Food Protection Trends - March/April 2025 - 115

of Salmonella Anatum in cattle lymph nodes is consistent
with other studies that have reported common isolation of
Salmonella Anatum from the cattle lymphatic system (2, 4,
15, 18).
Interestingly, the overrepresentation data among the
FSIS chicken isolates by source (i.e., ground, raw intact,
and young chicken) also provided initial insights that (i)
are consistent with previous data and (ii) may provide
important insight for future efforts to control targeted
serovars in chickens (17, 19). Most strikingly, Salmonella
Kentucky was overrepresented among young chicken and
underrepresented among ground chicken, which not only
could be seen as consistent with the reported virulence
attenuation of the predominant Salmonella Kentucky ST in
the United States (i.e., ST152) (20) but also could suggest
that existing interventions or practices may effectively
control transmission of this serovar from live birds to ground
chicken. On the other hand, Salmonella Infantis was found
to be highly overrepresented among ground chicken and
turkey and underrepresented in raw intact chicken, which
could indicate efficient transmission, with limited reduction,
from young chicken to ground chicken, which could indicate
Salmonella Infantis presence in deeper tissue.
We also find certain overrepresentation patterns that are
possibly related to biases in data collection. For example,
most serovars that showed evidence for overrepresentation in
a given turkey isolation source in NCBI represented serovars
that are historically less frequently isolated from turkey
(8). For example, Salmonella Montevideo was significantly
overrepresented in young turkey compared with other
serovars, despite its low isolation from turkey sources overall.
The overrepresentation of these less common turkey serovars
from these isolation sources in the NCBI data set may be an
artifact associated with many academic groups submitting
data to NCBI; these data may be a result of a specific study or
studies focused on one or more selected serovars.
Despite limitations of FSIS and NCBI databases,
analysis of them informs risk-based policy options and
intervention strategies
As detailed previously, possible limitations of the NCBI
and FSIS data sets include potential biases with biases
likely more frequent in the NCBI data set. For example, the
fact that many data in our NCBI data set analyzed (which
excluded FSIS submissions) are from academic sources may
lead to bias and associated over- or underrepresentation
of specific serovars, as certain serovars may be of higher or
lower interest to academia. For instance, Salmonella Infantis
has recently become of high interest among researchers,
which has resulted in numerous studies analyzing the
serovar. More broadly, academics might have had an
incentive to isolate certain serovars over others, depending
on research funding and grants, which may, for example,
lead to overrepresentation of antibiotic-resistant Salmonella,
115 Food Protection Trends March/April
representing a specific research priority. Second, the NCBI
data set may be biased and contain a moderate degree of
variability compared with the FSIS data set because samples
are collected by various researchers who use different
isolation, enrichment, and testing methodologies.
Although the FSIS data set indicates serovars identified
from positive samples during routine sampling and isolated
through standardized methodologies, there still exists a
potential for differences among samples because multiple
people are both sampling and handling products. Also,
these sampling and isolation techniques may select for
certain serovars and bias identification toward those.
Finally, the Salmonella serotyping protocol used by FSIS
does not identify the presence of more than one serovar
per sample. This means that the true Salmonella serovar
diversity in collected samples is not captured and that only
the most prevalent serovar (or the serovar that grows best in
enrichment media) is identified. It is, however, also likely that
in many academic studies only a single isolate per sample is
advanced to WGS. A specific limitation of our study relates
to the time frame covered by our data sets, which included
part of the severe acute respiratory syndrome coronavirus 2
(coronavirus disease 2019) pandemic. This may be why we
observed notable decreases in Salmonella-positive isolates
from the NCBI data set in 2020. It is likely that many
academic researchers were unable to collect samples this year.
The FSIS data set was likely not affected or at less affected
because FSIS continued to collect samples in 2020. Finally,
we found that swine-associated Salmonella data available
from FSIS are more limited than data available for beef,
chicken, and turkey.
Despite the limitations detailed in the prior paragraph,
our data support that the FSIS and NCBI databases,
particularly if used in combination, can be useful to support
the implementation of more risk-based control strategies
for Salmonella in raw meat and poultry. These databases are
particularly important as the U.S. Department of Agriculture
FSIS may transition into focusing on control of Salmonella
with the largest public health significance, with a reduced
focus on serovars with limited public health impact, even if
they are found in high frequency in certain animal sources,
such as Salmonella Kentucky in chicken. Although NCBI
and FSIS databases can allow for timely tracking of changes
in serovar prevalences and changes in associations of
serovars with different sources, it is important to consider
the various biases of FSIS and NCBI data and to also use
other foodborne pathogen data sets (e.g., poultry or meat
industry data), where possible, to minimize confounding
variables and provide a more comprehensive understanding.
Future regulatory approaches that will focus on Salmonella
serovars of public health relevance, as well as risk assessments
supporting these approaches, however, will also need
to consider other factors that affect risk and that cannot
(yet) be easily accessed through these databases, such as
Food Protection Trends - March/April 2025

Table of Contents for the Digital Edition of Food Protection Trends - March/April 2025
