Preventive Veterinary Medicine 73 (2006) 241–254

www.elsevier.com/locate/prevetmed

Prevalence of Salmonella spp. on Canadian pig

farms using liquid or dry-feeding
Abdolvahab Farzan a,*, Robert M. Friendship a,
Catherine E. Dewey a, Keith Warriner b,
Cornelius Poppe c, Kim Klotins d
a
Department of Population Medicine, Ontario Veterinary College, University of Guelph,
Guelph, Ont., Canada N1G 2W1
b
Department of Food Science, University of Guelph, Guelph, Ont., Canada N1G 2W1
c
OIÉ Reference Laboratory for Salmonellosis, Public Health Agency of Canada,
Guelph, Ont., Canada N1G 3W4
d
Ontario Ministry of Agriculture and Food, Guelph, Ont., Canada N1G 4Y2
Received 18 November 2004; received in revised form 12 August 2005; accepted 2 September 2005

Abstract

The objective of this study was to determine whether the shedding and antibody titre to
Salmonella was lower for pig herds provided liquid-feed compared to those on traditional dry
rations. Twenty liquid-feeding farms and 61 dry-feeding farms were selected. The amount of
antibodies to Salmonella in sera from 15 finisher pigs on each of 80 Ontario swine farms was
analyzed by means of enzyme-linked immunosorbent assay (ELISA). In addition, the presence of
Salmonella on the 20 liquid-feeding farms and 21 of the dry-feeding farms was assessed by culture of
15 fecal samples taken directly from finisher pigs and five pooled pen-fecal samples at each farm.
A cut-off of OD% 10 was used. The Salmonella sero-prevalence differed between the two groups of
farms. At least one pig tested sero-positive on 98% of the dry-feeding farms and 84% of the liquid-feeding
farms (P < 0.05). A multi-variable mixed linear regression model with the farm as a random variable and
farm factors as the fixed effects was fitted. Crude optical density (OD) of the individual pig was considered
as the continuous dependent variable. Dry-feeding and antimicrobial daily usage was associated with
crude OD (P < 0.05). In addition, crude OD increased with increasing herd size (P < 0.05).
Salmonella was isolated from 25 out of 420 fecal samples (6%) from dry-feeding farms compared
to three out of 400 samples (0.8%) from liquid-feeding farms. Eight of the dry-feeding farms (38%)

* Corresponding author. Tel.: +1 519 824 4120; fax: +1 519 763 3117.
E-mail address: afarzan@uoguelph.ca (A. Farzan).

0167-5877/$ – see front matter # 2005 Elsevier B.V. All rights reserved.
doi:10.1016/j.prevetmed.2005.09.003
242 A. Farzan et al. / Preventive Veterinary Medicine 73 (2006) 241–254

tested positive compared to only three of the liquid-feeding farms (15%). Salmonella was also
recovered from the pen environment on five dry-feeding farms but were not isolated from the
facilities using liquid-feeding. Salmonella Typhimurium was isolated from four farms in the dry-feed
group and on one farm with liquid-feeding. The one S. Typhimurium isolate from the liquid-feeding
farm exhibited no antimicrobial resistance, but those from dry-feeding farms were resistant to four or
more antimicrobial agents. The results of the logistic regression, with farm as a random effect showed
that dry-feeding [OR = 2.7 (1.1–15.1)] and continuous flow system [OR = 2.3 (1.2–12.7)] increased
risk of finding Salmonella in the individual pig.
These findings indicate that liquid-feeding and all-in all-out management of the grower–finisher
barns can reduce the Salmonella prevalence.
# 2005 Elsevier B.V. All rights reserved.

Keywords: Salmonella; Pig; Shedding; ELISA; Optical density; Liquid-feeding; Risk factors

1. Introduction

Salmonella enterica serovars (Salmonella) are important foodborne pathogens of public

health concern (Mead et al., 1999; Voetsch et al., 2004) and pork products play an
important role in the transmission of Salmonella to humans (Berends et al., 1998; Nielsen
and Wegener, 1997).
Control of Salmonella in pork can be conducted at all levels of production including pre-
harvest (farm level). In order to limit and control Salmonella occurrence in a national swine
herd, it is initially necessary to conduct epidemiological studies, first to determine the
prevalence of Salmonella and identify possible risk factors, and consequently to implement
and monitor control programs (Mousing et al., 1997).
To reduce the risk of pork contamination some countries have instituted monitoring
programs to identify pig farms with a high proportion of market hogs that are carrying
Salmonella spp. and then taking steps to reduce the on-farm prevalence (Christensen et al.,
2002; Mousing et al., 1997; van der Wolf et al., 2001a).
The major challenge in monitoring programs has been how to diagnose or monitor large
proportion of animals on a timely and repeated basis and do so with relatively moderate
costs. Enzyme-linked immunosorbent assay (ELISA) tests have been developed and used
to detect Salmonella antibodies in sera and meat juices (Alban et al., 2002; Nielsen et al.,
1995; Proux et al., 2000; Rostagno et al., 2004; van der Heijden, 2001; van der Wolf et al.,
2001b). The use of these tests has allowed researchers to investigate on farm risk factors for
infection and exposure rather than shedding (Beloeil et al., 2004; Blaha, 2004; Kranker
et al., 2001; Leontides et al., 2003; Lo Fo Wong et al., 2004; van Winsen et al., 2001; van
der Wolf et al., 2001a,c).
The interaction between Salmonella, host, and environment is complicated. The
environmental, feeding and management factors, upon which the transmission, shedding,
carrier state, and sero-prevalence are interpreted, are still being studied (Funk and
Gebreyes, 2004). On-farm intervention to reduce the prevalence of Salmonella is difficult.
‘Good hygiene’ and ‘all-in all-out management’ are not sufficient to reduce Salmonella
shedding in many circumstances. In addition to these steps, feed is an important component
in a pre-harvest Salmonella control program, in particular, the type of feed appears to be
 A. Farzan et al. / Preventive Veterinary Medicine 73 (2006) 241–254 243

