No Clear Effect of Initiating Vaccination against Common Endemic Infections on the Amounts of Prescribed Antimicrobials for Danish Weaner and Finishing Pigs during 2007–2013

Abstract

It is often stated that vaccines may help reduce antimicrobial use in swine production. However, limited evidence is available outside clinical trials. We studied the change in amounts of antimicrobials prescribed for weaners and finishers in herds following initiation of vaccination against five common endemic infections: Mycoplasma hyopneumoniae, Actinobacillus pleuropneumoniae, porcine circovirus type II, porcine reproductive and respiratory syndrome virus, and Lawsonia intracellularis. Comparison was made to the change after a randomly selected date in herds not vaccinating against each of the infections. Danish sow herds initiating vaccination during 2007–2013 were included (69–334 herds, depending on the analysis). Danish sow herds with no use of the vaccine in question were included as non-exposed herds (130–570 herds, depending on the analysis). Antimicrobial prescriptions for weaners in sow herds and for finishers in receiving herds were extracted from the VetStat database for a period of 12 months before and 6–18 months after the first purchase of vaccine, or random date and quantified as average animal daily doses (ADDs) per 100 animals per day. The herd-level difference between ADD in the period after and before vaccination was the outcome in linear regression models for weaner pigs, and linear mixed-effects models for finishing pigs, taking into account sow herds delivering pigs to two or more finisher herds. Three plausible risk factors (Baseline ADD, purchase of specific vaccine, purchase of other vaccines) and five confounders (herd size, export and herd health status, year and season) were initially considered in all 10 models. The main significant effect in all models was the Baseline ADD; the higher the Baseline ADD was for weaner and finishing pigs, the larger the decrease in ADD was following vaccination (or random date for non-vaccinating herds). Regardless of vaccination status, almost equal proportions of herds experienced a decrease and an increase in ADD resulting in no overall Change in ADD. Furthermore, only minor effects were found, when vaccinations were used in combination. In conclusion, this study provided little support for the hypothesis that vaccination against five common endemic diseases provides a plausible general strategy to reduce antimicrobial use in Danish pig herds.

Full text

January 2017 | Volume 3 | Article 1201
ORIGINAL RESEARCH
published: 16 January 2017
doi: 10.3389/fvets.2016.00120
Frontiers in Veterinary Science | www.frontiersin.org
Edited by:
Katharina Stärk,
SAFOSO AG, Switzerland
Reviewed by:
Gustavo Machado,
University of Minnesota, USA
Marcus G. Doherr,
Freie Universität Berlin, Germany
*Correspondence:
Amanda Brinch Kruse
amanda@sund.ku.dk
Specialty section:
This article was submitted to
Veterinary Epidemiology and
Economics,
a section of the journal
Frontiers in Veterinary Science
Received: 21July2016
Accepted: 19December2016
Published: 16January2017
Citation:
KruseAB, deKnegtLV, NielsenLR
and AlbanL (2017) No Clear Effect of
Initiating Vaccination against
Common Endemic Infections on the
Amounts of Prescribed Antimicrobials
for Danish Weaner and Finishing Pigs
during 2007–2013.
Front. Vet. Sci. 3:120.
doi: 10.3389/fvets.2016.00120
No Clear Effect of Initiating
Vaccination against Common
Endemic Infections on the Amounts
of Prescribed Antimicrobials for
Danish Weaner and Finishing Pigs
during 2007–2013
Amanda Brinch Kruse1*, Leonardo Víctor de Knegt1, Liza Rosenbaum Nielsen1 and
Lis Alban2
1 Department of Large Animal Sciences, Faculty of Health and Medical Sciences, University of Copenhagen, Frederiksberg,
Denmark, 2 Danish Agriculture & Food Council, Aarhus, Denmark
It is often stated that vaccines may help reduce antimicrobial use in swine production.
However, limited evidence is available outside clinical trials. We studied the change in
amounts of antimicrobials prescribed for weaners and finishers in herds following initiation
of vaccination against five common endemic infections: Mycoplasma hyopneumoniae,
Actinobacillus pleuropneumoniae, porcine circovirus type II, porcine reproductive and
respiratory syndrome virus, and Lawsonia intracellularis. Comparison was made to the
change after a randomly selected date in herds not vaccinating against each of the
infections. Danish sow herds initiating vaccination during 2007–2013 were included
(69–334 herds, depending on the analysis). Danish sow herds with no use of the vaccine
in question were included as non-exposed herds (130–570 herds, depending on the
analysis). Antimicrobial prescriptions for weaners in sow herds and for finishers in receiv-
ing herds were extracted from the VetStat database for a period of 12 months before
and 6–18months after the first purchase of vaccine, or random date and quantified as
average animal daily doses (ADDs) per 100 animals per day. The herd-level difference
between ADD in the period after and before vaccination was the outcome in linear
regression models for weaner pigs, and linear mixed-effects models for finishing pigs,
taking into account sow herds delivering pigs to two or more finisher herds. Three plausi-
ble risk factors (Baseline ADD, purchase of specific vaccine, purchase of other vaccines)
and five confounders (herd size, export and herd health status, year and season) were
initially considered in all 10 models. The main significant effect in all models was the
Baseline ADD; the higher the Baseline ADD was for weaner and finishing pigs, the larger
the decrease in ADD was following vaccination (or random date for non-vaccinating
herds). Regardless of vaccination status, almost equal proportions of herds experienced
a decrease and an increase in ADD resulting in no overall Change in ADD. Furthermore,

