The Importer Risk Assessment (IRA) model: the science behind it
International food trade has increased drastically over the past two decades in our country and continues to evolve. Consequently, potential sources of food safety risks have multiplied for the imported foods. In this context, the CFIA has committed to a better use of its data, reports and surveillance information to identify trends, allowing the agency to focus on risk and support program design, planning, compliance and enforcement efforts. In alignment with this intelligent oversight, there is a need for risk information to define the most effective controls. The Importer Risk Assessment (IRA) model will provide the required data for risk-informed decision making, supporting thereby the allocation of the agency's inspection resources according to the food safety risks. This risk assessment takes into consideration typical food safety hazards, and will be used to determine the level of oversight required to appropriately manage the risks.
The IRA model is being developed following the same scientific and transparent approach as those followed for the Establishment-based Risk Assessment models (ERA-Food, ERA- Hatchery, ERA-Feed Mill, ERA-Renderer).
The IRA model comprises 3 different groups of risk factors: inherent risk factors, mitigation factors and compliance factors. Data for these three components will be primarily extracted from CFIA databases, while some information is expected to be gathered directly from importers.
After a cycle of data extraction and analysis by the IRA model, the results obtained from this risk assessment will be used in the CFIA's food importer risk management strategy, including the prioritization of inspection activities, oversight priorities, and work planning.
The scientific approach and practical applications to develop the IRA model are presented below. Annex 1 presents the Scientific Advisory Committee members and the IRA technical working group. Most members of these committees have also been involved in the development of the ERA-Food model.
- Identification and selection of factors associated with the importer's food safety risk
- Evaluation of data availability for the IRA model
- Risk factors' criteria weighting for the IRA model
- Risk attribution at the commodity and sub-product levels
- Design of the IRA model algorithm
- Performance assessment of the IRA model outputs
- Licensed food importers' data extraction and analysis
Identification and selection of factors associated with the importer's food safety risk
The identification of risk factors included in the IRA model was based on the first scientific literature review and subsequent selection process conducted by the ERA team in 2013 (published in the Microbial Risk Analysis and the Food Microbiology Journal). The objectives of these steps were to identify and select the most important and significant risk factors contributing to the increase or decrease of the food safety risk.
Following the initial selection, the list of risk factors was evaluated in terms of applicability to importers, by considering the availability of potential data sources and current industry practices. At this stage, experts were also consulted to identify any additional risk factor (inherent, mitigation or compliance) not previously considered. The final list of risk factors was then reviewed and validated by the Scientific Advisory Committee (SAC) (figure 1).
Risk factors are grouped in three clusters: inherent risk factors, mitigation factors and compliance factors.
Inherent risk factors represent those that are intrinsic to a specific food commodity and product as determined by scientific data and expert opinion. These risk factors take into account the type of imported commodity(ies) and sub-product(s), their volume, the product's intended use (for example, either for direct consumption or for further processing by a domestic establishment) and the target audience (for example, vulnerable populations).
Mitigation factors are the practices and import strategies that an importer applies to reduce the inherent risk. These risk factors consider the various types of foreign food safety system recognition as established by Canada for other countries as well as the use of suppliers with international private certification scheme(s) (for example, ISO 22000, Global Food Safety Initiative (GFSI) benchmarked programs).
Compliance factors refer to an importer's track record with respect to how well he has complied with regulatory requirements. This is assessed using the importer's historical and current compliance data such as information on inspection reports, history of enforcement and control actions, confirmed complaints related to food safety and recalls. The elements related to inspection results are the preventive control plan (PCP) sub-elements applicable to importers as per the requirements established on the Safe Food for Canadian Regulations (SFCR).
The criteria used to assess each risk factor have been defined based on current industry practices and pre-established inspection criteria by the CFIA (for example, history of non-compliances with direct vs. potential impact on the food safety risk).
Figure 1: List of risk factors included in the IRA model
Evaluation of data availability for the IRA model
The objective of this step was to evaluate the availability of data for each individual importer using the CFIA's databases, and to study the import trends activities (for example, the volume and type of imported products, the frequency of importation for specific commodities). This evaluation was conducted by extracting 6 months of import data collected through the Integrated Import Declaration (IID) of Single Window Initiative system (SWI) from January to June 2020.
Results of this evaluation confirmed the availability of the data related to the importer's inherent risk factors that are included in the model, and identified gaps that are currently being addressed by the agency staff (for example, data sources for a mitigation factor). As a result of this exercise, for example, the CFIA is now collecting information on whether imported cheese is made with either pasteurized or unpasteurized milk, a key piece for evaluating the food safety risk of these products.