strongly associated with the presence of Salmonella. Many studies have shown that swine
farms that use dry-feeding have higher Salmonella sero-prevalence (Lo Fo Wong et al.,
2004; van der Wolf et al., 2001c; van Winsen et al., 2002) than farms with a liquid-feeding
system, possibly using a fermented diet (van Winsen et al., 2002; van der Wolf et al.,
2001a).
Liquid-feeding systems have been recently established in Ontario. These are relatively
large farms, and it has been estimated that liquid-feeding accounts for about 20% of pigs
sent to market. Liquid-feeding is a relatively recent innovation in Canada and its effect on
Salmonella prevalence in swine herds using Canadian feedstuffs and housing conditions is
unknown. The objective of our study was to use fecal culture and sera to determine whether
liquid-feeding is associated with a lower shedding and antibody titre to Salmonella than
dry-feeding in grower–finisher pig operations in Ontario. Liquid-feeding tends to be found
in newer facilities compared to dry-feeding. It is possible that liquid-feeding farms are
more likely to operate in an all-in all-out manner and carry out improved practices of
cleaning and rodent control than farms using dry-feeding. Therefore, we included other
risk factors in this study to control for possible confounding, to look at interaction, and to
assess the association between risk factors and the occurrence of Salmonella in grower–
finisher pigs.

2. Materials and methods

2.1. Study design, herd selection, and sample collection

The initial herd selection was based on a subset of 100 farms participating in a large
surveillance study of 5000 Ontario pig producers in 2000 (Friendship et al., 2002). These
100 farms had originally been chosen using a stratified random sample based on herd size.
To have balanced geographical representation, the sampling strategy was also used to
include farms in eastern Ontario and the Niagara region. Of the 100 farms in this survey,
grower–finisher pigs were present on 71 farms (10 liquid-feeding farms and 61 dry-feeding
farms), and so these herds were chosen. Additionally, nine liquid-feeding farms were
purposively added using a convenience sampling from the list of members of the Ontario
Swine Liquid-feeding Association (OSLFA). In order to be selected, the farms needed to
have grower–finisher pigs present. In total, 20 liquid-feeding farms and 61 dry-feeding
farms were selected.
In each herd, five pens with market weight pigs in the grower–finisher barn were
identified and three pigs per pen were chosen for the study. Blood samples were collected
from these 15 finishers on all farms except one liquid-feeding farm. Moreover, fecal
samples were collected from all 20 liquid-feeding farms and 21 of the 61 dry-feeding
farms. Rectal feces were obtained from these same 15 pigs. An additional pen-fecal sample
was collected from the fresh manure found on the floor in each of the five pens
(‘‘environment’’ sample). Fecal samples were collected into sterile containers and
transported on ice packs to the laboratory within 2–4 h and cultured immediately. By
sampling 15 pigs in finisher barns holding 300–1000 pigs each, there is a 95% confidence
that at least one pig would be identified if the minimum prevalence of disease was 20%. It
244 A. Farzan et al. / Preventive Veterinary Medicine 73 (2006) 241–254

assumes that the test has 100% specificity and 60% sensitivity and that the intracluster
correlation coefficient (ICC) is equal to 0.3.

2.2. Questionnaire

A questionnaire1 was completed at the time of the farm visit and information about
management such as pig flow, rodent evidence, antimicrobial daily usage, and herd size
(number of finishers at the site) was collected. Questions were asked the manager or owner
in a face-to-face interview and the information was recorded by the investigator. Feed tags
and drug use records were used to confirm the questionnaire responses. Questions focused
on: (i) either liquid or dry-feeding system used for grower–finisher pigs, (ii) mass
medication (a farm was classified as positive if an antimicrobial was added to feed and/or
water), (iii) number of grower–finisher pigs at the site at any one time (herd size), (iv) type
of pig flow on the basis of room or barn (‘all-in all-out’ or ‘continuous’), and (v) presence
of rodents or their feces seen in any area of the barn or food storage room.

2.3. Feeding system

We defined liquid-feeding as either ‘a mixture of feed and water’ or ‘fermented liquid-

feeding’. In the ‘‘fermented liquid-feeding’’ group the major co-products included whey
(acid or condensed), corn distiller soluble (CDS), brewer’s yeast, bakery and buttermilk
nutrient, and sugar syrup. The level of fermentation in the liquid-feeding was unknown.
This group of producers used high-pressure water spray to clean out their tanks or
equipment. Other producers mixed the traditional dry-feed with water before feeding it to
the pigs. Dry-feed used in participating farms was primarily corn based with a few farms
using barley as grain source. The protein supplement was almost exclusively soybean meal.
This dry-feed was presented as either mash or pellets.

