Task 4: Key Concepts about Identifying Analytic Implications of Different Types of Data

It is important to keep in mind that data from the dietary recalls, food frequency questionnaire, and supplement questionnaire each measure different aspects of dietary intake, cover different time periods, and are collected differently.  The fact that each instrument measures a different aspect of dietary intake affects whether or not it can legitimately be used alone in analyses.  The recall data and supplement questionnaire data can be used alone, but each one represents only a portion of nutrient intake.  Neither one alone is sufficient to estimate total nutrient intakes (from both foods and supplements). Similarly, food frequency data were not designed to be used alone to estimate absolute intakes of foods or nutrients even though some analytic applications allow them to be used alone.  In general, the NHANES food frequency data are meant to be used as supplementary (covariate) information in modeling data from the 24-hour recalls to estimate usual intakes when examining them in relation to some other variable of interest. 

The reference period also is different for each type of dietary data.  Dietary recalls cover intake for a given day—specifically, the previous 24 hours—although these data can be used to estimate usual intake as well.  Supplement data cover intakes from the previous month, and the food frequency questionnaire covers the previous year. 

Each type of dietary data is also gathered differently, which could lead to differential cognition (comprehension, recall, decisions & judgment, and response processes) and how individuals respond.  The recall data are gathered by a trained interviewer who probes about the previous day’s intake, capturing the details of the day’s eating using a multiple pass method.  The Day 1 recall data are collected by a personal interview in the Mobile Examination Center and the Day 2 recall are collected by a telephone interview.  The supplement data also are gathered by a trained interviewer asking a series of questions about type and amount.  The FFQ data are gathered by respondents completing a mailed questionnaire that asks about frequency of intake for a list of foods.

Because of these differences, the various types of dietary data lend themselves to different types of analyses and require different assumptions (see table below).  For example, a single 24-hour recall is sufficient for analyzing mean nutrient intakes from foods and beverages, whereas both days of data are required when estimating the distribution and prevalence of nutrient intake from foods and beverages.  There may be a sequence effect—that is, that the number and amount of foods is sometimes less on the first versus subsequent recalls—so when an analysis calls for both 24-hour recalls, you may want to control for this by adding a variable for recall day (first versus second) to the statistical analysis.

 

Components used in means and ratios of group-level means analyses

Assumptions and other issues using the 24HR and Supplement data in means and group-level means analyses:

• Assumption that the 24HR has no bias 
• Following that assumption, unadjusted mean of reported intakes can reflect mean of the population distribution of usual intake
•  Outliers may affect standard errors/confidence intervals

Components used in distribution and prevalence analyses

Assumptions and other issues using the 24HR and Supplement data in distribution and prevalence analyses:

• Assumption that the 24HR has no bias 
• Estimation of population distribution of usual intake requires statistical modeling (see Advanced Dietary Analyses Course)

Components used in correlation and regression analyses

Assumptions and other issues using the 24HR and Supplement data in correlation and regression analyses:

• Assumption that the 24HR has no bias 
• Estimation of population distribution of usual intake requires statistical modeling (see Advanced Dietary Analyses Course)