Risk factors' criteria weighting for the IRA model
The objective of this step was to quantify the relative importance of selected criteria used to measure the risk factors included in the IRA model, based on their impact on the food safety risk. This step was completed by conducting a Delphi approach-expert elicitation. In September 2021, a total of 49 assessment criteria were presented to 38 experts working in academia, government or industry, during a two-round expert elicitation to estimate their relative risk to human health. There was a good consensus on the relative risk given to most criteria. Respondent profile (working sector and years of experience) did not have a significant influence on the results. In addition, no expert expressed formal opposition to the inclusion of any criterion. Median values for each criterion is being used for the design of the IRA model algorithm. A scientific paper is currently being prepared for publication in a peer-reviewed journal.
Risk attribution at the commodity and sub-product levels
Leveraging on previous work completed for the development of the ERA-Food model, the IRA model algorithm relies on this model's data for source attribution at the commodity and sub-product levels. This step evaluates the contribution of different food commodities and sub-products to foodborne illnesses in the Canadian population, while considering the food availability as a measure of risk exposure. On April 2021, the estimates for source attribution at the commodity level were updated and a scientific paper covering this research is currently being prepared for submission to a peer-reviewed journal. A similar exercise to evaluate the source attribution at the sub-product level was completed in January 2022, and a scientific publication will follow shortly.
Design of the IRA model algorithm
Similar to the ERA-Food model, the principle of the IRA model algorithm is based on the allocation of risks to food importers according to their impact on the health of Canadian consumers. This health burden is expressed as Disability Adjusted Life years (DALYs), and the calculation of this value takes into consideration the yearly number of foodborne illness cases associated to each food safety hazard, their attribution to specific food commodities and sub-products, and the health impact per case of illness for each hazard. Hence, the health impact is first allocated to individual importers according to the volume and type of products they bring into Canada. Then, this health burden is adjusted considering the presence or absence of specific food safety risk factors and their relative risks (figure 2)
Figure 2: Structure of the Importer Risk Assessment (IRA) model
Performance assessment of the IRA model outputs
The objective of this exercise is to evaluate the performance of the IRA model outputs by estimating the agreement between the risk assessment results provided by the IRA model algorithm and those provided by CFIA senior inspectors, and to refine the model based on the identification of major discrepancies, if needed.
Following a similar approach as the one used for the other ERA models, a number of importers will be selected using a stratified random sampling approach, while considering a 1 year-period import data. For each importer, information related to the risk factors used as input in the IRA model will be summarized in a 1-page document and presented to the senior inspectors who will categorize them based on the food safety risk information provided. The correlation analysis obtained from this exercise will confirm the applicability of the IRA model before its full implementation. This step is expected to be completed in 2022/2023.
Licensed food importers' data extraction and analysis
In this step, data related to the inherent, mitigation and compliance risk factors of all licensed food importers under the SFCR will be extracted in order to run the IRA model algorithm. The CFIA's databases are the main data sources for the various criteria included in the 3 risk factor clusters. Investigations are currently being completed to evaluate the collection of additional information (for example, use of suppliers with international private certification schemes) via a voluntary-based questionnaire for importers.
Once these steps are completed, cycles of data extraction and analysis will be conducted by the IRA model. The risk results generated by the model will be used for inspection planning and will provide input into the agency's risk-informed approach to managing food safety risks.
Annex 1: The scientific advisory committee members and the technical working group for the Importer Risk Assessment model (as of February 2022)
|Sylvain Quessy, chief scientist||Université de Montréal|
|Julie Arsenault||Université de Montréal|
|Jeffrey Farber||University of Guelph|
|Rick Holley||University of Manitoba|
|Sylvain Charlebois||Dalhousie University|
|Aamir Fazil||Public Health Agency of Canada|
|Greg Paoli||Risk Sciences International|
|Martin Duplessis||Health Canada|
|Tom Gill||Dalhousie University|
|Anna Mackay||Canadian Food Inspection Agency|
|Romina Zanabria, lead developer||Canadian Food Inspection Agency|
|Manon Racicot||Canadian Food Inspection Agency|
|Alexandre Leroux||Canadian Food Inspection Agency|
|Elisabeth Mantil||Canadian Food Inspection Agency|
|Tamazight Cherifi||Canadian Food Inspection Agency|
|Raphael Plante||Canadian Food Inspection Agency|
|Genevieve Comeau||Canadian Food Inspection Agency|
|Suzanne Savoie||Canadian Food Inspection Agency|
|Virginie Lachapelle||Canadian Food Inspection Agency|
|Nassim Haghighi||Canadian Food Inspection Agency|
Other CFIA staff working under various branches (International Affairs, Programs and Policy, Operations) are also contributing to the development of the IRA model.
- Date modified: