Preventive Veterinary Medicine 98 (2011) 8187

Contents lists available at ScienceDirect

Preventive Veterinary Medicine

journal homepage: www.elsevier.com/locate/prevetmed

The prevalence of Salmonella enterica in Spanish feed mills and

potential feed-related risk factors for contamination
Gregorio J. Torres a , F. Javier Piquer a , Leonor Algarra a , Cristina de Frutos b ,
Odn J. Sobrino a,
a
Subdireccin General de Conservacin de Recursos y Alimentacin Animal, Ministerio de Medio Ambiente, y Medio Rural y Marino, C/Alfonso XII, 62, 28071
Madrid, Spain
b
National Reference Laboratory for Salmonellosis in Animals and Feed, Ministry of the Environment and Rural and Marine Affairs, Algete, Spain

a r t i c l e

i n f o

Article history:
Received 9 February 2010
Received in revised form 8 November 2010
Accepted 11 November 2010
Keywords:
Salmonella
Feed mills
Feed risk factors

a b s t r a c t
A cross-sectional study was conducted in Spain to estimate the prevalence of Salmonella
enterica in feed mills and to identify and evaluate potential risk factors associated with feed
contamination.
A total of 3844 samples were collected from 523 different feed mills using a stratied
sampling method. Samples were tested for the presence of Salmonella using conventional
culture methods. When the presence of Salmonella was detected, samples were further
characterised using serotyping at the National Reference Laboratory (NRL) for animal feed.
Additional data about the biosecurity and hygiene measures, feed material used and compound feed produced, were collected by ofcial veterinarians using a questionnaire in
situ.
In 144 of the feed mills visited (28%), Salmonella were present. However, it was only
isolated from 4.8% of samples taken from all of the feed mills (3.5% from feed materials, 3.2%
from compound feed and 12.5% from dust of the feed mill facilities). Salmonella serovars
of public health importance (Enteritidis, Typhimurium, Infantis, Virchow and Hadar), were
detected in only 2.7% of feed mills and in 0.3% of the samples studied.
Logistic regression was used to investigate potential feed-mill risk factors for the isolation
of Salmonella. Feed mill intake pits were demonstrated to have an increased risk of culturepositive dust samples (OR = 6.4; 95% CI: 2.715.1). The feed material used in the production
of compound feed was associated with recovery of Salmonella. Of the feed material used,
cotton seeds were identied as having the highest odds of contamination (OR = 3.8; 95%
CI: 1.78.3). Pelleting appears to reduce the chance of contamination because non-pelleted
compound feed is 8 times more likely to be contaminated than pelleted compound feed
(OR = 8.2; 95% CI: 2.526.6).
The role of the feed itself in the epidemiology of Salmonella seems to be of limited importance as compound feed is not frequently contaminated at the feed mill level. This should
not preclude Salmonella control measures from including all stages of feed production and
they should have a risk-based approach according to the ndings of this study.
2010 Elsevier B.V. All rights reserved.

1. Introduction

Corresponding author. Tel.: +34 91 347 82 55; fax: +34 91 347 66 71.
E-mail address: o.sobrino@marm.es (O.J. Sobrino).
0167-5877/$ see front matter 2010 Elsevier B.V. All rights reserved.
doi:10.1016/j.prevetmed.2010.11.009

Manufacturers of compound feed for livestock use

different types of feed materials. The origin, transport,
transformation and storage of feed materials may result