2.4. Salmonella isolation

Samples were cultured using 25 g of feces and a double enrichment step (Davies et al.,
2000). Twenty-five grams of feces were added to 225 ml of buffered peptone water (BPW)
and incubated for 24 h at 37 8C. A 0.1 ml sample was transferred to 9.9 ml Rappaport
Vassiliadis broth (RVB) and incubated for 24 h at 41.5 8C. Cultures were stored at room
temperature (20–22 8C) for 96 h after which 0.1 ml of the RVB culture was inoculated into
9.9 ml RVB and incubated at 41.5 8C for 24 h. Finally, a loopful of the delayed secondary
enrichment (DSE) broth was plated out on xylose–lysine–tergitol 4 (XLT-4) agar and
brilliant green (BG) agar and incubated at 37 8C for up to 72 h and examined every 24 h for
growth.
Five presumptive Salmonella spp. colonies were selected and tested on triple sugar iron
(TSI), lysine decarboxylase, citrate and urease agars, and examined with a slide
agglutination test employing polyvalent anti-Salmonella antisera to determine the presence
of Salmonella-specific somatic or O antigen. The isolates were stored at
70 8C in
1
Questionnaire can be obtained from Dr. Robert Friendship (rfriends@ovc.uoguelph.ca).
 A. Farzan et al. / Preventive Veterinary Medicine 73 (2006) 241–254 245

cryovials containing 0.3 ml tryptic soy broth (TSB), 0.3 ml glycerol, and 0.6 ml of a 2 h
culture of the isolate. Three isolates from each positive sample were submitted to the OIÉ
Reference Laboratory for Salmonellosis for serotyping, phage typing, and determination of
susceptibility to antimicrobials.
A herd was classified as positive when Salmonella spp. was isolated from at least one
fecal sample.

2.5. Resistance to antimicrobial agents

Antimicrobial susceptibility of Salmonella isolates was tested by the agar dilution

method (Poppe et al., 2001) and susceptibility breakpoint levels were those described by
the National Committee for Clinical Laboratory Standards (NCCLS) M31-A2 (2002) and
M100-S14 (2004).

2.6. ELISA

Sera were analyzed for the amount of antibodies against Salmonella at Biovet Canada
using a commercial ELISA test, Salmotype (Salmotype1 Labordiagnostik, Leipzig). The
indirect mixed ELISA test was designed to detect antibodies to Salmonella serogroups B,
C1, C2, and D (O-antigens 1, 4, 5, 6, 7 and 12). The crude optical density (OD) is directly
related to the amount of antibody to Salmonella present in the test sample. The presence or
absence of antibody to Salmonella in the sample is determined by relating the absorbance
value of the unknown sample to the positive control by calculating the sample to positive
(S/P) ratio. The S/P value is divided by a correlation factor (2.5) and therefore the results
are recalculated in OD% (adjusted OD). The basis for the calculation of this factor includes
German and Dutch reference samples and the Dutch Ring Test (van der Heijden, 2001).
The calculation can be summarized as:

S ðsample
negative controlÞ

¼
P ðpositive control
negative controlÞ
 
ðS=PÞ
OD% ðadjusted ODÞ ¼  100:
2:5

2.7. Data analysis

Data were entered into an Excel spreadsheet (Microsoft1 Excel 2000) and imported
into Stata (Stata 8 Intercooled for Windows 9x) in which data were analyzed. Descriptive
analysis was done in MiniTab version 14 (MiniTabR14b) and Excel (Microsoft1 Excel
2000).
We used a mixed model with two levels, pig and farm, because predictors were
measured at the farm level, and because shedding and antibody response of individual pigs
to Salmonella was correlated within farms.
Univariable analysis of the independent variables and their association with herd
Salmonella infection was performed using the single logistic regression. Variables with a
246 A. Farzan et al. / Preventive Veterinary Medicine 73 (2006) 241–254

P < 0.25 were considered for initial inclusion in a mixed effect multi-variable logistic
regression model which was modeled at the individual pig level with farm as a random
variable.
Using categorical OD% resulted in the loss of part of the data, so we used crude OD as a
continuous dependent variable in multi-variable mixed linear regression model with farm
as a random variable and farm factors as the fixed effects.
Herd size was scaled by subtracting 70 (the smallest herd size) and then dividing by 100
to improve interpretability of the coefficient.
The models were manually built using a backward elimination process and variables
with a P > 0.05 were excluded (Dohoo et al., 2003). The presence of confounding was
investigated by looking at the effect of each predictor on the coefficient of other variables in
the model. Variables were considered confounders if the change in coefficient for the feed
variable was 25% or greater (Dohoo et al., 2003). Two-way interactions were also tested
between the main effects in the model and the statistical significance in the model was
assessed by use of P < 0.05 in the likelihood ratio test.

