Modeling transfer of Escherichia coli O157:H7 and Staphylococcus aureus during slicing of a cooked meat product.
ABSTRACT Cross contamination is one of the most important contributing factors in foodborne illnesses originating in household environments. The objective of this research was to determine the transfer coefficients between a contaminated domestic slicing machine and a cooked meat product, during slicing. The microorganisms tested were Staphylococcus aureus (Gram positive) and Escherichia coli O157:H7 (Gram negative). The results showed that both microorganisms were able to transfer to all slices examined (20 successively sliced) and at different inoculum levels on the blade (10(8), 10(6) and 10(4)cfu/blade). The results also showed that the number of log cfu transferred per slice, during slicing, decreased logarithmically for both microorganisms at inoculum levels of 8 and 6log cfu/blade. The type of microorganism significantly influenced transfer coefficients (p<0.05) and there was an interaction between inoculum level and transfer coefficient for S. aureus (p<0.05), but not E. coli O157:H7. Finally, to describe bacterial transfer during slicing, two models (log-linear and Weibull) were fitted to concentration on slice data for both microorganisms (at 6 and 8 log cfu/blade), obtaining a good fit to data (R(2)⩾0.73).
- [Show abstract] [Hide abstract]
ABSTRACT: During a 10-year inspection survey (2001-2010), a microbiological study of ready-to-eat (RTE) foods and ready-to-bake frozen pastries from 15 canteens of the university campus was undertaken to determine their microbiological quality. The cumulative study revealed that the aerobic colony counts for the RTE product groups were as follows: from 10(6) to 10(8) CFU/g for 50% of sandwiches; under the detection limit (<10 CFU/g) for 88.6% of oven baked pastries; <10(5) CFU/g for 86.5% of desserts oven baked; from 10(3) to 10(9) CFU/g for desserts with dairy cream. The highest mean Enterobacteriaceae counts were recorded for desserts with dairy cream. The highest percentages of foodborne pathogens were: 20% Listeria monocytogenes and 12.5% Staphylococcus aureus in desserts with dairy cream; 17.5% Salmonella spp. and 8.5% presumptive Escherichia coli O157 in sandwiches; 14.6% Bacillus cereus in oven baked pastries. Aerobic colony counts were in the range 10(7)-10(8) CFU/g for 48.8% of frozen pastries; whereas Enterobacteriaceae counts between 10(3) and 10(4) CFU/g were detected in 35.3%. Foodborne pathogens prevalences for frozen pastries were as follows: B. cereus, 31.8%; Salmonella spp., 28.6%; presumptive E. coli O157, 25%; S. aureus, 8.7%; L. monocytogenes, 8.7%. Improved sanitary conditions in the processing plants and precautionary measures are necessary for consumer protection.Food Microbiology 06/2013; 34(2):337-43. · 3.41 Impact Factor
- [Show abstract] [Hide abstract]
ABSTRACT: The survival of Escherichia coli O157:H7 and Staphylococcus aureus on stainless steel surfaces with Saline Solution (SS), Tryptone Soy Broth (TSB) and meat purge was studied, and based on results, mathematical models describing survival of pathogens as a function of time were proposed. Results indicated that S. aureus was able to survive longer than E. coli O157:H7 in all substrates. The type of substrate had a greater impact on the survival of E. coli O157:H7. This microorganism only remained viable for 8 and 50 h (hours) on surfaces with SS and TSB, respectively while on meat purge, the microorganism could be recovered after 200 h. For S. aureus, SS and TSB led to similar survival times (250 h) whereas on meat purge, survival capacity increased to 800 h. Survival data for S. aureus could be well described by a log-linear model or a Weibull model depending on the type of substrate (R(2) > 0.85). E. coli O157:H7 displayed an evident tail zone which made the Weibull model more appropriate (R(2) > 0.94). These survival models may be used in quantitative risk assessment to produce more accurate risk estimates. Finally, the results highlight the importance of performing effective cleaning procedures to prevent cross contamination.Food Microbiology 04/2013; 33(2):197-204. · 3.41 Impact Factor
- [Show abstract] [Hide abstract]
ABSTRACT: Attachment of potential spoilage and pathogenic bacteria to food contact surfaces and the subsequent biofilm formation represent serious challenges to the meat industry, since these may lead to cross-contamination of the products, resulting in lowered-shelf life and transmission of diseases. In meat processing environments, microorganisms are sometimes associated to surfaces in complex multispecies communities, while bacterial interactions have been shown to play a key role in cell attachment and detachment from biofilms, as well as in the resistance of biofilm community members against antimicrobial treatments. Disinfection of food contact surfaces in such environments is a challenging task, aggravated by the great antimicrobial resistance of biofilm associated bacteria. In recent years, several alternative novel methods, such as essential oils and bacteriophages, have been successfully tested as an alternative means for the disinfection of microbial-contaminated food contact surfaces. In this review, all these aspects of biofilm formation in meat processing environments are discussed from a microbial meat-quality and safety perspective.Meat Science 05/2013; · 2.75 Impact Factor
Modeling transfer of Escherichia coli O157:H7 and
Staphylococcus aureus during slicing of a cooked meat product
F. Pe ´rez-Rodrı ´gueza,*, A. Valeroa, E.C.D. Toddb, E. Carrascoa,
R.M. Garcı ´a-Gimenoa, G. Zureraa
aDepartment of Food Science and Technology, University of Co ´rdoba, Campus de Rabanales, C-1, 14014 Co ´rdoba, Spain
bNational Food Safety and Toxicology Center, 165 Food Safety and Toxicology Building, Michigan State University, East Lansing, MI 48824-1314, USA
Received 6 October 2006; received in revised form 12 February 2007; accepted 12 February 2007
Cross contamination is one of the most important contributing factors in foodborne illnesses originating in household environments.
