Linking supermarket sales data to nutritional information: an informatics feasibility study

Area of Science:

• Public Health Informatics
• Nutritional Science
• Data Science

Background:

• Grocery sales data offers valuable insights for dietary assessment programs.
• A significant obstacle is the absence of a computable method to link food items to nutrient data.
• Sustainable linkage between point-of-sale (POS) data and nutrient databases is needed.

Purpose of the Study:

• To investigate the feasibility of sustainably linking supermarket POS data to the USDA's Survey of Food Intakes by Individuals (SFII) nutrient database.
• To develop and evaluate a method for this linkage using existing supermarket item category hierarchies.

Main Methods:

• Analysis of 2,009,533 de-identified grocery sales items from 32,785 customers over two weeks.
• Development of a mapping methodology utilizing the supermarket's item category hierarchy to connect sales items to USDA nutrient records.
• Assessment of mapping success rates and required additional effort.

Main Results:

• Approximately 70% of all grocery items were successfully mapped and linked to the USDA nutrient database.
• An additional effort could increase the mappable items to approximately 90%.
• All items were mapped to standard USDA food groups, demonstrating comprehensive categorization.

Conclusions:

• Mapping grocery sales data to detailed nutritional information is feasible and achievable.
• The developed methodology offers a sustainable approach for integrating POS data into dietary assessment.
• This linkage has significant implications for enhancing public health informatics and nutritional research.