(P-103) Methodologic Challenges and Considerations for the Collection of Dietary Intake Data in Real-world Populations

Background: A person’s diet is a modifiable risk factor for many conditions and can impact treatment outcomes. Dietary data are complex and can be collected via subjective or objective observation, with closed- or open-ended questions. Understanding the many considerations is key to tailoring dietary data collection methods for study designs and objectives.

Objectives: Describe the considerations and applicability in selecting appropriate dietary assessment methods for real-world studies.

Methods: Key study design considerations for selecting dietary data collection assessments are described, including patient burden, data accuracy, and study suitability. Strengths and limitations of common data collection approaches are highlighted. Technological innovations are described, including improvements in dietary data collection.

Results: Selection of a suitable dietary data collection approach hinges on a number of considerations: 1) study design; 2) study objectives; 3) planned uses of results; 4) population of interest; 5) data collection/analysis timelines; and 6) budget. These considerations are important as retrospective (> 24 hours) detailed data collection is challenging due to recall bias, whereas studies with rapid timelines, challenging populations, or small budgets may pose other restrictions (high cost and patient burden).
Common dietary data collection approaches include: short dietary assessments (limited dietary aspects), diet history or food frequency questionnaires (previous or average intake), and 24-hour dietary recalls or multi-day food records (detailed food intake diaries). Whereas less detailed approaches are generally less time-consuming and burdensome for study participants, detailed intake approaches allow greater insight into patient outcomes. Innovative solutions may help balance between collecting detailed, high quality data while achieving low patient burden. Such solutions include: 1) artificial intelligence-enabled platforms that recognize foods and portions from pictures; 2) web or mobile applications that use guided questions and animations to help participants enter food intakes; and 3) audio transcription software to lower the burden of patients writing detailed diaries.

Conclusions: Dietary data collection is complex and selecting the best approach must carefully consider real-world patient burden, data quality, and suitability of methods. Methodological decision-making should be led by experts with knowledge of dietary data collection and operational challenges of implementation in real-world settings.