Calf-Level Factors Associated with Bovine Neonatal Pancytopenia – A Multi-Country Case-Control Study
et al. (2013) Calf-Level Factors Associated with Bovine Neonatal Pancytopenia - A Multi-Country
Case-Control Study. PLoS ONE 8(12): e80619. doi:10.1371/journal.pone.0080619
Calf-Level Factors Associated with Bovine Neonatal Pancytopenia - A Multi-Country Case-Control Study
Bryony A. Jones 0
Carola Sauter-Louis 0
Joerg Henning 0
Alexander Stoll 0
Mirjam Nielen 0
Gerdien Van Schaik 0
Anja Smolenaars 0
Matthijs Schouten 0
Ingrid den Uijl 0
Christine Fourichon 0
Raphael Guatteo 0
Aure lien Madouasse 0
Simon Nusinovici 0
Piet Deprez 0
Sarne De Vliegher 0
Jozef Laureyns 0
Richard Booth 0
Jackie M. Cardwell 0
Dirk U. Pfeiffer 0
Martin Beer, Friedrich-Loeffler-Institut, Germany
0 1 Department of Population & Production Health, Royal Veterinary College , North Mymms, Herts , United Kingdom , 2 Department of Pathology and Infectious Diseases, Royal Veterinary College , North Mymms, Herts , United Kingdom , 3 Clinic for Ruminants, Faculty of Veterinary Science, University of Munich , Oberschleissheim, Germany , 4 School of Veterinary Science, University of Queensland , Gatton, Queensland , Australia , 5 Farm Animal Health , Faculty of Veterinary Medicine, Utrecht University , Utrecht , The Netherlands , 6 GD Animal Health Service , Deventer , The Netherlands , 7 UMR1300 Biology, Epidemiology and Risk Analysis in Animal Health, INRA, LUNAM Universite , Oniris , Ecole nationale ve te rinaire, agroalimentaire et de l'alimentation Nantes Atlantique Nantes, France, 8 Department of Internal Medicine and Clinical Biology of Large Animals, Faculty of Veterinary Medicine, Ghent University , Merelbeke , Belgium , 9 Department of Reproduction , Obstetrics and Herd Health , Faculty of Veterinary Medicine, Ghent University , Merelbeke , Belgium
Bovine neonatal pancytopenia (BNP), a high fatality condition causing haemorrhages in calves aged less than 4 weeks, was first reported in 2007 in Germany and subsequently observed at low incidence in other European countries and New Zealand. A multi-country matched case-control study was conducted in 2011 to identify calf-level risk factors for BNP. 405 BNP cases were recruited from 330 farms in Belgium, France, Germany and the Netherlands by laboratory confirmation of farmer-reported cases. Up to four calves of similar age from the same farm were selected as controls (1154 calves). Risk factor data were collected by questionnaire. Multivariable modelling using conditional logistic regression indicated that PregSureHBVD (PregSure, Pfizer Animal Health) vaccination of the dam was strongly associated with BNP cases (adjusted matched Odds Ratio - amOR 17.8 first lactation dams; 95% confidence interval - ci 2.4, 134.4; p = 0.005), and second or more lactation PregSure-vaccinated dams were more likely to have a case than first lactation vaccinated dams (amOR 2.2 second lactation; ci 1.1, 4.3; p = 0.024; amOR 5.3 third or more lactation; ci 2.9, 9.8; p = ,0.001). Feeding colostrum from other cows was strongly associated with BNP if the dam was not PregSure-vaccinated (amOR 30.5; ci 2.1, 440.5; p = 0.012), but the effect was less if the dam was PregSure-vaccinated (amOR 2.1; ci 1.1, 4.0; p = 0.024). Feeding exclusively dam's milk was a higher risk than other types of milk (amOR 3.4; ci 1.6, 7.5; p = 0.002). The population attributable fractions were 0.84 (ci 0.68, 0.92) for PregSure vaccination, 0.13 (ci 0.06, 0.19) for feeding other cows' colostrum, and 0.15 (ci 0.08, 0.22) for feeding dam's milk. No other calf-level factors were identified, suggesting that there are other important factors that are outside the scope of this study, such as genetics, which explain why BNP develops in some PregSure-colostrum-exposed calves but not in others.
Competing Interests: The study was funded by Pfizer Animal Health, the manufacturer of PregSure BVD vaccine. However, the funding agreement allowed us
to independently publish our findings, subject to giving Pfizer 30 days notice of the results. Three of the co-authors (GVS, AS, MS) are affiliated to GD Animal
Health Service, which is a private company that provides diagnostic services for livestock and carries out the national animal health surveillance in the
Netherlands. There is no conflict of interest with the work that led to this paper, given that GD has no interest in the vaccine or in the outcome of the study. This
does not alter the authors adherence to all the PLOS ONE policies on sharing data and materials.