The objective of this research was to determine the transfer coefficients between a contaminated domestic slicing machine and a cooked
meat product, during slicing. The microorganisms tested were Staphylococcus aureus (Gram positive) and Escherichia coli O157:H7 (Gram
negative). The results showed that both microorganisms were able to transfer to all slices examined (20 successively sliced) and at different
inoculum levels on the blade (108, 106and 104cfu/blade). The results also showed that the number of log cfu transferred per slice, during
slicing, decreased logarithmically for both microorganisms at inoculum levels of 8 and 6 log cfu/blade. The type of microorganism signif-
icantly influenced transfer coefficients (p < 0.05) and there was an interaction between inoculum level and transfer coefficient for S. aureus
(p < 0.05), but not E. coli O157:H7. Finally, to describe bacterial transfer during slicing, two models (log-linear and Weibull) were fitted to
concentration on slice data for both microorganisms (at 6 and 8 log cfu/blade), obtaining a good fit to data (R2P 0.73).
? 2007 Elsevier Ltd. All rights reserved.
Keywords: Quantitative risk assessment; Cross contamination; Predictive microbiology; Semi-logarithmic model; Weibull model; Transfer coefficient
In Spain, from 1993 to 1998, 49% of foodborne out-
breaks occurred in the household environment, Salmonella
being the main pathogen (?60%) followed by Staphylococ-
cus spp. (BgVV-FAO/WHO, 2000; Hernandez et al., 1998).
These data showed that consumers often failed to store,
handle and prepare food in a hygienic and safe manner.
Moreover, studies such as those of Gorman, Bloomfield,
and Adley (2002), Mattick et al. (2003) and Bloomfield
(2003) have demonstrated the ability of pathogens such
as Staphylococcus aureus, Escherichia coli, Salmonella and
Campylobacter to be transferred from contaminated food
to hand and food contact surfaces in the kitchen. Thus,
Beumer and Kusumaningrum (2003) and Bloomfield and
Scott (1997) stated that cross contamination in the kitchen
environment can be a significant factor in the household
Reij, Den Aantrekker, and ILSI Europe Risk Analysis
in Microbiology Task Force (2004) reviewed scientific
literature on recontamination routes and sources (e.g.
raw materials, food contact surfaces, food handlers, etc.)
and demonstrated their relevance to foodborne disease
outbreaks. They also suggested that such knowledge on
recontamination should be incorporated into exposure
assessments of quantitative microbiological risk assess-
ments (QMRA) to help in determining mitigation strategies
to reduce foodborne disease. Thus, the importance of per-
forming QMRA (Codex Alimentarius Commission, 1999)
has led to researchers to try and quantify cross contamina-
tion events. For instance, Zhao, Zhao, Doyle, Rubino, and
Meng (1998), Chen, Jackson, Chea, and Schaffner (2001),
0309-1740/$ - see front matter ? 2007 Elsevier Ltd. All rights reserved.
*Corresponding author. Fax: +34 957 212000.
E-mail address: email@example.com (F. Pe ´rez-Rodrı ´guez).
Meat Science 76 (2007) 692–699
Montville, Chen, and Schaffner (2001) determined the
transfer coefficient of Enterobacter aerogenes as a surrogate
microorganism for Salmonella through different handling
scenarios. The transfer coefficient is the proportion of cells
that is transferred between surfaces (food, equipments,
tools, hands, etc.) under one or more operations. The study
provided probability distributions of transfer coefficients
associated with different handling tasks, and also demon-
strated that the microorganism could be transferred from
hands to food even with the use of gloves. In addition,
Kusumaningrum, Riboldi, Hazeleger, and Beumer (2002)
noted that the transfer ability of S. aureus, Campylobacter
jejuni and Salmonella enteritidis between cutting board and
foods at different pressure levels was not dependent on the
type of microorganism.
Recent outbreaks caused by E. coli O157:H7 associated
with cross contamination between raw meat and raw vege-
tables prompted Wachtel, Mcevoy, Luo, William Camp-
bell, and Solomon (2003) to quantify the transfer of
E. coli O157:H7 from raw contaminated meat to hands
and cutting surfaces, and then to iceberg lettuce by contam-
inated hands and cutting boards. They studied the effect of
successive contacts (cutting lettuce leaves with a contami-
nated knife) on whether or not and how much was trans-
ferred. The result revealed the random nature of cross
contamination events when they used low inoculum. This
fact was also pointed out by Vorst, Todd, and Ryser
(2006) when slicing deli meats with a blade inoculated with
103cfu of Listeria monocytogenes. Also, in previous stud-
ies, we simulated the transfer of S. aureus and L. monocyt-
ogenes to a cooked meat product through a slicing
machine; the results indicated that bacteria were trans-
ferred logarithmically to slices during slicing (Pe ´rez-Rodrı ´-
McMasters, & Ryser, 2004).
In the present work, we determined the transfer coeffi-
cients at different inoculum levels of two pathogens, one
Gram positive (S. aureus) and the other Gram negative
(E. coli O157:H7), between a contaminated slicing machine
and a cooked meat product. Finally, two mathematical
models were fitted to concentration on slice data to enable
its inclusion in QMRA.