2
Kruse et al. Vaccination and Antimicrobials in Pigs
Frontiers in Veterinary Science | www.frontiersin.org January 2017 | Volume 3 | Article 120
INTRODUCTION
Due to its large production, the Danish pig sector accounts for
76% of the total amount of antimicrobial substances used for
livestock production per year in the country (1). Ocial focus
on reducing antimicrobial use has, therefore, been on the pig
production. e Danish Government and the swine industry
have put in place several initiatives to try to mitigate the poten-
tial risk related to the development of antimicrobial-resistant
bacteria. One of these initiatives, “the Yellow Card Scheme,”
which identies and warns livestock farmers using above a given
permitted limit of antimicrobials, was introduced in 2010 and
is managed by the Danish Veterinary Authorities. e antimi-
crobial use decreased aer the introduction of the Yellow Card
Scheme (2). From 2010 to 2014, there was a 14% reduction in
the antimicrobial treatment proportion, measured as dened
animal daily doses (ADDs) per 1,000 animals per day across
the total Danish pig production (1). e pig industry’s goal is
a further reduction by 10% before 2020 (3). To achieve this,
relevant and eective strategies that can minimize the need for
treatment with antimicrobials in the pig production are needed.
For animal welfare and productivity reasons, diseased
animals should be treated. However, an increased or improved
use of vaccination has been suggested as a potential strategy to
prevent specic diseases and/or secondary infections (4). Today,
the majority of Danish sows are being vaccinated against several
pathogens as a standard procedure. On the other hand, vaccina-
tion of ospring is not used to the same extent in Denmark as
in other EU countries with a similar pig production. ere may
be dierent explanations for this—one is the extended use of a
controlled program for specic pathogen-free (SPF) production
of piglets. However, the use of several vaccines has been on the
increase lately, especially since “the Yellow Card Scheme” was
adopted by the Danish Veterinary Authorities (2).
In Danish pig production, the majority of antimicrobials are
used for treatment of gastrointestinal and respiratory indications
in weaner pigs from 7–30kg, followed by treatment of gastroin-
testinal indications in nishing pigs (VetStat data, unpublished).
Mycoplasma hyopneumoniae (MYC), porcine circovirus type II
(PCV2), Actinobacillus pleuropneumoniae (APP), porcine repro-
ductive and respiratory syndrome (PRRS) virus, and Lawsonia
intracellularis (LAW) represent some of the most important
disease agents related to these indications, which are also prevent-
able through vaccination of breeding animals and/or ospring.
Mycoplasma hyopneumoniae is a bacterium causing enzootic
pneumonia in pigs. Enzootic pneumonia is most oen seen in
nishing pigs, where it is associated with productivity losses.
e bacteria are considered to be present in all Danish con-
ventional pig herds and in 67% of SPF herds (5). MYC in itself
does not necessarily cause disease problems in infected herds.
However, associated secondary infections may aggravate clinical
signs, and increase the need for treatment (6). Vaccination against
MYC would, therefore, be expected to reduce the need for anti-
microbial treatment. e eect of vaccination against MYC has
previously been shown to have a positive eect on growth and
reduced lung lesions (7–9). ere are several vaccines against
MYC available on the Danish market. MYC corresponded to 36%
of the vaccine dosages prescribed in 2013, being, therefore, the
most frequently used type of vaccine in pig production (10).
Porcine circovirus type II is a virus associated with several
dierent clinical signs in pigs. e virus is considered present in
nearly all Danish pig herds, without necessarily causing disease
problems. Previously, the main disease problem related to PCV2
in Danish pig production was postweaning multisystemic wast-
ing syndrome (PMWS) in weaner pigs. Nowadays, problems are
mainly related to reduced growth and increased mortality in
nishing pigs. PCV2 has an immuno-suppressive eect, which
potentiates the impact of other pathogens, resulting in a need
for antimicrobial treatment (11). erefore, vaccination against
PCV2 could potentially reduce disease occurrence and, conse-
quently, the use of antimicrobials. In fact, aer the adoption of
the Yellow Card Scheme, a 31% increase in the use of PCV2-
vaccines was seen over 1 year in Danish pig production (2).
Vaccination against PCV2 has been shown to result in increased
growth rate and reduced mortality in nishers (12), as well as
reduced antimicrobial use (13, 14). Vaccines against PCV2 are
the second most frequently used group of vaccines in Danish pig
production, representing 26% of the vaccine dosages prescribed
in 2013 (10).
Actinobacillus pleuropneumoniae is a bacterium causing
pleuropneumonia in pigs and is associated with reduced growth
and increased mortality, primarily in nishing pigs. ere are 15
dierent serotypes producing a combination of two or more of
the four toxins responsible for the pathology leading to disease in
pigs (15). e most prevalent serotypes in Denmark are serotypes
2, 5, 6, 5, and 7. Most SPF herds are free from APP. However,
serotype 6 is present in 26% of SPF herds, serotype 2 in 17%,
serotype 7 in 0.4%, and serotype 5 in 0.1% (5). Studies have shown
that vaccines against APP can reduce the prevalence of pleuritis
(16, 17). Vaccines against APP are the third most frequently used
type of vaccines in Danish pig production, representing 8% of the
vaccine dosages prescribed in 2013 (10). It could be expected that
preventing APP by using vaccination would reduce the treatment
of this bacterial infection.
only minor effects were found, when vaccinations were used in combination. In conclu-
sion, this study provided little support for the hypothesis that vaccination against five
common endemic diseases provides a plausible general strategy to reduce antimicrobial
use in Danish pig herds.
Keywords: antimicrobial, vaccination, pig production, Mycoplasma hyopneumoniae, porcine circovirus type II,
Actinobacillus pleuropneumoniae, porcine reproductive and respiratory syndrome virus, Lawsonia intracellularis