Bovine neonatal pancytopenia (BNP) was first reported in 2007
in Germany and subsequently observed at low incidence in several
other European countries and New Zealand . It affects calves
aged up to 4 weeks, causing skin and internal haemorrhages,
prolonged haemorrhage from wounds or orifices, and high case
fatality (90%) . No evidence of an infectious, toxic or genetic
aetiology has been found [2,57]. In 2010 an association was
suspected between affected calves and vaccination of their dams
with PregSureHBVD (PregSure, Pfizer Animal Health) . BNP
was induced experimentally by feeding calves colostrum from
unrelated PregSure-vaccinated dams that had previously given
birth to BNP calves . It was hypothesized that maternal
antibodies in colostrum were destroying calf blood and bone
marrow cells. PregSure is an inactivated bovine viral diarrhoea
(BVD) virus type 1 vaccine, which was first marketed in 2004 .
The manufacturer, Pfizer Animal Health, voluntarily stopped sales
to wholesalers in Europe in 2010, and New Zealand in 2011,
pending investigations into the cause of BNP. The marketing
authorisations in all concerned European Union (EU) Member
States were suspended in August 2010 following an EU
Commission decision based on recommendations from the
European Medicines Agency Committee for Veterinary Medicinal
Products (CVMP), with recall of product at wholesale level .
By the end of August 2012, 6913 suspected BNP cases had been
reported by farmers, veterinarians and laboratories via the
pharmacovigilance system in each European member state to
the marketing authorization holder, Pfizer Animal Health. The
numbers of reported suspected cases decreased in 2011 and 2012
compared with previous years. This case-control study was
conducted to identify potential risk factors for BNP occurrence
at the calf level.
Materials and Methods
All procedures on animals used in this study were in
concordance with the ethical conditions for animal
experimentation as mentioned in the European legislation (Directive 86/609/
EEC). The blood samples collected from calves on the farms were
taken for diagnostic purposes at the request of the owner as part of
clinical veterinary practice and therefore were not considered to be
experimental, so no formal approval of the protocol by an ethical
committee in any of the four countries was required.
The study was conducted between January and December 2011
in four countries that had experienced a high number of BNP
cases since 2007; Belgium, France, Germany and the Netherlands.
The target was to recruit 400 cases and match them to 24 control
calves of a similar age from the same farm. A multi-country design
was required to obtain sufficient cases and to recruit cases from
different management systems. A standard procedure for
recruitment of cases and controls was developed jointly by the country
Affected farms were identified from suspected cases reported
voluntarily by farmers and veterinarians to the research teams in
the respective country. Case-reporting was encouraged via notices
in the national veterinary and farming press in each country,
asking for cases of calves aged less than one month old with one or
more signs of bleeding from the skin (either spontaneously or from
injection or ear tag sites), mucosal petechial haemorrhages, blood
in diarrhoea, or death with internal or external bleeding. Free
diagnostic testing and/or post mortem examination were offered.
All farms that reported suspected cases were visited by a
veterinarian, either from the research team or their own
veterinarian on behalf of the research team, who conducted a
clinical examination. If the calf was less than 29 days of age
and had one or more clinical signs of BNP; multiple skin
haemorrhages, melena, petechiation of mucous membranes, or
sudden death with internal haemorrhage, then it was included
in the study as a suspected case. In order to confirm the
diagnosis a whole blood sample was collected if the calf was
alive, or if the calf was dead then a post mortem examination
and bone marrow histopathology were performed. During the
same visit, up to four calves without BNP clinical signs and
aged 1028 days were selected from the same farm to be
matched controls. Whole blood samples were collected to verify
that they did not have abnormal haematology. The veterinarian
collected data on the characteristics of suspected case and the
unaffected calves by face-to-face questionnaire-based interview
A case was therefore defined as a calf that had developed one or
more BNP clinical signs on or before 28 days of age and had bone
marrow depletion on histopathology and/or had
thrombocytopenia (,1506109/litre) and leucopenia (,56109/litre). A control
was defined as a calf on the same farm as a case, aged 1028 days
at the time of case reporting, no clinical signs of BNP up to 28 days
of age, and normal blood results (thrombocytes . = 3006109/
litre, leucocytes . = 56109/litre). Farmers were contacted again
when control calves were 28 days of age to confirm that they had
not developed BNP signs.
If the laboratory results subsequently indicated that a suspected
case did not meet the case definition then the case and its matched
controls were excluded from the study, and if a control calf did not
have normal blood results then it was excluded from the study.
The questionnaire was developed in English, translated into
French and German, and field-tested by researchers with
experience of BNP cases from the four countries. In the
Netherlands and Belgium the English questionnaire was used
and the interview conducted in Dutch. There were 91
questions, of which 36 collected descriptive data on calf, dam
and sire identification, calf characteristics, clinical signs and
laboratory results, and 55 collected data on potential risk factors
related to colostrum and milk feeding, dam and sire
characteristics, and dam vaccination history. Data were entered into an
internet-based form created in Open Source software
(LimeSurvey http://www.limesurvey.org/), exported to Microsoft
Excel and then to Stata IC 12.1 for coding, cleaning and
analysis. Thirty (6 descriptive, 24 potential risk factor) questions
were dropped due to a low number of responses or differences
in interpretation between countries. Five variables identified the
calf, dam and sire. Twenty two variables describing clinical and
post-mortem signs and laboratory results were used to define
cases and controls. Thirty four variables were used in the
statistical analysis and an additional 15 variables were created
by recoding, to give a total of 49 exposure variables (3
descriptive and 46 potential risk factor).