Todd, Pe ´rez-Rodrı ´guez,
2. Materials and methods
2.1. Inoculum preparation
Five strains of S. aureus (CCM 1484, ATCC 13565,
CCTM La 2812, ATCC 19095, ATCC 23235) and four
of E. coli O157:H7 (CCUG 20570, ATCC 35150, ATCC
43894, ATCC 43895) were obtained from the Spanish Type
Culture Collection (CECT). All strains were maintained at
?18 ?C in cryovials containing beads and cryopreservative
(Microbank?). Three days before the experiment, a bead
of each strain was transferred to a tube containing 10 ml
of Tryptone Soya Broth (TSB, Oxoid, UK) and incubated
at 37 ?C for 24 h. Then, 1 ml of the initial subculture was
pipetted into a tube containing 10 ml of TSB, and incu-
bated at 37 ?C for 24 h. Finally, a third subculture was
obtained in the same way, incubating 1 ml of inoculum in
a flask of 100 ml of TSB until the early stationary phase
was reached (16 and 18 h for E. coli O157:H7 and S. aur-
eus, respectively). Cocktails of strains for each pathogen
were generated by mixing 10-ml aliquots containing similar
numbers of each strain, diluted in TSB to approximately
104, 106and 108cfu/ml.
2.2. Slicing machine and artificial contamination
Before use, a domestic slicer and a polished stainless
steel blade (Demoka?, M-381 Zeta Plus Ø) were disinfected
with 70% (v/v) ethanol for 10 min. Then, both were washed
with hot water with anionic-active detergent and rinsed
with distilled water. The slicing machine surfaces and the
blade were sprayed with 70% (v/v) ethanol which was
allowed to evaporate prior to any contamination step.
The blade was inoculated at the different concentrations
(104, 106and 108cfu) with a pipette containing 0.5 ml of
TSB of the appropriate cocktail. The TSB medium was
used to simulate soiled conditions on the blade (Moore,
Sheldon, & Jaykus, 2003). The inoculum was allowed to
dry in a laminar flow cabinet for 50 min at 25 ?C and
50% relative humidity.
2.3. Transfer of S. aureus and E. coli O157:H7 from
contaminated blade at different inoculum levels via the slicer
to uninoculated cooked meat product by slicing
Large pieces of pork were used to simulate a delicatessen
meat product that could be contaminated with the inocu-
lated blade. Each piece of pork was trimmed to be uniform
so that all slices had a similar area and weight, and have a
plain and uniform surface for slicing. After drying, the inoc-
ulated blade was placed in the slicer, and then a pork piece
was sliced up to 21 times to yield slices (7–10 g each). Each
slice was picked up as it came off the machine using a pair of
sterilized pincers and placed in sterile stomacher bags.
Three repetitions for each inoculum level were performed.
The pressure applied on the pork piece was assumed to be
similar to that occurring on household slicing.
2.4. Estimation of the force or pressure applied during slicing
A sensor film provided by Sensor Products Inc was uti-
lized to estimate the force level applied during slicing. The
type of sensor film was Ultra Low Pressurex?, intended for
a pressure range of ?2–6 kg/cm2. The sensor film was
placed between the pork piece and the guard area of the sli-
cer. Next, the pork piece was pushed against the blade sim-
ulating a normal slicing process in a retail delicatessen.
Then, the sensor film was removed and prepared according
to manufacture’s instructions to be sent for exhaustive
image analysis; this analysis revealed the pressure distribu-
tion applied to the pork surface.
F. Pe ´rez-Rodrı ´guez et al. / Meat Science 76 (2007) 692–699
2.5. Microbiological analysis
2.5.1. Efficacy of the disinfection process
To prove the effectiveness of the disinfection process,
Plate Count Agar (PCA, Oxoid, UK), MacConkey Agar
(MCA, Oxoid, UK) and Baird Parker agar/medium (BP,
Oxoid, UK) with tellurite egg yolk supplement (Oxoid,
UK) Rodac plates were pressed on three different zones
in the slicing machine, including the blade, and incubated
at 30 ?C for PCA and 37 ?C for MCA and BP.
2.5.2. Chopped pork piece control
Before each experiment, two samples (10 g) of pork were
placed into sterile stomacher bags with peptone buffer 1%
and homogenized in a Stomacher (Seward Stomacher 400
Colworth House, England) for 2 min and tested for the
presence of S. aureus and E. coli O157:H7 using the enrich-
ment methods described below.
2.5.3. Quantification of transfer
To estimate the number of viable cells transferred during
the slicing process (up to 21 slices), slices were placed inde-
pendently into sterile stomacher bags and filled with 90 ml
of peptone buffer 1% and then homogenized in the Stom-
acher for 2 min. Serial decimal dilutions were prepared in
0.85% saline solution, cultured on agar plates and the col-
onies were enumerated. The selective medium and condi-
tions for S. aureus were BP with tellurite egg yolk
supplement incubated at 37 ?C for 48 h and for E. coli
O157:H7, MCA incubated at 37 ?C for 24 h was used. As
required, the samples giving no detectable counts by direct
plating were enriched by incubating the flasks containing
homogenized samples at 37 ?C for 24 h. In the case of the
S. aureus assay, the flasks were supplemented with NaCl
and sodium pyruvate to reach a concentration of 20%
and 1%, respectively. In the case of E. coli, the flasks were
supplemented with cefixime and vancomycin (Dodd et al.,
2003). Then, the enriched samples were examined for pres-
ence/absence by spreading 10 ll of enriched samples on BP
tellurite egg yolk in the case of S. aureus and Sorbitol Mac-
Conkey with potasium tellurite and cefixime for E. coli
O157:H7 (SMAC, Oxoid, UK) (37 ?C for 24 h). Any
growth on the SMAC plates which had shown characteris-
tic colonies of E. coli O157:H7 was considered to have been
transferred from the blade to the slices.
2.6. Data and statistical analysis
The detection limit of the microbiological analysis var-
ied depending on the slice weight, but was on average
around 30 cfu/slice for both pathogens. For data analysis,
we used this limit on those slices in which counting was not
possible, but were positive in the enrichment analysis.
The concentration of pathogen on the slices was
expressed per cm2of slice and logarithmically transformed
(log cfu/cm2) in Excel (Microsoft Corporation) spread-
sheet. Two different mathematical models were fitted to
the data to describe the concentration variation on the slice
as a function of the number of the slice taken (Nslice=
In this study, transfer coefficient is the logarithm of the
proportion (%) of bacterial cells transferred from the blade
to each slice (expressed per cm2) (1). The transfer coeffi-
cients were tested by analysis of variance (ANOVA), and
Duncan’s multiple-range tests using SPSS 12.0 software
(SPSS Inc., NC) for the factors, type of microorganism
and inoculum level on the blade
Tr ð%Þ ¼ log
where Tr (%) is the transfer coefficient; cfu=cm2
concentration on each slice; and cfubladeis the initial con-
centration on the blade.
2.7. Mathematical models
A log-linear model (log(B) = log(A) ? k Æ N) has been
used in previous studies to describe attachment strength
of bacteria on surfaces (ease of removal) (Midelet & Car-
pentier, 2002; Veulemans, Jacqmain, & Jacqmain, 1970).
The model assumes first-order kinetics. Such a model was
fitted to experimental data consisting of the number of col-
onies transferred to successive plates by contact or swab-
bing (Eginton, Gibson, Holah, Handley, & Gilbert, 1995;
Eginton, Holah, Allison, Handley, & Gilbert, 1998; Rich-
ard & Piton, 1986). The steeper the slope, the weaker the
microorganisms’ attachment strength. Based on this
approach, we propose to use the log-linear model (semi-
logarithmic model) described by Eq. (2) to determine the
bacterial concentration on the successive slices:
logðIsliceÞ ¼ logðIbladeÞ ? k ? Nslice=lnð10Þ
where Nsliceis the number of slice (Nslice= 1,2,...,20); Islice
is the concentration (log cfu/cm2) on the slice Nslice; Ibladeis
a regression parameter (the intercept); and k/ln (10) is a
regression parameter related to the slope through the
We decided to also use a non-linear model to model the
data. We chose the Weibull model because the underlying
principles of this model allows modeling of the processes
occurring during bacterial transfer between surfaces. Wei-
bull distributions are usually applied to objects with a large
number of links, each of which has a certain probability of
breaking. In many cases, when only one link breaks, the
object can experience failure (Olkin, Gleser, & Derman,
1980). For this reason, Weibull distributions are widely
used to model lifetime or failure time events (e.g. bulbs life-
time, fatigue of materials, etc.). Furthermore, Weibull
distributions are often used to model the time interval
between successive, random, independent events that occur
at a variable rate (Cullen & Frey, 1999; Vose, 2000). In pre-
dictive microbiology, Weibull models have been applied to
represent the logarithmic process of bacterial inactivation
F. Pe ´rez-Rodrı ´guez et al. / Meat Science 76 (2007) 692–699
(Ferna ´ndez, Collado, Cunha, Ocio, & Martı ´nez, 2002;
Peleg & Cole, 1998). In cross contamination events, the ini-
tial bacterial population on a surface includes a complex
net of interactions between the bacteria and the contact
substrate through attachment, in which each component
has a certain probability of breaking. To enable bacterial
transfer between surfaces it is necessary that these interac-
tions fail (attachment structures), and bacteria can travel
from one surface to another (Dickson, 1990). Therefore,
based on this hypothesis, contamination or recontamina-
tion between surfaces by contact could be described by
means of a probability curve of breaking probabilities,
i.e. the Weibull distributions. The other advantage of using
Weibull distributions, compared to others such as exponen-
tial or gamma distributions, is its high flexibility to fit to
many types of data (Vose, 2000). The Weibull model used
to fit the data is shown in Eq. (3), which has two parame-
ters, a and b. The parameter a is considered as a reaction
rate constant and b as a behaviour index. This model
reduces to a linear model when b = 1
logðIsliceÞ ¼ logðIbladeÞ ? ðNslice=aÞb
3. Results and discussion
The analysis of the control samples performed on the
pork pieces confirmed the absence of E. coli O157:H7
and S. aureus. The first slice of every assay was discarded
since the contaminated side of the blade did not come into
The image analysis obtained by Ultra Low Pressurex?
showed that pressure was variable along the slice surface
(Fig. 1). The highest pressure was around 6.75 kg/cm2,
but more typical values showed pressure on the slices
around 2.50 kg/cm2.
it wastherefore not
3.1. Low transfer ability
The summary of the experimental data are shown in
Table 1. The transfer values expressed as Tr (%) ranged
between 1.11 and – 4.34 log, values corresponding to S.
aureus at 8 log cfu/blade and E. coli O157:H7 at 6 log
cfu/blade, respectively. The magnitude of transfer data
was much lower than the levels found by Kusumaningrum
et al. (2002), who showed that 25–100% (expressed arith-
metically and per cm2) of S. aureus, Campylobacter jejuni
and S. Enteritidis on stainless steel boards was transferred
to roast chicken and cucumber. Similarly, Scott and
Bloomfield (1990) reported transfer coefficients of approx-
imately 20% and 21% for E. coli and S. aureus, respectively,
from contaminated laminate surfaces to stainless steel bowl
with low initial inoculum levels. This is consistent with the
findings of Moore et al. (2003) who demonstrated that the
mean transfer coefficients from contaminated stainless steel
to lettuce ranged from 2.3% to 66% for Salmonella and
from 16.3% to 38.4% for Campylobacter. In contrast, lower
transfer coefficients generally occur with non-uniform or
porous surfaces such as faucets, sponges and hands. An
example is the transfer of only 0.47% E. coli from a damp
cloth (inoculated with 107cfu/cm2) to the human hand
(Mackintosh & Hoffman, 1984). Similar transfer levels
were reported by Chen et al. (2001) for E. aerogenes trans-
fer between chicken, hands and spigots and by Wachtel
et al. (2003) for E. coli O157:H7 through fingers. Rusin,
Maxwell, and Gerba (2002) reported transfer coefficients
between 28% and 65% for non-porous surfaces such as fau-
cets or phone receivers with a high initial inoculum level of
108cfu/fomite, but <0.01% when porous surfaces were
used, e.g. sponge, dishcloth, laundry. This is understand-
able since most of the inoculum is no longer in contact with
Fig. 1. Distribution of pressures along the slice area observed during
simulation of slicing obtained by using Ultra Low Pressurex?.