3. Results

3.1. Bacteriology

3.1.1. Prevalence, serovars, phage types, and antimicrobial resistance

Salmonella spp. were isolated from 25 of 420 fecal samples (6%) from dry-feeding
farms compared to 3 of 400 samples (0.8%) from liquid-feeding farms. Furthermore,
eight farms (38%) using dry-feeding had at least one positive sample compared to only
three liquid-feeding farms (15%). Of the 25 positive samples from dry-feeding farms,
13 were from the samples taken from the floor of the pens (environment) and 12
directly from pigs. On a farm basis, of the eight dry-feeding farms where at least one
fecal sample was taken directly from a pig that tested positive, Salmonella was cultured
from the pen environment samples on five of these farms. On farms using liquid-
feeding, Salmonella was not isolated from the environment samples. The three positive
farms using liquid-feeding had only one positive pig fecal sample each, whereas the
eight positive farms from the dry-feeding has 2–6 positive samples each (pig or pen
environment).
From 28 positive fecal samples, 48 isolates were cultured. Of these, 45 originated from
farms using dry-feeding and three isolates from liquid-feeding farms. The isolates
belonged to 12 different serovars; the five most common were S. Typhimurium (29%), S.
enterica ssp. IV (IV:21:-:- or IV:Rough-O:-:-) (19%), S. London (10%), S. Infantis (10%),
and S. Havana (10%). All of the 12 serovars except one (S. enterica ssp. I:17:-:-) were
found on the farms using dry-feeding while only three serovars- S. Typhimurium, S.
Mbandaka and (I:17:-:-) were recovered from farms using liquid-feeding. S. Typhimurium
was the most common serovar (29%) isolated in four positive farms in the dry-feeding
group and one positive farm from liquid-feeding farms. Six of the eight (63%) positive
farms using dry-feeding yielded multiple serovars (2–4) while only a single serovar per
farm was found where liquid-feeding was used. The phage types of the S. Typhimurium
 A. Farzan et al. / Preventive Veterinary Medicine 73 (2006) 241–254 247

found in dry-feeding farms included DT104, DT104a, DT135, and DT69. The Salmonella
Typhimurium cultured from one pig on one of the liquid-feeding farms was identified as
DT135.
The Salmonella Typhimurium isolated from the liquid-feeding farm was susceptible to
all antimicrobials tested. In contrast, the S. Typhimurium isolates from dry-feeding farms
were resistant to at least four antimicrobial agents. The three positive liquid-feeding farms
were characterized by a single positive sample and only one of the isolates in these farms
was resistant to three antibiotics (florfenicol, nitrofurantoin, and tetracycline). On the other
hand, of 45 isolates, which were cultured from 25 positive samples from the dry-feeding
farms, 23 (51%) isolates displayed multiple resistances.

3.1.2. Risk factors for Salmonella shedding

The distribution of risk factors among 20 liquid and 61 dry-feeding swine farms is
shown in Table 1. Indicator variables were generated to categorize the continuous herd size
predictor to three dummy variables for herds with less than 500 grower–finishers (small
farms), 500–2000 grower–finishers (medium farms), and more than 2000 grower–finisher
pigs (large farms). In the univariable analysis, four risk factors (P < 0.25) were selected for
initial inclusion in the final mixed model analysis; feed, pig flow, presence of rodent, and
herd size.
In the final model, two risk factors were associated (P < 0.05) with an increased risk of
being Salmonella positive (Table 2). The odds of a fecal sample being Salmonella positive
in a herd using dry-feeding was four times higher than in a herd with liquid-feeding
(P = 0.03). Likewise, the chance of a sample being Salmonella positive in farms with a
continuous flow system was 3.9 times higher than in farms with an all-in all-out flow
system (P = 0.02).

Table 1
Distribution of risk factors and presence of Salmonella in fecal samples among 20 liquid and 21 dry-feeding swine
farms
Characteristics Liquid-feeding No. (%) Dry-feeding No. (%)
a
Salmonella Negative 17 (85) 13 (62)
Positive 3 (15) 8 (38)
Flow All-in all-out 12 (60) 30 (49)
Continuous 8 (40) 31 (51)
Rodentb No 14 (70) 30 (49)
Yes 6 (30) 31 (51)
Antimicrobial usage No 3 (15) 8 (13)
Yes 17 (85) 53 (87)
Herd size c <500 1 (5) 21 (34)
500–2000 8 (42) 31 (51)
>2000 10 (53) 9 (15)
a
Fecal samples from 20 liquid-feeding and 21 dry-feeding farms were cultured.
b
Evidence of rodents seen on farm.
c
Number of grower–finisher pigs at the site at any one time.
248 A. Farzan et al. / Preventive Veterinary Medicine 73 (2006) 241–254

Table 2
Farm-level factors associated with the culture of Salmonella at the pig level from 41 Ontario farms using a multiple
variable logistic regression model with farm as a random variable
Risk factor Odds ratio Standard error Confidence interval (95%) P-value
Feed Dry 4.1 2.7 1.1–15.1 0.03
Liquid 1
Pig flow Continuous 3.9 2.3 1.2–12.7 0.02
All in-All out 1

3.2. Serology

3.2.1. Optical density as categorical variable

Using cut-off of 10 OD%, 98% of dry-feeding farms and 84% of liquid-feeding farms
tested positive (P = 0.01). The median within herd sero-prevalence was 40% for liquid-
feeding farms and 60% for dry-feeding farms. At this cut-off of 10 OD%, 55% of pigs on
farms with dry-feeding tested positive compared to 41% positive pigs on farms using
liquid-feeding.
The distribution of sero-prevalence on farms using liquid-feeding differed from those
with dry-feeding (P = 0.03) (Fig. 1). On 6% of the dry-feeding farms, all 15 samples tested
positive while no liquid-feeding farm had 15 positive samples.

3.2.2. Optical density as a continuous variable

The crude OD ranged from 0.1 to 4.5 with the median of 0.74 (Q3
Q1 = 0.45). The
multi-variable mixed linear regression model included feeding type (dry versus liquid), pig
flow (continuous versus all-in all-out), presence of rodents (yes versus no), antimicrobial
daily usage (yes versus no), and herd size (continuous variable) as fixed effects. The plot of
residual in the form of a histogram and the Q–Q (quantile–quantile) demonstrated that the
residuals were not normally distributed and had a right skew. The Shapiro-Wilk W-test was
also significant (P < 0.001) suggesting that the data are not normally distributed.