Due to the matched design, conditional logistic regression with
farm as the matching variable was used for univariable analysis to
obtain matched odds ratios (mOR), 95% confidence intervals (ci)
and Wald test p values. It was first conducted on the dataset of
1559 calves, but due to missing observations the sample size for
each variable was different. Variables with greater than 30%
missing observations were excluded. The number of calves in the
final multivariable model was 1296 due to missing observations in
the retained variables, so univariable analysis was repeated with
the smaller dataset. Variables with p values greater than 0.2 in
univariable analysis were excluded from multivariable analysis.
Pair-wise associations between the exposure variables were
examined by the chi-squared test, and polychoric correlation
was used to check for collinearity.
Multivariable analysis was conducted using conditional logistic
regression with farm as the group variable. The model was built
using forward stepwise regression starting with the variable with
the highest odds ratio and lowest p value on univariable analysis.
Variables were retained if the likelihood ratio test indicated that
inclusion led to a better model fit (p,0.05). For variables from the
same risk factor group (e.g. colostrum management, dam
vaccination) that were likely to be collinear, the variable with
the highest number of observations was added first. It was then
removed and the other collinear variables were added and
removed one by one. If more than one of the collinear variables
improved the model then the one with the most observations and/
or that was most biologically relevant was retained, as described
The population attributable fraction, PAF (proportion of cases
in the total population that would be avoided if the exposure was
removed, assuming the exposure is causal), was estimated using the
punafcc package for Stata .
Data were collected for 502 suspected cases and 1583 potential
controls from 410 farms. However, 62 suspected cases did not fit
the case definition, 122 potential controls had incomplete blood
results, and 210 potential controls had atypical blood results
(thrombocytes ,3006109/litre and/or leucocytes ,56109/litre).
Exclusion of these led to the loss of 97 confirmed controls with no
matching case and 35 confirmed cases with no matching control.
The remaining 405 confirmed cases were matched to 1154
confirmed controls from 330 farms with an overall case control
ratio of 1:2.8 (Table 1). The proportions of dairy, beef and mixed
farms that were included in the study from each country
approximately reflected the proportions of those types of farms
in each country. In the Netherlands, dairy farms predominate so
almost all suspected calves came from dairy farms. In Germany a
high proportion of cases recruited were from the south where beef
farms are few so the high proportion of dairy farms reflects the
regional distribution. The number of cases per farm varied from 1
to 3 and the number of controls from 1 to 9 (Table S1). The
proportion of males amongst cases (41%) was higher than amongst
controls (34%). In the Netherlands, cases were more likely than
controls to be male probably due to the sale of male calves before
one month of age so fewer were available to be selected as
controls. This difference was not observed in the other countries.
The most common breed of calf was Holstein-Friesian or Red
Holstein-Friesian (56.0%), followed by Fleckvieh (13.7%), Belgian
Blue (7.0%), Charolais (5.9%), other pure beef or dairy breeds
(7.8%), and crossbreeds (9.6%) (Table S2). There was no evidence
of a difference in breed distribution between cases and controls.
Colostrum management (Table 2, Table S3). There was
no evidence of an association between case/control status and the
following variables: suckled dam within 12 hours of birth, time to first
colostrum administration, number of times colostrum received in first 24 hours,
artificial colostrum used. There were increased odds of exposure of
cases compared with controls for total colostrum received, colostrum
obtained from cow(s) different from dam, pooled colostrum, pooled colostrum
includes colostrum of dam, and frozen colostrum. These variables were
associated with each other so colostrum obtained from cow(s) different
from dam was included first in the multivariable model from this
group because it had the highest number of observations, and
pooled colostrum, pooled colostrum includes colostrum of dam and frozen
colostrum were nested within it.
Belgium France Germany Netherlands Total
Milk feeding (Table 2, Table S3). There were increased
odds of exposure of cases compared with controls for feeding milk
powder, raw milk from dam, raw milk from dam only, and type of milk fed.
There was no evidence of an association between case/control
status and feeding raw milk, bulk milk, milk from cows with high somatic
cell count/clinical mastitis or withdrawn/discarded milk. Type of milk fed
was included first in the multivariable model from this group
because it provided the most information.
Dam and sire characteristics (Table 3, Table S4). There
was no evidence of an association between case/control status and
dam breed, dam was born on farm, dam was reared at another farm, or source
of bull. Case dams (dams of cases) had increased odds of being in
second or more lactation rather than first lactation compared with
control dams (dams of controls). Case dams had 12 times the odds
of previously giving birth to a BNP calf compared to control dams
(mOR 12.02; ci 5.44, 26.57; p = ,0.001) Since previously giving birth
to a BNP calf was considered to be on the causal pathway between
other exposures and the outcome, it was not included in the
multivariable model. Lactation number was correlated with many of
the vaccination variables and was a potential confounder of the
association between dam vaccination and case/control status.
