Supermarket Transaction Records In Dietary Evaluation: the STRIDE study: validation against self-reported dietary...

Area of Science:

• Nutritional Science
• Public Health
• Data Science

Background:

• Accurate population dietary monitoring is crucial for public health.
• Scalable methods are needed to assess dietary intake across large populations.
• Supermarket transaction data offers a potential scalable data source.

Purpose of the Study:

• To compare dietary estimates derived from supermarket transactions with those from an online Food Frequency Questionnaire (FFQ).
• To evaluate the utility of the Supermarket Transaction Records In Dietary Evaluation (STRIDE) study for population dietary monitoring.

Main Methods:

• Recruited 1788 participants from a UK retailer's loyalty database across four UK regions.
• Collected 1 year of purchase data during and 1 year prior to the study.
• Calculated Bland-Altman agreement and limits of agreement (LoA) for key nutrients (absolute and relative).

Main Results:

• Analysis included 686 participants (72% female, mean age 56).
• Absolute nutrient intake estimates from purchases showed wide LoA, with significant over- and under-estimation.
• Relative (energy-adjusted) nutrient estimates demonstrated smaller LoA, indicating better agreement.

Conclusions:

• Agreement between supermarket purchases and actual dietary intake is highly variable.
• Relative nutrient values from purchases show stronger agreement, particularly for smaller, loyal households.
• Relative nutrient purchases may serve as a reasonable proxy for dietary composition in population-level research.