82

G.J. Torres et al. / Preventive Veterinary Medicine 98 (2011) 8187

in contamination of the feed with several food-borne

pathogens (Durand et al., 1990; Whyte et al., 2002).
Salmonella enterica has been identied as the major microbiological hazard in feed, but the implication of this for
animal and human health remains unclear (Davies et al.,
2001; EFSA, 2006). In countries with a low Salmonella
prevalence in animals, feed is considered to be the major
source of this pathogen. However, it is difcult to quantify the public health risk associated with the presence of
Salmonella in feed in countries with high prevalence rates
(EFSA, 2008).
Salmonella contamination levels in compound feed
ranging from 0% to 6% have been reported by some Eureopean Union (EU) countries (EFSA, 2006). Data on feed
materials are variable but mainly restricted to Northern
European countries (Weirup, 2006; EFSA, 2008).
The distribution of Salmonella in contaminated feed is
not uniform. Sampling protocols must be designed according to an appropriate statistical analysis (Legan et al., 2001).
The ISO 6579:2002 method, with a sensitivity of up to 65%
and a specicity of 97% (Koyuncu and Haggblom, 2009),
was adopted by the EU as the standard for monitoring the
presence of Salmonella in feed samples (EFSA, 2008).
At the feed mill level, contamination with Salmonella
may be related to different factors such as dust contamination, the presence of vectors and poor hygiene conditions
(EFSA, 2008).
Unloading the ingredients into the intake pit creates
large amounts of dust that might carry Salmonella into
the premises (Davies and Wray, 1996; Whyte et al., 2002).
Salmonella contamination can also occur if the facilities are
easily accessible to vectors such as insects, birds or rodents
(Maciorowski et al., 2006). Handling and processing of
ingredients can also spread Salmonella contamination, particularly in farm mill facilities where poor hygiene practices
during feed processing are commonly observed (EFSA,
2008).
Manufacturing conditions such as moisture and temperature also have an effect on the level of bacterial
contamination in the compound feed. Pelleting may not
be able to completely eliminate microbiological contaminations but markedly reduces the bacterial load in feed
(Hacking et al., 1978; McCapes et al., 1989).
The public health risks associated with the presence
of Salmonella in compound feed might depend on which
serovars are involved (Herikstad et al., 2002; EFSA, 2008).
The current EU control target for livestock species includes
Salmonella serovars Enteritidis, Typhimurium, Infantis, Virchow and Hadar since they are responsible for most of the
human infection cases (EFSA, 2009a).
In this context, the aim of this study was to assess the
prevalence of S. enterica in the Spanish feed mills and to
identify and evaluate potential risk factors associated with
feed mill contamination.
2. Material and methods
2.1. Study design and sampling frame
A cross-sectional study was conducted in Spain from
October 2007 to December 2008. The sampling frame

included the 1328 feed mills registered according to

Regulation (EC) 183/2005. There were limited epidemiological data in the literature regarding the prevalence
of Salmonella in Spanish feed mills. The sample size was
calculated to estimate 50% prevalence in feed mills with
99% condence and 5% accuracy (Kirkwood and Sterne,
2003).
The calculated sample size was then increased by a correction factor of 20% owing to the low sensitivity of the
diagnostic test resulting in a sample size of around 530 feed
mills.
The sample population was selected using a proportional stratied random sampling method. The sample size
was proportionally allocated to each autonomous region
taking into account the number of feed mills and the compound feed produced monthly in each of them.
Within each stratum the feed mills were randomly selected using a computer-generated random
table (Microsoft Ofce Excel 2003 ). All the samples
were taken under the supervision of ofcial veterinarians.
The sites and type of samples to be investigated and
the number of samples collected from each site or type of
samples were decided according to expert opinion and supported by unpublished results of a qualitative risk analysis
conducted prior to the design of the study. Seven samples
of 500 g each were collected in each feed mill using sterilized/disposable material following the general provisions
of European Directive 71/250/EEC specically adapted for
this study. The sampling protocol included one dust sample
from the feed mill facilities, three samples of feed materials and three samples of compound feed from each of
the feed mills investigated. All sample were accompanied
by a closed-question questionnaire previously designed
and validated by the authors in a pilot study involving
eight feed mills with different monthly production. The
questionnaire (available upon request) included questions
relating to risk factors derived from the literature review
and divided into ve major groups (feed materials, feed
mill facilities and building, storage conditions, staff and
biosecurity).
2.2. Bacteriological analysis
Samples were sent to regional ofcially designated laboratories within 48 h after collection. Samples were kept
at <7 C and were processed within seven days of collection. Each sample was split in two portions (250 g
each). An aliquot of 25 g of one of the portions was
used as a nal sample for Salmonella isolation according to the ISO 6579:2002/Amd 1:2007 method. One
single Salmonella-conrmed isolate from each positive
sample was forwarded to the National Reference Laboratory (NRL) for typing following the WhiteKauffmannLe
Minor scheme (Grimont and Weill, 2007). Salmonella
isolates were preserved in a tryptone, skim milk and
glycerine mixture and frozen at 80 C which ensures
the integrity of the strains for a minimum of two
years.
The information gathered from the questionnaires and
the microbiology results were stored in a database.