Cases had increased odds of having a Fleckvieh rather than a
Holstein-Friesian or Red Holstein-Friesian sire, compared with
controls. It was not possible to obtain estimates for all breeds
because of low numbers of observations.
Dam BVD vaccination (Table S5). Case dams had
increased odds of being vaccinated against BVD compared with
control dams, and increased odds of having received more doses of
BVD vaccine. There was no evidence of any difference in the
timing of last BVD vaccination before calving between case and
control dams. BVD vaccination variables (all types of vaccine)
were correlated with each other, as well as with specific BVD
vaccine variables. They were not used in the multivariable model
because the specific BVD vaccine variables provided more
Dam PregSure vaccination (Table 4, Table S5). Case
dams had increased odds of being PregSure-vaccinated compared
with control dams. Case dams were more likely to have received
their last dose of PregSure longer before calving, and to have
received more doses of PregSure, compared with control dams.
PregSure vaccination variables were correlated with each other so
they were added to the model separately. The responses to dam
PregSure-vaccinated were more reliable than no. doses PregSure because
some farmers reported the initial two doses as a single dose, and
for some dams the number of doses was unknown. First lactation
cows were less likely to have received PregSure (49%) compared
with second (14%) and third or more lactation cows (12%), and
were more likely to have received fewer doses of PregSure than
second or third or more lactation cows (Table S6). The polychoric
correlation coefficient between PregSure doses and lactation
number was 0.61.
Other dam BVD vaccinations (Table 4, Table S5). After
PregSure, Bovilis BVD (MSD Animal Health) was the most
commonly used BVD vaccine, followed by Rispoval 3 (containing
BVD, Respiratory Syncytial Virus RS, and Parainfluenza type 3,
Pfizer Animal Health) and Bovidec BVD (Novartis Animal
Health). Case dams were more likely to have received Bovilis
BVD, Bovidec BVD and Mucosiffa (Merial) than control dams.
There was no evidence of an association between case/control
status and Rispoval BVD, Rispoval RS-BVD, Rispoval 3 or
Mucobovin (Merial) vaccination of dams. Some of these variables
were correlated with BVD vaccination (all types of vaccine) and
PregSure variables. There were 26 BVD vaccine combinations
and 34 BVD vaccination sequences (Tables S7, S8). The most
Total colostrum received by
Colostrum obtained from
cow(s) different from dam (1296)
Pooled colostrum (from
multiple cows) (1070)
Pooled colostrum includes
colostrum of dam (1061)
Frozen colostrum (1133)
Milk powder (1296)
Raw milk from dam (1296)
Type of milk fed to calf
Raw milk from dam only
Dam had previous
Raw milk from dam
Raw milk mixed & milk powder
Raw milk dam only
Raw milk dam only & milk powder
common combination was PregSure and Bovilis BVD (22%) and
the most common sequence of BVD vaccines was PregSure
followed by Bovilis BVD (21%). When the BVD vaccine
combinations were regrouped into five categories (Table S5), case
dams were more likely to have received PregSure in combination
with other BVD vaccines than PregSure only, compared with
control dams. When the BVD vaccine sequence variable was
regrouped into eight categories (Table S5), case dams had
increased odds of having received one or more other BVD
vaccine then PregSure, or receiving one or more other BVD
vaccine, then PregSure, then one or more other BVD vaccine,
rather than PregSure only, compared with control dams. These
Matched odds ratio (mOR) 95% confidence interval
Wald test p value
Holst Friesian/Red HF
Other pure breeds*
*Brown Swiss, Limousin, Limpurger, Pinzgau, Charolais, MRIJ, Montbeliarde, Abondance, Scandinavian Roodbont, Aubrac, Angus, Blanc Bleu, Maine Anjou, Normande,
Prog Federat Eur Pie, Blonde dAquitaine, Aure et St Girons Ca.
No. doses PregSure
Bovidec BVD (1284)
Mucosiffa BVD (1290)
variables were correlated with the other BVD vaccination
variables so they were added separately to the model.
Other dam vaccinations (Table 4, Table S5). Case dams
had increased odds of having received bluetongue or IBR
vaccination, and being vaccinated against more diseases,
compared with control dams. There was no association between case/
control status and dams receiving rota/coronavirus or other
vaccines. These variables were correlated with each other and with
some of the other vaccine variables.
Four variables were retained in the final model: dam
PregSurevaccinated, colostrum from different cow(s), lactation number, and raw milk
from dam only (Table 5, Table S9). No. doses PregSure, and
combinations and sequences of BVD vaccines were alternatives
to dam PregSure-vaccinated, and frozen colostrum, pooled colostrum and
pooled colostrum including dams were alternatives to colostrum from
different cow(s), but they had fewer observations.
There was evidence of interaction between dam
PregSurevaccinated and lactation number, between dam PregSure-vaccinated and
colostrum from different cow(s), and between lactation number and
colostrum from different cow(s). Including two-way interactions
between dam PregSure-vaccinated and lactation number and between
dam PregSure-vaccinated and colostrum from different cow(s) was a better
fit than the models with single interactions or all three interactions.
