalence. Predictions were most accurate for upper-range BTM-ELISA antibody titres, while they were less accurate at higher and lower antibody titres. This is likely a result of the overall high within-herd seroprevalence. In herds with similar BTM-ELISA results 82% of the paired-herd EDF comparisons were significantly different. In herds with similar within-herd seroprevalence and in herds with similar mean-herd serology ELISA results, 46% and 47% of the paired-herd EDF comparisons were significantly different, respectively. Abbreviations: SBV, Schmallenberg virus; Spp, Species; ELISA, Enzyme Linked Immunosorbent Assay; EDF, Empirical Distribution Functions; BVD, Bovine Viral Diarrhoea Virus; BDV, Border Disease Virus; BLV, Bovine Leukaemia Virus; OD, Optical Density; S/N%, Sample-to-negative percentage ratio; S/P%, Sample-to-positive percentage ratio; PI, Prediction Interval; BHV, Bovine Herpes Virus. ∗ Corresponding author. E-mail addresses: aine.collins@teagasc.ie (Á.B. Collins), jim.grant@teagasc.ie (J. Grant), damien.barrett@agriculture.gov.ie (D. Barrett), michael.doherty@ucd.ie (M.L. Doherty), john.mee@teagasc.ie (J.F. Mee). http://dx.doi.org/10.1016/j.prevetmed.2017.04.013 0167-5877/© 2017 Elsevier B.V. All rights reserved.