G.J. Torres et al. / Preventive Veterinary Medicine 98 (2011) 8187

83

Table 1
Summary of variables derived from the questionnaire and investigated as potential risk factors for Salmonella contamination.
Exposure variables

Description (number categories)

p-Value#

Autonomous region
Season
Level of production
Feed materialsa
Dust from feed mill facilitiesb
Presentation of compound feedc
Intended livestock destinationc

Autonomous region where the feed mill is located (16)

3 month periods (4)
Tons per month (4)
Raw material used in the production of compound feed (7)
Sites within the feed mill where dust was investigated (12)
Pellets versus not pelleted
Livestock species of destination of the compound feed (5)

0.008
0.240
0.792
<0.001
0.034
<0.001
0.964

#
a
b
c

Univariable statistical analysis.

Only in the feed material data subset.
Only in the dust from feed mill facilities data subset.
Only in the compound feed data subset.

2.3. Statistical analysis

The prevalence of Salmonella was estimated considering that a feed mill was classied as Salmonella positive
when at least one Salmonella colony was isolated from any
of the samples investigated. The prevalence of Salmonella
and condence intervals were estimated based on a normal
approximation by the Wilson method (Altman et al., 2000).
The association between the categorical exposure factors and the outcome presence of Salmonella was
investigated in a single level contingency table analysis
(signicance determined using the chi-square statistic). All
exposure variables with a p value lower than 0.25 were
selected for further multivariable analysis (Table 1).
The multivariable analysis involved the stratication of
the dataset by the sample origin into three data subsets
due to the biological differences of each group of samples. A
logistic regression model was built between each sampling
specic-data subset using the statistical package Stata V.10
(Stata Corporation , Timberlake). The rst model exclusively considered data obtained from samples taken from
dust of the feed mill facilities (n = 607), the second model
was built with the data corresponding to the samples of
feed materials (n = 1579) and the third model included data
from the compound feed samples (n = 1658).
The model-building strategy involved a forward selection of variables with two-tailed Walds test signicant
values (p 0.05). The most frequent category of each exposure variable was selected as the reference category. The
models built were compared for a better tting with the
likelihood ratio test. The Hosmer and Lemesshow test
was used to check the goodness of t of the nal model
(Kirkwood and Sterne, 2003). Confounding was assessed
every time a variable was dropped from the model. A variable was considered as a confounder factor if it modied
the crude OR for the independent variable by 10% or more
(Kirkwood and Sterne, 2003).
All the parameters were estimated with 95% condence
intervals.

Table 2
Salmonella serovars most frequently isolated.
Serovars

Number of positive
samples (%)

Mbandaka
Anatum
Senftenberg
Typhimurium
Agona

14 (7.6)
13 (7.0)
12 (6.5)
10 (5.4)
8 (4.3)