The odds of a case having a second lactation dam were five times
the odds of having a first lactation dam, if the dam was
PregSurevaccinated (interaction term adjusted matched odds ratio - amOR
4.8; ci 1.1, 20.7; p = 0.034), and the odds of having a vaccinated
third or more lactation dam were 7 times the odds of having a
vaccinated first lactation dam (interaction term amOR 7.4; ci 1.9,
28.9; p = 0.004). The odds of a case having received colostrum
from different cow(s) were 90% lower than the odds of not having
received colostrum from other cows, if the dam was
PregSurevaccinated (interaction terms amOR 0.1; ci 0.01, 0.95; p = 0.046).
There was very strong evidence of an association between case/
control status and having a PregSure-vaccinated dam. A case had
18 times the odds of being born to a dam that was
PregSurevaccinated rather than unvaccinated compared with a control, if
the dam was first lactation and the calf did not receive colostrum
from other cows, adjusting for type of milk fed (amOR 17.8; ci.
2.4, 134.4; p = 0.005). For calves that received colostrum from
Dam PregSure vaccination,
no colostrum from different cow(s), dam is first lactation
Dam PregSure vaccination,
calf received colostrum from different cow(s), dam is first lactation
no colostrum from different cow(s), dam is second lactation
no colostrum from different cow(s), dam is third or more lactation
Dam PregSure vaccination,
Dam PregSure vaccination,
when dam is unvaccinated
when dam is PregSure vaccinated
Colostrum from different cow(s)
when dam is unvaccinated
Colostrum from different cow(s)
when dam is PregSure vaccinated
Raw milk from dam only
odds ratio (amOR)
Exposure variables are dam PregSure vaccination, lactation number, colostrum from different cow(s) and raw milk from dam only. Interactions were included between
dam PregSure vaccination and lactation number, and between dam PregSure vaccination and colostrum from different cow(s). The interaction terms have been
multiplied with the baseline odds ratios to obtain odds ratios for each variable category above baseline.
other cows, there was no evidence of a difference in the odds of a
case having a PregSure-vaccinated dam rather than an
unvaccinated dam (amOR 1.2; ci 0.3, 5.6; p = 0.79), if the dam was first
lactation, adjusting for type of milk fed. Compared with a control,
a case with a second lactation dam that did not receive colostrum
from other cows had 86 times the odds of having a
PregSurevaccinated dam rather than an unvaccinated dam (amOR 86.0; ci
7.4, 995.3; p = ,0.001), and a case with a third or more lactation
dam that did not receive colostrum from other cows had 132 times
the odds of having a PregSure-vaccinated dam rather than an
unvaccinated dam (amOR 132.0; ci 9.9, 1764.7; p = ,0.001),
adjusting for type of milk fed.
For calves with unvaccinated dams, there was no evidence of a
difference in the odds of a case being born to a dam in first, second
or third or more lactation compared with a control, adjusting for
source of colostrum and type of milk fed (second lactation amOR
0.5; ci 0.1, 1.6; p = 0.23, third or more lactation amOR 0.7; ci 0.2,
2.4; p = 0.60). However, if the dam was PregSure-vaccinated, a
case had twice the odds of having a second lactation dam (amOR
2.2; ci 1.1, 4.3; p = 0.024) and 5 times the odds of having a third
lactation dam (amOR 5.3; ci 2.9, 9.8; p = ,0.001) compared with
a first lactation dam, adjusting for source of colostrum and type of
A case had 30 times the odds of having received colostrum from
another dam compared with a control, if its dam was not
PregSure-vaccinated (amOR 30.5; ci 2.1, 440.5; p = 0.012), but
had only twice the odds of having received colostrum from another
dam if its dam had been PregSure-vaccinated, adjusting for
lactation number and type of milk fed (amOR 2.1; ci 1.1, 4.0;
p = 0.024).
A case had 3 times the odds of having been fed raw milk only
from its dam rather than other types of milk (with or without dams
milk) compared with a control (amOR 3.4; ci 1.6, 7.5; p = 0.002)
when adjusting for dam PregSure-vaccination, lactation number
and source of colostrum.
The population attributable fraction (PAF) for dam PregSure
vaccination was 0.84 (ci 0.68, 0.92) indicating that if no
PregSurevaccinated cows had been used for breeding then 84% of cases
would have been avoided. If calves had been fed colostrum from
their own dams only, rather than colostrum from other cows, then
12% of cases would have been avoided (PAF 0.13; ci 0.06, 0.19). If
calves had not been fed exclusively on their dams milk, then 15%
of cases would have been avoided (PAF 0.15; ci 0.08, 0.22). These
estimates are based on the assumption of a causal relationship
between each variable and case/control status.
Of the 440 confirmed BNP cases in this study, 20 cases (4.5%)
had dams that were not PregSure-vaccinated and had not
previously had a BNP calf, had received only dams colostrum,
and came from farms with no history of PregSure vaccination or
Our results show that PregSure vaccination of a cow was
strongly associated with her having a BNP calf, and that older
PregSure-vaccinated cows were more likely to have a BNP calf
than younger vaccinated cows. This was partly explained by the
increased odds of BNP with increasing doses of PregSure and
correlation between PregSure doses and lactation number.
