Schmallenberg virus in Dutch dairy herds: Potential risk factors for high within-herd seroprevalence and malformations in calves, and its impact on productivity

Veterinary Microbiology 168 (2014) 281–293

Contents lists available at ScienceDirect

Veterinary Microbiology journal homepage: www.elsevier.com/locate/vetmic

Schmallenberg virus in Dutch dairy herds: Potential risk factors for high within-herd seroprevalence and malformations in calves, and its impact on productivity A.M.B. Veldhuis a,*, S. Carp-van Dijken b, L. van Wuijckhuise b, G. Witteveen b, G. van Schaik a a b

GD Animal Health Service, Department of Research and Development, Epidemiology Group, PO Box 9, 7400 AA Deventer, The Netherlands GD Animal Health Service, Department of Ruminant Health, PO Box 9, 7400 AA Deventer, The Netherlands

A R T I C L E I N F O

A B S T R A C T

Article history: Received 26 July 2013 Received in revised form 8 November 2013 Accepted 13 November 2013

In November 2011, the new orthobunyavirus Schmallenberg virus (SBV) was identified in dairy cows that had induced fever, drop in milk production and diarrhoea in the Netherlands (Muskens et al., 2012. Tijdschrift voor Diergeneeskunde 137, 112–115) and a drop in milk production in cows in Northwestern Germany (Hoffmann et al., 2012. Emerging Infectious Diseases 18 (3), 469–472), in August/September 2011. This study aimed at quantifying risk factors for high within-herd prevalence of SBV and SBV-induced malformations in newborn calves in dairy herds in the Netherlands. Additionally, the within-herd impact of SBV infection on mortality rates and milk production was estimated. A case-control design was used, including 75 clinically affected case herds and 74 control herds. Control herds were selected based on absence of malformations in newborn calves and anomalies in reproductive performance. SBV-specific within-herd seroprevalences were estimated. Risk factors for high within-herd SBV seroprevalence (>50%) and the probability of malformed newborn calves in a herd were quantified. In addition, within-herd impact of SBV with regard to milk production and mortality was estimated. Animal-level seroprevalence was 84.4% (95% confidence interval (CI): 70.8–92.3) in case herds and 75.8% (95% CI: 67.5–82.5) in control herds. Control herds that were completely free from SBV were not present in the study. Herds that were grazed in 2011 had an increased odds (OR 9.9; 95% CI: 2.4–41.2)) of a high seroprevalence (>50%) compared to herds that were kept indoors. Also, when grazing was applied in 2011, the odds of malformations in newborn calves tended to be 2.6 times higher compared to herds in which cattle were kept indoors. Incidence of malformations in newborn calves at herd level was associated with both within-herd seroprevalence and clinical expression of the disease in adult cattle. The rate of vertical transmission of SBV to the fetus once a dam gets infected seemed low. A total of 146 stillborn or malformed calves were submitted by 65 farmers during the study period, of which 19 were diagnosed as SBV-positive based on pathological investigation and/or RT-qPCR testing of brain tissue. Based on these results combined with calving data from these herds we roughly estimated that at least 0.5% of the calves born between February and September 2012 have been infected by SBV.

Keywords: Schmallenberg virus Dairy cattle Seroprevalence Risk factors Impact

* Corresponding author. Tel.: +31570660408; fax: +31570660354. E-mail address: [email protected] (A.M.B. Veldhuis). 0378-1135/$ – see front matter ß 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.vetmic.2013.11.021

282

A.M.B. Veldhuis et al. / Veterinary Microbiology 168 (2014) 281–293

A drop in milk production was observed between the end of August 2011 and the first half of September (week 35–36), indicating the acute phase of the epidemic. During a 4week period in which SBV infection was expected to have occurred, the total loss in milk production in affected dairy herds was around 30–51 kg per cow. SBV had no or limited impact on mortality rates which was as expected given the relatively mild expression of SBV in adult cows and the low incidence of malformations in newborn calves. ß 2013 Elsevier B.V. All rights reserved.

1. Introduction From the third week of August 2011 onwards, numerous veterinary practitioners contacted the veterinary consulting service (GD Veekijker) of GD Animal Health to report suddenly decreased milk production, watery diarrhea and sometimes fever in dairy herds (Muskens et al., 2012). Several affected farms were visited and faecal and serum samples from clinically affected animals and bulk milk samples were investigated for numerous endemic and exotic pathogens at GD Animal Health and the national reference laboratory (CVI). No causal agent was found until November 18th 2011, when the Friedrich Loeffler Institut (FLI, Germany) reported the isolation of a novel orthobunyavirus in German cattle, Schmallenberg virus (SBV) (Hoffmann et al., 2012). Next, the first malformed newborn lambs were reported in the Netherlands at the end of November (Van den Brom et al., 2012), followed by the first malformed calf on December 13th. Around the same time, presence of the virus was confirmed in the Netherlands in brain tissue of deformed lambs and serum samples of dairy cows which had shown clinical signs in August, using a real-time PCR developed by FLI. Subsequently, birth of malformed calves and lambs with the arthrogryposis–hydranencephaly syndrome was made notifiable between December 20th, 2011 and July 6th, 2012. SBV rapidly infected a large fraction of the Dutch ruminant population (Elbers et al., 2012; Veldhuis et al., 2013). Little is known however about factors that determine the level of seroprevalence and morbidity at herd level. Also, impact on key performance indicators such as milk production and mortality rates is unknown. The objectives of this study were (i) to explore and quantify the risk factors for a high within-herd seroprevalence of SBV in Dutch dairy herds, (ii) to identify herdlevel risk factors for SBV-induced malformations in newborn calves, and (iii) to describe the effects of SBV infection on mortality rates and milk production. 2. Materials and methods 2.1. Selection of herds A sample size of 75 case herds and 75 control herds was chosen in order to detect a risk factor exposure odds ratio (OR) of 2–3, with 95% confidence and 80% power (WinEpiscope 2.0—De Blas et al., 2000). First, 26 dairy herds for which veterinary practitioners had reported clinical signs that were likely due to SBV infection from August 20th and early September were selected as case herds. The other 49 case herds were selected as follows: out of all cattle herds that had notified malformations in newborn calves from