Summary of the transfer data for S. aureus and E. coli O157:H7 from an inoculated slicing machine blade to pork pieces at inoculum levels of 8, 6 and 4 log
Inoculum (log cfu/blade)
Staphylococcus aureusEscherichia coli O157:H7
1.76 ± 0.07a
1.37 ± 0.27a
1.65 ± 0.32a
Tr (%) TrT(%)
0.62 ± 0.67a
0.60 ± 0.90a
5.34 ± 0.58
2.78 ± 0.73
0.99 ± 0.73
?1.25 ± 0.58a
?1.62 ± 0.79a,b
1.48 ± 0.77b
3.43 ± 0.83
1.41 ± 0.90
?2.77 ± 0.90a
?2.67 ± 0.95a
± indicates the standard deviation.
aandbindices show the different homogenous groups reported by Duncan’s test.
ATr (%) = log[(cfusliceper cm2/cfu initial on blade) · 100].
BTrT(%) = log[(cfu transferred in 20 slices/cfu initial on blade) · 100].
F. Pe ´rez-Rodrı ´guez et al. / Meat Science 76 (2007) 692–699
the surface of the transfer material. Even though our data
indicated low transfer rates, this can be explained because a
slicing machine is different from the static contact experi-
ments in the other studies. Transfer during slicing involves
horizontal movement between surfaces when cooked meat
surface runs over the blade, and at the same time the latter
revolves rapidly. Thus, food particles with the inoculum
would tend to be spun off during the slicing process onto
other parts of the slicing machine and also into the environ-
ment. This is what Vorst et al. (2006) found for Listeria
monocytogenes in deli meats on a similar slicing machine.
3.2. Logarithmic decrease during slicing
In Fig. 2, the transfer results for S. aureus expressed as
cells transferred per slice (log cfu/cm2) are represented
against number of slice. At inoculum levels of 8 and 6 log
cfu/blade, cells transferred per slice decreased logarithmi-
cally from the first documented slice to the last slice in
increments of 1.68 and 1.74 log cfu/cm2, respectively. In
the case of 4 log cfu/blade, the results showed a high degree
of variability as shown in Fig. 2, where bars representing
the standard deviation were larger. Moreover, transferred
cells per slice did not seem to follow a clear trend, probably
due to the low precision of the microbiological method at
low concentrations. Nevertheless, cells transferred per slice
decreased logarithmically from the first documented slice
to the last slice at 0.80 log cfu/cm2which might be consid-
ered within the experimental error range (Mossel, Corry,
Struijk, & Baird, 2005).
The results for E. coli O157:H7 in Fig. 3 show that for 8
and 6 log cfu/blade, the cells transferred per slice decreased
logarithmically from the first documented slice to the last
slice at 1.44 and 2.03 log cfu/cm2, respectively, showing a
linear-logarithmic trend as in the case of S. aureus. For
the lowest inoculum of 4 log cfu/blade, no counts were
obtained in any of the slices in the three repetitions. How-
ever, the analysis by enrichment demonstrated that the 20
slices were contaminated by E. coli O157:H7.
The highest quantity of cells transferred per slice did not
always correspond to the first slice. For example, at 6 log
cfu/blade, although the first slice had 3.71 ± 0.27 log cfu/
cm2, the highest concentration was marginally higher for
the second and third slices, 3.87 ± 0.26 and 3.80 ±
0.41 log cfu/cm2, respectively. This same observation can
be observed for E. coli O157:H7 at 8 log cfu/blade, in
which the first and second slice had 3.97 ± 1.10 and
4.66 ± 0.72 log cfu/cm2, respectively. These facts can be
seen visually in Figs. 2 and 3.
3.3. Influence of microorganism type and inoculum size on
The analysis of variance of the transfer coefficients (Tr
(%)) at 8 and 6 log cfu/blade, according to Eq. (1), showed
(p = 0.014). Also, the analysis of variance of the TrT(%)
in 20 slices (log [cfu transferred in 20 slices/cfu initially
on blade] · 100) at 8 and 6 log cfu/blade revealed signifi-
(p = 0.003). A possible explanation for this is the different
susceptibility of both microorganisms to stressful environ-
mental conditions, which is higher for E. coli O157:H7.
Kusumaningrum et al. (2002) reported no significant differ-
ences in transfer coefficients among their inoculated strains
(S. aureus, C. jejuni and S. Enteritidis) when roasted
chicken and cucumber were contaminated under pressure.
However, without pressure, the analysis of variance did
indicate significant differences for the three organisms. In
this scenario, in addition to the absence of pressure, the
shorter drying time prior to contact (615 min), and rela-
tively high resistance under stressful conditions for the test
microorganisms (the most susceptible microorganism to
slow-air-drying, C. jejuni, was detectable on contaminated
surface until 4 h) could explain the divergence in results
between the studies. Moreover, Eginton et al. (1995) did
not find significant differences in the degree of attachment
to surfaces between the microorganisms, which may affect
their transfer ability.
In respect to the inoculum level effect, the analysis of
variance and Duncan’s test performed on the first six slices
(with the highest transfer coefficients) showed significant
05 10 15 20
Fig. 2. Transfer (log cfu/cm2) of S. aureus from an inoculated slicer blade
with 8 (r), 6 (m) and 4 (j) log cfu/blade. Data reported are means ±
standard deviation for three replicates. The log-linear (----) model and
Weibull (—-) models fitted to transfer data at 8 and 6 log cfu/blade.