1000 tons per month, 159 (30.4%) produced between 1000

and 5000 tons per month, 56 (10.7%) produced between
5000 and 10,000 tons per month and 68 (13.0%) produced
more than 10,000 tons per month.
A total of 3844 samples were analysed. Salmonella was
isolated from 185 samples (4.8% 95% CI: 4.25.5). Stratifying the positive samples by their origin, Salmonella was
detected in 3.5% (95% CI: 2.54.4) of the samples from feed
material, 3.3% (95% CI: 2.44.1) of the compound feed samples and 12.5% (95% CI: 9.815.2) of the dust samples taken
from the feed mill facilities.
Salmonella was isolated in samples from 144 different
feed mills; thus the estimated prevalence of Salmonella
contaminated feed mills was 27.5% (95% CI: 23.931.5).
When only samples of compound feed were considered 47
(8.9% 95% CI: 7.111.2) of all the feed mills investigated had
at least one positive sample.
Sixty ve different serovars were identied; Mbandaka (n = 14, 7.6%) and Anatum (n = 13, 7.0%) were the
most common serovars. Among the ve serovars most
frequently detected, the only serovar with public health
signicance, as stated by the EU, was Typhimurium (n = 10,
5.4%) (Table 2).
Strating by type of sample, Salmonella serovars Anatum (n = 9, 10.7%) and Mbandaka (n = 6, 7.1%) were the most
common serovars in dust from feed mill facilities. Those
serovars were also most frequently isolated in samples
from compound feed (n = 5, 8.5% in both cases). In samples from feed material Salmonella ser. Senftenberg (n = 8,
13.7%) was the most frequent one.

3. Results
3.2. Results of the univariable and multivariable analysis
3.1. Prevalence of Salmonella
Five hundred and twenty-three feed mills were individually investigated. Of these, 240 (45.9%) produced less than

The univariable analysis revealed that the variables seasonality (p = 0.239), level of monthly production (p = 0.792)
and the intended compound feed livestock destination

84

G.J. Torres et al. / Preventive Veterinary Medicine 98 (2011) 8187

Table 3
Numerical description and percentage of positive samples for each of the categories of the exposure variables not associated with presence of Salmonella.
Exposure variable

Categories

Feed mill production tons/month

<1000
10005000
500010,000
>10,000

1649
1202
504
489

4.7
5.3
4.8
4.3

Seasonality

JanuaryMarch
AprilJune
JulySeptember
OctoberDecember

986
998
795
1083

4.6
3.8
5.8
5.2

489
149
25
602
393

3.3
4.0
4.0
3.3
2.8

Pigs
Bovine
Ovine
Poultry
Unknown

Intended livestock of destination

Positive samples (%)

N = number of samples from each categories.

a
Only in the compound feed data subset.

(p = 0.964) were not associated with the presence of

Salmonella. In the multivariable analysis, the variable
autonomous region (p = 0.008) was considered a potential confounding factor for the association between the
exposure factors and the outcome presence of Salmonella
(Tables 1 and 3).
The rst logistic regression model exclusively included
sampling site as an exposure variable (Table 4). In this case,
the variable autonomous region did not improve the model
goodness-of-t. The intake pit was the site with the highest
probability of Salmonella isolation, considering the odds of
isolated Salmonella in dust from the coolers of the feed mill
as the baseline (OR = 6.4; 95% CI: 2.715.1). The Hosmer

and Lemesshow statistic showed a good t of the model

(p = 0.875).
The second logistic regression model was built using
material used in feed production as the exposure variable. The variable autonomous region was not retained
in the nal equation. This model revealed that cotton
seeds had four times the odds of being contaminated with
Salmonella than soybean meal (OR = 3.8; 95% CI: 1.78.3). It
also showed that barley grain (OR = 0.1; 95% CI: 0.010.8)
and corn (OR = 0.1; 95% CI: 0.010.5) were less likely to
be contaminated than soybean meal (Table 5). The Hosmer and Lemesshow test indicated good t of the model
(p = 0.909).