December 20th, 2011 until February 20th 2012 (n = 350), excluding the 26 herds that were already selected as case herd, 185 herds were classified as ‘likely SBV infected’ based on pathological findings found in the malformed calf. From these, 174 herds were dairy herds. Out of these dairy herds, 130 were randomly selected as candidate case herd for our study. Farmers were contacted by phone in random order and the first 49 farmers that agreed to participate were included as case herd. About 25% of the contacted herds refused to participate. The veterinary practice of each case herd was contacted and requested to select a control herd located in the same geographical area as the case herd, matching case and control by region in that manner. The following inclusion criteria needed to be met for a control herd: (i) no clinical signs potentially related to a SBV infection had been observed in adult cattle in 2011, (ii) no remarkable change in reproductive performance had been observed in 2011 compared to previous years, and (iii) no malformations in newborn calves had been observed until the date of selection that could be a result of SBV infection. A total of 74 control herds were selected. 2.2. Data collection and diagnostic procedures All 149 herds were visited once by a veterinarian of GD Animal Health (n = 4) between February 22nd and August 3rd, 2012. During the herd visit, the veterinarian interviewed the farmer about general farm characteristics, grazing management, insect control, animal health and malformations in calves (if observed) and vaccination strategies. To estimate within-herd seroprevalences, serum of 70 adult cattle were sampled by the private veterinarian of the farm to estimate a within-herd seroprevalence of 50% with 95% confidence and 5% error, given an average dairy herd size of 85 lactating cows in the Netherlands (WinEpiscope 2.0—De Blas et al., 2000). Samples needed to be collected before September 1st 2012 by the veterinarian of the farm. If less than 70 adult animals were present at the farm at time of blood collection, veterinarians were allowed to include blood samples from youngstock. Samples were investigated for presence of antibodies against SBV by means of an inhouse indirect whole virus ELISA (Van der Heijden et al., 2013) with a relative sensitivity of 98.8% (95% CI: 93.3– 99.8) and a relative specificity of 98.8% (95% CI: 97.5–99.6) (using a SBV virus neutralization test (Loeffen et al., 2012) as the reference test). Test outcomes were expressed as positive, negative or non-specific. Test outcomes were considered to be non-specific if the serum sample responded to the control material (without viral antigen), irrespective of the gross optical density of the sample.

A.M.B. Veldhuis et al. / Veterinary Microbiology 168 (2014) 281–293

To study the impact of SBV infection on the productivity of the herd, milk production recordings, mortality data and herd health status information for 2011 were used for detailed data analyses. Data from the National Identification and Registration database were used to obtain each animal’s birth date, sex, ID numbers of offspring (if present), date of arrival in the herd and date of removal. Herd health information with regard to BVD, IBR, Salmonellosis and Paratuberculosis were derived from GD Animal Health, based on the status of the herd in herd health programs in the last quarter of 2011. Farmers were requested to submit any stillborn, aborted or malformed newborn calf and a blood sample of the aborting dam during the study period. Submitted calves were investigated pathologically and tested for presence of SBV in brain tissue by RT-qPCR (Dijkman et al., 2012). In the Netherlands, fallen stock is collected at the farm for rendering. The rendering plant distinguishes three different age categories for cattle: 0–3 days (not ear tagged yet), 3 days—1 year, and >1 year. Mortality rates in adult cows, calves and newborn calves were derived from data provided by Rendac. Numbers of collected animal cadavers per week were available for 148 out of 149 herds. Bulk milk collection data for 2011 (2–3 records per week per herd) were provided by two dairy enterprises (Royal FrieslandCampina and PartiCo). Milk production data on test-day level (4–6 week interval per herd) were provided by the Royal Dutch Cattle Syndicate (CRV). An average milk production per animal per day was calculated by dividing the total amount of collected bulk milk on a given day by the number of animals that joined the testscheme that day. As the latter was only available on a 4–6 weekly interval per herd, this number was interpolated between test-days. Milk production records on test-day level were used to calculate an average milk production per animal per day for herds where no bulk milk recording were available (n = 6). For another 6 herds, no average milk production per animal could be calculated due to missing number of animals that joined the test-scheme. The combined dataset consisted of 16,904 daily milk production records of 143 herds. Extreme values of average daily milk production per cow (>50 kg) were excluded (n = 75; 0.44%). The average milk production per cow per day was aggregated at week level for each herd prior to analysis. 2.3. Data analysis During the study period (February 1st–September 1st, 2012) it appeared that SBV had rapidly spread throughout the Netherlands in 2011, affecting a large fraction of the ruminant population (Elbers et al., 2012; Veldhuis et al., 2013). Therefore, it was likely that the control herds in our study were also infected by SBV at some level. As a result, a number of analyses in this study have been carried out on subsets of herds, or herds were reclassified based on seroprevalence, morbidity (in adult animals) or malformations in calves. Statistical analyses were performed using STATA/SE version 12.1 software (StataCorp, 2011) and SAS/STAT1 version 9.3 software (SAS Institute Inc., 2013), reclassify herds based on seroprevalence, morbidity (clinical signs in adult cattle) and malformations in calves.

283

2.3.1. Seroprevalence estimation For herds with more than 70 submitted serum samples, 70 samples were randomly selected for seroprevalence estimations. Overall animal level apparent prevalences and within-herd prevalences were estimated using an intercept-only logistic regression model (logit) adapted to the survey design characteristics of the dataset (svyset herd_id [pweight = weight]), and svy prefix). By this method, both clustering of animals (or test results) within herds and sampling weights are taken into account in the estimation procedure. Sampling weights were calculated as the inverse of the animal’s sampling probability, which was calculated as the proportion of sampled animals per herd in relation to the herd size (at January 1st, 2012). The sampling probability of herds was considered to be equal across herds. Apparent prevalences in adult cows and youngstock were estimated by adding the variable ‘age category’ to the model. Then, true animal prevalences based upon the imperfect performance of the serological test were calculated according to Rogan and Gladen (1978) using the estimated prevalences. 2.3.2. Identification of risk factors Due to lack of normality in residuals following linear regression, within-herd seroprevalences were categorized as ‘low’ or ‘high’, using a cut-off of 50% seroprevalence, to be able to apply logistic regression analysis. Multivariable logistic regression analyses (logit) were conducted at herd level to describe the relationship between potential risk factors and (i) the within-herd SBV seroprevalence and (ii) the birth of malformed calves between in January 1st, 2011 and the date on which the interview was conducted. All variables obtained from the questionnaire and other data sources were first subjected to univariable testing of significance followed by multivariable analysis. Only variables showing a p-value less than 0.20 were included in a multivariable model. Final multivariable models were obtained by a backward selection procedure, removing step by step each variable with a p-value greater than 0.10. Confounding of variables was monitored during this procedure by the change in coefficient values. If the relative change exceeded 25% or more, or by 0.1 when the value of the coefficient was between 0.4 and 0.4, the removed variable was considered a potential confounder and re-entered in the model. Several questions with regard to grazing management were included in the questionnaire. To prevent collinearity, only one grazing-management variable was included in final multivariable models. Multivariable models with the highest R2-value are shown. In the final model, all two–way interactions were tested. The goodness of fit of logistic models was assessed using Pearson’s goodness of fit test (estat gof). 2.3.3. Within-herd impact of SBV infection on milk production and mortality For the analysis of daily milk production per animal in 2011, a linear regression model with a correlated residual error structure (PROC MIXED, repeated week/subject = herd_ID, type = cs) was conducted to describe the average loss in milk production per animal per day during the assumed acute phase of SBV infection. In order to apply

284

A.M.B. Veldhuis et al. / Veterinary Microbiology 168 (2014) 281–293

a model that best captured the correlation between the repeated milk production records per herd, several covariance structures (compound symmetry, first order auto-regressive structure and variance components) have been applied and model fit was monitored by Akaike’s information criterion (AIC). The compound symmetry covariance structure was applied in the final milk production model as it gave the best results (AIC closest to zero). The model that was used can be described as: Y i j ¼ m þ b1 SBV j þ b2 quarti j þ b3 h size j þ ei j

