Food Protection Trends - November/December 2017 - 424

TABLE 2. Reduction in Enterobacteriaceae and E. coli in the confirmatory experiment by
using median regression model. There were three sampling techniques (highest
counts in the main study, A, B and D), two sampling times (2 and 24 h), and
one inoculum level (104). The regression coefficients described the change in
log10 CFU per cm2 for either Enterobacteriaceae or E. coli as an explanatory
variable changed by one unit, the other variables being held constant. Baseline
levels for the categorical variables were sampling technique A and 2 h period
between inoculation and sampling.

Variable

Enterobacteriaceae (log10)
Coefficient (95% CI)

P-value

E. coli (log10)
Coefficient (95% CI)

P-value

Time = 2 hours
Destructive method (A)

0.00 (-)1

-

0.00 (-)2

-

Gauze cloth swab (B)

-1.07 (-1.55/-0.58)

< 0.001

-0.29 (-0.39-0.19)

< 0.001

Sponge-stick (D)

-0.79 (-1.78/-0.20)

0.01

-0.19 (-0.42-0.04)

0.11

Time = 24 hours
Destructive method (A)

0.00 (-)3

-

0.00 (-)4

-

Gauze cloth swab (B)

-0.44 (-1.16/-0.28)

0.21

-0.15 (-0.36-.0.05)

0.14

Sponge-stick (D)

-0.44 (-1.11/-0.28)

0.21

-0.14 (-0.35-.0.05)

0.14

Enterobacteriaceae median recovered by destructive method after 2 h: 0.56 log CFU/cm2 and
2
after 24 h: -0.16 log CFU/cm2.
3
E. coli median recovered by destructive method after 2 h: 0.08 log CFU/cm2 and 4) after 24 h:-0.70 log CFU/cm2.
1

time (P = 0.30) were of lesser importance. Corresponding
numbers for E. coli were for inoculum (P < 0.001), sampling
technique (P = 0.52), and sampling time (P = 0.25). There
were no significant differences between the results at 2 h and
12 h for either Enterobacteriaceae or E. coli, nor between 12 h
and 24 h. Sampling at 12 h was therefore not included in the
main experiment. No interaction effects were found. The initial inoculation level was 2.1 log CFU/cm2 in the main study
and 2.2 log CFU/cm2 in the confirmatory study.
In the main study, the destructive sampling method (A)
with a circular excision of 5 cm2 area resulted in higher
mean recovery of Enterobacteriaceae than the swabbing
techniques, with the exception of the gauze cloth swab (B),
both before and after chilling (P < 0.05 by ANOVA) (Fig.
3). The numbers obtained using the destructive method (A)
2 h after inoculation was taken as the reference level (Table
1 and Table 2). The coefficients in the regression equation
quantify the expected change in Enterobacteriaceae/E. coli
log per cm2 from the changes in the explanatory variables;
sampling techniques (B-E) and period after inoculation

424

Food Protection Trends November/December

(24 h) (Table 1 and Table 2). In the confirmatory study, the
destructive sampling method (A) gave a higher recovery
of Enterobacteriaceae than either the gauze cloth swab
method (B) or the sponge-stick (D) method, both before
and after chilling (P < 0.05 by ANOVA) (Fig. 4). Swabbing
before chilling (2 h after inoculation) with a gauze cloth
(B) and sponge-stick (D) resulted in the second highest
Enterobacteriaceae numbers in both the main study and
confirmatory study, with a median that was 0.5 log/cm2
lower than that of the destructive method. Sponge swab
(C) resulted in a median value that was 0.7 log/cm2 lower
than that from excision (A), and the wet-dry double-swab
technique (E) had 0.8 log/cm2 lower median than excision
(A) and its median was lower than that of all the other
techniques (P < 0.05). The recovery of Enterobacteriaceae
was lower for all techniques when sampling was carried
out after chilling rather than before (P < 0.05). All the
non-destructive methods had lower median recoveries of
Enterobacteriaceae than the destructive method, by 0.1 - 0.2
log/cm2 (P < 0.05).
Table of Contents for the Digital Edition of Food Protection Trends - November/December 2017