Feeding colostrum from other cows, in addition to or instead of
dam colostrum, was strongly associated with BNP if the dam had
not been PregSure-vaccinated, but its effect was less if the dam had
been vaccinated. When calves received colostrum from other
cows, most farmers were unable to identify which cows the
colostrum came from, so their PregSure vaccination status was
unknown. These findings suggest that if a dam had been
PregSurevaccinated and the calf received colostrum from its dam then there
was an increased odds of BNP. But if the calf received some
colostrum from other cows (which may or may not have been
vaccinated) then the dilution effect of receiving some non-BNP
colostrum reduced the odds. However, for calves of unvaccinated
cows, feeding colostrum from other cows was strongly associated
with BNP, presumably because feeding colostrum from multiple
cows increased the chance of the calf ingesting some colostrum
from a PregSure-vaccinated cow. The strong association between
BNP and PregSure vaccination, and consumption of colostrum
from PregSure-vaccinated dams, was consistent with
singlecountry case-control studies with small sample sizes conducted in
the UK and Germany [13,14]. We also found that exclusively
feeding dams raw milk was associated with an increased odds of
BNP, compared to feeding milk from one or more other sources.
Early lactation milk contains some antibodies that can still be
absorbed up to 48 hours after birth . Feeding milk powder or
bulk tank milk instead of, or to supplement, dams milk will reduce
the amount of antibody ingested. Alternatively the observed effect
could be due to an association between feeding dams milk and
other management factors that affect the risk of BNP. In our study,
calves that suckled their dam within 12 hours of birth were more
likely to be fed only dams raw milk. We hypothesize that calves
that suckle could receive a larger volume of colostrum earlier in life
leading to higher levels of antibody absorption, which could
increase the risk of BNP.
The estimated population attributable fractions indicate that the
most effective intervention would have been to avoid breeding
from PregSure-vaccinated cows (86% case avoided). Not
exclusively feeding dams milk would have avoided 15% cases and not
feeding colostrum from other cows would have avoided 12%.
Twenty cases in this study had no apparent exposure to the
identified risk factors. They could have been misclassified with
respect to dam vaccination status, colostrum feeding or dams BNP
history due to incorrect farmer recall, or they could have
developed pancytopenia due to other causes. Prior to the
identification of BNP there were sporadic cases of unexplained
pancytopenia in young calves , so some of our 20 calves
could represent a background incidence of pancytopenia that is
unrelated to ingestion of colostrum from PregSure-vaccinated
cows. Further research is required to determine whether sporadic
unexplained pancytopenia cases have the same pathogenesis as
PregSure-associated BNP cases, and therefore whether the
introduction of PregSure vaccination has increased the incidence
of an existing but rare syndrome.
Despite widespread use of PregSure vaccine prior to its
withdrawal, BNP incidence has been low, suggesting that
consumption of colostrum from a PregSure-vaccinated dam is
not a sufficient cause of BNP. We did not identify any other
important calf or management-related risk factors. Research into
BNP pathogenesis is on-going with a prevailing hypothesis that it is
a neonate-maternal incompatibility phenomenon related to
PregSure-induced maternal alloantibodies against bovine cell
surface molecules due to bioprocess-related impurities from the
cell line used for virus propagation . Bastian et al. 
showed that sera of dams that had previously had a BNP calf
contained alloantibodies that bound to bovine leucocytes, and
Foucras et al  reproduced BNP in healthy calves by
transferring serum antibodies from PregSure-vaccinated dams.
Animals vaccinated with three doses of PregSure had higher
alloantibody titres compared with animals receiving a single
PregSure dose or other BVD vaccines . This supports our
finding that the odds of BNP increase with the number of
PregSure doses given to the dam. Bridger et al  showed that
maternal alloantibodies to surface antigens of neonatal leucocytes
are transferred via colostrum from BNP dams to neonatal calves,
and higher antibody titres in the dam led to more severe clinical
signs in the calf. Various proteins found in both PregSure vaccine
and the cell line used to produce the vaccine have been implicated
as possible alloantigen candidates [2123,25]. Bell et al. 
suggest that feeding BNP colostrum from multiple cows increases
the likelihood that the colostrum will contain antibodies that will
react with most calf allotypes, which fits with our finding that
calves from unvaccinated dams were at increased risk of BNP if
they were fed colostrum from other cows. They suggest that the
unique adjuvant in PregSure could amplify production of
alloantibody against vaccine antigens as well as boosting
pregnancy-induced maternal alloantibodies against
paternallyderived foetal MHC antigens . The latter mechanism may
explain the aetiology of some of the sporadic cases of unexplained
pancytopenia that are not linked to the ingestion of colostrum
from PregSure-vaccinated cows.