(1)

where Yij is the average milk production per animal per day in week i in herd j; m is the intercept; b1SBVj is the SBV infection period of herd j, defined as the week of notification of clinical symptoms and the three subsequent weeks; b2quartij is the quarter of 2011 (1, 2, 3, 4); b3hsizej is the herd size of herd j: the number of adult animals in the herd on January 1st, 2012; eij is the random error in week i for herd j. The variables ‘quarter’ and ‘herd size’ were forced in the model to account for herd size and seasonal differences in milk production. The analysis was performed on the 26 case herds that had initially reported clinical signs to GD Animal Health in August/September 2011. As the 26 case herds might not have been representative for all herds infected with SBV, another model was fitted for 128 herds with a high SBV withinherd seroprevalence (>50%). For these herds it was certain that SBV infection had occurred, whether the farmer had reported or observed clinical symptoms or not. In this model, the ‘SBV infection period’ was defined as the four weeks between August 15th and September 11th, 2011. Normality of linear model residuals were checked using normal probability plots. Mortality rates per age category per herd were calculated for the third and fourth quarter of 2011, based on the assumption that an effect of SBV on mortality rates in 2011 could be expected in those quarters. Mortality rates in cows (>1 year) were calculated as the total number of collected cow cadavers per quarter divided by the number of adult cows registered in the I&R-database in that quarter. For calves (3 days—1 year), mortality rates were calculated as the total number of collected calf cadavers divided by the number of calves registered in the I&R-database in that quarter. Newborn calf (0–3 days) mortality was calculated as the total number of collected newborn calf cadavers divided by the total number of adult cows (>2 years) registered in the I&R-database. This denominator was chosen because the number of non-ear tagged calves cannot be obtained from the I&R-database. Assuming that each adult cow (>2 year) generally gives birth to calf once a year, the number of cows (>2 year) was used as proxy for the number of non-ear tagged calves. For both quarters, differences in mortality rates per age category were tested by Wilcoxon rank-sum testing, in (i) case and control herds, (ii) herds with or without SBV-induced malformations in newborn calves, and (iii) herds with low or high within-herd seroprevalence (>50%).

3. Results 3.1. Descriptive results 3.1.1. Herds From the 149 selected herds, 39 herds were located in the north of the Netherlands, 102 in the central region and eight in the south. Location of the selected herds is displayed in Fig. 1. Table 1 provides an overview of herd characteristics of case and control herds. Case herds selected on acute symptoms in August/September 2011 and case herds selected on malformations in calves in 2011/2012 are displayed separately. During the farm visits it appeared that in five control herds, malformations in newborn calves had been detected after the initial selection of the herd as control herd. Also, in 28 control herds, the farmer stated to have seen clinical symptoms such as diarrhea, drop in milk production and/ or fever in adult cattle in 2011, potentially indicating acute SBV infection. Malformations in newborn calves were reported in two out of these 28 herds. 3.1.2. Seroprevalence In total, serum samples from 148 out of 149 herds were collected from February 13th 2012 to September 7th 2012. Samples collected after September 1st 2012 (n = 197) were excluded from further analysis to avoid seroprevalence estimates being influenced by seroconversions that occurred following infection in 2012. Consequently, three herds were not included in the seroprevalence estimation. Only samples with a negative or positive outcome were used for prevalence estimations, excluding 424 nonspecific samples. The average number of non-specific test results per herd was 2.77. The number of samples submitted per herd ranged from 30 to 90 (median 68.9). After randomly selecting a maximum of 70 samples per herd, a total of 9516 samples were used for further analysis. The mean age of the animals in the sample was 53.2 months. The age of sampled animals in case herds (54.2 months) was higher than in control herds (52.2 months) (p < 0.001). A total of 89 herds submitted one or more samples from animals that were less than 24 months of age at sampling. The number of samples of youngstock per herd ranged from 0 to 34 (median = 1). This resulted in a total number of 512 samples from youngstock (5.4%) in the dataset. True and apparent seroprevalences were fairly equal due to the high test sensitivity and specificity, therefore true prevalences are not shown. An overall animal-level seroprevalence was estimated at 79.9% (95% confidence interval (CI): 72.9–85.5). In adult animals, a seroprevalence of 80.1% (95% CI: 72.8–85.9) was found, compared to 75.5% (95% CI: 67.7–81.9) in youngstock. Overall animal-level seroprevalence in case herds was estimated at 84.4% (95% CI: 70.8–92.3) and did not differ from control herds in which a seroprevalence of 75.8% (95% CI: 67.5–82.5) was found (Table 2). A within-herd prevalence based on adult cattle (>2 year old) only was calculated for each herd (range: 13.1–100%), for which the frequency distribution within case and control herds is given in Fig. 2. The overall

A.M.B. Veldhuis et al. / Veterinary Microbiology 168 (2014) 281–293

285

Table 2 Overall estimated animal seroprevalences for Schmallenberg virus determined on samples collected from February 13th to September 1st, 2012.

Fig. 1. Location of selected case herds (black dots) and control herds (grey dots).

within-herd prevalence was not higher in case herds (90.6%) than in control herds (80.6%) (Wilcoxon rank-sum test p = 0.641). 3.1.3. Clinical symptoms attributed to SBV infection In total, SBV-related clinical symptoms in adult cattle have been observed by 85 out of 149 farmers in 2011. Whether these symptoms were actually related to acute SBV infection is unknown. From the 70 farmers who stated to have observed malformations in calves, 50 stated to

Category

N animals

Seroprevalence (95% CI)

Overall animal prevalence Animals > 24 months of age Youngstock Case herds (n = 75) Control herds (n = 74)

9516 9004 512 4786 4730

79.9% 80.1% 75.5% 84.4% 75.8%

(72.9–85.5) (72.8–85.9) (67.7–81.9) (70.8–92.3) (67.5–82.5)

have seen a drop in milk production in adult cattle in 2011, of which most (38%) stated that these symptoms had started in the middle/end of August 2011. Fourteen percent of the farmers stated that the symptoms started in June or July 2011, and 16% stated that the symptoms had started in September or later in 2011. Thirty-eight out of the 70 farmers stated to have seen diarrhea in adult cattle in 2011, of which the majority (53%) stated that these symptoms had started in the middle/end of August 2011. Thirteen percent of the farmers stated that the diarrhea symptoms started in June or July 2011, and 21% stated that the symptoms had started in September or later in 2011. These figures are in agreement with the time period in which farmers without malformations in newborn calves saw diarrhea and drop in milk production in adult cows. 3.1.4. Postmortem examination of submitted calves A total of 207 samples of aborted fetuses (n = 61), stillborn or malformed newborn calves, or calves that died within three days after birth (n = 146) have been submitted for postmortem examination and investigation of brain tissue for presence of SBV (Fig. 3). Calves were submitted between February 7th and September 1st, 2012, by 65 farmers. Aborted fetuses were submitted between February 14th and August 8th, 2012, from 41 herds. In total, 19 calves and two fetuses were considered SBV-positive either on postmortem examination or presence of SBV in

Table 1 Median and interquartile range (25th and 75th percentile) of herd characteristics of the dairy herds included in the study. Characteristic

Case herds

*

Herd size: cows >2 years of age Milk production per animal (kg/day)# Within-herd seroprevalence of SBV (%)