Fig. 1. Distribution of 19 liquid-feeding and 61 dry-feeding farms in Ontario into five categories (three categories
were created based on distribution of farms (quartiles)) based on the prevalence (%) of Salmonella positive pigs
tested by ELISA (OD% 10) in 2003 (Fisher’s exact P = 0.03).
 A. Farzan et al. / Preventive Veterinary Medicine 73 (2006) 241–254 249

Table 3
Predicted effect of feeding style and other risk factors on the log of crude OD obtained by Salmonella-ELISA for
1200 samples taken from 80 Ontario swine farms in the multi variable mixed linear regression model
Predictor Coefficient Standard Confidence P-value
error interval (95%)
Fixed effects
Dry-feeding 1.20 1.08 1.04–1.40 0.016
Antimicrobial usage 1.24 1.08 1.06–1.46 0.007
Herd sizea 1.00 1.00 1.00–1.01 0.003
Constant 0.53 0.91 0.44–0.63 0.000
Random effects
Herd variation = 1.28
The combined within-herd
variation and measurement error = 1.58
a
Number of grower–finisher pigs at the site at any one time (was rescaled by subtracting of 70 and dividing the
result by 100).

Therefore, the model was fitted with the log transformed crude OD in that outcome and
continuous predicator (herd size) had a linear relationship and the residuals had a constant
variance at different levels of the predictors. When the model was fitted, the coefficients
were back transformed into the original scale in order to make their explanations easier.
However, this back transformation resulted in asymmetrical confidence intervals for the
coefficients.
Dry-feeding and antimicrobial daily usage were predicted to increase the value of crude
OD by 1.2 (1.04, 1.4) units (P = 0.016) and 1.24 (1.06, 1.46) units (P = 0.007), respectively
(Table 3). In addition, for each additional 100 grower–finisher pigs, the crude OD increased
by one unit (P = 0.003). The variation of the value of crude OD due to farm effect was 40%
of the total random variance (Table 3). The ‘antimicrobial daily usage’ predictor changed
the coefficient of dry-feeding by 35% suggesting that it was a confounder. No interaction
was observed at the level of P < 0.05.

4. Discussion

4.1. Liquid-feeding

Liquid-feeding was associated with a decreased risk of being culture positive for
Salmonella compared to dry-feeding. Also, using either crude OD (multi-variable mixed
linear regression model), or categorical OD%, farms using dry-feeding were more likely to
test sero-positive than farms using liquid-feeding. The association between liquid-feeding
and lower Salmonella prevalence in swine herds has been reported previously (Lo Fo Wong
et al., 2004; van Winsen et al., 2002; van der Wolf et al., 1999, 2001a,b,c). Pelleted feed is
also a common risk factor for Salmonella occurrence (Kranker et al., 2001; Leontides et al.,
2003; Lo Fo Wong et al., 2004; Mikkelsen et al., 2004).
The protective effect of liquid-feeding on Salmonella can be a result of low pH, the
presence of lactic acid and acetic acid produced by fermentation (van Winsen et al., 2002).
250 A. Farzan et al. / Preventive Veterinary Medicine 73 (2006) 241–254

The organic acids lower the pH of the liquid-feed to 4 or lower in liquid fermented feed and
therefore inhibit multiplication of Salmonella (van Winsen et al., 2002). Modifying the
microflora of the gastro-intestinal tract with lactic acid-producing bacteria has been
proposed as a mechanism by which Salmonella is excluded (Canibe and Jensen, 2003; van
Winsen et al., 2002; van der Wolf et al., 2001a,b,c). In addition, it is possible that the
ingredients present in a liquid-feeding ration, such as whey or distiller’s grain, are more
likely to support the proliferation of lactic acid-producing bacteria by acting in a
‘‘prebiotic’’ manner.
Low-prevalence herds can test negative over a longer period of time as a result of
providing liquid-feed (van der Wolf et al., 2001b). In our study, farms with liquid-feeding
tended to be more likely classified as either negative or low sero-prevalence than dry-
feeding farms. The potential of liquid-feeding to reduce within herd sero-positivity is of
importance for future Salmonella control programs.
The Salmonella isolated from farms using liquid-feeding showed little antimicrobial
resistance in contrast to some of the isolates from the dry-feeding farms. Around 60% of
the liquid-feeding farms using antimicrobials in feed, compared to 87% of dry-feeding
farms (Table 3). Therefore, the reasons for the low prevalence of multi-drug resistant
Salmonella associated with pigs fed liquid-feed remains unclear and further studies need to
be carried out to investigate the association between liquid-feeding and antimicrobial
resistance in Salmonella spp.