Table 4
Frequency of Salmonella in the different feed mill locations investigated and output of the rst logistic regression model.
Feed mill facilities dust sampled

% Positive (N)

OR

95% CI

Cooler
Dust from collectors
Dust from gridding
Dust from elevators
Dust from the intake pit
Storage bin for nal product
Dust from pellet press
Dust from mill
Dust from corridors
Dust from packaging
Dust from mixer
Dust from scales

5.3 (131)
9.6 (83)
10.0 (10)
28.6 (7)
26.6 (128)
9.9 (43)
0.01 (6)
13.7 (73)
9.3 (43)
0.0 (15)
4.6 (43)
8.3 (12)

1.0
1.9

6.4
1.3
2.2
2.8
1.8

0.9

0.6

2.7
0.3
0.2
0.8
0.5

0.2

p-Value

5.4

15.1
5.4
20.2
7.7
6.5

4.3

0.237

<0.001
0.400
0.700
0.060
0.361

0.859

N = number of samples taken from each feed mill facilities.

Table 5
Frequency of Salmonella in the feed materials investigated and output of the second logistic regression model.
Feed materials

% Positive (N)

OR

95% CI

Soybean meal
Barley grain
Sunower meal
Fish meal
Corn
Wheat bran
Cotton seeds
Wheat our
Others

5.2 (497)
0.6 (177)
5.7 (35)
8.3 (24)
0.4 (247)
4.08 (147)
17.2 (58)
2.3 (86)
1.7 (288)

1.0
0.1
1.1
0.9
0.1
0.8
3.8
0.4
0.3

0.01
0.2
0.2
0.01
0.3
1.7
0.1
0.1

N = number of samples taken of each ingredient.

p-Value

0.8
4.8
3.8
0.5
1.9
8.3
1.8
1.1

0.026
0.902
0.844
0.011
0.574
0.001
0.258
0.221

G.J. Torres et al. / Preventive Veterinary Medicine 98 (2011) 8187

85

Table 6
Frequency of Salmonella in the two types of compound feed presentation investigated and output of the third logistic regression model.
Exposure variable

% Positive (N)

Autonomous regionsa
Andaluca
Aragn
Castilla-La Mancha
Castilla y Len

Cataluna
Galicia
Murcia
Navarra
Pas Vasco
Valencia

3.5 (143)
4.8 (165)
3.8 (185)
0.6 (159)
4.9 (327)
3.0 (135)
6.0 (83)
3.7 (54)
2.0 (51)
5.3 (114)

Presentation
Pellets
Non pelleted

0.6 (476)
4.31 (1.182)

B (SE)

OR

95% CI

p-Value

0
0.1 (0.589)
0.3 (0,604)
1.9 (1.105)
0.1 (0.529)
0.7 (0.689)
0.2 (0.656)
0.1 (0.860)
1.1 (1.112)
0.1 (0.652)

1.0
1.1
0.7
0.1
1.1
0.5
1.2
0.9
0.3
1.0

0.3
0.2
0.01
0.4
1.1
0.3
0.2
0.04
0.3

3.4
2,3
1.2
3.1
2.0
4.2
4.5
3.0
3.7

0.903
0.559
0.075
0.878
0.327
0.803
0.862
0.331
0.952

0 (0)
2.1 (0.601)

8.2

2.5

26.6

<0.001

N = number of samples taken for each category.

a
There were seven autonomous regions (Asturias, Baleares, Canarias, Cantabrra, Extremadura, La Rioja and Madrid; N = 242) where not positive samples
were found.

The third model included the autonomous regions as

a confounder factor and the variable presentation of the
compound feed (pellets versus non pelleted) as an exposure variable. Non pelleted compound feed had 8 times the
odds of being contaminated as pelleted compound feed
(OR = 8.2; 95% CI: 2.526.6) (Table 6). The Hosmer and
Lemesshow test indicated that the model tted the data
adequately (p = 0.869).
4. Discussion
Food safety is essential to preserve public health. Implementation of a holistic approach from the farm to the fork
throughout the food chain is necessary to guarantee that
food intended for human consumption is safe (EFSA, 2006).
Salmonella contaminated feed is an important source of
bacteria for the food chain (EFSA, 2008). However, there
is controversy about the role that feed mills play in the
epidemiology of Salmonella (Usera et al., 2001; Hggblom,
personal communication).
To our knowledge, this study is the rst large-scale survey to estimate the prevalence of Salmonella in feed mills in
Spain. The sampling method used, together with the large
number of samples investigated provided high sensitivity
in the detection of Salmonella contamination in feed mills.
In this study one quarter of the investigated mills produced at least one positive sample. When only compound
feed samples were used to estimate prevalence, the percentage of positive feed mills was much lower. Similar
results were obtained in previous studies (Whyte et al.,
2002; EFSA, 2008). Other authors also pointed out that the
risk of contamination of the compound feed is low at the
mill level and only becomes relevant once the feed arrives
at the farm (Hernndiz, personal communication).
The level of Salmonella contamination and the serovars
cultured from vegetable based feed materials were similar
to those published in the literature (EFSA, 2008). However, the prevalence in these protein sources was lower
than the results obtained in studies carried out in Sweden
and Norway, where contamination levels of up to 30%
were found (Weirup, 2006). Salmonella serovars with pub-

