Food Protection Trends - November/December 2017 - 420

Two classes of methods are used for sampling carcasses,
destructive and non-destructive. The reference method of
sampling for microbiological criteria regarding total bacterial
counts and Enterobacteriaceae before chilling is a destructive
method (26). However, regulation EC No 2073 (2005)
states that food business operators may use other methods, if
they can demonstrate that these procedures provide at least
equivalent performance (17). In the U.S., swabbing is the
common sampling technique. In Norway also, the national
guidelines for good hygienic slaughter practices recommend
swabbing of carcasses for sampling for Escherichia coli after
chilling (1). Regulation 2073/2005 refers to the standard
International Organization for Standardization (ISO) 17604
"Microbiology of the food chain - Carcass sampling for
microbiological analysis," which lists the different destructive
and non-destructive sampling techniques, sampling sites,
and rules for sample storage and transport (17, 26). The
diagnostic value (i.e., sensitivity, specificity, precision, and
predictive value) of the still widely-used sampling techniques
are not documented in the available literature (25), and an
official quantitative conversion factor between destructive
and non-destructive methods has not yet been established
(4). A wide array of sampling techniques is currently used in
the European meat industry, and it is therefore difficult both
to evaluate their value in use and to make comparisons of
results from different abattoirs and countries. The destructive
method has been compared with swabbing techniques (nondestructive methods) in other studies; before chilling of pig
carcasses (29) and after chilling of beef (9) and pig carcasses
(11). These studies resulted in approval for use of a swabbing
technique in Denmark by the Danish Veterinary and Food
Administration (37) and subsequent approval from the
Norwegian Food Safety Authority for use in Norway (33).
The destructive method of excision harvests almost all
bacteria on the surface but reduces the commercial value of
the carcass. Excision further requires more equipment (sterile
knife, forceps, etc.) and skills/experience and is more time
consuming, and it therefore might be less practical for routine
carcass sampling (8). Swabbing is a preferred technique in
many abattoirs because it is non-destructive, enables sampling of larger areas of the carcass than excision, and might
be more reliable when the level of total contamination is low
and heterogeneously distributed on the carcass (4). However,
swabbing is subject to several possible sources of error due to
operator variability. Bacterial recovery with swabbing increases
with swab material abrasiveness (4, 8), and coordination of the
methodology and experimental design (swabbing materials,
stage and time of sampling, size and location of the sampled
area, microbiological analyses, etc.) is recommended in order
to evaluate the performance of different carcass sampling
techniques (4). Published studies have mainly focused on
individual steps and have been based on randomly selected,
naturally contaminated carcasses. Carcasses are big, expensive,
and difficult to handle; furthermore, natural contamination is

420

Food Protection Trends November/December

distributed heterogeneously. Natural contamination of carcasses with E. coli is normally very low, and it is therefore difficult
to study the recovery from sampling procedures, especially on
chilled carcasses. The work described here is part of a research
project (Hygenea), in which the aim is to ensure yield and payback from hygiene investments and achievements by developing tools for risk-based assessment of hygiene in abattoirs.
The aim of this study was to evaluate five sampling techniques
for quantitative microbiological characterization of slaughter
hygiene of carcasses, before and after chilling, and to identify
an acceptable and applicable non-destructive technique to be
used in a subsequent baseline study.
MATERIALS AND METHODS
General experimental design
The study was implemented in three stages: a pilot experiment, a main experiment, and a confirmatory experiment.
For all stages, samples were randomized.
In the pilot experiment, two levels of inoculum, two sampling techniques (destructive method (A) and gauze cloth
swab (B)), and three sampling times were used (12 combinations). Each combination was run in triplicate, resulting in a
total of 36 samples.
The (second) main experiment focused on the sampling
techniques and used one inoculum level, five techniques and
two sampling times (10 combinations), with six replicates.
The main dataset thus consisted of 60 samples.
The (third) confirmatory experiment concentrated on the
three sampling techniques that had the highest recovery of
bacteria in the main experiment, one inoculum level and two
sampling times (six combinations). Each combination was
run in six replicates, resulting in a total of 36 samples.
For each experimental combination in all three sub-studies, non-inoculated controls were also analyzed. All three
experiments (pilot, main, and confirmatory study) were
conducted in one laboratory (at Animalia), and the preparation of inocula and the microbial analyses were all conducted
at the Norwegian Veterinary Institute.
Assembly of the carcass model
A polypropylene drainage pipe with an outside diameter
of 160 mm was cut into 4 cm lengths that were used as meat
support frames for the carcass model. Flanks from sheep
(M. obliquus externus abdominis) were removed less than
1 h post mortem at a local abattoir, put in plastic-covered
crates and transported to the laboratory within an hour. To
make the carcass model, a flank was placed over one end
of the pipe frame, with the outer surface of the meat facing
outward and fastened tightly in place with the aid of two
or three cable ties (Fig. 1). The flanks had an approximate
thickness of 3-5 mm. The meat tissue carcass model was then
inoculated and sampled at the designated time as described
under "Preparation of inocula." For each technique, a noninoculated control sample was used.

Table of Contents for the Digital Edition of Food Protection Trends - November/December 2017