4.2. Other risk factors

Different management risk factors for Salmonella occurrence have been reported before
(Beloeil et al., 2004; Berends et al., 1997; Funk and Gebreyes, 2004; Funk et al., 2001; Lo
Fo Wong et al., 2004; Nollet et al., 2004; van der Wolf et al., 1999, 2001c). The positive
association between herd size and prevalence of Salmonella has been shown in other
studies (Carstensen and Christensen, 1998; Christensen et al., 2002; van der Wolf et al.,
1999, 2001c). In our study, large herd size was associated with a higher amount of
antibodies to Salmonella. Conversely, no association between herd size and Salmonella
shedding was found. We defined herd size as the number of grower–finisher pigs at the site
at any one time. However, herd size might be influenced by other factors such as the size of
the production site, number of workers, number of sows, nursery pigs or gilts may also be
important when considering the herd size. Liquid-feeding systems were more likely to be
found on larger farms (P = 0.03) (Table 1) or on farms that were part of a large production
system. The small number of farms included in the bacteriological part of our study may
have obscured the true association between herd size and Salmonella shedding.
All-in all-out pig flow, which is known as a practice to reduce infectious disease in swine
in general was also associated with lower Salmonella shedding in our study. Several studies
have shown that rodents play a significant role in transmitting Salmonella into animal
production facilities (Funk et al., 2004). In the present study no significant association
between the presence of rodents and Salmonella infection status was observed. This can be
interpreted as either inaccuracy in measuring rodent variable, insufficient statistical power,
or lack of a true association. It should be noted that liquid-feeding system represents a
relatively new technology in pig production in Ontario and therefore is adopted by farms
 A. Farzan et al. / Preventive Veterinary Medicine 73 (2006) 241–254 251

implementing good agricultural practice (GAP). In this respect it is likely that such farms
exercise more effective rodent control compared to traditional dry-feeding farms. This is
supported by the finding that 30% of liquid-feeding farms compared to 51% of dry-feeding
farms had an evidence of rodent (Table 1). The presence of rodents might indicate an
indirect effect in that it is possible that the type of farms that have more rodents are older
facilities that are difficult to clean. Additional investigations should be carried out to
determine the role of rodents in transmitting Salmonella on Ontario swine farms.
The relatively small number of herds in this study greatly limits the examination of risk
factors for Salmonella prevalence. Likewise, inaccuracy in measuring the risk factors or the
absence of important risk factors could result in bias. For instance, liquid-feeding barns are,
as stated earlier, in general newer than dry-feeding facilities, although some of the liquid-
feeding systems are installed in older barns. The age of facilities was not recorded in our
study.

4.3. Serology and optical density

The crude optical density is directly related to the amount of antibodies to Salmonella
present in serum. However, categorized OD% with different cut-offs (10, 20, 30, and 40)
have been used in Salmonella ELISA tests (Alban et al., 2002; Blaha, 2004; Hurd et al.,
2004; Leontides et al., 2003; Lo Fo Wong et al., 2004; Mousing et al., 1997; Proux et al.,
2000; van Winsen et al., 2001). The issue of OD% cut-off in Salmonella ELISA tests
depends on the objective of the study. A lower cut-off could be used for screening
objectives and a higher cut-off might be chosen for monitoring purposes. The OD% 10 is an
experimental cut-off value by which negative and positive controls were identified. Based
on an ELISA test with OD% 10 about 95% of the farms had at least one positive pig.
However, the sero-prevalence was lower (54%) if higher cut-off (OD% 40) was chosen. In
the present study, the sero-prevalence differed in liquid-feeding and dry-feeding farms at
the level of P < 0.05 only at cut-off of OD% 10. However, at this cut-off the sensitivity and
specificity of the ELISA test were 0.94 and 0.49, respectively. The question is which OD%
cut-off could represent the‘true’ Salmonella status in liquid- and dry-feeding farms.
Despite the fact that the coating antigens in this ELISA test represent only 48% of the
antigens of Salmonella isolated in our study, crude OD may more likely mirror the ‘true’
Salmonella status than if the categorized OD% was used.

4.4. Bacteriology

Amount of feces, secondary and delayed enrichment can markedly influence the
sensitivity of bacterial culture (Davies et al., 2000; Funk et al., 2000). In this study we used
25 g of the fecal sample and applied two enrichment steps in order to increase the
sensitivity of the test. However, the overall prevalence of Salmonella, 27% of the herds and
3% of all samples, obtained in our study is probably an underestimate of true Salmonella
prevalence in Ontario. In other words, herd prevalence of Salmonella on farms with dry-
feeding type (38%) may be closer to true Salmonella prevalence in Ontario swine farms
because they were a sub-sample from a random sample of 100 farms chosen to represent
the Ontario industry. More importantly, our estimates of prevalence of Salmonella on these
252 A. Farzan et al. / Preventive Veterinary Medicine 73 (2006) 241–254

particular farms may be low because we obtained only a small number of samples and
visited the herds only once. In addition only market weight pigs were sampled. It is
possible that the grower pigs in the herd may have been shedding at the time of our visit
so that if different age groups had been examined a higher prevalence would have been
found.

5. Conclusion

The results of this study suggest that liquid-feeding is associated with lower Salmonella
prevalence in finisher pigs. Liquid-feeding is a new technology for the swine industry in
Canada. Liquid-feeding installation is relatively expensive and tends to be found in newer
and larger facilities compared to dry-feeding. However, there may be other risk factors or
confounding factors, such as cleaning and disinfection or biosecurity practices. Therefore,
it will be necessary to do further studies to determine that liquid-feeding per se is an
effective method of controlling Salmonella.
The 80 farms in this study represented a wide range of production types and were well
distributed geographically over southern Ontario. However, caution is needed in
extrapolating these findings to the entire Ontario pig population in that these farms
were not a true random sample.
We also concluded that crude OD can be used instead of categorised OD% to investigate
the risk factors for Salmonella exposure and to determine Salmonella sero-status in swine
farms.