It was anticipated that BNP incidence would decline during the
study period due to withdrawal of PregSure from distribution in
2010, and advice given to some farmers to avoid feeding colostrum
from cows that had previously had BNP calves. A multi-country
study was therefore necessary to obtain sufficient BNP cases to
have the statistical power to detect important risk factors. A
casecontrol study is the most appropriate design to investigate a rare
disease, and for the investigation of calf-level risk factors it was
necessary to match by farm. This meant that most management
factors were the same for both cases and controls on the same
farm, and, even where there were differences, the farmer could
have reported routine practices rather than what had happened to
individual calves, leading to misclassification and an
underestimation of effect. The multi-country design meant that there was
variation between countries in case-reporting, farm visits, sample
collection, post-mortem examination, laboratory testing and
questionnaire interpretation. There were also differences between
countries in production methods, cattle breeds, and policies on
BVD control and vaccination programmes. The matched design
minimised the effect of these country differences on our results.
Case recruitment relied on passive reporting by farmers and
veterinarians, strengthened by a communication campaign to
encourage reporting of cases and provision of free post mortem
examination and laboratory testing. The reasons for a farmer or
veterinarian not reporting a case might include; being unaware of
the invitation to report cases or too busy to report, or having
reported cases prior to the study and therefore seeing no benefit in
reporting further cases. It is possible that farmers who report
disease are more likely to take preventive measures such as
vaccination. There is therefore a potential bias in selection of case
farms in that the study population might not be representative of
all BNP-affected farms in the four countries. But even if these
biases were present, they are unlikely to have biased the particular
inferences reported here.
Strict definitions for case and control calves were applied due to
variations between countries in time elapsed before blood analysis
and use of multiple laboratories, to minimise the inclusion of false
positive cases. Using the thrombocyte and leucocyte values of the
Netherlands control calves, a 95% reference interval was
calculated based on the mean and standard deviation, where
values below the lower limit of the interval were considered
abnormal for the current study. Using these results the blood
values for controls were set at .3006109/litre thrombocytes and
.56109/litre leucocytes. For case calves, the thrombocyte value
was set at ,1506109/litre, the lower threshold of the reference
range in the Netherlands.
The number of questions in the questionnaire was high, to
capture management practices in the diverse farming systems of
the four countries, which increased the risk of detecting an
association by chance when there was no true association. As the
questionnaire had been translated into three languages and
administered by a range of veterinarians and researchers, prior
to data analysis the research team discussed in detail how the
questionnaire had been administered and differences in
interpretation, and there was regular consultation during analysis to
inform the interpretation of results.
In conclusion, this multi-country study provides strong evidence
that receiving colostrum from a PregSure-vaccinated cow is a major
risk factor for BNP. If calves are only given colostrum from
unvaccinated cows then it is highly unlikely that a calf will develop
BNP. The study design and sample size provided adequate
statistical power to investigate many hypotheses related to farming
practices such as colostrum management, calf feeding and
vaccination, but no other important calf management-related risk
factors were identified. This suggests that there are other important
factors, such as genetics, that were outside the scope of this study,
which explain why BNP develops in some
PregSure-colostrumexposed calves but not in others. These require further investigation.
Table S1 Number of cases and controls per farm: number
(percentage) of farms with each case:control ratio.
Numbers of cases and controls by breed.
Number of PregSure doses by lactation number of
Combinations of BVD vaccines.
Sequences of BVD vaccines.
Multivariable model Stata output.
The authors gratefully acknowledge the contributions of the farmers and
veterinarians who participated in this study.
Conceived and designed the experiments: BAJ CSL JH MN GVS A. Stoll
MS CF RG PD SDV RB JMC DUP. Performed the experiments: CSL JH
A. Smolenaars A. Stoll MS RG AM PD JL. Analyzed the data: BAJ MN
GVS IDU CF RG AM SN SDV JMC DUP. Wrote the paper: BAJ CSL
MN GVS IDU CF RG AM PD SDV JL RB JMC DUP.
1. Pardon B , Stuyven E , Stuyvaert S , Hostens M , Dewulf J , et al. ( 2011 ) Sera from dams of calves with bovine neonatal pancytopenia contain alloimmune antibodies directed against calf leukocytes . Vet Immunol Immunopathol 141 : 293 - 300 .
2. Krappmann K , Weikard R , Gerst S , Wolf C , Kuhn C ( 2011 ) A genetic predisposition for bovine neonatal pancytopenia is not due to mutations in coagulation factor XI . Vet J 190 : 225 - 229 .
3. Promed ( 2011 ) Bovine neonatal pancytopenia (03): New Zealand Archive Number: 20110828.2636.
4. Klee W ( 2009 ) Proceedings of satellite symposium on haemorrhagic diathesis in calves , Marseille, France; Marseille, France.
5. Friedrich A , Rademacher G , Weber B , Kappe E , Carlin A , et al. ( 2009 ) ( Increased incidence of haemorrhagic diathesis as a result of bone marrow damage in young calves) . Tierarztliche Umschau 64 : 423 - 431 .
6. Kappe EC , Halami MY , Schade B , Alex M , Hoffmann D , et al. ( 2010 ) Bone marrow depletion with haemorrhagic diathesis in calves in Germany: characterization of the disease and preliminary investigations on its aetiology . Berl Munch Tierarztl Wochenschr 123 : 31 - 41 .
