Table 2.

Methods of clinical randomized controlled trials using the NOVA system to categorize UPFs.

References	Food Collection Method and Frequency	Nutritional Program for Data Entry	Classification Method	Discrepancy Resolution	Examples of ‘Difficult’ Food Categorization/UPF Brands Used in Menus/Comments
Capra, 2024 [31]	N/A (Plan to collect three 24 h dietary recalls of habitual diet, then study food will be provided)	NDS-R 2022, Nutrition Coordinating Center, University of Minnesota	Nutrition label for each food item was used to classify menu foods manually using NOVA. Recipes for non-UPFs were developed to provide alternatives for commercial items like bread. Ingredient and menu examples provided in original article.	Not described	UPF breakfast menu contains Eggo waffles vs. non-UPF menu contains homemade waffles UPF snack menu contains apple slices with peanut butter vs. non-UPF menu contains natural fruit licorice candy Most common food additives (eaten ≥ 10 times per week) in the UPF menus: high-fructose corn syrup, soy lecithin, citric acid, sodium citrate, annatto color, artificial flavors, sorbic acid
Fagherazzi, 2021 [65]	Two 24 h recalls administered during the third and fifth appointments (6–8 and 12–14 weeks of intervention)	Microsoft Office Excel® spreadsheet validated by Campos et al. [75]	Foods were classified according to NOVA and Dietary Guidelines for the Brazilian Population. When inadequate details provided, foods were categorized based on the typical form in which they are consumed.	Not described	Processed fruit juices and yogurts categorized as UPFs when brands were not provided
Fangupo, 2021 [67]	FFQ completed by parent on at least one of three occasions: 12, 24, and 60 months of age	N/A	Foods were classified based on the NOVA system. Product/recipe ingredients taken into consideration. Less straightforward items were disaggregated when able or discussed.	Consensus reached by researchers regarding how to disaggregate and categorize unclear foods	Categorized bacon, peanut butter, and cheese as NOVA 3 Categorized bread, commercial hummus, chocolate as NOVA 4 Items requiring disaggregation or discussion: porridge, canned fruits, pasta or tomato sauce, other fresh or canned fish, yogurt, Subway sandwich, kebabs or wraps, sushi, etc.
Gonzalez-Palacios, 2023 [68]	FFQ collected at baseline and 6 and 12 months	N/A	Specialized working group of experts in nutritional epidemiology and dieticians classified all FFQ items using NOVA. Supplementary Table S1 of original article shows classification of the 143 items in FFQ into each NOVA group, 36 of which were classified as UPFs. UPFs were further subdivided into six subgroups.	Not described	Coffee classified as NOVA 1, but decaffeinated coffee classified as NOVA 3 Items classified as NOVA 3: bacon or similar, homemade potato chips, homemade pastries, jams, dessert wine Items classified as NOVA 4: breakfast cereal, pastries or similar, chocolates and chocolate, cocoa powder
Hall, 2019 [62]	Study-designed diets provided for two weeks each (inpatient) without a washout period	ProNutra software (version 3.4, Viocare, Inc., Princeton, NJ, USA)	Food and beverages categorized according to NOVA. Detailed 7-day rotating menus with food brands provided in supplement.	Not described	UPF snack menu contains baked potato chips (Lay’s), dry roasted peanuts (Planters) and applesauce (Lucky Leaf) vs. non-UPF menus contain raisins (Monarch), fresh fruits, and raw nuts (Giant & Diamond)
Konieczna, 2021 [69]	FFQ collected at baseline, 6 and 12 months	N/A	Two dietitians independently classified all FFQ items using NOVA, then reviewed by nutritional epidemiologists.	Discrepancies in categorizations of food and drinks were discussed and consensus reached	The FFQ does not differentiate between plain, sweetened, or flavored yogurts and whole-grain cereals so they were grouped together as NOVA 1 Fruit juices, milkshakes, meatballs, hamburgers, and pizza, regardless of whether they are artisanal or industrial, were categorized as NOVA 4
O’Connor, 2023 [63] Refers to Hall, 2019 [62]	Study-designed diets provided for two weeks each (inpatient) without a washout period	ProNutra software (version 3.4, Viocare, Inc., Princeton, NJ)	Food and beverages categorized according to NOVA. Detailed 7-day rotating menus with food brands provided in supplement.	Not described	Refer to Hall, 2019 [62], above
Phillips, 2021 [70]	Smartphone app (myCircadianClock) used to record food and drink, and upload photos of food, drink, and medications daily	myCircadianClock entries categorized using Python scripts	Text entries classified by 4 independent reviewers. Food collected in German was classified by one reviewer due to language barriers. Some foods categorized by assumptions on base recipes and ingredients. Foods were assumed homemade unless stated otherwise or when processing was more common. Mixed dishes were classified to the highest NOVA group based on base recipe. Added new categories for beverages grouped into “Alcohol-containing drinks” (A), “Caffeinated drinks” (C), “Sweet drinks” (S), and “Other drinks” (D). Each drink could be assigned to multiple categories (e.g., soda Coca-Cola was ultra-processed, caffeinated, and sweet, abbreviated NOVA4-CS).	Consensus was reached for entries by at least 3 of 4 reviewers	Foods were assumed to be homemade with limited exceptions (i.e., chocolate-containing food and drinks, biscuits, toast and soft bread, croissants, pizza, burgers, plant-based drinks)
Rego, 2023 [64]	Study-designed diets provided (breakfast eaten in lab daily, remaining meals provided in portable cooler) Habitual diet determined using three 24 h dietary recalls	Open Food Facts app and NDS-R 2022, Nutrition Coordinating Center, University of Minnesota	Menus developed by a research dietician to meet UPF and other nutritional requirements and reviewed by a second dietician. Habitual diet UPF intake determined manually by trained evaluators using NDS-R output files and recall forms.	Not described	Breakfast cereal in UPF (Lucky Charms cereal) vs. non-UPF (Nature’s Path Organic Fruit Juice Corn Flakes Cereal) diet Snacks in UPF (Pringles, plain; Keebler Old Fashioned Sugar Cookie) vs. non-UPF (Cape Cod Kettle Cooked Chips; homemade sugar cookie) diet
Sneed, 2023 [66]	Three 24 h recalls each collected at baseline, 12, 24, and 36 months	NDS-R, Nutrition Coordinating Center, University of Minnesota	Some foods categorized by one expert coder to start, then six pairs of trained coders using NOVA and a set of decision rules adapted by the study team. Discrepancies resolved by defaulting to the higher processing level. Classification of mixed dishes were based on the processing level of the main ingredient contributing the highest calorie content and/or the methods used to prepare the food such as frying and not disaggregated.	Weekly meeting to discuss and resolve questions; study team made final decision to resolve coding discrepancies	Fast-food items typically considered minimally processed (e.g., 2% milk, apple slices, white rice, etc.) were further evaluated using ingredient label for industrial processing/food additives Difficulty distinguishing processed fruits (e.g., canned with added sugar) vs. ultra-processed fruits (e.g., canned with high-fructose corn syrup or sweeteners) Breads were generally classified as “industrial” and labeled as UPF unless explicitly noted as homemade or artisanal

Abbreviations: UPF = ultra-processed food; FFQ = food frequency questionnaire; NDS-R = Nutrition Data System for Research; N/A = not applicable.