Herds that grazed cattle in 2011 Herds that purchased animals in 2011§ Herd health status 2011 IBR free status (certified) BVD free status (certified) Paratuberculosis (unsuspected) Salmonellosis (unsuspected)

Control herds (74 herds)

Herds that reported acute clinical signs in Aug/Sept’11 (26 herds)

Herds that reported malformations in newborn calves during notifiable period (49 herds)

Median (25–75th percentile)

Median (25–75th percentile)

Median (25–75th percentile)

91.5 (80–106) 28.4 (26.8–29.5) 98.1 (95.7–99.3)

80 (64–99) 26.3 (24.5–27.6) 97.0 (85.5–99.2)

99 (75–120) 27.7 (25.6–29.7) 90.5 (73.4–97.7)

Frequency (%) 22 (84.6%) 9 (34.6%)

Frequency (%) 42 (85.7%) 12 (24.5%)

Frequency (%) 49 (66.2%) 19 (25.7%)

6 (23.1%) 8 (30.8%) 25 (96.2%) 10 (38.5%)

18 22 36 27

22 28 64 34

(36.7%) (44.9%) (73.5%) (55.1%)

* At January 1st, 2012. Milk production per cow from week 1 to week 30 2011 (assumed SBV-free period). Purchased from within the Netherlands.

# §

(29.7%) (37.8%) (86.5%) (46.0%)

286

A.M.B. Veldhuis et al. / Veterinary Microbiology 168 (2014) 281–293

Fig. 2. Frequency distribution of within-herd seroprevalence of adult animals (>2 years) in case and control herds.

brain tissue. From the 19 calves, 10 were considered positive based on macroscopic pathological findings (gross lesions in the central nervous system, signs of arthrogryposis, etc.). From these 10 calves, presence of the virus in brain tissue was confirmed in three calves. The majority of the submitted calves (n = 136) did not show any clinical signs indicating SBV infection. Nevertheless, the virus was detected in brain tissue in nine of them. In the 65 herds of which a calf was submitted, a total of 4050 calves were born between February 7th and September 1st, 2012 of which 317 did not receive the compulsory ear tag, according to the I&R database. In general, it can be assumed that non-ear tagged calves either died within three days after birth or were stillborn. The cause of death of 171 out of the 317 calves is unknown as they were not submitted by the farmer for autopsy. Based on the calves that have been submitted, it can be concluded that at least 19 out of the 4050 calves born between February 7th and September 1st, 2012 (0.47%) were infected with SBV.

3.2. Risk factors for high SBV prevalence and malformations in calves Results of univariable analysis of farm management, housing and herd health variables prior to each multivariable analysis are shown in Table 3 (continuous variables) and Table 4 (categorical variables). Only pvalues less than 0.20 are displayed. Variables that were excluded from all multivariable analyses, based on univariate p-value >0.20, were roof opening (in cm), opening of stable doors (during day and/ or night), type of housing (loose housing, tie stall, deep litter stable), fly control in the stable between July and November, fly control applied at cows between July and November 2011, level of lighting in the stable throughout the night, purchase of animals, participation in cattle shows, presence of deer, presence of alpacas or lamas, presence of sheep, herd health status with regard to Paratuberculosis, Salmonellosis, IBR and BVD, and vaccination against BVD, BTV or neonatal diarrhea in 2011.

Fig. 3. Results of pathological investigation of aborted fetuses, stillborn or malformed calves, submitted between February and September 2012.

A.M.B. Veldhuis et al. / Veterinary Microbiology 168 (2014) 281–293

287

Table 3 Univariable results of all continuous variables of farm management, housing and herd health offered to multivariable analyses (if overall p-value <0.20). Variable

Median (25–75th percentile)

Probability of ‘high’ within-herd prevalence, p-value

Probability of malformations in calves, p-value

Herd size# Within-herd SBV seroprevalence (%) Mean daily milk production/cow (kg)* Duration of grazing (consecutive months)

90 (73–110) 95.7 (81.6–99.1) 27.3 (25.5–29.2) 5 (0.25–6)

<0.001 n.a. >0.20 <0.001

>0.20 0.069 0.156 0.010

#

Number of cows >2 years present on January 1st, 2012. * Milk production per cow from week 1 to week 30 2011 (assumed SBV-free period).

3.2.1. Herd-level risk factors for high within-herd seroprevalence In the final multivariable model of risk factors for a high within-herd seroprevalence, the factors ‘grazing in 2011’ and ‘herd size’ remained significant after the backwards selection process (Table 5). These two factors were the only variables included in the full model (Table 4), therefore model fit statistics of the final model remained unchanged (Table 5). With grazing, the odds of a herd to have a seroprevalence of >50% increased 9.9 fold (95% CI: 2.37– 41.2) compared to herds that were not grazed in 2011. Correcting for grazing, the odds of a high seroprevalence decreased slightly with increasing herd size (OR 0.98, p = 0.021). 3.2.2. Herd-level risk factors for malformations in calves In the multivariable model of risk factors for the birth of malformed calves in 2011/2012, the factors ‘clinical symptoms observed’ and ‘grazing in 2011’ were significant in the final model (Table 6). An interaction between ‘clinical symptoms observed’ and SBV-specific ‘within-herd seroprevalence’ was found (p = 0.04). This indicates that the effect of clinical symptoms in adult

Table 5 Results of multivariable logistic regression analysis of risk factors for high within-herd SBV-specific seroprevalence in dairy herds after the 2011 epidemic, with risk factors and their categories, odd ratios (OR), 95% confidence intervals, significance (p-value) and model fit statistics (Pseudo R2 and Akaike’s information criterion (AIC)) (n = 145). Variable

Category

OR

95% CI

p-value

Grazing in 2011 Herd size

Yes No Continuous

9.89 Reference 0.98

2.37; 41.2 – 0.97; 1.00

0.002 – 0.021

Pseudo R2 AIC

0.33 67.69

cattle on the birth of malformed calves is influenced by the seroprevalence in the herd. Therefore, it was decided to re-introduce ‘within-herd seroprevalence’ in the model as a new variable combining both within-herd seroprevalence (‘high’ vs. ‘reduced’, using a cut-off of 70% seroprevalence) and ‘clinical symptoms’. The cut-off of 70% within-herd seroprevalence was chosen to obtain sufficient numbers of observations in each of the four categories:

Table 4 Univariable results of all categorical farm management, housing and herd health variables offered to multivariable analyses (if overall p-value <0.20). Variable Herd size# 25% smallest herds (<74 cows) 50% average sized herds (74–109 cows) 25% largest herds (>109 cows) Farmer observed clinical symptoms in adult cattle in 2011* Yes No Cattle were grazed in 2011 Yes No Horizontal ventilation openings at the side of the stable <30 cm/no opening 30–100 cm without windbreak curtain 30–100 cm with windbreak curtain >100 cm without windbreak curtain >100 cm with windbreak curtain Presence of goats Yes No IBR vaccination in 2011 Yes No # *

At January 1st, 2012. Diarrhea, drop in milk production and/or fever.

Freq.

Probability of ‘high’ within-herd prevalence, p-value

Probability of malformations in calves, p-value

<0.001

>0.20

n.a.

