This modeling study is based on the framework and data developed by USDA for food pattern modeling. USDA established an approach for food pattern modeling to develop dietary patterns that meet nutrient goals for the U.S. population ages 2 years and older [19]. The approach has been used to develop the HEP, and also the first food patterns for infants and toddlers introduced in the 2020–2025 DGA [3, 19,20,21]. The established method for developing dietary patterns is a 5-step process defined as follows: Step 1) establish energy levels, Step 2) establish nutritional goals, Step 3) establish food groupings and food group amounts, Step 4) determine nutrients obtained from foods within each group, and Step 5) evaluate nutrient levels compared to nutrient goals. The underlying data used by USDA in the development of dietary patterns for the 2020–2025 DGA are two days of food intake as reported in the National Health and Nutrition Examination Survey (NHANES) 2015–2016, and the Standard Reference (SR) Legacy nutrient composition data used by USDA to process nutrient intakes in the survey. In the current study, modifications were made only to the percentage of energy from macronutrients necessary to achieve the objective of the study (step 2 of the 5-step process), and these modifications in turn required revision of food group amounts (step 3 of the 5-step process) to reflect the adjustment in macronutrient goals.

The defining characteristic of the patterns modeled in this study is a higher level of energy from protein relative to the HEP. In this study, the modified HEP models with higher protein are referred to as “M-HEP”. Modified patterns were constructed with 20% (M20-HEP), 25% (M25-HEP), and 30% (M30-HEP) energy from protein, or 100 g, 125 g, and 150 g protein, respectively, in a 2,000 kcal diet. Food Patterns developed by USDA represent 12 energy levels from 1,000 to 3,200 kcal to address the range of energy needs for most individuals ages 2 years and older. Given the exploratory nature of this study, patterns were developed for the 2,000 kcal diet as a representative pattern. The 2,000 kcal diet is appropriate for energy needs of adolescents and adults over a range of ages and activity levels, including older sedentary men and older active women. Consistent with the approach used in development of dietary patterns in the DGA, nutrient requirements for patterns at the 2,000 kcal energy recommendation were designed to meet the DRIs for females ages 19–30 years and males ages 51–70 years [22].

The nutrient profiles for all food groups and subgroups used in this modeling are the consumption-weighted nutrient-dense food averages for the U.S. population ages 2 years and older as established by the food pattern modeling team for the 2020–2025 HEP [22]. Recommended amounts of foods from each food group in the HEP for a 2,000 kcal diet provided the starting values for all food groups and subgroups in these models, and consumption-weighted data for the U.S. population ages 2 years and older were selected for representative nutrient profiles of all food groups for the U.S. population. The HEP was modified to meet the nutrient goals of 20% (M20-HEP), 25% (M25-HEP), and 30% (M30-HEP) energy from protein by adjusting contributions of protein foods to meet the protein targets. The constructed diets largely align with the HEP, with modifications necessary only to provide higher levels of protein while still meeting nutrient goals.

The protein subgroups modified to develop the HEP in this analysis are subgroups for meat, poultry, eggs, seafood (with separate groups for seafood with higher and lower concentrations of omega-3 fatty acids, i.e., high n-3 and low n-3), nuts/seeds, and soy products. The beans and peas group, which counts towards either the vegetable or protein foods group in USDA food patterns, was also a potential source of protein in the models. Consistent with the DGA, in this study the meat group (referenced in some literature as red meat [23]) includes beef, pork, lamb, goat, and game meat, all in either unprocessed (fresh) or processed forms, while the poultry group includes chicken, turkey, Cornish hens, duck, goose, and game birds, likewise all in unprocessed (fresh) or processed forms. Lean meat or poultry therefore includes both fresh and processed forms.

In USDA’s development of the HEP, the nutrient profile of each food group represents a composite of the population weighted average nutrient profiles of similar foods in the constituent subgroups referred to as item clusters. Representative data selected by USDA for the nutrient profiles typically reflect nutrient data for a food with the least amount of added sugars, sodium, and solid fat. Examples of item clusters for protein foods include lean meat, poultry and fish cooked without added fat or salt, canned fish without added salt, hard-boiled whole eggs, reduced-fat hot dogs, reduced-fat turkey sausage (representative of all sausage), unsalted nuts, tofu, and soy protein isolate.

