Whether or not a country’s food businesses have largely achieved food safety certification is the second most important predictor of instances of foodborne illness, according to a recent study funded by the U.S. Department of Agriculture’s National Institute of Food and Agriculture (USDA’s NIFA).
To determine the relationship between certification against food safety standards and foodborne illness outbreaks, researchers collected data on foodborne illnesses in the U.S. spanning 2015–2020 from the U.S. Centers for Disease Control and Prevention’s National Outbreak Reporting System (CDC’s NORS). Certification data for the same time period was gathered from the websites of standards holders, specifically, SQF, PrimusGFS, BRC, USDA Good Agricultural Practices (GAP), GlobalG.A.P., and FSSC22000. Data for ISO 22000 and IFS were omitted because ISO 22000 adoption in the U.S. is minimal, and IFS does not disclose certification information. Therefore, the study encompasses certification adoption information for five of the seven major private food safety standards, plus one government standard (USDA GAP).
European data was obtained from the European Food Safety Authority (EFSA) Dashboard for 2015–2020. For food safety standards, the researchers collected data on national-level adoption of GlobalG.A.P., ISO 22000, and FSSC 22000 standards.
Using regression models, food safety certification along with select economic variables (i.e., gross domestic product [GDP], farm income, and food manufacturing employment) were used to explain the number of foodborne illnesses. For the U.S., at the state level, the researchers found certifications for SQF, PrimusGFS, BRC, or FSSC 22000 to be negatively associated with cases of foodborne illness. For Europe, at the national level, certifications to ISO 22000 or FSSC 22000 were shown to be negatively associated with the instances of foodborne illness.
The researchers also used machine learning techniques to examine how well food safety certification data can predict foodborne disease outbreaks. After applying several algorithms to the U.S. data, the researchers found that models using food safety certification adoption can predict the number of U.S. foodborne illnesses or deaths with a relatively high degree of precision (testing accuracy at around 70 percent or better).
Additionally, feature importance analysis allowed the researchers to inspect the relative importance of each variable (GDP, farm income, and food manufacturing employment) for making accurate predictions of foodborne illness or death numbers. Through ranking the importance of explanatory variables, the study suggests that certification against food safety standards could be the second most important variable (after GDP) explaining foodborne illness outbreaks.