Food Protection Trends - November/December 2021 - 558

irrelevant and was therefore removed. Within each of these
major bones, there are minor bones specifying more specific
qualities of a produce operation. As with the major bones,
these were modified, removed, or added based on relevance;
for instance, forklifts and forklift stops were added as minor
bones under the " facilities " major bone.
Based on the RCA, the following interventions were tested:
(i) increasing cleaning and sanitation from once a week to
twice a week, (ii) weekly (applied on Monday of each week)
use of quat powder around forklift stops and floor cracks,
(iii) a site-specific niche (a dead-end pipe) removal, and (iv)
implementing a deep cleaning protocol in the drains.
Sample collection
To test the effectiveness of intervention implementation,
we performed sampling of the packinghouse environment
(Fig. 1) in addition to previous pre-intervention sampling
reported by Sullivan and Wiedmann (33). Five samplings
were performed from September 1st, 2020 to October
30th, 2020, including one sampling before and 4 samplings
after intervention implementation. All sampling events
were performed on Fridays to test if the second cleaning
and sanitation event in a given week (performed on
Wednesdays) was associated with a lower log odds of a
sample testing positive for Listeria; in pre-intervention
sampling a higher percent of Listeria positive samples were
seen in end-of-week samples as compared to early-in-week
samples. On each visit, 35 samples were collected from
the " wet-area " of the packing house (i.e., the area with
the bin dump, flume, brush beds and waxing equipment);
only 32 samples were collected on the final sampling
event (October 30th) because 3 sampling sites were no
longer present due to removal of the dead-end pipe. All
samples were collected from zones 2 and 3 (i.e., no food
contact surface samples were collected). The sampling sites
included: (i) 11 PVC pipe samples, (ii) 9 drain samples,
(iii) 2 forklift stop samples, (iv) 1 floor crack sample, and
(v) 12 equipment frame samples (Table S1). Samples were
collected using sponges hydrated with Dey-Engley broth
(3M, Saint Paul, MN). Sampling was performed at least 2 h
into production. All samples were transported back to the
lab on ice, then stored at 4°C until processing.
Listeria enrichment and isolation
All samples were processed within 24h of collection
using a modified version of the FDA BAM method (18).
Briefly, 90 mL of buffered Listeria enrichment broth
(BLEB, BD, Franklin Lakes, NJ) was added to each sponge
sample, followed by stomaching at 230 RPM for 1 minute
and incubation at 30°C for a total of 48h. After the initial
4h of incubation, 360 µL of Listeria selective enrichment
supplement (LSES, Oxoid, Basingstoke, UK) was added. At
24h and 48h into incubation, 50 µL of the enriched samples
are streaked for isolation onto modified Oxford agar (MOX,
558 Food Protection Trends November/December
BD) and Listeria monocytogenes plating medium (LMPM,
Biosynth International, Itasca, IL). The MOX plates were
incubated at 30°C for 48h and the LMPM plates were
incubated at 35°C for 48h. After incubation, characteristic
Listeria colonies were sub-streaked from the MOX and
LMPM plates onto brain heart infusion agar plates (BHI,
BD); characteristic Listeria colonies on MOX are dimpled
and pewter and characteristic L. monocytogenes colonies on
LMPM are round and blue. Up to 16 characteristic colonies
per sample were sub-streaked onto BHI, such that up to 4
colonies were selected from 24h MOX plates, 24h LMPM
plates, 48h MOX plates, and 48h LMPM plates. The BHI
plates were incubated at 37°C for 24h. sigB sequencing and
allelic typing
PCR and subsequent sequencing of a 660 bp fragment of
sigB was performed on all characteristic Listeria colonies substreaked
to BHI for species identification; allelic type (AT)
assignment based on the sigB sequence was performed for a
preliminary assessment of the Listeria subtypes present in the
packinghouse (23). PCR and sequencing were performed according
to the protocol described by Sullivan and Wiedmann
(33). All isolates confirmed as Listeria were stored as 15%
glycerol stocks at -80°C.
Statistical analysis
All data visualization, cleaning, and analyses were performed
in R version 4.0.0 (26). Logistic regression was
performed to determine if there were significant differences
in the percent of samples positive for Listeria before and after
increasing cleaning and sanitation frequency to twice a week.
Date of sampling and if the sample was collected before or
after increased cleaning or sanitation were tested for inclusion
in the models as potential explanatory factors. The model
outcome was the percentage of samples positive for Listeria.
To identify which of the explanatory factors were associated
with the outcome, univariable logistic regression was first
performed. Any variable with P < 0.1 was then included in
a multivariable regression model. To identify which explanatory
factors should be included in the final multivariable
logistic regression model, backwards selection was performed
to identify the model with the lowest Bayesian Information
Criterion (BIC) value. For a model to be selected as the final
model, its BIC value had to be at least 2 less than the next
simplest model.
In addition, the relationship between the increased
cleaning and sanitation frequency and the percent of Listeria
positive samples was also assessed using the sampling data
collected over multiple seasons before intervention implementation
to determine if the interventions were effective
compared to historical data. To do so, the dataset in the
current study was combined with the Listeria presence/
absence data reported by Sullivan and Wiedmann (33) for
the same packinghouse. However, the sites from Sullivan and
Wiedmann (33) included in the analyses in the current study
Food Protection Trends - November/December 2021

Table of Contents for the Digital Edition of Food Protection Trends - November/December 2021