3
Kruse et al. Vaccination and Antimicrobials in Pigs
Frontiers in Veterinary Science | www.frontiersin.org January 2017 | Volume 3 | Article 120
Porcine reproductive and respiratory syndrome virus
multiplies in macrophages in the lungs, thereby making pigs
more susceptible to bacterial infections, such as infections
with Streptococcus suis (18). ere are two dierent strains; the
United States (US) strain and the European strain, which are
both present in the Danish national pig herd. Among SPF herds,
27% are infected with the US strain, whereas 20% are infected
with the European strain. e two strains are causing similar
clinical signs in pigs, mainly reproductive failure and respiratory
distress (5). Vaccines against PRRS virus represent 2% of the vac-
cines dosages prescribed (10). e use of vaccines against PRRS
in Denmark has been on a more or less constant lower level,
compared to the use of vaccines against MYC, PCV2, and APP
(VetStat data, unpublished). ere are currently two modied-
live vaccines and two inactivated vaccines against PRRS on the
Danish market (19).
Lawsonia intracellularis is one of the predominant agents
responsible for the development of porcine proliferative enter-
opathy, resulting in diarrhea in weaner and nishing pigs (20).
e herd-level prevalence of LAW in Danish herds is above
90% (21), but the infection does not necessarily lead to clinical
disease. Still, substantial amounts of antimicrobials are used
for the treatment of gastrointestinal indications in Danish pigs.
erefore, prevention of diarrhea caused by LAW is likely to lead
to reduced use of antimicrobials. ere is currently only one
vaccine available for prevention of LAW, and it accounted for
3% of the vaccine dosages prescribed for pigs in 2013 (10). is
implies that it is used to nearly the same extent as vaccines against
PRRS, and the purchase of this vaccine has increased since 2010
(VetStat data, unpublished). Previous studies have found reduced
mortality and increased growth followed by vaccination against
LAW ( 22), as well as reduced antimicrobial use (23).
Initiating vaccination in a herd represents an extra produc-
tion cost. Hence, it is important for the farmer to know the
cost-eciency of vaccines in dierent situations. Previous studies
testing the eect of vaccines against MYC, PCV2, APP, PRRS, or
LAW have mainly investigated the eect on production or health
parameters. Few studies have investigated the eect of vaccina-
tion on antimicrobial usage, and usually, the eect has only been
assessed in one farm at a time, not allowing for any generalization
of results. A recent study by Temtem etal. (10) evaluated the eect
of vaccines against MYC, PCV2, and LAW on antimicrobial use
in 1,513 Danish pig farms, using a cross-sectional study design to
compare the total amount of antimicrobials prescribed in 2013 in
herds with and without vaccination against one or more of MYC,
PCV2, and LAW. However, the date of initiation of vaccination
and other possible important risk factors and confounders were
not taken into account.
e objective of the present study was to estimate the eect
of initiating vaccination against MYC, PCV2, APP, PRRS, and
LAW on the change in antimicrobial prescription in weaner and
nishing pigs at herd-level, while taking into account plausible
associated risk factors and confounders. To evaluate whether
eventual detected eects were actually related to vaccination
rather than general trends in the target population, the change
in antimicrobial prescription in vaccinating herds was compared
to the change in antimicrobial prescription in randomly selected
comparable periods for herds not vaccinating against the specic
vaccine in question.
MATERIALS AND METHODS
Herd Enrollment
Full-line conventional pig herds were identied using yearly data
extractions from the Central Husbandry Register (CHR) and
quarterly extractions of movements between pig herds from the
Danish Pig Movement Database. e following types of herds
were identied and included (Figure1):
• Type 1: farrow-to-nisher herds, which contained age groups
Sows, Wean ers , and Finishers registered under one CHR
number (implying one geographical location). In order to be
considered a farrow-to-nisher production, the number of
Weaner and Finisher pen places had to be at least 1.5 times the
number of sow pen places, indicating that at least part of the
ospring remained in the herd until the nishing stage.
• Type 2: herds with age groups Sows, Wea ner s, and Finishers
registered under one CHR number (source herd) and with
movement of growing pigs to herds with age group Finishers
(receiving herd).
• Type 3: herds with age groups Sows and Weaner s registered
under one CHR number (source herd) and with movement
of growing pigs to herds with age group Finishers (receiving
herd).
• Type 4: herds with age groups Sows, Wea ner s, and Finishers
registered under one CHR number (source herd) and with
movement of growing pigs to herds with age groups Wea ner s
and Finishers (receiving herd).
• Type 5: herds with age groups Sows and Weaner s registered
under one CHR number (source herd) and with movement of
growing pigs to herds with age groups Wea ner s and Finishers
(receiving herd).
It was a requirement that each receiving herd only received
pigs from one source herd in each quarter of a year, whereas
source herds could deliver pigs to more than one nisher pig
herd. Moreover, only source herds with a minimum of 100
sow pen places and a minimum number of weaner pen places
equal to or higher than 1.5 times the number of sow pen places
were included, to make sure these herds were not sow-only or
weaner-only. For receiving herds to be included, a minimum of
100 nisher pen places was required.
Purchase of Vaccination
To study the eect of initiating vaccination against MYC, PCV2,
APP, PRRS, and LAW, 10 dierent longitudinal studies were car-
ried out on historical data; for each vaccine group, one study was
made on the eect on antimicrobial use in weaner and nishing
pigs. Purchase of vaccines for Danish pig herds is recorded in
the Danish VetStat Database (VetStat). Data used in this study
were raw historical data from VetStat, retrieved on June 1, 2015.
Purchase of all vaccine products against the ve disease agents
was extracted for the source herds. All herds with their rst
purchase of a vaccine under study between January 1, 2007, and

FIGURE 1 | Illustration of the types of source herds included in the study, characterized by herd composition and types of pigs moved to other
herds.
4
Kruse et al. Vaccination and Antimicrobials in Pigs
Frontiers in Veterinary Science | www.frontiersin.org January 2017 | Volume 3 | Article 120
December 31, 2013, were included. A sow herd was considered to
have initiated vaccination in this period and was included in the
analyses if it fullled the following criteria:
1. No prior purchase of the vaccine in question either in the
source herd or receiving herds from January 1, 2005.
2. Purchase of the vaccine for at least 1year following the rst
purchase.
3. Purchase of more than a minimum threshold of vaccine doses
per year per sow. e minimum of doses was determined
based on evaluation of a histogram showing the distribution of
vaccine coverage. e vaccine coverage was calculated as the
number of doses in the rst year of vaccination divided by the
number of sows registered in the individual herd multiplied
with 25, representing the average number of weaned pigs per
sow per year in Denmark. For MYC and PCV2, the threshold
was set at >0.5 and <1.5 corresponding to herds with a vaccine
coverage between 50 and 150%. is large a margin around
the 100% coverage was needed in order to include those herds
representing the area on the histogram with the majority of
the observations. For LAW, APP, and PRRS, the threshold was
lowered in order to avoid too many herds being excluded due
to this criterion. erefore, only an upper threshold was set at
<1.5.
Non-vaccinating herds were identied based on extraction of
all active source herds in CHR with no purchase of the vaccine
in question, at any time within the period between January 1,
2005, and April 30, 2015. A herd was considered active when
having recordings of antimicrobial prescriptions in VetStat for all
quarters, in a period of 12months before and 18months aer the
rst vaccine purchase.
For each analysis, information about purchase of the remain-
ing four types of vaccines within the period studied for each
herd was included. For this, data on the general purchase of the
vaccines in the period from January 1, 2005, until April 30, 2015,
were extracted and summarized by quarter of the year, to be able
to match correctly with the selected study period for each herd
included. For a herd to be considered using other vaccines, the
herd had to have purchased the given vaccine for at least 1year
in total and within at least one quarter of the study period for the
individual herd.
Antimicrobial Use
Herd-level antimicrobial prescription data extracted from the raw
VetStat data were used in this study as a proxy for antimicrobial
use. All prescriptions of antimicrobials, irrespective of indication
for weaner and nishing pigs, were included in the study. For
farrow-to-nisher herds, data on antimicrobial prescription
for weaner and nishing pigs were extracted for the individual
herd. For full-line herds identied using movement data, data on
antimicrobial prescription for weaner pigs were extracted from
the source herd, while for nishing pigs data were extracted from
the receiving herds (Figure1).
Antimicrobial data for weaner and nishing pigs were
extracted for a period of 2.5years. is consisted of data from
1year before vaccination was initiated, until 1.5years aer vac-
cination was initiated. e period of the rst 6months aer vacci-
nation was initiated was considered a transition period, in which
not all weaner and nishing pigs entering the stables had been
vaccinated yet. Data from this period were excluded for the data
analyses. Each prescription was converted into a number of ADD,
using standardized doses per antimicrobial product developed by
the Danish Veterinary and Food Administration. e ADD takes