Acknowledgements

We would like to thank the Public Health Agency of Canada, the Ontario Ministry of
Agriculture and Food, Ontario Pork, The Canadian Research Institute for Food Safety
(CRIFS), the Swine Liquid-feeding Association, Biovet, the participating swine producers,
and the research technicians who collected samples.

References

Alban, L., Stege, H., Dahl, J., 2002. The new classification system for slaughter-pig herds in the Danish
Salmonella surveillance-and-control program. Prev. Vet. Med. 53, 133–146.
Berends, B.R., Van Knapen, F., Mossel, D.A., Burt, S.A., Snijders, J.M., 1998. Impact on human health of
Salmonella spp. on pork in The Netherlands and the anticipated effects of some currently proposed control
strategies. Int. J. Food Microbiol. 44, 219–229.
Beloeil, P.A., Fravalo, P., Fablet, C., Jolly, J.P., Eveno, E., Hascoet, Y., Chauvin, C., Salvat, G., Madec, F., 2004.
Risk factors for Salmonella enterica ssp. enterica shedding by market-age pigs in French farrow-to-finish
herds. Prev. Vet. Med. 63, 103–120.
Berends, B.R., Van Knapen, F., Snijders, J.M., Mossel, D.A., 1997. Identification and quantification of risk factors
regarding Salmonella spp. on pork carcasses. Int. J. Food Microbiol. 36, 199–206.
Blaha, T., 2004. The present state of the German QS Salmonella monitoring and reduction programme. Deut.
Tierarztl. Wochenschr. 8, 324–326.
 A. Farzan et al. / Preventive Veterinary Medicine 73 (2006) 241–254 253

Canibe, N., Jensen, B.B., 2003. Fermented and nonfermented liquid-feed to growing pigs: effect on aspects of
gastrointestinal ecology and growth performance. J. Anim. Sci. 81, 2019–2031.
Carstensen, B., Christensen, J., 1998. Herd size and sero-prevalence of Salmonella enterica in Danish swine herds:
a random-effects model for register data. Prev. Vet. Med. 34, 191–203.
Christensen, J., Baggesen, D.L., Nielsen, B., Stryhn, H., 2002. Herd prevalence of Salmonella spp. in Danish pig
herds after implementation of the Danish Salmonella Control Program with reference to a pre-implementation
study. Vet. Microbiol. 88, 175–188.
Davies, P.R., Turkson, P.K., Funk, J.A., Nichols, M.A., Ladely, S.R., Fedorka-Cray, P.J., 2000. Comparison of
methods for isolating Salmonella bacteria from faeces of naturally infected pigs. J. Appl. Microbiol. 89, 169–
177.
Davies, R., Breslin, M., 2001. Environmental contamination and detection of Salmonella enterica serovar
enteritidis in laying flocks. Vet. Rec. 149, 699–704.
Dohoo, I., Martin, W., Stryhn, H., 2003. Veterinary Epidemiologic Research. AVC Inc., University of Prince
Edward Island, Canada, pp. 273–316, 317–334, 335–372 and 473-498.
Friendship, R., Dewey, C., Blackwell, T., 2003. The Ontario Swine Sentinel Herd Project. Centralia Swine
Research Update, Kirkton, 18-19.
Funk, J.A., Gebreyes, W., 2004. Risk factors associated with Salmonella prevalence on swine farms. J. Swine
Health Prod. 12, 1–6.
Funk, J.A., Davies, P.R., Gebreyes, W.A., 2001. Risk factors associated with Salmonella enterica prevalence in
three-site production system in North Carolina, USA. Berl. Munch. Tierarztl. Wochenschr. 114, 335–338.
Funk, J.A., Davies, P.R., Nichols, M.A., 2000. The effect of fecal sample weight on detection of Salmonella
enterica in swine feces. J. Vet. Diagn. Invest. 12, 412–418.
Hurd, H.S., McKean, J.D., Griffith, R.D., Rostagno, M.H., 2004. Estimation of the Salmonella enterica prevalence
in finishing swine. Epidemiol. Infect. 132, 127–135.
Kranker, S., Dahl, J., Wingstrand, A., 2001. Bacteriological and serological examination and risk factor analysis of
Salmonella occurrence in sow herds, including risk factors for high Salmonella seroprevalence in receiver
finishing herds. Berl. Munch. Tierarztl. Wochenschr. 9–10, 350–352.
Leontides, L.S., Grafanakis, E., Genigeorgis, C., 2003. Factors associated with the serological prevalence of
Salmonella enterica in Greek finishing swine herds. Epidemiol. Infect. 1, 599–606.
Lo Fo Wong, D.M.A., Dahl, J., Stege, H., van der Wolf, P.J., Leontides, L., von Altrock, A., Thorberg, B.M., 2004.
Herd-level risk factors for subclinical Salmonella infection in European finishing-pig herds. Prev. Vet. Med.
62, 253–266.
Mead, P.S., Slutsker, L., Dietz, V., McCaig, L.F., Bresee, J.S., Shapiro, C., Griffin, P.M., Tauxe, R.V., 1999. Food-
related illness and death in the United States. Emerg. Infect. Dis. 5, 607–625.
Mikkelsen, L.L., Naughton, P.J., Hedemann, M.S., Jensen, B.B., 2004. Effects of physical properties of feed on
microbial ecology and survival of Salmonella enterica serovar Typhimurium in the pig gastrointestinal tract.
Appl. Environ. Microbiol. 70, 3485–3492.
Mousing, J., Jensen, P.T., Halgaard, C., Bager, F., Feld, N., Nielsen, B., Nielsen, J.P., Bech-Nielsen, S., 1997.
Nation-wide Salmonella enterica surveillance and control in Danish slaughter swine herds. Prev. Vet. Med. 29,
247–261.
National Committee for Clinical Laboratory and Standards (NCCLS), 2002. Performance standards for anti-
microbial disk and dilution susceptibility tests for bacteria isolated from animals; approved standard, 2nd ed.
NCCLS document M31-A2. NCCLS, Wayne, PA.
National Committee for Clinical Laboratory and Standards (NCCLS), 2004. Performance standards for antimicrobial
susceptibility testing; 14th informational supplement. NCCLS document M100-S14. NCCLS, Wayne, PA.
Nielsen, B., Baggesen, D., Bager, F., Haugegaard, J., Lind, P., 1995. The serological response to Salmonella
serovars Typhimurium and infantis in experimentally infected pigs. The time course followed with an indirect
anti-LPS ELISA and bacteriological examinations. Vet. Microbiol. 47, 205–218.
Nielsen, B., Wegener, H.C., 1997. Public health and pork and pork products: regional perspectives of Denmark.
Rev. Sci. Tech. 16, 513–524.
Nollet, N., Maes, D., De Zutter, L., Duchateau, L., Houf, K., Huysmans, K., Imberechts, H., Geers, R., De Kruif,
A., Van Hoof, J., 2004. Risk factors for the herd-level bacteriologic prevalence of Salmonella in Belgian
slaughter pigs. Prev. Vet. Med. 65, 63–75.
254 A. Farzan et al. / Preventive Veterinary Medicine 73 (2006) 241–254

