Food Protection Trends - July/August 2020 - 233

used to identify various correlations between environmental
factors and health outcomes, such as foodborne illnesses and
socioeconomic status (1, 38), geographic factors affecting
utilization of health services (2), and built the environment
and chronic disease (25).
Routine food premise inspections are an important public
health activity for providing snapshots of, as well as for
improving the performance and safety of, food premises.
Inspection reporting methods are standardized across local
government jurisdictions but may vary slightly among countries, provinces, and states. The Toronto, Ontario, food safety
inspection program, called DineSafe, has had its inspection
results publicly available since its inception in 2001 (37).
Inspections identify minor, significant, or crucial infractions
observed at the time of inspection. The type of infractions
observed affect the premises' risk categorization (low,
moderate, high) and establishment status (pass, conditional
pass, closed) (7). Risk categorization determines how often
the restaurant should be inspected per year (1-3 times), and
establishment statuses are issued as a green (pass), yellow
(conditional pass), or red (closed) sign that must be publicly
displayed near the restaurant's front entrance (7). Restaurants are permitted to continue business operations if they
pass inspection (zero infractions or only minor infractions
observed), but if they receive a conditional pass (at least one
significant infraction observed), they must correct the infractions within 24-48 hours in order to receive a pass upon
re-inspection (7). However, if they receive a closed result
(one or more crucial infractions observed), they must cease
all operations and correct all crucial infractions for re-inspection before being allowed to resume operation (7).
Many restaurant characteristics affect food inspection
results. For example, previous studies have investigated the
impact of operator and restaurant ethnicity on the number
of critical health violations (15), the chain status of restaurants on violation type (22), and food-handler certification
of kitchen managers on the presence and type of violation
(3). Other studies have investigated whether inspections are
influenced by the characteristics of the inspectors themselves,
such as the gender of the inspector (19) or potential inspector biases that are knowingly or unknowingly present (20,
26). However, few researchers have investigated whether
spatial characteristics such as location or neighborhood attributes are associated with inspection outcomes in restaurants.
Lee et al. investigated locational factors in Miami hotel
restaurants and reported that high proximity to beaches
and downtown cores was associated with increased rates
of critical food safety violations, whereas high proximity to
airports was associated with decreased rates (21). Darcey and
Quinlan investigated neighborhood socioeconomic factors in
Philadelphia restaurants and reported that a higher frequency
of inspections was associated with high poverty rates and
high concentrations of minority populations (9). They also
found increased rates of critical health violations in specific

groups of ethnic restaurants located in census tracts with
higher proportions of minority populations, a trend observed
in food stores (grocery stores) as well (9).
The purpose of this study was to investigate the spatial
distribution of restaurant inspection results in Toronto,
Ontario, and to investigate possible associations between
census tract demographic information and the rate of
conditional pass/closed (CP/C) inspection results. To
achieve these objectives, we performed a simple descriptive
analysis of inspection outcomes in 2017 and 2018; a spatial
autocorrelation test to determine whether substandard
inspection outcomes were clustered or randomly distributed
across Toronto; a hot spot analysis to identify the locations
of substandard outcome clusters; and negative binomial
regression modeling to identify associations with CP/C
inspection rates. This study demonstrates the potential for
GIS to be used as a tool to inform public health and food
safety efforts and planning.
METHODS
Data acquisition and organization
The DineSafe database, obtained from the City of Toronto's Open Data Catalogue (8) on January 3, 2019, contained
89,757 entries from November 28, 2016 to January 3, 2019.
The database is updated in real time, but contains results dating back only approximately 2 years, depending on the date
of access. It contains inspection details of any food premises
within the boundaries of the City of Toronto and includes
the address, GIS coordinates, infraction details, infraction
severity, risk categorization, and inspection outcomes.
The DineSafe database records at least one entry for each
inspection. If one inspection has multiple violations, each
unique violation is recorded as a separate entry with the
same inspection details (i.e., date, location, outcome). For
the purposes of this study, we were only interested only
in the inspection outcome. Therefore, duplicate entries
containing extra infraction details (n = 37,186) were dropped
by retaining only the first entry for each inspection. Next,
any food premises that were not labeled as "restaurant" were
removed (n = 26,619), as were any inspections performed in
2016 (n = 4,022). Finally, if a restaurant had more than one
inspection performed within each year, entries on all of the
subsequent inspections for that year were removed (2017:
n = 7,108; 2018: n = 6,437) to ensure inclusion of results
of only one inspection per restaurant per year. This left an
analytic dataset consisting of 5,950 inspection results for
2017 and 6,457 results for 2018.
Data mapping
Census tracts were selected as the geographic unit of
analysis. One census tract typically contains between 2,500
and 8,000 residents, and the boundaries are rarely changed
(34). Boundaries are based on road networks and natural
features, and in large urban centers with high population

July/August Food Protection Trends

233

Food Protection Trends - July/August 2020

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