0510 15 20
Fig. 3. Transfer (log cfu/cm2) of E. coli O157:H7 from an inoculated slicer
blade at 8 (r) and 6 (m) log cfu/blade. Data reported are means ± stan-
dard deviation for three replicates. The log-linear (----) model and
Weibull (—-) models are fitted to transfer data at 8 and 6 log cfu/blade.
F. Pe ´rez-Rodrı ´guez et al. / Meat Science 76 (2007) 692–699
differences for S. aureus (p = 0.027) but not for E. coli
O157:H7 (p = 0.894). The analysis of variance including
20 slices gave the same result. Thus, for S. aureus, there
was significant difference (p = 0.034) and Duncan’s test
indicated the existence of two homogeneous groups, the
first one formed by the transfer data of 4 and 6 log cfu/
blade and the second one formed by 6 and 8 log cfu/blade
Other variance analysis performed by comparing the
TrT(%) indicated no significant difference between inocu-
lum levels for S. aureus and E. coli O157:H7 (p = 0.327
and 0.982, respectively) (Table 2). Therefore, it can be
concluded, for S. aureus, the 8 and 6 log cfu/blade inocu-
lum level group exerted a statistically different effect on
Tr (%) from the 6 and 4 log cfu/blade group. However,
in the case of E. coli O157:H7 it was clear that the inocu-
lum size did not influence the transfer capability of the
microorganism. These conclusions did not agree with those
reported by Montville and Schaffner (2003) who stated that
‘‘When the population of bacteria on the source surface
was high, the log 10% transfer was relatively low. Where
the population on the source surface was lower, the log
10% transfer tended to be higher’’. Our results are more
consistent with the findings of Kusumaningrum et al.
(2002) who did not find significant differences between dif-
ferent inoculum levels for S. aureus, C. jejuni and S. Ente-
ritidis, andof Dickson (1990),
L. monocytogenes and Salmonella typhimurium.
3.4. An example of modeling transfer data for the QMRA
Transfer data obtained for the levels 6 and 8 log cfu/
blade were used to fit the two mathematical models
described previously, which could be applied in QMRA
to describe cross contamination scenarios. We only used
the data at high inoculum levels for S. aureus and E. coli
(log 6 and 8 cfu/blade), because the transfer trends at
4 log cfu/blade were not obvious and much of the transfer
data were generated through enrichment. To perform the
model, we preferred to use, as transfer unit, the concentra-
tion transferred onto the slice (log cfu/cm2) since this could
be easily interpreted and used by modelers in QMRA. Nev-
ertheless, the models represented by Eqs. (2) and (3) could
be log-back-transformed easily resulting in Eqs. (4) and (5)
for the log-linear and the Weibull model, respectively:
Islice=Iblade¼ expð?k ? NsliceÞ
Eqs. (4) and (5) return the transfer coefficients measured as
a fraction of 1, but it could also be expressed as % cfu/cm2
if multiplied by 100.
Parameters and goodness-of-fit indices for the log-linear
and Weibull model are shown in Table 2, and the fitting
lines are plotted in Figs. 2 and 3. With regard to the log-lin-
ear model, S. aureus and E. coli presented a logarithmic
trend at 8 and 6 log cfu/blade (R2= 0.73–0.95), though
S. aureus showed the best fit as shown by the values of stan-
dard error of prediction (SEP) and R2(Table 2). These
findings agree with the results of Vorst et al. (2006) who
reported that with 8 log/blade, the number of L. monocyt-
ogenes cells transferred to ready-to-eat turkey and bologna
decreased logarithmically during slicing (R2> 0.92). At 8
and 6 log cfu/blade, the model showed higher slopes for
E. coli O157:H7 (k/ln(10) = 0.11 and 0.10, respectively)
than for S. aureus (k/ln(10) = 0.08 and 0.09, respectively).
Therefore, when using a log-linear model to describe the
transfer, these results would indicate that the logarithmic
decrease was slightly more rapid for E. coli O157:H7 at
8 log cfu/blade.
Parameters and goodness-of-fit indices for the Weibull
model are shown in Table 2 and the fitting lines are plotted
in Figs. 2 and 3. In comparison, the Weibull model pre-
sented a better fit than the log-linear for both microorgan-
isms (Table 2). In the case of E. coli at 8 log cfu/blade, the
Weibull model almost described a log linear model since
the value for the parameter b was practically 1. Hence, both
curves overlap as can be seen in Fig. 3. Though the good-
ness-of-fit indices showed little difference between the mod-
els, we believe that the Weibull model overall best models
the experimental data, especially where the first slices
showed much higher concentration than the rest (two
phases in the curve), e.g. at 6 log cfu/blade for E. coli
O157:H7 (Fig. 3). These values are crucial in risk, so a bet-
ter fit in this zone would lead to more accurate risk estima-
tion in QMRA. In this respect, the application of the
Regression parameters and goodness-of-fit indices for both semi-logarithmic and Weibull models obtained from transfer data of S. aureus and E. coli
O157:H7 at 8 and 6 log cfu/blade
log cfu/bladelog-linear model Weibull model
7 · 10?5
E. coli O157:H7
Ak is the slope of the semi-logarithmic model logðIsliceÞ ¼ logðIbladeÞ ? k ? Nslice=lnð10Þ.
BStandard error of prediction (SEP): SEP ¼100
Ca and b are parameters of the Weibull model.
F. Pe ´rez-Rodrı ´guez et al. / Meat Science 76 (2007) 692–699
Weibull model would be more relevant for S. aureus due to
the fact that production of enterotoxin is mainly associated
with relatively high concentrations of pathogen, i.e.