5
Kruse et al. Vaccination and Antimicrobials in Pigs
Frontiers in Veterinary Science | www.frontiersin.org January 2017 | Volume 3 | Article 120
into account a standard average weight of a weaner pig (15kg)
and a nishing pig (50kg), as well as the total amount and dose
of the antimicrobial prescribed. All prescriptions in ADD were
divided into quarterly prescriptions for each herd, over the given
period of 2.5years selected for analysis. Each herd had to have
prescriptions in each of the quarters within this study period, in
order to be considered an active producing herd and, therefore,
be included in the study. e number of weaner and nishing
pigs in each herd at the time of vaccination was provided by data
from the CHR. ese numbers were used in the calculation of
the average ADD per 100 weaners per day, and of ADD per 100
nishers per day, in the 1-year period before (Baseline ADD) and
6–18months aer vaccination was initiated (ADD Aer) in each
herd. e change in amount of prescribed antimicrobials (Change
in ADD) following initiation of vaccination was calculated for
each herd by subtracting Baseline ADD from the ADD Aer.
For herds not vaccinating against a specic pathogen (e.g., APP,
MYC, etc.), a random date was chosen in the same period as for
the vaccinating herds (between January 1, 2007, and December
31, 2013), and the same set-up was used around this date; anti-
microbial prescription during 1year before the random date was
included, followed by 6months of data, which were excluded,
and then data on antimicrobial prescription covering 1year were
included. Also, Change in ADD was calculated for these herds,
which acted as a comparison group. It should be noted that, in the
analyses, it was still taken into account whether these herds that
were not exposed to the specic vaccine in question (e.g., MYC)
were vaccinated against any of the other vaccines (i.e., PCV2,
APP, PRRS, and/or LAW).
On rare occasions, prescription entry errors occur in the
VetStat database, resulting in either negative or extreme values,
when calculating ADD over a selected time period. When nega-
tive values of ADD were identied, the corresponding herd was
excluded from the study. Also, a few herds with extremely high
ADD values (>60 ADD/100 weaners/day and >20 ADD/100 n-
ishers/day) were excluded, as these most likely reected recording
errors or dramatic unregistered changes in the herd population.
Description of Variables
In all models, the outcome variable was the Change in ADD, and
this variable was included as a continuous variable aer checking
for linearity. ree variables were tested as potential explanatory
variables for Change in ADD:
• Vaccination: categorical variable with two levels; “Yes” and
“No,” representing vaccinating (exposed) and non-vaccinating
(non-exposed) herds, respectively.
• Baseline ADD: continuous variable, representing a baseline
measure of antimicrobial use. It was estimated as the anti-
microbial prescription 1year before vaccination (or random
date for the non-vaccinating group), in average ADD per 100
animals per day, for weaned and nishing pig.
• Other vaccines: four categorical variables, one for each of the
other four vaccines than the one under investigation in each
study, with two levels; “Yes” and “No,” representing source
herds that were or not using another vaccine when vaccination
with the study-specic vaccine started.
In addition, ve variables were included as potential
confounders:
• Sows: the number of sows in the individual source herd was
included as a continuous variable, representing the herd size.
• Year : the year of the rst purchase of vaccine (or random
date) was included as a categorical variable, representing the
dierent changes and levels of antimicrobial use, which has
been seen in Danish pig production between 2007 and 2013.
Aer initial analyses with individual years, this variable was
further grouped into “Before 2010” (<2010) and “Aer 2010”
(≥2010), representing the period before and aer the Yellow
Card Scheme was implemented in Denmark.
• Season: because it is known that antimicrobial use can uc-
tuate by season and a farmer may be more likely to initiate
vaccination during seasons with high antimicrobial use, the
season might confound the estimate of the eect of initiating
vaccination. erefore, the quarter of the year of the rst pur-
chase of vaccine (or random date for non-vaccinating herds)
was included, to account for season as a categorical variable
with four levels; “1,” “2,” “3,” and “4,” representing the rst to
the fourth quarter of the year.
• SPF: categorical variable with two levels; “SPF” and “Non-SPF,”
representing source herds enrolled or not in the SPF system
at the time of vaccination (or random date), respectively.
Information about enrollment of each herd in the Danish SPF
system was provided by SEGES Pig Research Centre (SEGES).
e SPF variable was used as a measure of herd health status
and considered as a potential confounder, as herds in the SPF
system might have a better health status than non-SPF herds,
which may inuence the antimicrobial use in those herds.
• Export: categorical variable with two levels; “Yes” and “No,”
representing source herds with and without export of growing
pigs, respectively. For each herd, information about exporting
of animals was assessed using the Danish Pig Movement
Database, which contains all movements of animals from
Danish herds to other countries. Besides having moved
animals from the source to the receiving herd, some of the
herds had also exported either 7 or 30 kg weaner pigs to
other countries. is information was included as a potential
confounder in the analyses, as importers might demand that
pigs are vaccinated against specic diseases, even if the Danish
herd was not infected with the given disease agent.
Continuous variables were plotted against the outcome
variable and against each other, to visually check for linearity and
correlations. For categorical variables, distributions were checked
for a reasonable number of observations in each variable level.
Statistical Analyses
All statistical analyses and data management were carried out
using the soware R version 3.1.3. Separate data analysis was
conducted for each of the 10 models, representing the eect
of the ve types of vaccines in weaner and nishing pig herds.
General linear regression models were used to model the Change
in ADD for weaner pigs as a function of the potential explanatory
variables and covariates.

