Table 3.
Clinical observational studies published within the last year using the NOVA system to categorize UPFs.
| References | Food Collection Method and Frequency | Nutritional Program Used | Classification Method | Discrepancy Resolution | Examples of ‘Difficult’ Food Categorization |
|---|---|---|---|---|---|
| Ashraf, 2024 [89] | 24 h dietary recall using ASA24 |
ASA24-Canada-2016, Canadian Nutrient File 2015 and FNDDS |
Food items were classified according to the NOVA system manually using primarily the “Food Description” variable within the ASA24. The “Food Source” variable (e.g., fast food or vending machine) was also used to identify UPFs. In cases of ambiguity, the least processed category was chosen. Zero kcal foods (e.g., water) not classified and excluded from analysis. | Not described | Cheese was considered NOVA 3, but cheese products categorized as NOVA 4 Mass-produced bacon called NOVA 4 |
| Bonaccio, 2023 [81] | 188-item FFQ | Specifically designed software linked to Italian Food Tables | Two researchers independently coded each food into one of four categories. Conservative classification was used for challenging items. Only unequivocal foods were classified as NOVA 4 (e.g., margarine, sweet or savory packaged snacks, etc.). Some uncertain foods were classified using the most common brands in the Italian Food composition Database with the Open Food Facts database. | Discrepancies in classification were discussed with a third researcher and conservative classification was used | Bread was categorized as NOVA 3 Breakfast cereal and biscuits classified using the most consumed brands in the Italian Food composition Database with the Open Food Facts database |
| Cho, 2024 [82] | 103-item FFQ | N/A | Three study researchers classified food items on the FFQ into NOVA categories. The senior author supervised and checked for accuracy. Limited information was available to determine if some items were UPFs, so in this case, they were called non-UPFs and then sensitivity analysis was performed with them as UPFs. | Not described | Items called non-UPFs then UPFs in sensitivity analysis: chicken (e.g., drumstick and wing), canned tuna, dumpling, yogurt, coffee, and soy milk Another sensitivity analysis excluded pizza/hamburgers from the UPF category since they can be made without UPF ingredients |
| Cordova, 2023 [76] Referred to Huybrechts, 2022 [93] |
Country-specific FFQ; combination of FFQ and 7- and 14-day food records were used in Sweden and the UK, respectively | EPIC database | Generic or multi-ingredient foods were decomposed into ingredients. Because data collection started in the 1990s and the food environment has changed over the years, “middle-bound” scenario or the most likely environment was used for food processing. | Not described | Bread in Italy: lower and middle bound assumed NOVA 3—bakery; upper bound assumed NOVA 4—commercial Bread in UK: lower bound assumed NOVA 3—bakery; middle and upper bound assumed NOVA 4—commercial Cooked tomato (as pizza ingredient in Italy): lower and middle bound assumed NOVA 1—fresh; upper bound assumed NOVA 4—commercial pizza |
| García-Blanco, 2023 [83] | 147-item FFQ | N/A | Two researchers independently coded each food into one of four categories based on the NOVA system. | Discrepancies resolved by consensus | Foods that were unknown if they are homemade or industrialized (e.g., pizza, popcorn, lasagna) were classified as UPFs because most traditional foods have been replaced by industrial food products in supermarkets |
| Houshialsadat, 2023 [90] Referred to Machado, 2019 [96] |
Two 24 h dietary recalls, second recall was ≥8 days after the first | Australian Food Composition Database | Two expert evaluators classified foods into one of four categories based on the NOVA system, then a second set of two experts checked classifications. Decisions were made based on lists of ingredients from food packages or company websites. Homemade recipes were disaggregated and classified by underlying ingredients. | Discrepancies were discussed until consensus reached among all researchers |
When classification not clear (e.g., cake or cupcake, honey, commercial or homemade), the conservative alternative was chosen (e.g., homemade and disaggregated) In Australia, many commercially produced breads are processed rather than ultra-processed, so coded two commercial white breads as NOVA 4 and the rest as NOVA 3 |
| Kityo, 2023 [84] Referred to methods by Khandpur, 2021 [97] |