7. Pardon B , Steukers L , Dierick J , Ducatelle R , Saey V , et al. ( 2010 ) Haemorrhagic diathesis in neonatal calves: an emerging syndrome in Europe . Transbound Emerg Dis 57 : 135 - 146 .
8. Promed ( 2010 ) Bovine neonatal pancytopenia - Germany: vaccinal etiology suspected . Archive Number: 20100409.1142.
9. Friedrich A , Buttner M , Rademacher G , Klee W , Weber BK , et al. ( 2011 ) Ingestion of colostrum from specific cows induces Bovine Neonatal Pancytopenia (BNP) in some calves . BMC Vet Res 7 : 10 - 20 .
10. Kasonta R , Sauter-Louis C , Holsteg M , Duchow K , Cussler K , et al. ( 2012 ) Effect of the vaccination scheme on PregSure(R) BVD induced alloreactivity and the incidence of Bovine Neonatal Pancytopenia . Vaccine 30 : 6649 - 6655 .
11. EMA ( 2010 ) Opinion following an Article 78 procedure for Pregsure BVD and associated names . Committee for Veterinary Medicinal Products, European Medicines Agency , 7 /10/2010.
12. Newson RB ( 2013 in press) Attributable and unattributable risks and fractions and other scenario comparisons . The Stata Journal.
13. Lambton SL , Colloff AD , Smith RP , Caldow GL , Scholes SF , et al. ( 2012 ) Factors associated with bovine neonatal pancytopenia (BNP) in calves: a casecontrol study . PLoS One 7 : e34183 .
14. Sauter-Louis C , Carlin A , Friedrich A , Assad A , Reichmann F , et al. ( 2012 ) Case control study to investigate risk factors for bovine neonatal pancytopenia (BNP) in young calves in southern Germany . Prev Vet Med 105 : 49 - 58 .
15. Stott GH , Marx DB , Menefee BE , Nightengale GT ( 1979 ) Colostral immunoglobulin transfer in calves I. Period of absorption . J Dairy Sci 62 : 1632 - 1638 .
16. Fukunaka M , Toyoda Y , Kobayashi Y , Furuoka H , Inokuma H ( 2010 ) Bone marrow aplasia with pancytopenia and hemorrhage in a Japanese Black calf . J Vet Med Sci 72 : 1655 - 1656 .
17. Ammann VJ , Fecteau G , Helie P , Desnoyer M , Hebert P , et al. ( 1996 ) Pancytopenia associated with bone marrow aplasia in a Holstein heifer . Can Vet J 37 : 493 - 495 .
18. Shimada A , Onozato T , Hoshi E , Togashi Y , Matsui M , et al. ( 2007 ) Pancytopenia with bleeding tendency associated with bone marrow aplasia in a Holstein calf . J Vet Med Sci 69 : 1317 - 1319 .
19. Gosselin VB , Fecteau G , Nichols S ( 2011 ) Presumptive bovine neonatal pancytopenia in a Holstein calf in Quebec . Can Vet J 52 : 788 - 790 .
20. Bastian M , Holsteg M , Hanke-Robinson H , Duchow K , Cussler K ( 2011 ) Bovine Neonatal Pancytopenia: is this alloimmune syndrome caused by vaccineinduced alloreactive antibodies? Vaccine 29 : 5267 - 5275 .
21. Deutskens F , Lamp B , Riedel CM , Wentz E , Lochnit G , et al. ( 2011 ) Vaccineinduced antibodies linked to bovine neonatal pancytopenia (BNP) recognize cattle major histocompatibility complex class I (MHC I) . Vet Res 42 : 97 - 106 .
22. Euler KN , Hauck SM , Ueffing M , Deeg CA ( 2013 ) Bovine neonatal pancytopenia - Comparative proteomic characterization of two BVD vaccines and the producer cell surface proteome (MDBK) . BMC Vet Res 9 : 18 - 26 .
23. Foucras G , Corbiere F , Tasca C , Pichereaux C , Caubet C , et al. ( 2011 ) Alloantibodies against MHC class I: a novel mechanism of neonatal pancytopenia linked to vaccination . J Immunol 187 : 6564 - 6570 .
24. Bridger PS , Bauerfeind R , Wenzel L , Bauer N , Menge C , et al. ( 2011 ) Detection of colostrum-derived alloantibodies in calves with bovine neonatal pancytopenia . Vet Immunol Immunopathol 141 : 1 - 10 .
25. Laming E , Melzi E , Scholes SF , Connelly M , Bell CR , et al. ( 2012 ) Demonstration of early functional compromise of bone marrow derived hematopoietic progenitor cells during bovine neonatal pancytopenia through in vitro culture of bone marrow biopsies . BMC Res Notes 5 : 599 - 608 .
26. Bell CR , Rocchi MS , Dagleish MP , Melzi E , Ballingall KT , et al. ( 2013 ) Reproduction of bovine neonatal pancytopenia (BNP) by feeding pooled colostrum reveals variable alloantibody damage to different haematopoietic lineages . Vet Immunol Immunopathol 151 : 303 - 314 .