6
Kruse et al. Vaccination and Antimicrobials in Pigs
Frontiers in Veterinary Science | www.frontiersin.org January 2017 | Volume 3 | Article 120
e same variables were used to model the Change in ADD
for nishers in a linear mixed-eects model, using the lme4
package in R (24). Source herd was included as a random eect
to account for clustering of antimicrobial use patterns in herds
with animals originating from the same source herd, and all other
explanatory factors and covariates were included as xed eects.
First, univariable models between each explanatory variable and
the outcome were assessed, and if associations showed P<0.20,
the variable was included in the multivariable model. However,
to prevent misinterpretation in the face of poor data availability,
if one of the stratied groups in a categorical variable contained
fewer than ve observations, the variable was not included in the
multivariable model. e nal multivariable model was identied
by backwards-stepwise elimination of non-signicant predictors,
using the drop1-function in R. Signicant two-way interactions
of all main eects were checked one by one. e criterion for
keeping a predictor or a two-way interaction in the model was
P<0.05, and models were compared using Akaike’s Information
Criteria (AIC), with the AIC closest to 0 indicating the best
model. Confounding was assessed by evaluating the models with
and without each of the potential confounders. A variable was
considered a confounder, if it changed the parameter estimates
of any of the other signicant variables by >20%. When an
interaction between vaccination with the study-specic vaccine
and one of the other vaccines was statistically signicant, the two
were recoded as a four-way variable, to be included in the nal
multivariable model, allowing for the estimation of the eect and
signicance of each category dened by the pairwise combina-
tion of vaccination statuses (Yes/Yes, No/No, Yes/No, No/Yes).
e statistical signicance of dierences observed between the
four categories was assessed with Tukey’s “Honest Signicant
Dierence” method, using the multcomp package in R. e nal
models were presented with parameter estimates including SE
and P-values, as well as the explanatory degree of the model.
e distribution of residuals was checked for normality using
residual plots. e explanatory degree of the general linear regres-
sion models was assessed using the R2. ‘e explanatory degree’
of the mixed-eects models was assessed using the marginal
pseudo-R2: (Re−Rfm)/Re, where Re is the residual variance of the
model, only containing the random eect of source herd, and Rfm
is the residual variance of the nal model (25).
RESULTS
Descriptive Statistics
ere were small, characteristic dierences between vaccinat-
ing and non-vaccinating herds, when looking at the descriptive
statistics; for the majority of the studies, the Change in ADD was
lowest and the Baseline highest in the vaccinating group. Also,
the vaccinating groups consisted of larger herds, represented by
the number of sows in the source herd, when compared to non-
vaccinating herds (Tables S1–S10 in Supplementary Material).
ere were no substantial dierences between the mean and
median values for Change in ADD, Baseline ADD, and Sows.
erefore, it was chosen to present the mean. Overall, the mean
Change in ADD was close to 0, but with a large range, especially
for weaner pigs. As expected, the largest Baseline ADD was found
among weaner pigs, with a nearly ve times higher Baseline
ADD than observed for nishers. e mean number of sows
in the farrow-to-nisher and source herds only diered 1–7%
between the studies on eect in weaner and nishing (Table S11
in Supplementary Material). is happened because nearly the
same source herds were used in the analyses for weaners and
nishers for each type of vaccine.
For each of the 10 studies, there were between 71 and 334
vaccinating source herds delivering pigs to between 89 and 365
receiving herds. For the group of non-vaccinating herds, there
were between 130 and 570 source herds delivering pigs to between
158 and 662 receiving herds, in each of the studies.
Analytical Statistics
A summary of the main ndings from each of the nal regression
models is presented in Tab le 1. Detailed results from the models
can be found in Tables S12–S16 in Supplementary Material.
No study-specic vaccinations were found to have a signi-
cant impact on the antimicrobial consumption, when analyzed
independently. e baseline antimicrobial consumption before
initiation of vaccination (Baseline ADD) was the only consist-
ently signicant independent variable in all models, indicating
that herds with a higher Baseline ADD obtained a larger reduc-
tion in ADD, when compared to herds with a lower Baseline
ADD. e interaction between Baseline ADD and vaccination
status was non-signicant in all models, meaning that the eect of
Baseline ADD was the same for herds initiating vaccination as for
non-vaccinating herds. For those reasons, plots presented in this
manuscript are based on the eect of Baseline ADD on Change
in ADD, added with specic variables of interest, depending on
the case (Figures2 and 3).
For weaners, all ve nal models showed an eect of year,
either as a direct eect or as interacting eect with Baseline ADD,
while for nishers this was only the case for two of the ve models
(Tabl e 1 ). e eect of year in all seven models showed that a
larger decrease in ADD was seen aer 2010, when compared to
the period before 2010. e eect was the same, irrespective of
vaccination status, since no interaction between year and vac-
cination was found in any of the models. e eect of Baseline
ADD on Change in ADD according to year (when signicant) is
shown for all models in Figure2.
e eect of initiating study-specic vaccinations was only
signicant in an interaction with the use of another vaccine in
the same period for LAW and PRRS vaccination in nishers
(Tabl e 1 ). An increase in ADD was observed for herds using
both LAW and PRRS, when compared to non-vaccinating herds
(Figure 3; Table S16 in Supplementary Material). In addition,
the use of APP vaccines had a positive eect on the Change in
ADD for weaners in the PRRS-model, implying that a larger
increase in ADD is seen in herds with use of APP vaccines. For
the latter, there were no signicant interaction between initiating
vaccination against PRRS and using APP vaccines, which is why
this eect was general for both vaccinating and non-vaccinating
herds (Table1; Table S15 in Supplementary Material).
e number of sows, representing herd size, was signicantly
associated with an increase in ADD in the PCV2 model for nish-
ers, as well as in the LAW and APP models for weaners. For the

TABLE 1 | Summarizeda results of the final linear regression models (weaners) and linear mixed models with random effects of potential confounders
(finishers) predicting the change in antimicrobial consumption after initiation of vaccination in selected Danish swine herds between 2007 and 2013.
Vaccine Age group Statistically significant effects R2
Vaccine Baseline Other individual variables Interactions
Mycoplasma hyopneumoniae Weaners No Yes Actinobacillus pleuropneumoniae (APP)
vaccine
Baseline×year 0.25
MYC Finishers No Yes Specific pathogen-free No 0.27
Porcine circovirus type II (PCV2) Weaners No Yes Year No 0.27
Export
PCV2 Finishers No Yes Year No 0.24
Number of sows
APP Weaners No Yes Number of sows Number of
sows×year
0.26
Baseline×year
APP Finishers No Yes No No 0.29
Porcine reproductive and respiratory syndrome
(PRRS)
Weaners No Yes Year No 0.21
APP vaccine
PRRS Finishers No Yes No Baseline×year 0.24
Lawsonia intracellularis Weaners No Yes Number of sows Baseline×year 0.21
LAW Finishers YesbYes No Vaccine×PRRS
vac.
0.30
aThe complete list of variables, coefficients, SE, and p-values for all models are shown in Tables S12–S16 in Supplementary Material.
bNon-significant as an isolated variable but significant as an interaction with another vaccine.
7
Kruse et al. Vaccination and Antimicrobials in Pigs
Frontiers in Veterinary Science | www.frontiersin.org January 2017 | Volume 3 | Article 120
latter model, the number of sows was only associated with an
increase in ADD in the period before 2010, which was shown
by the signicant interaction between year and Sows (Table1).
As for variables Export and SPF, the model estimates revealed
that exporting herds had a larger increase in ADD for weaners
than non-exporting herds in the PCV2-model (Tab le1 ; Table S13
in Supplementary Material). In the MYC model, SPF herds had a
larger increase in ADD for nishers, when compared to Non-SPF
herds (Table S12 in Supplementary Material). Again, these eects
were the same, irrespective of vaccination status.
ere was no confounding eect of any of the variables, and no
eect of Season in any of the models. Each one of the nal models
explained 21–30% of the variation in the outcome variable.
DISCUSSION
Effect of Baseline and Initiation
ofVaccination
e objective of this study was to determine the eect of initiating
vaccination against MYC, PCV2, APP, PRRS, and LAW on the
Change in ADD for weaner and nishing pigs at herd level. We
found that the Baseline ADD level had a persistent impact, being
signicant in all models and with more or less the same degree of
impact on the Change in ADD for weaners as for nishing pigs.
e eect of Baseline ADD in the models showed that, the higher
the Baseline ADD in a herd, the larger the decrease in ADD seen
over time. is was a general eect, regardless of vaccination, as
the eect did not dier between vaccinating and non-vaccinating
herds.
Change in Antimicrobial Use in Danish Pig
Herds
Overall, we found an average change in antimicrobial use around
0, for both weaner and nishing pigs, regardless of vaccination
status. Some herds experienced a decrease in ADD and, in gen-
eral, these herds had a high Baseline ADD. Other herds remained
at a more or less constant level of ADD, or even increased in ADD
over time. e latter were herds with an average or low Baseline
ADD. erefore, it seems reasonable that the Baseline ADD is an
important variable to include, when looking at Change in ADD
in individual pig herds. Still, the level of Baseline ADD and the
level of Change in ADD in a herd are inuenced by many other
factors. As we see from the nal models in this study, these two
variables, although important, only explained between 1/4 and
1/3 of the total variation of the observed Change in ADD in the
herds included. is illustrates that Change in ADD in pig herds
is a multifactorial and very complex measure.
One important factor to determine the Change in ADD must
be disease occurrence. Producers experiencing disease problems
will, in collaboration with the veterinarian, put in place measures
to reduce disease and its consequences. is could result in a
decrease in antimicrobial use over time. For vaccinating herds,
the decrease observed in the present study was most likely due
to the eect of vaccination. For non-vaccinating herds, other
measures may have been used, such as type of feed, internal
biosecurity (including ways of immunizing sows), and way of
purchasing breeding animals.
Herds with no disease problems are at a constant risk of
getting disease outbreaks. is is either due to the presence or