>105cfu/g (Anunciac ¸a ¨o, Linardi, do Carmo, & Bergdoll,
1995; Ro ¨rvik & Granum, 1996). However, for E. coli
O157:H7, it is well-recognized that the infectious dose is
well under 100 organisms (Tilden et al., 1996).
Firstly, the results show that all the pork slices could be
contaminated by both pathogens at the inocula of log 4, 6
and 8 log cfu/blade, even though the transfer was lower
than in other transfer scenarios. As expected, the first slices
had the highest contamination levels, and the risk of illness
would be higher after eating these compared to the subse-
quent slices in a product being sliced. This fact is even more
relevant in those pathogens that should be ingested at high
dose to produce illness, such as L. monocytogenes or S. aur-
eus (but not E. coli O157:H7). However, the risk of the lat-
ter slices causing illness would be increased if growth of the
pathogens were allowed during storage. This would be less
likely for S. aureus unless the slices were kept at ambient
temperatures for several hours to produce enterotoxin;
nevertheless this has happened with other foods in causing
mass casualities (Do Carmo et al., 2004). Therefore, the
stage in slicing where the cross contamination takes place
(i.e. first and last slices) could have a significant impact,
especially on individual risk. Other factors such as strain
persistence in the environment, disinfection practices, stor-
age temperature and times, should also be included in a
QMRA. Clearly, the pathogen type does affect Tr (%)
and TrT(%). Moreover, the inoculum level on the blade
influenced significantly the transfer of S. aureus, but not
E. coli O157:H7. An analysis of the most suitable model
for incorporating transfer data into QRMAs was not clear
since both log-linear and Weibull models fitted the data
describing the log-decrease of cells during slicing. However,
the Weibull model showed a better fit for the first few slices
where the risk of illness upon consumption is greatest.
Inoculum levels used here were relatively high compared
to the concentrations usually found in reality. Ideally, they
should be similar if transfer coefficients are influenced by
inoculum level. In this sense, future research should con-
sider the use of microbiological analysis with lower detec-
tion limits to better imitate the real world. Besides, other
factors like the type of medium used to inoculate the blade
or the contaminated area should be investigated as to their
effect on the transfer coefficient.
This work has been performed in the framework of the
collaboration between the National Food Safety and Tox-
icology Center at Michigan State University and the Uni-
versity of Co ´rdoba, and was partly financed by the
MCYT AGL2005-119 and the Research Group AGR-170
HIBRO of the Plan Andaluz de Investigacio ´n (PAI).
Anunciac ¸a ¨o, L. L., Linardi, W. R., do Carmo, L. S., & Bergdoll, M. S.
(1995). Production of Staphylococcal enterotoxin A in cream-filled
cake. International Journal of Food Microbiology, 26, 259–263.
Beumer, R. R., & Kusumaningrum, H. (2003). Kitchen hygiene in daily
life. International Biodeterioration and Biodegradation, 51(4), 299–302.
BgVV-FAO/WHO Collaborating Centre for Research and Training in
Food Hygiene and Zoonoses. (2000). WHO Surveillance Programme
for Control of Foodborne Infections and Intoxications in Europe:
Spain. Seventh Report 1993–1998. <http://www.bfr.bund.de/internet/
7threport/CRs/spa.pdf> Accessed 20.06.2006.
Bloomfield, S. F. (2003). Home hygiene: a risk approach. International
Journal of Hygiene Environmental Health, 206, 1–8.
Bloomfield, S. F., & Scott, E. (1997). Cross-contamination and infection
in the domestic environment and the role of chemical disinfectants.
Journal of Applied Microbiology, 83, 1–9.
Chen, Y., Jackson, K. M., Chea, F. P., & Schaffner, D. W. (2001).
Quantification and variability analysis of bacterial cross contamination
Codex Alimentarius Commission (CAC). (1999). Principles and Guide-
lines for the Conduct of Microbial Risk Assessment CAC/GL-30.
030e.pdf> Accessed 25.02.2006.
Cullen, A. C., & Frey, H. C. (1999). Probabilistic techniques in exposure
assessment. New York: Plenum Press, pp. 73–74.
Dickson, J. S. (1990). Transfer of Listeria monocytogenes and Salmonella
typhimurium between beef tissue surfaces. Journal of Food Protection,
De Sena, M. J., et al. (2004). A case study of a massive Staphylococcal
food poisoning incident. Foodborne Pathogens and Disease, 4, 241–246.
Dodd, C., Sanderson, M., Sargeant, J., Nagaraja, T., Oberst, R., Smith,
R., et al. (2003). Prevalence of Escherichia coli O157 in cattle feeds in
Midwestern feedlots. Applied of Environmental Microbiology, 69,
Eginton, P. J., Gibson, H., Holah, J., Handley, P., & Gilbert, P. (1995).
Quantification of the ease of removal of bacteria from surfaces. Journal
of Industrial Microbiology, 15, 305–310.
Eginton, P. J., Holah, J., Allison, D. G., Handley, P. S., & Gilbert, P.
(1998). Changes in the strength of attachment of micro-organisms to
surfaces following treatment with disinfectants and cleansing agents.
Letters of Applied Microbiology, 27, 101–105.
Ferna ´ndez, A., Collado, J., Cunha, L. M., Ocio, M. J., & Martı ´nez, A.
(2002). Empirical model builiding based on Weibull distribution to
describe the joint effect of pH and temperature on the thermal
resistance of Bacillus cereus in vegetable substrate. International
Journal of Food Microbiology, 77, 147–153.
