Food Protection Trends - November/December 2017 - 426

pling techniques. When sampling was performed after chilling, the median values for all the swabbing techniques were
0.1-0.2 log/cm2 lower than for the destructive method (P
< 0.05). The efficacy of the sampling methods, in order according to recovery, were: destructive method > gauze cloth
swab ≥ sponge-stick ≥ sponge swab > wet-dry double-swab.
In the main experiment, one inoculum level was used for
all samples. As the results between the two sampling times
(2 h and 24 h) differed in the main study, the regression
results are presented separately for each sampling time (Table
1). The effect of sampling technique was more marked for
samples taken at 2 h.
Corresponding results for the confirmatory experiment, in
which the three best techniques from the main experiment
were used; destructive (A), gauze cloth swab (B), spongestick (C), are presented in Fig. 4, where two samples taken at
24 h yielded results below the detection limit for Enterobacteriaceae and E. coli for technique A. As for the main experiment, the statistical results for each sampling time are presented separately for the confirmatory study (Table 2). The
control samples (without inocula) in all experiments showed
low contamination of the meat models (range of 0-36 CFU/
cm2 Enterobacteriaceae and E. coli).
DISCUSSION
The results confirm that the destructive sampling method, recommended by the EU (17) produces the highest
recovery of E. coli from a specific, small area when 5 cm2 of
inoculated excised flank meat is tested, compared with four
non-destructive swab sampling techniques. The difference
between the recoveries obtained for Enterobacteriaceae using
the destructive method and the gauze cloth swab was lower.
This result is in accordance with other studies that have
found higher numbers of total aerobic counts recovered by
excision than swabbing (13, 22, 24, 40). However, Gill and
Jones found that swabbing with gauze and sponge retrieved
numbers of coliforms and E. coli that were higher than those
obtained by sampling by excision (22). Our experiments further demonstrated differences between the non-destructive
techniques used, with only two (B; gauze cloth swab and D;
sponge-stick) of them assessed as being potentially appropriate for use in our subsequent baseline studies in a number of
European abattoirs. Successful removal of bacteria from the
surfaces of carcasses and onto the swabs depends on many
factors associated with the carcass surfaces, such as moisture,
temperature, hardness, and structure. Factors associated with
the swab, including abrasiveness, swab type, pressure applied,
moisture, number of strokes, area of swabbing, and operator-related differences, are also of relevance (2, 7, 36). All
these factors result in large variations among results. However, an important point is that swabbing covers larger areas
than excision and may be more reliable for sampling carcasses
with low numbers of bacteria that are spread unevenly over
the carcass surface (4). Lindblad et al. claimed that swabbing

426

Food Protection Trends November/December

a larger (400 cm2) with gauze pads met the same process
hygiene criteria as excision of a small area (20 cm2) (30).
Although efficacy of recovery of bacteria is important when
selecting a sampling technique for carcasses, other factors
also influence the choice. Swabbing is a quick, non-destructive method that is easy to use in commercial studies at
operational chain speeds (3). Swabbing is considered practical and cost efficient for the meat industry because it is less
laborious than excision sampling and does not compromise
meat quality.
Hutchison et al. sampled more than 1,700 carcasses
during 70 separate visits to commercial abattoirs (24). The
results showed that variation in contamination was so large
that calculation of a conversion factor between the different
sampling techniques used on the natural contaminated
carcasses was not possible (24). Therefore, our comparative
study was performed in a laboratory, using pre-defined
inocula for contamination, in order to standardize the
bacterial levels on the meat models. This approach was also
intended to minimize confounding effects and ensure that
bacterial recovery was sufficiently high that the performance
of the different sampling techniques could be evaluated. The
meat tissue model was artificially contaminated to mimic
fecal contamination that may occur during the slaughtering
process. The level of contamination was chosen in order
to achieve quantifiable results from testing the model. The
results from the pilot study showed that the high inoculum
(106 CFU/mL) was unnecessarily high, while the low
inoculum (103 CFU/mL) was too low, resulting in many
samples below the detection limit. As inoculated meat
models, rather than naturally contaminated carcasses were
used, and since investigation of the differences between
techniques was the main objective of the study, we have
not presented the mean values. Although the inoculum size
used in the main and the confirmatory study was rather high
(104 CFU/mL), this was considered necessary to ensure
bacterial recovery, as natural contamination levels of E.
coli are normally very low, especially after carcass chilling.
Furthermore, similar results have been observed from
analysis of sheep carcasses in different European abattoirs
(unpublished data).
The meat models used in our experiments were made using
flanks from sheep, but the results are also valid for other meat
species. Dorsa et al. evaluated sampling methods for recovery
of bacteria inoculated onto beef carcass surfaces and found
that tissue type (lean/adipose) did not alter the efficacy of
the sampling methods (13). Gallina et al. studied microbial
recovery from naturally contaminated carcasses from several
species and reported differences in recovery between samples
from ruminants and horses (19). This observation might
be attributable to variations in the microbiota on carcasses
from different species, the microorganisms' abilities to
attach to the meat, the extent to which the different bacteria
were stressed, and the need for bacterial resuscitation in the

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