Development of a nutrient-based method for classifying foods to be consumed in moderation

This study developed and validated a nutrient-based method for classifying foods to be consumed in moderation according to the 2020-2025 Dietary Guidelines for Americans, demonstrating that the classification effectively distinguishes lower-quality foods and correlates negatively with overall diet quality.

The Gist

Imagine you're walking through a massive, chaotic supermarket. The aisles are packed with thousands of different items, from fresh apples to sugary sodas, from plain rice to greasy pizza. The "Dietary Guidelines" (the rulebook for healthy eating) tell you: "Eat more veggies, eat less sugar, and watch out for too much salt and bad fats."

But here's the problem: The rulebook doesn't give you a shopping list. It doesn't say, "If you see a cookie, put it in the 'Eat in Moderation' cart." It just gives you numbers for your whole day. So, how do you know which specific items to grab and which to put back?

That's exactly what this research team set out to solve. They built a smart sorting machine (called the Moderation Food Classification Method, or MFCM) that acts like a super-accurate barcode scanner for food.

The Problem: The "Gray Area" of Eating

Think of the Dietary Guidelines as a traffic light system for your diet:

• Green Light: Eat as much as you want (veggies, fruits, whole grains).
• Red Light: Stop or slow down (sugar, salt, bad fats).
• Yellow Light (The Problem): The guidelines say "limit" these bad things, but they don't tell you exactly which foods are the "Red Light" ones. Is a slice of bread okay? What about a slice of pizza? What about a carrot cooked with a little butter? Without a clear rule, it's hard to know what to put in your cart.

The Solution: The "Nutrient Scanner"

The researchers created a set of strict, math-based rules to scan every single food item in the database. They looked at four main "villains" that the guidelines warn against:

1. Added Sugar (The sweet trap)
2. Saturated Fat (The clogging fat)
3. Sodium (The salty sneaker)
4. Refined Grains (The empty carb)

They set up a "Tripwire" system. If a food trips any of these wires, it gets flagged as a "Moderation Food."

Here is how their scanner works, using some fun analogies:

• The Sugar Trap: If a food gets more than 20% of its energy from added sugar, the scanner screams, "Stop! This is a candy bar!"
• The Fat Filter: They realized that some healthy foods (like veggies) get a little fat from cooking, but fried foods get too much. So, they created a special rule: If a snack or a veggie dish has more than 9% of its weight as fat, it's flagged. It's like a bouncer checking if the "fat content" is too high for the party.
• The Salt Shaker: They used the FDA's rule: If a single serving has more than 460mg of salt (about 20% of your daily limit), it gets flagged.
• The Grain Gadget: If a bread or pasta is mostly made of white flour (refined) and lacks fiber, it trips the wire.

The "Whole Food" Safety Net:
There was one catch. Some healthy foods, like avocados or Brazil nuts, are naturally high in fat. The scanner almost flagged them as "bad" by mistake! The researchers fixed this by adding a "Whole Food Exemption": If it's a whole plant food (fruit, veg, nut, seed), it stays Green, no matter what.

Did the Machine Work?

They tested their scanner on thousands of foods and real people's diets. Here's what happened:

1. The "Snack" Test: The scanner correctly identified that 97% of snacks (chips, cookies) and 97% of sweets were "Moderation Foods." It also correctly said that plain broccoli and fresh apples were not moderation foods.
2. The "Nutrient Density" Check: They compared the "Moderation Foods" to the "Good Foods." The "Moderation Foods" were like empty calories—they had very few vitamins and minerals but were packed with sugar, salt, and fat. The "Good Foods" were nutrient powerhouses.
3. The "Real Life" Test: They looked at what real Americans ate. They found that people who ate a lot of "Moderation Foods" had much lower overall diet quality scores. It was like a direct line: More "Moderation Foods" = Less Healthy Diet.

Why This Matters

Before this study, if you wanted to tell someone to "eat less junk," you had to guess what junk was. Now, we have a standardized, scientific definition.

Think of this method as a universal translator for nutrition.

• For Researchers: It lets them compare diets across different studies without confusion.
• For Doctors: It gives them a clear way to tell patients, "Try to swap these specific 'Moderation Foods' for these 'Green Light' foods."
• For Policy: It could help governments put warning labels on food (like the black octagon labels in Chile) that actually match what scientists say is unhealthy.

The Bottom Line

This paper didn't just invent a new diet; it built a rulebook for the rulebook. It gave us a clear, scientific way to say, "This specific food is high in the things we need to limit."

Instead of guessing whether a bag of chips is "okay," we now have a tool that says, "Yes, that's a Moderation Food. Enjoy a little, but don't make it your main course." It turns the confusing world of nutrition into a simple game of Green Light, Yellow Light, and Red Light.

Technical Summary

1. Problem Statement

The 2020–2025 Dietary Guidelines for Americans (DGA) recommend limiting the intake of added sugars, saturated fat, refined grains, and sodium. However, the guidelines lack specific, quantitative criteria for identifying individual foods high in these "nutrients to limit." Existing tools have significant limitations:

• Diet Quality Indices (e.g., HEI-2020): These assess overall diet quality at the population or individual level but do not provide food-level thresholds to identify specific "moderation foods."
• Nutrient Profiling (e.g., NRF, Food Compass): These rank foods by overall nutritional value but use heterogeneous criteria and often lack standardized cut-offs for categorizing foods as "to be limited."
• Regulatory Labeling (e.g., FDA "Healthy" claims): These rely on fixed serving sizes and daily values (e.g., 20% DV) that may misclassify individual foods (e.g., labeling plain sugar as not "high in added sugar" due to small serving sizes) and do not cover non-packaged or mixed foods.

