Portion sizes have increased substantially over the past several decades and are thought to contribute to rising rates of overweight and obesity.(1) People eat more when provided with larger portions, and over time, consider larger portions as normal.(2) Providing descriptive labeling and visual cues can be helpful in guiding consumers to appropriate portion selection.(3) Globally, few countries regulate servings sizes for packaged foods. Instead, Nutrition Facts are reported per 100g or 100mL, and/or per amount declared by the manufacturer. These variations lead to inconsistencies in on-pack portion sizes and may cause consumer confusion.(4) Our objective was to develop a methodology to establish globally relevant portion size recommendations for both nutrient-dense and discretionary foods that could be implemented in countries where permitted.

A stepwise and systematic approach incorporated values from regulated serving sizes (n = 10), portion recommendations from food-based dietary guidelines (FBDG; n = 90, aggregated into 6 regions), and food consumption data from Europe (n = 24 countries) and Australia for a total of 18 possible data inputs. These inputs were used to derive Global Portion Values (GPVs) as the median of all values for 50 food and beverage categories including both nutrient-dense and discretionary foods. To better understand the consistency of input values, 25th and 75th percentiles, and dispersion of input values, assessed as percent of median absolute deviations (MAD)/median, were calculated.

GPVs were calculated for milk and dairy (n = 6 food groups), protein foods (n = 5), mixed dishes (n = 5), grains (n = 7), fruits and vegetables (n = 5), snacks and sweets (n = 11), sauces and condiments (n = 5), fats, oils and sugars (n = 4) and beverages (n = 2). The number of inputs for each portion value ranged from 6 (Appetizers/hors d’oeuvres) to 18 (juices). Only 5 of 50 food groups had fewer than 10 inputs;

21 had ≥15. Dispersion ranged from 0-33%; 88% of GPVs were considered “consistent” (dispersion

<25%) and 8 categories had 0% variation, indicating high consistency across all inputs. Example

GPVs for nutrient-dense foods include: 240mL for milk (16 inputs, 8% dispersion); 30g for cheese (17 inputs, 0% dispersion); 90g for main dishes (meat, poultry, fish) without sauce (16 inputs, 13% dispersion); 50g for bread and rolls (17 inputs, 20% dispersion); 130g for canned vegetables (10 inputs, 2% dispersion). Example GPVs for discretionary foods include: 30g for chocolate (15 inputs, 17% dispersion); 30g for potato chips and crisps (12 inputs, 17% dispersion); and 30mL for salad dressing (13 inputs, 0% dispersion).

This novel method incorporating regulated serving sizes, portions from FBDGs and actual intakes can be used to derive portion recommendations for commonly consumed foods and beverages. If adopted, these values could improve consistency in portion labeling in countries where they are permitted and therefore guide consumers to more appropriate portions.