lic health signicance, according to the EU regulation (EFSA,

2009a), were observed in a small proportion of the samples
investigated, indicating the limited potential signicance
for public health.
Salmonella ser. Enteritidis, the dominant serovar in Spanish laying hen farms and in the Spanish outbreaks of human
salmonellosis (EFSA, 2007), was only isolated in one dust
sample from the feed mill facilities. This sample was taken
from the pit intake area of a home-mixer mill which was
located on a Salmonella ser. Enteritidis positive laying hen
farm. Similarly, Salmonella ser. Typhimurium was isolated
from the intake pit of in a home-mixer mill integrated in a
Salmonella-positive pig farm. Feed mill surroundings might
be an important source of Salmonella and feed mills should
not be located close to other agricultural activities (EFSA,
2008).
There is scarce information about the risk of introducing Salmonella to livestock as a result of home-produced
compound feed (EFSA, 2008). In our study, most of the
feed-mills that produced less than 1000 tons/month were
on-farm facilities but the statistical analysis failed to identify associations with Salmonella contamination. Season of
sampling could not be statistically associated with the presence of Salmonella in feed, despite human salmonellosis
outbreaks being strongly linked to summer months (Kinde
et al., 2005).
The autonomous region where the feed mill was located
was considered as a confounder variable in the multivariable analysis of the three sub-datasets. The differences
observed in the univariable analysis related to presence of
Salmonella might be explained by technological differences
of the feed industry among regions. Nevertheless this variable was nally retained only in the equation that modeled
the compound feed data subset.
The type of sample was a key issue when investigating the presence of Salmonella. The frequency of nding
Salmonella was higher in samples taken from the dust of
the feed mill facilities than those taken from feed materials or from compound feed. This might be due to the test
being more sensitive when the samples include dust and
ne particles (EFSA, 2008). Similar results were obtained by