There is a critical need for a standardized, evidence-based, nutrient-based method to classify individual foods as "moderation foods" that can be applied across diverse dietary databases and research studies.

2. Methodology

The authors developed the Moderation Food Classification Method (MFCM) using a three-principle framework and validated it using two major datasets: the 2017–2018 Food and Nutrient Database for Dietary Studies (FNDDS) ($n=6,909$ foods) and the 2017–2018 National Health and Nutrition Examination Survey (NHANES) ($n=6,180$ participants).

A. Development of Criteria (MFCM)

The method was guided by three principles: alignment with DGA recommendations, consistency with literature-based food categorizations (e.g., "junk" vs. "nutrient-dense"), and parsimony/applicability to existing data variables.

Final Thresholds: A food is classified as a "moderation food" if it meets at least one of the following criteria:

1. Added Sugars: $\ge$ 20% of total kcal (twice the DGA daily limit of 10%).
2. Saturated Fat: $\ge$ 20% of total kcal (applied to all groups except vegetables and snacks/sweets).
3. Total Fat (Specific Groups): $\ge$ 9% total fat by weight (applied only to vegetables and snacks/sweets to capture fried items and high-fat snacks that may be low in saturated fat but high in energy density).
4. Sodium: $\ge$ 460 mg per serving (aligned with FDA "high sodium" definition).
5. Refined Grains: $\le$ 50% whole grains AND a carbohydrate-to-fiber ratio $>$ 10:1 (applied only to foods with $\ge$ 0.5 ounce-equivalents of grain).
6. Alcohol: All alcoholic beverages.

Note: Whole plant foods (fruits, vegetables, whole grains, legumes, nuts, seeds) were explicitly excluded from being classified as moderation foods, with manual recoding for three specific high-fat items (Brazil nuts, fresh coconut, avocado).

B. Validation Strategy

• Content Validity: Assessed by comparing MFCM classifications against literature-based lists of foods to limit (e.g., sweets, salty snacks) and nutrient-dense foods.
• Convergent Validity (Food Level): Compared nutrient density (using the NRF 9.3 index) and energy density between moderation and non-moderation foods.
• Convergent Validity (Diet Level): Analyzed NHANES data to correlate the percentage of daily energy intake from moderation foods with the Healthy Eating Index-2020 (HEI-2020) scores (Total, Adequacy, and Moderation components).

3. Key Contributions

• Standardized Classification: The MFCM provides the first quantitative, nutrient-based algorithm to classify individual foods as "moderation foods" consistent with the 2020–2025 DGA.
• Algorithmic Flexibility: The method is designed to be calculated using standard variables available in the FNDDS and ASA24 (Automated Self-Administered 24-hour dietary recall), making it highly applicable for large-scale surveillance and research.
• Refined Fat Criteria: The introduction of a dual-threshold approach for fat (saturated fat % kcal for most foods; total fat % weight for vegetables/snacks) successfully resolved misclassifications where fried vegetables or high-fat snacks were previously overlooked or where low-fat vegetables were falsely flagged.
• Ordinal Scoring: Beyond a binary classification, the method allows for an ordinal score (number of thresholds met), enabling a gradient analysis of food quality.

4. Key Results

• Content Validity:
  • The method correctly classified 97% of snacks and 96% of sweets as moderation foods.
  • Conversely, it classified only 19% of vegetables and 23% of fruits as moderation foods.
  • All milkshakes, frankfurters, pizza, nachos, potato chips, cookies, doughnuts, and soft drinks were classified as moderation foods.
  • Low-fat milk, grapes, melon, broccoli, and spinach were correctly classified as non-moderation foods.
• Convergent Validity (Food Level):
  • Moderation foods had significantly lower nutrient density (NRF 9.3) than non-moderation foods (Mean difference: -74.8; 95% CI: -70.6 to -78.9).
  • Nutrient density decreased linearly as the number of thresholds met increased (1 vs. 2 vs. $\ge$3 thresholds).
• Convergent Validity (Diet Level):
  • In the NHANES sample, moderation foods contributed 77% of total daily energy intake.
  • They accounted for 94% of added sugar intake, 78% of sodium, 83% of saturated fat, and 94% of refined grains.
  • There was a strong negative correlation between moderation food intake (% kcal) and overall diet quality:
    • HEI-2020 Total Score: $r = -0.72$
    • HEI-2020 Adequacy Score: $r = -0.67$
    • HEI-2020 Moderation Score: $r = -0.56$

5. Significance and Implications

• Research Utility: The MFCM enables consistent identification of "moderation foods" across studies, facilitating the investigation of diet-disease relationships and the comparison of findings across diverse populations.
• Public Health & Policy: The method can inform public health interventions, such as:
  • Labeling: Providing a basis for front-of-package warning labels (similar to Chile's policy) to help consumers identify foods to limit.
  • Procurement: Assisting charitable food organizations and institutions (schools, hospitals) in selecting healthier food options.
  • Surveillance: Monitoring the prevalence of moderation foods in the food supply and tracking changes in population intake over time.
• Clinical Application: The tool offers a concrete, evidence-based framework for dietary counseling, allowing clinicians to move beyond abstract nutrient limits to specific food-based recommendations.

Limitations: The study relied on self-reported dietary data (subject to bias) and generic food representations in FNDDS (lacking brand-specific recipe variations). However, the authors argue these limitations affect all nutrient-based classifications equally and do not undermine the method's validity for relative comparisons.

Conclusion: The MFCM is a valid, robust, and practical tool for classifying foods high in nutrients to limit. It bridges the gap between high-level dietary guidelines and actionable, food-level guidance, supporting improved diet quality assessment and intervention strategies.