Gorman, R., Bloomfield, S., & Adley, C. C. (2002). A study of cross-
contamination of food-borne pathogens in the domestic kitchen in the
Republic of Ireland. International Journal of Food Microbiology, 76,
Hernandez, G., Mangas, I., Mateo, S., de Liacer, A., Eiman, M., &
Sanchez-Serrano, L. P. (1998). Foodborne disease outbreaks in Spain,
1988–1995: impact of an intervention measure. Proceedings of the 4th
world congress foodborne infections and intoxications 7–12 June 1998
(Vol. 1, pp. 279–283). Berlin: Federal Institute for Health Protection of
Consumers and Veterinary Medicine.
Kusumaningrum, H. D., Riboldi, G., Hazeleger, W. C., & Beumer, R. R.
(2002). Survival of foodborne pathogens on stainless steel surfaces and
cross-contamination to foods. International Journal of Food Microbi-
ology, 85, 227–236.
Mackintosh, C. A., & Hoffman, P. N. (1984). An extended model of
transfer of microorganisms via the hands: differences between organ-
F. Pe ´rez-Rodrı ´guez et al. / Meat Science 76 (2007) 692–699
isms and the effect of alcohol disinfection. Journal of Hygiene,
Cambridge, 92, 345–355.
Mattick, K., Durham, K., Domingue, G., Jørgensen, F., Sen, M.,
Schaffner, D. W., et al. (2003). The survival of foodborne pathogens
during domestic washing-up and subsequent transfer onto washing-up
sponges, kitchen surfaces and food. International Journal of Food
Microbiology, 85, 213–226.
Midelet, G., & Carpentier, B. (2002). Transfer of microorganims,
including Listeria monocytogenes, from various materials to beef.
Applied and Environmental Microbiology, 68, 4015–4024.
Montville, R., & Schaffner, D. W. (2003). Inoculum size influences
bacterial cross contamination between surfaces. Applied of Environ-
mental Microbiology, 69, 7188–7193.
Montville, R., Chen, Y. H., & Schaffner, D. W. (2001). Glove barriers to
bacterial cross contamination between hands to food. Journal of Food
Protection, 64, 845–849.
Moore, C. H., Sheldon, B. W., & Jaykus, L. A. (2003). Transfer of
Salmonella and Campylobacter from stainless steel to romaine lettuce.
Journal of Food Protection, 66, 2231–2236.
Mossel, D. A. A., Corry, J. E. L., Struijk, C. B., & Baird, R. M. (2005).
Essentials of the microbiology of foods. Wiley, p. 241.
Olkin, I., Gleser, L. J., & Derman, C. (1980). Probability models and
applications (pp. 276–285). New York: Macmillan Publishing Co., Inc.
Peleg, M., & Cole, M. B. (1998). Reinterpretation of microbial survival
curves. Criteria Review Food Science, 38, 353–380.
Pe ´rez-Rodrı ´guez, F., Fuentes, J. M., Valero, A., Carrasco, E., Garcı ´a-
meat slicing. In Paper presented at the COST 920, Pamplona, Spain.
Reij, M. W., Den Aantrekker, E. D., & ILSI Europe Risk Analysis in
Microbiology Task Force. (2004). Recontamination as a source of
pathogens in processed foods. International Journal of Food Microbi-
ology, 91, 1-11.
Richard, J., & Piton, C. (1986). Semi-log model for interpreting the results
of swabbing surfaces naturally contaminated. Journal of Applied
Bacteriology, 60, 243–249.
Ro ¨rvik, L. M., & Granum, P. E. (1996). Staphylococcus aureus. In:
Smittsamme sjudkdommer fra mat, Granum PE (Red), Ho ¨yskoleforla-
get, Kristiansand S, Norge (pp. 166–175) (in Norwegian).
Rusin, P., Maxwell, S., & Gerba, C. (2002). Comparative surface-to-hand
and fingertip-to-mouth transfer efficiency of gram-positive bacteria,
gram-negative bacteria, and phage. Journal of Applied Bacteriology, 93,
Scott, E., & Bloomfield, S. F. (1990). The survival and transfer of
microbial contamination via cloths, hands, and utensils. Journal of
Applied Bacteriology, 68, 271–278.
Tilden, J. J., Young, W., McNamara, A. M., Custer, C., Boesel, B.,
Lambert-Fair, M. A., et al. (1996). A new route of transmission for
Escherichia coli: infection from dry fermented salami. American
Journal of Public Health, 86, 1142–1145.
Veulemans, A., Jacqmain, E., & Jacqmain, D. (1970). Etude d’une
me ´thode simple pour la de ´termination du degre ´ de pollution des
surfaces et la comparaison du pouvoir de ´sinfectant de divers produits
d’entretien. Revue des Fermentations et des Industries Alimentaires, 25,
Vorst, K. L., Todd, E. C., & Ryser, E. T. (2006). Transfer of Listeria
monocytogenes during mechanical slicing of turkey breast, bologna,
and salami. Journal of Food Protection, 69, 619–626.
Vorst, K. L., Todd, E. C., Pe ´rez-Rodrı ´guez, F., McMasters, R. L., &
Ryser, E.T. (2004). Transfer of Listeria monocytogenes from a
delicatessen slicer to ready-to-eat meat products. In Paper presented
at the IAFP’s 91st annual meeting, Phoenix, AZ.
Vose, D. (2000). Risk analysis: A quantitative guide. New York, NY:
Wiley, p. 326.
Wachtel, M. R., Mcevoy, J. L., Luo, Y., William Campbell, A. M., &
Solomon, M. B. (2003). Cross-contamination of lettuce by Escherichia
coli O157:H7 via contaminated ground beef. Journal of Food Protec-
tion, 66, 1176–1183.
Zhao, P., Zhao, T., Doyle, M. P., Rubino, J. M., & Meng, J. (1998).
Development of a model for evaluation of microbial cross-contami-
nation in the kitchen. Journal of Food Protection, 61, 960–963.
F. Pe ´rez-Rodrı ´guez et al. / Meat Science 76 (2007) 692–699