Poppe, C., Ayroud, M., Ollis, G., Chirino-Trejo, M., Smart, N., Quessy, S., Michel, P., 2001. Trends in
antimicrobial resistance of Salmonella isolated from animals, foods, of animal origin, and the environment
of animal production in Canada, 1994–1997. Microb. Drug. Resist. 7, 197–212.
Proux, K., Houdayer, C., Humbert, F., Cariolet, R., Rose, V., Eveno, E., Madec, F., 2000. Development of a
complete ELISA using Salmonella lipopolysaccharides of various serogroups allowing to detect all infected
pigs. Vet. Res. 31, 481–490.
Rostagno, M.H., Hurd, H.S., Mckean, J.D., 2004. Bacteriological and serological Salmonella prevalence in
finishing pigs. In: Proceedings of the 18th International Pig Veterinary Society (IPVS) Congress, vol. 2,
Hamburg, Germany, June 27–July1, p. 649.
van der Heijden, H.M., 2001. First international ring trial of ELISAs for Salmonella-antibody detection in swine.
Berl. Munch. Tierarztl. Wochenschr. 114, 389–392.
van der Wolf, P.J., Bongers, J.H., Elbers, A.R.W., Franssen, F.M.M.C., Hunneman, W.A., van Exsel, A.C.A.,
Tielen, M.J.M., 1999. Salmonella infections in finishing pigs in The Netherlands: bacteriological herd
prevalence, serogroup and antibiotic resistance of isolates and risk factors for infection. Vet. Microbiol. 67,
263–275.
van der Wolf, P.J., Elbers, A.R., van der Heijden, H.M., van Schie, F.W., Hunneman, W.A., Tielen, M.J., 2001a.
Salmonella seroprevalence at the population and herd level in pigs in The Netherlands. Vet. Microbiol. 80,
171–184.
van der Wolf, P.J., Lo Fo Wong, D.M., Wolbers, W.B., Elbers, A.R., van der Heijden, H.M., van Schie, F.W.,
Hunneman, W.A., Willeberg, P., Tielen, M.J., 2001b. A longitudinal study of Salmonella enterica infections in
high-and low-seroprevalence finishing swine herds in The Netherlands. Vet. Quart. 23, 116–121.
van der Wolf, P.J., Wolbers, W.B., Elbers, A.R.W., van der Heijden, H.M.J.F., Koppen, J.M.C.C., Hunneman,
W.A., van Schie, F.W., Tielen, M.J.M., 2001c. Herd level husbandry factors associated with the serological
Salmonella prevalence in finishing pig herds in The Netherlands. Vet. Microbiol. 78, 205–219.
van Winsen, R.L., Keuzenkamp, D., Urlings, B.A.P., Lipman, L.J.A., Snijders, J.A.M., Verheijden, J.H.M., van
Knapen, F., 2002. Effect of fermented feed on shedding of Enterobacteriaceae by fattening pigs. Vet.
Microbiol. 87, 267–276.
van Winsen, R.L., van Nes, A., Keuzenkamp, D., Urlings, H.A., Lipman, L.J., Biesterveld, S., Snijders, J.M.,
Verheijden, J.H., van Knapen, F., 2001. Monitoring of transmission of Salmonella enterica serovars in pigs
using bacteriological and serological detection methods. Vet. Microbiol. 80, 267–274.
Voetsch, A.C., Van Gilder, T.J., Angulo, F.J., Farley, M.M., Shallow, S., Marcus, R., Cieslak, P.R., Deneen, V.C.,
Tauxe, R.V., 2004. FoodNet estimate of the burden of illness caused by nontyphoidal Salmonella infections in
the United States. Clin. Infect. Dis. 38, 127–134.