86

G.J. Torres et al. / Preventive Veterinary Medicine 98 (2011) 8187

the Danish Surveillance Services on feed mills (Brondsted

and Rask, 2008). These ndings highlight the importance
of monitoring plans for cleaning and sanitation systems
in the framework of rigorous HACCP program (EFSA,
2008).
The intake pit was the sampling site with an increased
chance of isolating Salmonella compared to the other
sampling sites investigated. It is a place where dust accumulation is easily found and also a very attractive place for
vectors such as insects, rodents, or birds (EFSA, 2008). We
were not able to investigate the inuence of pest control
in Salmonella contamination due to the lack of variation
in our study population. Further studies should consider
the relationship between the effectiveness of pest control
programs and presence of Salmonella in the intake pit area.
Several studies have identied soybean meal as the feed
material with the highest risk of being contaminated with
Salmonella (EFSA, 2008). This study investigated a large
number of feed material samples and cotton seeds were
identied as the feed material with the highest probability
of being Salmonella contaminated. Hygroscopic properties
of cotton seeds which increase water content may explain
this result as water content favours Salmonella growth. It
is also possible that rodents were more attracted to cotton seeds than to other feed materials and therefore there
was a higher chance of Salmonella contamination due to the
presence of rodents.
Susceptibility of animals to Salmonella varies according
to the animal species and the Salmonella serovars involved
(EFSA, 2008). In the ofcial feed control programs the EFSA
advises that the intended species to which the compound
feed will be fed must be considered (EFSA, 2008). The
results from our study indicated that the intended livestock
species was not associated with the presence of Salmonella.
These results might conrm that hygiene standards implemented in the Spanish feed mills do not vary depending
on the livestock species of destination. However, the
high number of samples with no information regarding the intended livestock destination could bias these
results.
Pelleting has been recommended as an effective method
for feed decontamination (Hacking et al., 1978; McCapes
et al., 1989; EFSA, 2008). Temperatures of 80 C will
be sufcient to eliminate Salmonella from compound
feed (Blankenship et al., 1984) except when there is an
extremely bacterial high load (EFSA, 2008). Our results
were consistent with the opinion that pelleting is able to
reduce Salmonella contamination in compound feed. However, the efcacy of pelleting as a treatment to decrease the
spread of Salmonella at farm level is questioned by some
authors. Pelleted feed has been identied as the main risk
factor associated with faecal shedding of Salmonella in the
Spanish pig farms (Garca et al., 2009).
A critical issue that remains unsolved is the extent to
which Salmonella contaminated feed at the feed mill level
can contribute to animal salmonellosis. Our data show
that the most prevalent serovars found in animals and in
humans in Spain (Echeita et al., 2007; EFSA, 2009b) were
seldom found in the samples investigated in the present
study, and therefore, it is unlikely that feed mills could act
as a relevant source of Salmonella contamination.

5. Conclusion
Our study shows that Salmonella is frequently isolated
from the Spanish feed mills but it is less likely to be isolated
from compound feed at the mill level. Furthermore, the role
that feed mills play in the epidemiology of animal or human
salmonellosis may be limited, particularly considering the
serovars habitually found in animal or human outbreaks.
Salmonella control measures at the feed mill level should
consider all stages of feed production. According to the
ndings of this study, to achieve a more sensitive and costeffective control measure a risk-based approach must be
adopted.
Acknowledgements
Data have been obtained from the Spanish Salmonellain-feed Research Program coordinated by the Spanish
Ministry of the Environment and Rural and Marine Affairs.
The authors are grateful to the administrative ofcials veterinarian of the Spanish Autonomous Regions for their
contribution to this work and, particularly, to those ofcials
veterinarian who worked in the monitoring surveillance
scheme and those responsible for analysis of sampling
in the Regional Laboratories and in the National Reference Laboratory for Salmonella in animals and feed (Algete,
Madrid, Spain). They are also grateful to Prof. Juan Miguel
Rodrguez Gmez and Keith Hamilton for their assistance
with the review of the manuscript.
References
Altman, D., Machin, D., Bryant, T., Gardner, M., 2000. Statistic with Condence, 2nd ed. WileyBlackwell.
Blankenship, L.C., Shackfelford, D.A., Cox, N.A., Burdick, D., Dailey, J.S.,
1984. Survival of Salmonella as a function of poultry feed processing
conditions. In: International Symposium on Salmonella, New Orleans.
Brondsted, T., Rask, K., 2008. Control of Salmonella in feed in Denmark. The
Danish Plant Directorate.
Davies, R., Wray, C., 1996. Persistence of Salmonella Enteritidis in poultry
units and poultry food. British Poultry Science 37, 589596.
Davies, R., Breslin, M., Corry, J., Hudson, W., Allen, V., 2001. Observations
on the distribution and control of Salmonella species in two integrated
broiler companies. Veterinary Record 149, 227232.
Durand, A., Giesecke, W., Barnard, M., van Der Walt, M., Steyn, H., 1990.
Salmonella isolated from feeds and feed ingredients during the period
19821988: animal and public health implications. Journal of Veterinary Research 57, 175181.
A., de la Fuente, M., Gonzlez-Sanz, R., Dez, R.,
Echeita, M., Aladuena,
Arroyo, M., Cerdn, F., Gutirrez, R., Herrera, S., 2007. Anlisis de las
cepas de Salmonella aisladas de muestras clnicas de origen humano
Anos