FIGURE 2 | Continued
8
Kruse et al. Vaccination and Antimicrobials in Pigs
Frontiers in Veterinary Science | www.frontiersin.org January 2017 | Volume 3 | Article 120
introduction of various infectious disease agents, which can
result in a sudden increase in antimicrobial use. e estimated
change in antimicrobial use in pig herds over time, regardless
of vaccination status, showed signs of regression toward the
mean. is makes sense as, in population-based distributions,
the conditional expectation of values located in the tails are

FIGURE 2 | Model-predicted associations between Baseline ADD (horizontal axis) and Change in ADD (vertical axis), before (continuous line, based
on black dots) and after (dashed line, based on grey dots) 2010, in groups of Danish swine herds that initiated vaccination or not against the five
different endemic agents under study in 2007–2013. Each graph illustrates one model derived from vaccine- and production type-specific dataset. In models
with only one (continuous) line, there was no significant effect of year, i.e. before vs. after 2010, in which the Yellow Card Scheme was initiated.
9
Kruse et al. Vaccination and Antimicrobials in Pigs
Frontiers in Veterinary Science | www.frontiersin.org January 2017 | Volume 3 | Article 120
typically closer to the overall mean, than to its observed value.
erefore, herds with high consumption are more likely to
present a decrease, and herds with low consumption are more
likely to increase.
An increase in ADD for vaccinating herds should not be
interpreted as a missing eect of the vaccines. e increase may
be a result of the increased occurrence of diseases other than
the one being vaccinated against. In this study, we used the total
ADD, since the validity of the dierent disease indications var-
ies, and some prescriptions do not have an indication assigned
to it (VetStat data, unpublished). ey would, therefore, not be
included, if the ADDs were split into disease indications.
Effect of Restrictions from Authorities
A reason for reducing antimicrobial use, other than mitigat-
ing disease in a pig herd, could be demands from authorities.
Ingeneral, there has been much focus on reducing antimicrobial
use in the general pig population in Denmark, as also seen in
many other EU countries. Demands from Danish authorities
increased aer 2010, when the Yellow Card Scheme was imple-
mented, and an eect on the antimicrobial level was seen already
from mid-2010, when permitted limits were announced (2). is
eect can also be conrmed by the data included in this study,
showing that the Change in ADD was aected by year (before and
aer 2010), especially in interaction with ADD Baseline. In the
same period when the antimicrobial use decreased in Danish pig
herds, an increased purchase of vaccines was observed, especially
against MYC, PCV2, and APP. is probably reects that some
swine producers increased their use of vaccines, as an alternative
to antimicrobial treatment.
ere was no signicant eect of interactions between initia-
tion of vaccination and year in any of the models, indicating

FIGURE 3 | Model-predicted associations between Baseline ADD per 100 finishers per day (horizontal axis) and Change in ADD (vertical axis) in
groups of Danish swine herds that initiated Lawsonia intracellularis vaccination or not, while also vaccinating or not against porcine reproductive
and respiratory syndrome in 2007–2013.
10
Kruse et al. Vaccination and Antimicrobials in Pigs
Frontiers in Veterinary Science | www.frontiersin.org January 2017 | Volume 3 | Article 120
that the decrease in ADD seen aer 2010 was not aected
by initiation of vaccination. Still, some herds probably suc-
ceeded in reducing their antimicrobial consumption followed
by initiation of vaccination (2). Other producers possibly
adjusted their herd management, quality of feed, or level of
biosecurity, in order to comply with the new restrictions. Some
farmers probably also reduced their antimicrobial use of solely
psychological reasons, by reducing their probability of getting
a Yellow Card, if they were close to the Yellow Card limits.
is kind of impact is dicult to determine and is out of the
scope of this study.
Effect of Combination of Vaccines
It is generally believed that preventing several diseases through
vaccination can have a larger eect, than the eect of preventing
the sum of each of them. In this study, there was a signicant eect
of the interaction between initiating vaccination in combination
with existing vaccine programs. is eect on ADD was observed
for initiation of vaccination against LAW with concurrent use of
PRRS vaccines in nishers (Figure3). It was not expected that
using two vaccines in combination would result in an increased
Change in ADD, implying an increased use of ADD. is prob-
ably reects that there was clinical disease due to these agents
prior to the initiation of vaccination compared to the herds not
vaccinating, resulting in an apparent missing eect of the vac-
cines due to reverse causality.
Explanations for Lack of Effect of
Vaccines
e lack of signicant eect of vaccines on the antimicrobial use
should not be interpreted as an indication of the vaccines not
being eective. is study paid attention to the eect of initiation
of vaccination, but it did not include long-term eects. Moreover,
herds included in this study could have initiated vaccination for
various reasons, which were unknown at the time of the study.
Vaccines should prevent disease in individual animals but can
also be used as a control measure at herd level.
Register data—as used here—include both herds with
severe problems related to disease and herds in which vaccina-
tion is used as a preventive measure or required by the buyer.
Producers who export 7 or 30kg pigs to other countries sell
their pigs for a higher price, if the pigs are vaccinated according
to the requirements of the purchasing farmer. For farmers in
Germany, this scenario is applicable, since it is a requirement
that the pigs are vaccinated against PCV2. Exporting of pigs was
included as a variable in each model but was only signicant
in the model testing the eect of initiating vaccination against
PCV2 on the antimicrobial use in weaned pigs. It was shown that
herds exporting weaned pigs had a larger increase in consump-
tion of antimicrobials than herds which do not. No signicant
eect of the interaction between Export and Vaccination was
observed, so the eect was general for both vaccinating and
non-vaccinating herds.
Vaccinating herds were compared to a group of herds, which
did not use the vaccines in question during the study period. e
reason for not using that vaccine could be that there was no need
for it, meaning that non-vaccinating herds could be herds with
better animal health, and therefore, it would be more dicult to
obtain an eect of vaccination. In line with this, the antimicrobial
use was likely a measure of disease, to some extent. However, high
use in a Danish context (above 28 ADD per 100 weaners per day,
and six ADD per 100 nishers per day) is not necessarily equal
to substantial disease problems and poor pig health. erefore,
a positive eect of vaccination may be more dicult to see in
the country, with a more visible increase in antimicrobial use,
since the Baseline is already at a low level, when compared to

11
Kruse et al. Vaccination and Antimicrobials in Pigs
Frontiers in Veterinary Science | www.frontiersin.org January 2017 | Volume 3 | Article 120
many other EU countries (26). In the study by Raith etal. (14),
a signicant decline in the use of antimicrobials was found aer
initiation of the PCV2 vaccine program in Austrian pig herds.
However, herds in that study reduced their antimicrobial use
to a level that would be considered high, in a Danish context
(0.56 ADDkg/kg/year, corresponding to 7.7 ADD/100 animals
per day, for nishing pigs). Again, this illustrates the impact of
the Baseline ADD, which should always be taken into account,
when looking at Change in ADD and the eect of vaccination on
antimicrobial use.
Temtem et al. (10) found the same lack of eect, or even
apparent reverse causality, between vaccination and antimi-
crobial use. Even though we have taken more information into
account, resulting in models with a higher explanatory degree,
there is still some variance that cannot be explained with the
available variables and models. is variance could reect some
psychologically based reasons, which are not directly measurable.
Using vaccines in a herd may represent a dierent mindset of
the producer or the responsible veterinarian, compared to herds
not using vaccines. Producers using vaccines may have a general
higher perception of the necessity to control and prevent diseases.
For a producer, the use of vaccines and antimicrobials, probably
in combination, might be the best way of having a successful
production with fewer and less sick animals. is could also be
inuenced by some veterinarians being more likely to suggest the
use of vaccines than others.
Data Availability
Denmark is a country with a large pig production. Only a propor-
tion of Danish pig herds were included in this study, due to dier-
ent reasons and criteria. Some herds initiated vaccination before
the beginning of the study period in 2007 and were, therefore,
not included in the study. Only herd types allowing us to follow
the pigs from vaccination in the sow herd, through weaning and
nishing, were selected for this study. Other types of herds, for
example herds with only one age group registered, also vaccinated
against the ve agents of interest within the study period. But for
these types of herds, it would have been impractical to trace back
and forth the vaccine dosages and antimicrobial prescriptions
to include. e sample of herds included here covered around
50–70% of all Danish vaccinating herds, depending on the study
(VetStat data, unpublished).
Some herds were excluded due to extreme or negative values.
ese could have been further investigated, but that would
require a large amount of time-consuming manual work, to
identify and correct the reason for these outliers. Only a few
percent of the total number of herds were excluded in each study,
due to extreme or negative values. Hence, this should not have
biased the results.
In the present study, it was necessary to loosen up on the
criterion regarding number of vaccine doses purchased. Not
many herds would have been included in the study testing eect
of APP, PRRS, and LAW, if the same criteria were used as for the
study testing eect of MYC and PCV2. is may reect that these
vaccines are applied dierently in Danish pig herds. is could,
for example, reect the use of vaccines for other age groups than
what they are licensed for. Another explanation could be the use
of half dosages, or not using the vaccine continuously throughout
the year. We knew that herds included in the study had purchased
vaccines for at least 1year, but it was dicult to get more informa-
tion than that, besides the number of doses purchased and the
number of animals expected to be vaccinated. Again, this could
also explain the missing eect of initiating vaccination against
these agents.
CONCLUSION
is study provided little support for the hypothesis that vaccina-
tion against ve common endemic diseases provides a plausible
strategy to reduce antimicrobial use in Danish pig herds, overall
speaking. Still, vaccination can be an asset in some situations.
AUTHOR CONTRIBUTIONS
AK contributed to the design and interpretation of the work,
performed the data analyses, and draed the manuscript. LK
contributed by re-evaluating and re-running the analyses for
the models containing interactions between vaccines, produc-
ing the plots and gures, and collaborated in the interpretation
and new text for the revised version. LN and LA contributed
to the design and interpretation of the work and revised the
manuscript. All the authors did the nal approval of the ver-
sion to be published and hereby agreed to be accountable for all
aspects of the work.
ACKNOWLEDGMENTS
e authors would like to thank Nana Dupont (University of
Copenhagen), Mette Fertner, and Anna Camilla Birkegård
(Technical University of Denmark, National Veterinary Institute)
for contributing with R-codes for analyzing and managing
data from VetStat and CHR. Also, we would like to thank e
Danish Veterinary and Food Administration for allowing us
access to VetStat and CHR data, especially Erik Jacobsen for help
with questions regarding VetStat data, and Elisabeth Okholm
Nielsen for academic discussions throughout the process. SEGES
provided SPF data, and especially Charlotte Sonne Kristensen
is acknowledged for the help with that. Furthermore, we thank
Margit Andreasen (VIF) for input to the study design and sug-
gestions for relevant literature.
FUNDING
is study was funded by UC-care (http://uc-care.ku.dk/english)
and the Faculty of Health and Medical Sciences, University of
Copenhagen, Denmark.
SUPPLEMENTARY MATERIAL
e Supplementary Material for this article can be found online at
http://journal.frontiersin.org/article/10.3389/fvets.2016.00120/
full#supplementary-material.

12
Kruse et al. Vaccination and Antimicrobials in Pigs
Frontiers in Veterinary Science | www.frontiersin.org January 2017 | Volume 3 | Article 120
REFERENCES
1. Danish Integrated Antimicrobial Resistance Monitoring and Research
Programme [DANMAP]. DANMAP 2014: Use of Antimicrobial Agents and
Occurrence of Antimicrobial Resistance in Bacteria from Food Animals, Food
and Humans in Denmark. (2015). Available from: http://www.danmap.
org/~/media/Projekt%20sites/Danmap/DANMAP%20reports/DANMAP%
202014/Danmap_2014.ashx
2. Alban L, Dahl J, Andreasen M, Petersen JV, Sandberg M. Possible impact
of the “yellow card” antimicrobial scheme on meat inspection lesions
in Danish nisher pigs. Prev Vet Med (2013) 108:334–41. doi:10.1016/
j.prevetmed.2012.11.010
3. Anonymous. Reduction in Use of Antibiotic for Pigs – And of Tetracycline
Especially. (In Danish: Reduktion i forbruget af antibiotika til grise – og af
tetracyklin i saer). SEGES Danish Pig Research Centre (2015). Available from:
http://vsp.lf.dk/Aktuelt/Nyheder/2015/02/110215_Reduktion%20af%20
forbruget%20i%20antibiotika.aspx
4. Postma M, Stärk KDC, Sjölund M, Backhans A, Beilage EG, Lösken S,
et al. Alternatives to the use of antimicrobial agents in pig production:
a multi-country expert-ranking of perceived eectiveness, feasibility and return
on investment. Prev Vet Med (2015) 118:457–66. doi:10.1016/j.prevetmed.
2015.01.010
5. Kristensen CS, Vinther J, Kristiansen MG, Pedersen KS. e Importance of SPF
Diseases for Productivity, Antibiotic Use and Health. (In Danish: Betydningen
af SPF-sygdomme for produktivitet, antibiotikaforbrug og sundhed). SEGES
Danish Pig Research Centre (2015). Meddelelse Nr. 1039. Available from:
http://vsp.lf.dk/~/media/Files/PDF%20-%20Publikationer/Meddelelser%
202015/Meddelelse_1039.pdf
6. Maes D, Segales J, Meyns T, Sibila M, Pieters M, Haesebrouck F. Control
of Mycoplasma hyopneumoniae infections in pigs. Vet Microbiol (2008)
126:297–309. doi:10.1016/j.vetmic.2007.09.008
7. Jensen CS, Ersbøll AK, Nielsen JP. Meta-analysis comparing the eect of
vaccines against Mycoplasma hyopneumoniae on the daily weight gain
in pigs. Prev Vet Med (2002) 54:265–78. doi:10.1016/S0167-5877(02)
00005-3
8. Moreau IA, Miller GY, Bahnson PB. Eects of Mycoplasma hyopneumoniae
vaccine on pigs naturally infected with M. hyopneumoniae and porcine
reproductive and respiratory syndrome virus. Vaccine (2004) 22:2228–333.
doi:10.1016/j.vaccine.2003.10.041
9. Kristensen CS, Vinther J, Svensmark B, Baekbo P. A eld evaluation of two
vaccines against Mycoplasma hyopneumoniae infections in pigs. Acta Vet
Scand (2014) 56:24. doi:10.1186/1751-0147-56-24
10. Temtem C, Kruse AB, Nielsen LR, Pedersen KS, Alban L. Comparison of
the antimicrobial consumption in weaning pigs in Danish sow herds with
dierent vaccine purchase patterns during 2013. Porcine Health Management
(2016) 2:23. doi:10.1186/s40813-016-0042-1
11. Kristensen CS, Larsen LE, Hjulsager CK. Investigation of PCV2 Status in Two
Danish Herds – Two-Year Follow-Up. (In Danish: Undersøgelse af PCV2-status
i to danske besaetninger – to års opfølgning). SEGES Danish Pig Research
Centre (2012). Meddelelse nr. 933. Available from: http://vsp.lf.dk/~/media/
Files/PDF%20-%20Publikationer/Meddelelser%202012/Meddelelse%20
933_Unders%C3%B8gelse%20af%20PCV2-status%20i%20to%20danske%20
bes%C3%A6tninger%20-%20to%20%C3%A5rs%20opg.pdf
12. Kristensen CS, Baadsgaard NP, To N. A meta-analysis comparing the
eect of PCV2 vaccines on average daily weight gain and mortality rate in
pigs from weaning to slaughter. Prev Vet Med (2011) 98:250–8. doi:10.1016/
j.prevetmed.2010.11.015
13. Aerts R, Wertenbroek N. Implementing PCV2 vaccination resulting in
reduction of antibiotic use on Dutch farrow-to-nish farm. Proc. SafePork.
(2011). p. 339–40. Available from: http://lib.dr.iastate.edu/cgi/viewcontent.
cgi?article=1483&context=safepork
14. Raith J, Trauer M, Firth CL, Lebl K, Schleicher C, Köfer J. Inuence of por-
cine circovirus type 2 vaccination on the level of antimicrobial consumption
on 65 Austrian pig farms. Vet R ec (2016) 178:20. doi:10.1136/vr.103406
15. Kristensen CS. Pneumoniae – Pleuropneumoniae/Actinobacillus pleuropneumo-
niae/Ap. (In Danish: Lungesyge – Ondartet/Actinobacillus pleuropneumoniae/
Ap). SEGES Danish Pig Research Centre (2013). Available from: http://vsp.
lf.dk/Viden/Sygdom%20og%20behandling/Lusvejsystemet/Lungesyge%20
ondartet%20Actinobacillus%20pleuropneumoniae%20Ap.aspx?full=1
16. Sjölund M, Wallgren P. Field experiment with two dierent vaccination strat-
egies aiming to control infections with Actinobacillus pleuropneumoniae in a
fattening pig herd. Acta Vet Scand (2010) 52:23. doi:10.1186/1751-0147-52-23
17. Sacristán RDP, Michiels A, Martens M, Haesebrouck F, Maes D. Ecacy
of vaccination against Actinobacillus pleuropneumoniae in two Belgian
farrow-to-nish pig herds with a history of chronic pleurisy. Vet R ec (2014)
174:302. doi:10.1136/vr.101961
18. Halbur P, anawongnuwech R, Brown G, Kinyon J, Roth J, acker E, etal.
Ecacy of antimicrobial treatments and vaccination regimens for control of
porcine reproductive and respiratory syndrome virus and Streptococcus suis
coinfection of nursery pigs. J Clin Microbiol (2000) 38:1156–60.
19. Hoelstad BE. Introduction of Replacement Gilts to PRRS-Positive Sow Herds.
Master’s thesis, Copenhagen, Denmark: Technical University of Denmark
(2016). Available from: http://vsp.lf.dk/~/media/Files/PDF%20-%20
Publikationer/Notater%202016/speciale_til%20notat%201609.pdf
20. Smith SH, McOrist S. Development of persistent intestinal infection and
excretion of Lawsonia intracellularis by piglets. Res Vet Sci (1997) 62:6–10.
doi:10.1016/S0034-5288(97)90171-5
21. Stege H, Jensen TK, Møller K, Baekbo P, Jorsal SE. Prevalence of intestinal
pathogens in Danish nishing pig herds. Prev Vet Med (2000) 46:279–92.
doi:10.1016/S0167-5877(00)00148-3
22. Kolb J, Sick F. Summary of eld trials implementing Enterisol® Ileitis against
ileitis. Proc Am Assoc Swine Vet (2003):243–4.
23. Bak H, Rathkjen PH. Reduced use of antimicrobials aer vaccination of pigs
against porcine proliferative enteropathy in a Danish SPF herd. Acta Vet Scand
(2009) 51:1. doi:10.1186/1751-0147-51-1
24. Bates D, Maechler M, Bolke B, Walker S. Fitting linear mixed-eects models
using lme4. J Stat Sow (2015) 67:1–48. doi:10.18637/jss.v067.i01
25. Dohoo I, Martin W, Stryhn H. Veterinary Epidemiologic Research. 2nd ed.
Charlottetown, PEI, Canada: VER Inc. (2010).
26. European Medicines Agency [EMA]. Sales of Veterinary Antimicrobial
Agents in 26 EU/EEA Countries in 2013. European Medicines Agency (2015).
Available from: http://www.ema.europa.eu/docs/en_GB/document_library/
Report/2015/10/WC500195687.pdf
Conict of Interest Statement: e authors declare that the research was con-
ducted in the absence of any commercial or nancial relationships that could be
construed as a potential conict of interest.
Copyright © 2017 Kruse, de Knegt, Nielsen and Alban. is is an open-access article
distributed under the terms of the Creative Commons Attribution License (CC BY).
e use, distribution or reproduction in other forums is permitted, provided the
original author(s) or licensor are credited and that the original publication in this
journal is cited, in accordance with accepted academic practice. No use, distribution
or reproduction is permitted which does not comply with these terms.