0.001

<0.001

0.057

>0.20

0.101

>0.20

0.155

>0.20

0.086

38 73 38 85 61 113 36 17 42 20 47 22 7 138 23 123

A.M.B. Veldhuis et al. / Veterinary Microbiology 168 (2014) 281–293

288

Table 6 Results of multivariable logistic regression analysis of risk factors for malformations in newborn calves in dairy herds after the 2011 epidemic, with risk factors and their categories, odd ratios (OR), 95% confidence intervals, significance (p-value) and model fit statistics (Pseudo R2 and Akaike’s information criterion (AIC)) (n = 142). Variable

Category

OR

95% CI

p-value

Grazing in 2011

yes no 1 2 3 4

2.64 Reference 0.10 3.61 0.48 Reference

0.99; – 0.01; 0.58; 0.22; –

0.053 – 0.037 0.170 0.056

Prevalence and clinical signs category

Pseudo R2 AIC

*

7.06 0.87 22.6 1.02

0.09 187.3

* (1) ‘Reduced’ within-herd prevalence & no clinical symptoms observed in 2011. (2) ‘Reduced’ within-herd prevalence & clinical symptoms observed in 2011. (3) ‘High’ within-herd prevalence & no clinical symptoms observed in 2011. (4) ‘High’ within-herd prevalence & clinical symptoms observed in 2011.

(1) herds with ‘reduced’ within-herd prevalence & no clinical symptoms observed in 2011 (n = 13), (2) herds with ‘reduced’ within-herd prevalence & clinical symptoms observed in 2011 (n = 7), (3) herds with ‘high’ within-herd prevalence & no clinical symptoms observed in 2011 (n = 47), (4) herds with ‘high’ within-herd prevalence & clinical symptoms observed in 2011 (n = 75).

Grazing in 2011 tended to be associated with a 2.6 times higher odds on malformations in newborn calves (95% CI: 0.99–7.06). Malformations in newborn calves were less likely in herds in which the farmer did not observe SBV-related clinical symptoms in adult cattle in 2011, for both herds with reduced (OR 0.10) and high seroprevalence (OR 0.48), compared to herds with a high seroprevalence and clinical signs in adult cattle. Variables that were included in the full model however excluded during the model selection procedure were opening of stable doors (during day and/or night), milk production level, presence of goats and vaccination against

IBR in 2011. Model fit statistics of the full model (Pseudo R2 of 0.1435/AIC of 183.5) were slightly better than the final model (Table 6), yet log likelihood ratio testing indicated a non-significant improvement (p = 0.359). 3.3. Effects of SBV infection on milk production and mortality On average, 118 daily milk production records were available per herd for 2011 from 143 herds (range: 8–167). In case herds, the average milk production per cow per day in 2011 (25.9 kg, s.e.m 0.05) was slightly lower than in control herds (26.6 kg, s.e.m. 0.05) (p < 0.001). An overview of the mean daily milk production per cow in case and control herds, averaged by week, is shown in Fig. 4. A clear absolute minimum in milk production can be detected around week 35 (August 29th to September 4th, 2011), in both case and control herds. In case herds, the production level remained low for several weeks after the absolute minimum in week 35–36. The results of multivariable analysis of weekly milk production per animal per day in the 26 herds that reported clinical symptoms to GD Animal Health in

Fig. 4. Mean milk production (kg) per cow per day for week 1 to 52, 2011, of 149 dairy herds.

A.M.B. Veldhuis et al. / Veterinary Microbiology 168 (2014) 281–293 Table 7 Results of multivariable linear regression analysis on the weekly milk production per cow per day in 2011, for 26 herds that reported clinical symptoms in August/September 2011, with independent variables, estimates, 95% confidence intervals, significance (p-value) and model fit statistics (R2 and Akaike’s information criterion (AIC)) (n = 1287). Variable

Category

Regression coefficient

95% CI

SBV period

Yes No 4 3 2 1 Continuous

1.70 Reference 1.70 0.87 0.54 Reference 0.01

3.06; – 3.00; 2.18; 0.76; – 0.01;

Quarter

Herd size 2

R AIC

0.34 0.41 0.45 1.83 0.03

Variable

Category

Regression coefficient

95% CI

p-value

0.015 – 0.011 0.194 0.416 – 0.433

SBV period

Herd size

Yes No 4 3 2 1 Continuous

0.99 Reference 1.54 0.99 0.41 Reference 0.019

1.38; -0.60 – 1.79; 1.28 1.28; 0.70 0.16; 0.66 – 0.02; 0.02

<0.0001 – <0.0001 <0.0001 0.0012 – <0.0001

R2 AIC

0.07 34,365

0.12 5952

4. Discussion This study aimed at quantifying herd-level risk factors for high within-prevalence of SBV and SBV-induced

Wilcoxon rank-sum p-value.

Table 8 Results of multivariable linear regression analysis on the weekly milk production per cow per day in 2011, for 128 herds with a SBV within-herd prevalence of >50%, with independent variables, estimates, 95% confidence intervals, significance (p-value) and model fit statistics (R2 and Akaike’s information criterion (AIC)) (n = 6372).

p-valuej

August/September 2011 showed that in the four weeks that were defined as SBV-infection period, the average loss in daily milk production per cow was 1.71 kg per day (p = <0.001) (Table 7). Thus, during that month, the total loss per cow was approximately 51 kg of milk. Results of the equivalent analysis on 128 herds with a relatively high within-herd seroprevalence (>50%) showed that in the assumed SBV-infection period, the average loss in daily milk production per cow was 0.99 kg per day (p = 0.000) (Table 8). In those herds, the total loss per cow was approximately 30 kg of milk in that month. Mortality records were obtained for 148 herds, resulting in 269 quarterly observations of the third (2011q3) and fourth (2011q4) quarter of 2011. An overview of the median mortality rates per age category per group of herds, in 2011q3 and 2011q4, is provided in Table 9. No differences in newborn calf (0–3 days) mortality rates in 2011q3 were found between case and control herds, neither between herds with or without malformations in newborn calves, nor between herds with low or high within-herd seroprevalence. In 2011q4, a difference in newborn calf mortality was found between herds with low or high within-herd seroprevalence (p = 0.0211), but not between the other groups of herds. Calf (3 days—1 year) mortality rates were not different between herds in 2011q3. In 2011q4, calf mortality rates appeared to be higher in case herds (p = 0.044 (see footnote 1)) compared to control herds. Also, calf mortality rates in 2011q4 appeared to be higher in herds with malformed calves compared to herds without malformed calves (p = 0.011 (see footnote 1)). Cow (>1 year) mortality rates in 2011q3 were higher in herds with a low within-herd seroprevalence compared to herds with a high within-herd seroprevalence (p = 0.005 (see footnote 1)). No differences in cow mortality in 2011q3 were found between the other groups of herds. In 2011q4, no differences in cow mortality rates were found.

1

289

Quarter

clinical signs in newborn calves. Additionally, an attempt was made to estimate within-herd impact of SBV infection with regard to mortality rates and milk production. SBV has rapidly emerged throughout the Netherlands in 2011, affecting a large fraction of the ruminant population (Veldhuis et al., 2013). Little is known about this novel virus and how it affects cattle health at herd level. During the epidemic of 2011/2012 it became clear that the extent of the within-herd impact was diverse, varying from subclinical expression of the disease to acute clinical signs in cattle and malformations in newborn calves. In this study we focused on identifying herd-level risk factors that could explain differences in seroprevalence and incidence of malformations in newborn calves in dairy herds. For this purpose, case herds were selected based on SBV-related clinical signs in adult cattle in the late summer of 2011, or malformations in newborn calves during the notifiable period in the Netherlands (December 2011–July 2012). Control herds were selected based on absence of both malformations in newborn calves and anomalies in reproductive parameters. A matched casecontrol design by region was chosen, assuming that herds within the same region could have been exposed to the (infectious) vector in the same period. During the study period it appeared that certain control herds were misclassified, as SBV-induced malformations in newborn calves were observed in these herds after they were selected as controls. Also, seroprevalence appeared to be high in both case and control herds. Therefore, it was decided not to conduct analyses in a matched approach, but to reclassify herds based on seroprevalence, morbidity (clinical signs in adult cattle) and malformations in calves. By doing so, influence of the exposure of control herds to SBV, potentially weakening effect sizes, has been avoided. As a consequence, power of risk-factor analyses is somewhat reduced as the number of herds per group (for example high and low within-herd seroprevalence) was suboptimal. The SBV-specific seroprevalence was high in both case herds (84.4%; 95% CI: 70.8–92.3) and control herds (75.8%; 95% CI: 67.5–82.5). In the control herds, serum samples have been collected with a slight time delay compared to sampling in case herds, as the selection of control herds

290

A.M.B. Veldhuis et al. / Veterinary Microbiology 168 (2014) 281–293

Table 9 Median mortality rates per herd in newborn calves (0–3 days), calves (3 days—1 year) and cows (>1 year), in the third and fourth quarter of 2011, for 148 herds. Mortality rates are displayed for case and control herds, herds with and without malformations in newborn calves and herds with low and high within-herd seroprevalence (>50%). Significant differences in mortality rates between herds following Wilcoxon rank-sum testing (p-value <0.05) are displayed in bold (n = 296 observations). Group

Newborn calf mortality (median %) 2011q3

2011q4

2011q3

2011q4

2011q3

2011q4

Case herds (n = 74) Control herds (n = 74) Herds with malformed calves (n = 69) Herds without malformed calves (n = 79) Herds with high within-herd seroprevalence (n = 133) Herds with low within-herd seroprevalence (n = 15)

1.51 2.07 1.47 2.13 1.94 1.56

1.42 1.81 1.59 1.54 1.37 2.58

0.00 1.54 0.00 1.59 1.49 0.00

2.02 0.00 2.22 0.00 0.00 2.33

0.74 0.24 0.71 0.53 0.00 1.10

0.58 0.55 0.71 0.00 0.56 0.58

depended on information from the veterinary practitioner of the farm. Visiting of case herds started immediately at the beginning of the study period. As a consequence, about 40% of the serum samples from control herds were collected in the second half of the study period: in July 2012 (from 24 herds) and in August 2012 (from 5 herds). In contrast, 20.3% of the samples from case herds have been collected in July/August 2012 (n = 15 herds). When the rate of transmission of SBV has increased during 2012 (i.e. SBV has overwintered), the probability of seroconversion in naı¨ve animals may have increased towards the end of our study period. This could have increased the observed seroprevalences, in particular in control herds. To examine this, animal level seroprevalences in control herds were reestimated using two subsets; samples collected before July 1st, 2012 and samples collected in July and August 2012. Seroprevalence in samples from July/August 2012 (72.7%; 95% CI: 56.8–84.4) appeared no to be different from samples collected before July 1st 2011 (78.2%; 95 CI: 69.9– 84.7). 4.1. Risk factors for high within-herd seroprevalence and malformations in calves The results of this study showed that malformations in newborn calves were more likely in herds in which the farmer observed SBV-related clinical symptoms in adult cattle, compared to herds in which no clinical symptoms were observed. However, the magnitude of this effect was influenced by the seroprevalence in the herd, i.e. malformations in calves were most likely to occur in herds in which clinical signs were observed in combination with a high seroprevalence. Clinical disease related to the acute phase of SBV infection (drop in milk production, acute diarrhea, and/or fever) have been observed in both case and control herds. Whether these symptoms were actually caused by SBV infection is unknown. The acute clinical signs observed in adult cattle were not specific, and might also be caused by other infectious diseases (BVD, Salmonellosis). Also, the clinical signs could be associated with feeding disorders and the wet and cold weather conditions in the summer of 2011. Thus, some level of bias in terms of misclassification of herds with regard to morbidity is expected. In addition, due to recall bias, farmers that have observed malformations in calves may

Calf mortality (median %)

Cow mortality (median %)

be more likely to incorrectly state the presence of clinical signs in adult cattle than those that did not observe malformations. This may have resulted in an overestimation of the effect of SBV-related clinical signs in adult cattle on malformations in newborn calves. Herds in which the farmer stated to have observed SBV-related clinical signs had characteristics that indicated the presence of other infectious diseases (open farming system, suspected paratuberculosis herd health status, etc). Nevertheless, it is possible that herds coping with (non-SBV) infections are more susceptible for SBV infection (and thus malformations in calves) as a result of reduced immunological resistance. Alternatively, morbidity following SBV infection might be more severe when the infection occurs in parallel with other infections. Either way, the rate of transplacental transmission to the fetus might be higher in animals with decreased resistance prior to SBV infection. When modeling the incidence of malformed newborn calves, 26 herds were included that were initially selected as case herd due to their reporting of SBV-related clinical signs in the late summer of 2011. The inclusion of these herds might bias the estimates of the model as they were selected based on a risk factor of interest. Therefore, the model was rerun without the 26 case herds, which resulted in similar results regarding the combined effect of clinical signs and seroprevalence on the probability of malformations in calves (results not shown). When grazing was applied in 2011, the odds on malformations in newborn calves tended to be 2.6 times higher compared to herds in which cattle were kept indoors. This effect slightly increased when excluding the 26 case herds with clinical disease in the late summer of 2011 (OR 3.2, p = 0.043). With regard to the probability of a high within-herd seroprevalence (>50%) of a herd, grazing increased the odds considerably (OR 9.9, p = 0.002). Orthobunyaviridae are generally transmitted by arthropod vectors. SBV has been detected in several Culicoides species in different parts of Europe in 2011 (Rasmussen et al., 2012; De Regge et al., 2012; Elbers et al., 2013). It is expected that animals kept outdoors are more exposed to vectors than animals kept indoors, which is also shown by Baylis et al. (2010). They found that the risk of cattle receiving bites from the Culicoides obsoletus subspecies is reduced by housing in spring as well as in autumn. In addition, grazing has been identified as a risk factor for another vector-borne disease

A.M.B. Veldhuis et al. / Veterinary Microbiology 168 (2014) 281–293

(BTV-8) in Dutch cattle in the past (Santman-Berends et al., 2010). Nevertheless, the exceptionally wet and cold weather conditions in the Netherlands during the late summer of 2011 might have reinforced the impact of infectious diseases in cattle that were grazed. This also might have facilitated the diarrhea and drop in milk production during the acute phase of SBV infection, as to our knowledge these symptoms are not common for orthobunyavirus infections. Hence, the adverse weather conditions might have increased the magnitude of the effect of grazing we found. In addition, keeping animals indoors year-round did not prevent SBV infection as in our study an average within-herd prevalence of 63.9% was found in herds that stated not to have applied grazing at all in 2011 (n = 36). Adapting grazing management in a way that naı¨ve cattle, in particular youngstock, are exposed to the vector and thus acquire immunity before they become pregnant might prevent transplacental transmission to the fetus and limit the impact of the disease. Alternatively, preventing cows from being in the susceptible stage of gestation during the active season of the vector might as well limit the incidence of malformations in newborn calves. In the Netherlands however, a yearround calving pattern is applied, thus the feasibility of such a preventive measure is limited. No effect of stable design with regard to ventilation, such as keeping stable doors open and horizontal ventilation openings at the side of the stable, could be detected. Such characteristics of stable design might inhibit entrance of Culicoides through increased air circulation and thus prevent increase in seroprevalence, as suggested with regard to BTV-8 infection by SantmanBerends et al. (2010). This could not be confirmed in our study. An additional preventive measure focusing on control of the vector is protection of animals by repellents (in the stable or directly on animals). In our study, this did not seem to reduce exposure to SBV, as no effect of insect control strategies could be found. Also, Culicoides are generally known to be attracted by light and are most active during evening hours. However, in our study, no effect of lighting in the stable throughout the night could be found. 4.2. Impact The results of our study indicated a drop in milk production between the end of August 2011 and the first half of September (week 35–36), which is in agreement with our previous suggestions with regard to the start of epidemic (Veldhuis et al., 2013). In case herds, the average milk production per cow per day in 2011 was slightly lower than in control herds. This could be an effect of grazing, as 85% of the case herds applied grazing in 2011, compared to 66% of control herds. Milk production per animal is known to be negatively influenced by grazing, as a more constant ration, not influenced by weather conditions, can be provided when animals are kept indoors. To estimate the total loss in milk production per cow as a result of SBV infection, we defined a 4-week period in which SBV infection was expected to have occurred in herds with either clinical disease in adult cattle or a high within-herd seroprevalence. During this period,

291

the total loss in milk production was around 30 kg per animal (in herd with a high within-herd seroprevalence) to 51 kg per animal (in herds that reported SBV-related clinical signs in August/September 2011). Completely attributing this decrease to SBV infection is probably incorrect. The adverse weather conditions may have partly affected milk production levels. Also, it is expected that other (herd-level) factors are responsible for variation in milk production levels, given the low coefficients of determination (R2) of the two models. Milk production in the Netherlands is generally lower in summer and autumn due to increased grazing in these seasons. However, taking grazing into account in our model did not alter the estimate. Proper control herds without SBV infection were not available. From our data, no decrease in milk production per animal could be shown in herds with a relatively low within-herd seroprevalence (<50%; n = 15), although this might be due to a lack of power. Whether the period we defined as SBV-infection period was indeed the period in which SBV infection occurred in all study herds is uncertain. Nevertheless, it is very likely that SBV emerged in the Netherlands in August 2011, as numerous reports were made about clinical disease in cattle from the third week of August 2011 onwards (Muskens et al., 2012). Also, in our previous work we estimated a SBV seroprevalence of 4% in sheep in the Netherlands in August 2011, which increased to 60% in September 2011 (Veldhuis et al., 2013). Milk production recordings were provided for a single year (2011), thereby no within-herd comparisons could be made with the same period in previous years. It has been suggested that SBV infections (i.e. malformations) in newborn calves and lambs are the result of transplacental infection, in analogy with the pathogenesis of related Orthobunyaviruses such as Akabane virus (Parsonson et al., 1981) and Aino virus (Uchinuno et al., 1998). We roughly estimated that at least 0.46% of the calves born between February and September 2012 have been infected by SBV, based on malformed, aborted or stillborn calves submitted for autopsy during the study period. All 149 farmers were requested to submit all aborted, stillborn or malformed calves during the study period at no costs. Based on our finding that some calves were diagnosed as SBV-infected without showing any deformities, it cannot be excluded that a proportion of the calves that was not submitted – although perhaps born without deformities – was infected with SBV as well. Also, it is unlikely that no calves were aborted or stillborn in the herds from which no calves were received. Nevertheless, a low probability of infection of newborn calves is expected. Garigliany et al. (2012) stated, although not supported by results, that infection of pregnant cows by SBV often results in transmission to the fetus across the placenta, but this transmission does not automatically lead to abortion, stillbirth, or congenital deformities. In accordance, 9 out of the 19 stillborn calves that were considered SBV-positive in our study after post mortem investigation did not show any clinical signs indicating SBV infection (i.e. gross lesions in the central nervous system or signs of arthrogryposis). This suggests that the impact of the epidemic with regard to the total number of affected offspring might be

292

A.M.B. Veldhuis et al. / Veterinary Microbiology 168 (2014) 281–293

somewhat underestimated if calves are diagnosed on macroscopic pathological findings alone. In addition, it is expected that during the period in which notification of malformations in newborn ruminants was mandatory in the Netherlands, a number of SBV-infected newborn calves has not been noticed by the farmer due to absence of clinical signs. From the available mortality data of 2011, we focused on mortality rates in the third and fourth quarter of 2011, based on the assumption that an effect of SBV on mortality rates in 2011 could be expected in those quarters. It was hypothesized that mortality rates in newborn calves (0–3 days) might have been higher as a result of the SBV epidemic, in particular in 2011q4, although the number of newborn calves with malformations in the Netherlands was relatively low (Anonymous, 2012). Indeed, no differences in newborn calf mortality rates in 2011q3 were found between case and control herds, nor between herds with or without malformations in newborn calves, nor between herds with low or high within-herd seroprevalence. In the fourth quarter of 2011, a difference in newborn calf mortality was found between herds with low or high within-herd seroprevalence, but not between the other groups of herds. Since most malformed calves in the Netherlands were born between the end of January and the end of April 2012 (Bouwstra et al., 2013), a more profound effect of SBV on mortality in newborn calves specifically might be detectable in mortality rates of the first quarter of 2012. The increase in calf (3 days—1 year) mortality rates in the fourth quarter of 2011 we found might indicate mortality during the acute phase of the infection in calves. To our knowledge, such an effect of SBV infection in young ruminants has not reported by others. No differences in cow (>1 year) mortality rates were found in the third and fourth quarter of 2011, except for low and high prevalent herds in the third quarter of 2011. This lack of profound difference in cow mortality rates in the second half of 2011 is not unexpected given the relatively mild expression of SBV in adult cows during the acute phase of infection.