en Espana.
2004 y 2005 (I). Boletn epidemiolgico 15 (13),
145156.
EFSA, 2006. The community summary report on trends and sources of
zoonoses, zoonotic agents, antimicrobial resistance and foodborne
outbreaks in the European Union in 2005. The EFSA Journal, 96.
EFSA, 2007. Report of the task force on zoonoses data collection on the
analysis of the baseline study on the prevalence of Salmonella in holdings of laying hen ocks of Gallus gallus. The EFSA Journal, 97.
EFSA, 2008. Microbiological risk assessment in feedingstuff for foodproducing animals. The EFSA Journal, 720.
EFSA, 2009a. Scientic opinion of the panel on biological hazards on a
request from European Commission on quantitative estimation of the
impact of setting a new target for the reduction of Salmonella in breeding hens of Gallus gallus. The EFSA Journal, 1036.
EFSA, 2009b. Trends and sources of zoonoses and zoonotic agents in the
European Union in 2007. The EFSA Journal, 223.
Garca, C., Carvajal, A., Collazos, A., Rubio, P., 2009. Herd-level risk factors
for faecal shedding of Salmonella enterica in Spanish fattening pigs.
Preventive Veterinary Medicine 91, 130136.

G.J. Torres et al. / Preventive Veterinary Medicine 98 (2011) 8187

Grimont, P., Weill, F., 2007. Antigenic Formulae of the Salmonella Serovars,
9th ed. WHO Collaborating Centre for Reference and Research on
Salmonella.
Hacking, W., Mitchell, W., Carlson, H., 1978. Salmonella investigation in
an Ontario feed mill. Canadian Journal Compendium Medicine 42,
400406.
Herikstad, H., Motarjemi, Y., Tauxe, R., 2002. Salmonella surveillance: a
global survey of public health serotyping. Epidemiology and Infection
129, 18.
Kinde, H., Castellan, D.M., Kerr, D., Campbell, J., Breitmeyer, R.,
Ardans, A., 2005. Longitudinal monitoring of two commercial layer ocks and their environments for Salmonella enterica
serovar Enteritidis and other Salmonellae. Avian Diseases 49, 189
194.
Kirkwood, B.R., Sterne, J.A., 2003. Medical Statistics, 2nd ed. Blackwell
Science Ltd.
Koyuncu, S., Haggblom, P., 2009. A comparative study of cultural methods for the detection of Salmonella in feed and feed ingredients. BMC
Veterinary Research 5, 6.

87

Legan, J., Vandeven, M., Dahms, S., Cole, M., 2001. Determining the concentration of microorganisms controlled by attributed sampling plans.
Food Control 12, 137147.
Maciorowski, K., Herrera, P., Kundinger, M., Ricke, S., 2006. Animal feed
production and contamination by foodborne Salmonella. Journal of
Consumer Protection and Food Safety 1, 1972009.
McCapes, R., Ekperigin, H., Cameron, W., Richie, W., Slagter, J., Stangeland,
V., Nagaraja, K., 1989. Effect of a new pelleting process on the level
of contamination of poultry mash by Escherichia coli and Salmonella.
Avian Diseases 33, 103111.
A., Daz, R., de la Fuente, M., Cerdn, P., Gutirrez,
Usera, M., Aladuena,
R., Echeita, A., 2001. Anlisis de las cepas de Salmonella spp. aisladas
en el ano
2000. Boletn
de muestras de origen no humano en Espana
epidemiolgico 9, 281.
Weirup, M., 2006. Salmonella Contamination of Feed an assessment on
behalf of Swedish Board of Abricurrtury of risks in Sweden.
Whyte, P., McGill, K., Collins, J.D., 2002. A survey of the prevalence of
Salmonella and other enteric pathogens in a commercial poultry feed
mill. Journal of Food Safety 23, 1324.