In the HEP, item clusters for meat and poultry include fresh and processed (i.e., cured) forms of these protein foods. For the current study, item clusters corresponding to fresh and processed forms of meat/poultry were disaggregated into the proportions of ounce-equivalents (ounce-eq) used by USDA in development of the 2020–2025 HEP (Table 1). The meat and poultry groups combined account for 68% of all ounce-eq in the protein foods group, with 57% as meat and 43% as poultry. Approximately one-third (34%) of all meat is represented by processed meat, while 12% of all poultry is represented by processed poultry. Across the combined subgroups of meat and poultry, 24% of these protein foods are represented by processed products and the balance (76%) is represented by fresh products. The proportionally weighted nutrient profiles for the disaggregated fresh and processed item clusters within the meat and poultry subgroups developed for use in this analysis, along with the nutrient profiles for all protein foods subgroups, are shown in Table 2. All values in Table 2 represent the nutrient profile per 1 ounce-eq of these protein foods, defined as 1 ounce lean meat, poultry, or seafood, 1 egg, ¼ cup cooked beans or tofu, 1 tablespoon peanut butter, or ½ ounce nuts/seeds.

A series of food pattern models was developed using stepwise reductions in the number of ounce-eq of processed meat/poultry relative to levels in the HEP. At each target level of energy from protein (i.e., 20%, 25%, and 30%), a pattern that maintained the HEP ratio of processed to fresh meat and poultry ounce-eq (i.e., 34:66 for processed: fresh meat and 12:88 for processed: fresh poultry) and proportionately increased all protein foods was first attempted. Patterns that proportionately increased all protein foods other than processed meat and poultry were then modeled, thus maintaining current allowances for processed meat and processed poultry in the HEP, namely 4.5 and 1.0 ounce-eq per week, respectively, based on weighted consumption data for the U.S. population. Contributions of processed meat and poultry were then decreased to 2 and 0.5 ounce-eq per week, respectively; an additional model was developed in which all processed meat/poultry was eliminated. If the M-HEP could not be achieved with proportional increases in all protein food subgroups, the subgroups were individually adjusted. With the exception of eggs, minimum levels of weekly ounce-eq from each protein subgroup were maintained at levels no lower than levels in the HEP. The weekly allowance for eggs, which are a concentrated source of cholesterol, was reduced to offset the cholesterol provided by additional ounce-eq of meat, poultry, and seafood.

To accommodate increased energy from protein foods, levels of refined grains and starchy vegetables were first decreased and energy allowed from the solid fats and added sugars components of the “calories for other uses” was reduced. To maintain some flexibility in other dietary choices and consistency with the HEP, the M-HEP retained total calories for other uses in the range of 75% to 100% of calories allocated for this use in the HEP (i.e., approximately 180 to 242 kcal). The percent allocation of solid fat and added sugars was maintained approximately in the ratio used in the HEP of 16 g solid fat and 27 g added sugars.

In developing the M-HEP in this study, nutrient goals consistent with those used by the USDA were used to assess nutrient adequacy. Nutrient goals for the patterns are diets within the macronutrient AMDRs, at least 90% of the Recommended Dietary Allowance (RDA) or Adequate Intake (AI) for micronutrients, fiber, and fatty acids as established by the IOM [2, 24, 25], below the Chronic Disease Risk Reduction (CDRR) for sodium [25], and within quantitative recommendations in the 2020–2025 DGA for saturated fat [1]. In the current analysis, M-HEP model diets likewise were considered nutritionally sufficient if the levels of nutrients were at least 90% of the RDA/AI for both males ages 51–70 years and females ages 19–30 years which are representative populations consuming a 2,000 kcal diet and the nutrient goals used by USDA to assess compliance [22]. The HEP developed by USDA does not meet the specified nutrient goals for vitamin D, vitamin E, choline, iron (females) and magnesium (males) [22]. If the level of these nutrients in a model was marginally less than the specified RDA/AI target (i.e., 90% of the RDA/AI), the model was considered sufficient, which is consistent with the approach utilized by the USDA [20]. Although the 2020–2025 DGA does not specify a limit on dietary cholesterol, Americans are encouraged to consume only as much as necessary within a nutritionally adequate diet [1]. The patterns modelled in this study were designed to provide less than 300 mg cholesterol, which is the limit used in food pattern modeling exercises to support the 2020–2025 DGA [22].

In developing the M-HEP, food group recommendations for underconsumed foods were maintained within the median and 95th percentiles of Usual Intakes (UI) to maintain feasible dietary patterns while overconsumed components were maintained between 5th percentile and median intakes [22]. Given the exploratory nature of this modeling exercise, this constraint was not consistently enforced and deviations are noted.

The patterns generated following this stepwise approach were further reviewed and modified by manually adjusting food group amounts to create food patterns translatable for communicating dietary guidance, which is consistent with the refinements detailed by USDA in development of the HEP. These modifications included rounding food groups to the nearest half unit (e.g., 2.4 units was rounded to 2.5 units) and minor adjustments to minimize variations within a food group across food patterns for a given protein level.