5. Conclusion An association was found between the probability of SBV infection and grazing. Herds that were grazed had an increased probability of a high seroprevalence compared to herds that were kept indoors in 2011. Also, the probability of SBV-induced malformed offspring in a herd increased with grazing. Incidence of malformations in newborn calves at herd level was associated by both clinical expression of the disease in adult cattle and within-herd seroprevalence. Nevertheless, the rate of vertical transmission of SBV to the fetus once a dams gets infected seems low. The potential relationship between an animal’s health status prior to SBV infection and the subsequent probability of infection (or, the probability of deformities in offspring) needs further investigation. The within-herd impact of SBV infection with regard to milk production and mortality rates appears to be limited.

Acknowledgments This study was financially supported by the Dutch Ministry of Economic Affairs and the European Commission (DG SANCO). We thank all participating dairy farmers for providing data and the laboratory staff at the Animal Health Service. References Anonymous, 2012. Nederlandse Voedsel- en Warenautoriteit: Voortgangsrapportage Schmallenberg, Peildatum 10-07-2012. Report in Dutch. retrieved October 23rd, 2012 from http://www.vwa.nl/ actueel/bestanden/bestand/2201855. Bouwstra, R.J.R.J., Kooi, E.A., de Kluijver, E.P., Verstraten, E.R.A.M., Bongers, J.H., van Maanen, C., Wellenberg, G.J., van der Spek, A.N., van der Poel, W.H.M., 2013. Schmallenberg virus outbreak in the Netherlands: routine diagnostics and test results. Veterinary Microbiology 165, 102–108. Dijkman, R., Mars, M.H., Wellenberg, G.J., 2012. Detection of Schmallenberg virus in bovine semen by one-step real-time PCR. In: Proceedings Second EAVLD Congress 1–4 July 2012. pp. S3–P19. Baylis, M., Parkin, H., Kreppel, K., Carpenter, S., Mellor, P.S., McIntyre, K.M., 2010. Evaluation of housing as a means to protect cattle from Culicoides biting midges, the vectors of bluetongue virus. Medical and Veterinary Entomology 24, 38–45. Van den Brom, R., Luttikholt, S.J.M., Lievaart-Peterson, K., Peperkamp, N.H.M.T., Mars, M.H., van der Poel, W.H.M., Vellema, P., 2012. Epizootic of ovine congenital malformations associated with Schmallenberg virus infection. Tijdschrift voor Diergeneeskunde 137, 106–111. De Blas, N., Ortego, C., Frankena, K., Noordhuizen, J., Thrusfield, M., 2000. Win Episcope 2.0. University of Zaragoza, Spain, Wageningen University and Utrecht University, The Netherlands, and University of Edinburgh, Scotland http://www.clive.ed.ac.uk/winepiscope/ De Regge, N., Deblauwe, I., de Deken, R., Vantieghem, P., Madder, M., Geysen, D., Smeets, F., Losson, B., van den Berg, T., Cay, A.B., 2012. Detection of Schmallenberg virus in different Culicoides spp. by realtime RT-PCR. Transboundary and Emerging Diseases 59, 471–475. Elbers, A.R.W., Loeffen, W.L.A., Quak, S., De Boer-Luijtze, E., van der Spek, A.N., Bouwstra, R., Maas, R., Spierenburg, M.A.H., De Kluijver, E.P., van Schaik, G., van der Poel, W.H.M., 2012. Seroprevalence of Schmallenberg virus antibodies among dairy cattle, The Netherlands, winter 2011–2012. Emerging Infectious Diseases 18 (7) 1065–1071. Elbers, A.R.W., Meiswinkel, R., van Weezep, E., Sloet van OldruitenborghOosterbaan, M.M., Kooi, E.A., 2013. Schmallenberg virus in Culicoides spp. biting midges, the Netherlands, 2011. Emerging Infectious Diseases 19 (1) 106–109. Garigliany, M.M., Bayrou, C., Kleijen, D., Cassart, D., Desmecht, D., 2012. Schmallenberg virus in domestic cattle, Belgium, 2012. Emerging Infectious Diseases 18 (9) 1512–1514. Van der Heijden, H.M.J.F., Bouwstra, R., Mars, M.H., van der Poel, W.H.M., Wellenberg, G.J., van Maanen, C., 2013. Development and validation of an indirect enzyme-linked immunosorbent assay for the detection of antibodies against Schmallenberg virus in blood samples from ruminants. Research in Veterinary Science, http://dx.doi.org/ 10.1016/j.rvsc.2013.04.022. Hoffmann, B., Scheuch, M., Ho¨per, D., Jungblut, R., Holsteg, M., Schirrmeier, H., Eschbaumer, M., Goller, K.V., Wernike, K., Fischer, M., Breithaupt, A., Mettenleier, T.C., Beer, M., 2012. Novel orthobunyavirus in cattle, Europe, 2011. Emerging Infectious Diseases 18 (3) 469–472. Loeffen, W., Quak, S., de Boer-Luijtze, E., Hulst, M., van der Poel, W., Bouwstra, R., Maas, R., 2012. Development of a virus neutralisation test to detect antibodies against Schmallenberg virus and serological results in suspect and infected herds. Acta Veterinaria Scandinavica 54 (1) 44. Muskens, J., Smolenaars, A.J.G., van der Poel, W.H.M., Mars, M.H., van Wuijckhuise, L., Holzhauer, M., van Weering, H., Kock, P., 2012. Diarree en productiedaling op Nederlandse melkveebedrijven door het Schmallenbergvirus. Tijdschrift voor Diergeneeskunde 137, 112– 115. Parsonson, I.M., Della-Porta, A.J., Snowdon, W.A., 1981. Akabane virus infection in the pregnant ewe. 2. Pathology of the foetus. Veterinary Microbiology 6, 209–224. Rasmussen, L.D., Kristensen, B., Kirkeby, C., Rasmussen, T.B., Belsham, G.J., Bødker, R., 2012. Culicoids as vectors of Schmallenberg virus [letter]. Emerging Infectious Diseases 18 (7) 1204–1206.

A.M.B. Veldhuis et al. / Veterinary Microbiology 168 (2014) 281–293 Rogan, W., Gladen, B., 1978. Estimating prevalence from the results of a screening test. American Journal of Epidemiology 107 (1) 71–76. Santman-Berends, I.M.G.A., Bartels, C.J.M., van Schaik, G., Stegeman, J.A., Vellema, P., 2010. The increase in seroprevalence of bluetongue virus (BTV) serotype 8 infections and associated risk factors in Dutch dairy herds, in 2007. Veterinary Microbiology 142 (3–4) 268–275. SAS Institute Inc., 2013. SAS/STAT 9.3 User’s Guide. SAS Institute Inc., Cary, NC, USA.

293

StataCorp, 2011. Stata Statistical Software: Release 12. StataCorp LP, College Station, TX. Uchinuno, Y., Noda, Y., Ishibashi, K., Nagasue, S., Shirakawa, H., Nagano, M., Ohe, R., 1998. Isolation of Aino virus from an aborted bovine fetus. Journal of Veterinary Medical Science 60 (10) 1139–1140. Veldhuis, A.M.B., van Schaik, G., Vellema, P., Elbers, A.R.W., Bouwstra, R., van der Heijden, H.M.J.F., Mars, M.H., 2013. Schmallenberg virus epidemic in the Netherlands: spatiotemporal introduction in 2011 and seroprevalence in ruminants. Preventive Veterinary Medicine 112, 35–47.