Questions: Dietary Pattern Assessment and Diet Quality Indices

Question 1

A researcher compares two approaches: (1) a meta-analysis of randomized trials on vitamin C supplementation and cardiovascular risk, and (2) a prospective cohort study using the Healthy Eating Index to predict cardiovascular mortality. Which approach is more likely to capture synergistic dietary effects?

• A. The vitamin C meta-analysis, because randomized trials have higher internal validity
• B. The Healthy Eating Index approach, because it captures co-occurring dietary exposures that act together
• C. Both approaches are equally capable of capturing dietary synergy
• D. Neither approach can capture synergy — only controlled feeding trials can do that

Answer: (B)

Single-nutrient analyses, even high-quality RCTs, isolate one variable at a time and cannot capture the co-occurring exposures that characterize real dietary patterns. The Healthy Eating Index scores alignment with an overall dietary pattern — vegetables, whole grains, fish, sodium, saturated fat — where foods co-occur and interact. The core insight is that nutrients are packaged in foods consumed in culturally structured combinations, and their combined effects may differ from the sum of individual effects.

Question 2

A nutritional epidemiologist wants to identify a dietary pattern that best explains variation in circulating inflammatory biomarkers across a cohort. Which statistical method is most appropriate?

• A. Factor analysis, because it identifies all food groups that tend to be consumed together
• B. A priori Mediterranean Diet Score, because the Mediterranean diet has established anti-inflammatory effects
• C. Reduced-rank regression, because it derives patterns that maximally explain variation in specified biological intermediates
• D. The Healthy Eating Index, because it provides a comprehensive assessment validated against dietary guidelines

Answer: (C)

Reduced-rank regression derives patterns from the data by maximizing explanation of specified response variables (here, inflammatory biomarkers), making derived patterns more directly mechanistically linked to the outcome of interest. Factor analysis identifies co-consumption patterns but without reference to any particular biological outcome. A priori indices like the Mediterranean Diet Score are based on existing nutritional knowledge and may miss data-specific patterns more strongly associated with the biological pathway being studied.

Question 3

Two studies using different dietary indices — the Mediterranean Diet Score and the Alternate Healthy Eating Index — can both find associations with mortality even if their high-scoring foods partially differ, because what matters is the overall pattern rather than any specific food item.

• T. True
• F. False

Answer: True

Overall diet quality, by any of several measures, predicts mortality and chronic disease risk more consistently than any single nutrient or food. The consistency of findings across different measurement instruments strengthens causal inference. This cross-index consistency supports the pattern-level insight: no single food drives the association, but the overall dietary structure does.

Question 4

A posteriori dietary patterns derived by factor analysis in one population generalize well to other populations, making them the preferred tool for international dietary comparisons.

• T. True
• F. False

Answer: False

A posteriori patterns are derived statistically from the data of a specific population and may not transport to another. A 'Western' or 'prudent' pattern identified in a US cohort may not characterize the same foods or behaviors in a South Asian or African population. A priori indices like the Mediterranean Diet Score are more comparable across populations because they use predefined scoring templates, though they may miss locally relevant patterns.

Question 5

Why do researchers studying diet and disease risk prefer pattern-level analyses over single-nutrient analyses, even when they have precise measurements of individual nutrient intakes?

Model answer: Nutrients are consumed in foods, and foods are consumed in culturally and economically structured combinations. The combined biological effects of these co-occurring exposures may differ from the sum of their individual effects (synergy or antagonism). Statistically, nutrients within common food sources are highly correlated, making it mathematically problematic to adjust for one while holding others constant. Pattern-level analyses capture the full dietary exposure as it actually occurs.

Single-nutrient studies have often produced inconsistent or null findings for nutrients that pattern studies find clearly protective, precisely because isolating a single nutrient severs it from the dietary context in which its effects operate. This is the core motivation for the entire field of dietary pattern research.