Food Protection Trends - July/August 2020 - 238

in Ontario, and restaurants are particularly attracted to
neighborhoods with high levels of tourist activity (41).
However, because high real estate prices may deter smaller
businesses from settling in the core downtown area, it may
contain a higher ratio of high-end or chain restaurants
compared with other parts of Toronto. Chain restaurants
have been found to have lower rates of violations than
independently owned restaurants, because of their own
quality control and food safety monitoring programs (14).
Thus, further research regarding the distribution of chain vs.
non-chain restaurants should be conducted in Toronto to
investigate whether an association exists between chain status
and food safety violations.
Our hot spot analysis also showed that although some
clusters were quite large and occurred in similar areas
across 2017 and 2018, only one hot spot in 2018 directly
overlapped with a hot spot in 2017. Inspector biases may
influence the clustering locations of high violation counts
(26). Medeiros and Wilcock performed a qualitative survey
of public health inspectors to determine possible factors
that may influence their judgment during inspections and
found that confirmation bias, length of relationship with
operators, intimidation effect, and availability bias played
more definitive roles than other types of biases (26).
Pothukuchi et al. found that for critical violations, the
gender of the inspector was significant; females found more
critical violations than males (30). Kramer also found that
male inspectors found fewer non-critical, critical, and total
violations than female inspectors did (19). Future research
could investigate whether inspector bias is associated with
inspection outcomes or number and type of violations in the
study region.
Our regression analysis showed that CP/C violations were
correlated with the percentage of immigrants and percentage
of individuals who did not use English or French as their
home language. Both factors were very highly correlated
(r = 0.92), making it difficult to determine the primary
association. Increased adverse inspection outcomes or critical
violations have shown some association with increased risk
of foodborne illnesses (16, 17, 29, 31), which may put the
surrounding neighborhood populations at risk. Immigrants
with poor command of the official language already face
considerable challenges in accessing the healthcare system,
while also facing higher rates of poverty and unemployment
(33, 42). Additionally, in Toronto, seniors (age 65 and over)
make up a disproportionate part of the non-official-language
speaking immigrants (44.6%), since seniors account for only
15.6% of the city's overall population (33).
We also found that census tracts with higher rates of
household poverty scores (measured by LIM-AT) were
associated with more CP/C results, immigration, and nonEnglish home language. Neighborhoods experiencing higher
rates of poverty may struggle to maintain costly buildings
and equipment. An analysis of restaurant inspections in Las

238

Food Protection Trends July/August

Vegas found that 37.2% of violations were attributable to nonhuman factors (e.g., equipment failure, missing equipment,
inaccessible sinks) in restaurants (5). Neighborhoods with
lower socioeconomic status and more minority groups have
also been found to have differential access to safe food in some
settings, with increased rates of microbiological contamination
in certain foods sold at food retailers in these neighborhoods
(18, 32). However, it is uncertain if those living in lower
socioeconomic status neighborhoods actually experience more
foodborne illness from all food sources, because of potential
reporting biases (27). Future research should investigate the
microbiological quality and food safety practices in restaurants
in different socioeconomic status neighborhoods to identify
which populations are at highest risk for consuming unsafe
foods and experiencing foodborne illness.
This study has several limitations. For example, the
analysis was restricted to the first annual inspection results
in restaurants, so that only 12,407 (24.6%) inspections
out of 52,571 unique inspections across 2017 and 2018
were included. This was done to reduce the dataset to one
inspection per premise per year and to reduce bias from
the follow-up inspections designed to require restaurants
to achieve a pass. Further research could also investigate
clustering of inspections within premises over time. In
addition, we focused only on restaurants, and future work
could examine differences in CP/C rates between other food
premise types.
In addition, the published DineSafe data does not
include the identities of the inspectors; therefore, when the
inspection results were analyzed, it was assumed that the data
was completely objective and not subject to inspector bias.
However, as previously mentioned, existing research suggests
that inspectors are not completely free from subjective and
gender biases during inspections (19, 26). Finally, drawing
associations between census tract characteristics and CP/C
outcomes assumes that only the respective populations
within each census tract are at risk of poor food safety
practices of those restaurants. Liu et al. (23) found that only
25% of sit-down restaurants to which participants traveled
were located within their respective census tracts, and that
on average, individuals traveled 3.3 miles to visit them. This
may suggest that a wider scope of population demographics
should be observed in future studies.
There are also potential challenges to comparing this
study to similar research. Local inspection outcome type
was analyzed, making comparison to other food safety
inspection systems difficult, as the type of scores or outcomes
assigned may vary across different jurisdictions (4, 37, 40).
Additionally, many of the studies referenced in this article
compared neighborhood characteristics to number and type
of violations, as opposed to overall inspection outcomes,
making comparisons between findings more difficult.
Furthermore, although the use of equal interval classification
and Jenks natural breaks have their own merits, the defined

Food Protection Trends - July/August 2020

Table of Contents for the Digital Edition of Food Protection Trends - July/August 2020