106-item FFQ | N/A | A nutritionist classified each FFQ item using the NOVA system with slight modification developed by Khandpul et al., then a registered dietitian validated each classification. Mixed dishes or aggregated foods were disaggregated and weights were applied using Korean food recipe information. | When a consensus was not reached, the nutritionist visited stores and websites to verify food labeling information and manufacturing processes and/or referred to previous publications | Most loaf bread (‘sikppang’), toast bread, and buns consumed in Korea are mass-produced, packaged, contain additives, and are commonly sold in convenience stores/marts, so categorized as NOVA 4 The major brand of yogurt consumed in Korea is ‘Yoplait’, which is sweetened, flavored, colored, and has artificial additives according to the labeling information, so categorized as NOVA 4 Dumplings, black bean and spicy seafood noodles were disaggregated into basic ingredients and called NOVA 1 or 3 |
| Kong, 2024 [77] | Two 24 h dietary recalls | FNDDS and NNDSR | NHANES food codes were obtained which categorized foods according to NOVA. Homemade dishes with unknown ingredients were classified according to their expected components. Foods lacking sufficient information to determine the degree of processing was usually solved by selecting a lower degree of processing. | Not described | “Yogurt, NFS” was classified as NOVA 1 “Restaurant, Chinese, Sesame Chicken” was coded as “Orange chicken” and classified as “meat” and NOVA 1 |
| Lane, 2023 [85] Referred to methods by Machado, 2019 [96] |
121-item FFQ | Nutrient Data Table for Use in Australia 1995 | Two authors with Australian food and dietary intake knowledge classified all FFQ food items into NOVA categories. For items that could not be discriminated (e.g., ‘bread’, ‘pasta or noodles’, ‘low fat cheese’, ‘yoghurt’, ‘fruit juice’), the authors referred to the National Nutrition Survey 1995-96 and NNPAS 2011-12 for comparison and decision making. When lacking details, foods were disaggregated and the conservative alternative was chosen (i.e., homemade or processed vs. UPF). | Not described | When classification not clear (e.g., cake or cupcake, honey, commercial or homemade), the conservative alternative was chosen (e.g., homemade and disaggregated) NOVA 3 breads: focaccia, ciabatta, baguette, pane di casa, sour dough, flats (naan, paratha, chapatti, roti, injera, and pita), pumpkin bread, corn bread and tortillas NOVA 4 breads: bagel, breadcrumbs, hot dog breads, fast-food breads, pizza bases, all light breads and with addition of fiber, vitamins, and minerals |
| Morales-Bernstein, 2024 [78] | Country-specific FFQ; combination of FFQ and 7- and 14-day food records were used in Sweden and the UK, respectively | N/A | Food items were categorized using the NOVA system. Food preparations using traditional methods (e.g., homemade) were disaggregated using standardized recipes. | Not described | Preserved vegetables, legumes and fruits categorized as NOVA 3 Potato products, vegetable spreads and fizzy drinks were categorized as NOVA 4 |
| Pant, 2023 [86] Referred to Machado, 2019 [96] and Lane, 2023 [85] |
101-item FFQ | N/A | Food items from the FFQ were classified into one of the four NOVA groups and cross-checked between two independent reviewers. If classification was unclear, the NNPAS 2011-12 was consulted or lesser degree of processing was selected. | Discrepancies were resolved by group consensus | Pizza and peanut butter were classified as NOVA 1 Tomato sauce and tomato paste were classified as NOVA 4 |
| Park, 2024 [91] | One 24 h dietary recall | Standard Food Composition Table by the National Institute of Agricultural Sciences | Two researchers classified each food item using the NOVA system. Product names, manufacturer, and nutritional information used to classify food as accurately as possible. | Items with discrepancies were discussed and resolved by consensus | Most or all fruit jams and canned fruits categorized as NOVA 3 Most or all bread and bakery products categorized as NOVA 4 |
| Price, 2024 [79] | Two 24 h dietary recalls | NHANES Nova 2015–18 database Food coded for NHANES using FNDDS and NNDSR |
Food classifications made using underlying ingredients. Foods were categorized using NOVA as UPFs in three ways: (1) using original NOVA methods, (2) excluding ≥25% whole grains from UPFs, and (3) excluding ≥50% whole grains from UPFs. | Not described | Commercial whole-grain bread and ready-to-eat cereals categorized as NOVA 4 reanalyzed as non-UPFs |
| Samuthpongtorn, 2023 [87] Referred to Hang, 2023 [95] |
One FFQ every 4 years between 2003 and 2017 | N/A | Three researchers independently assigned each food item to a NOVA group. Foods lacking consensus were discussed with an expert group and additional resources (research dieticians, cohort-specific documents, and online grocery store scans) were used. | Items lacking consensus were discussed with an expert group and additional resources used | Foods lacking sufficient detail (i.e., “popcorn”; “soy milk”; “pancakes or waffles”; “pie, home-baked or ready-made”; “beef, pork, lamb sandwich”; “tomato sauce”) were assigned to a non-UPF group, then later to a UPF group for sensitivity analysis |
| Sullivan, 2023 [88] | 124-item FFQ completed at baseline, year 2, and year 4 | Diet History Questionnaire nutrient and food group database; Diet*Cal Analysis Program (version 1.4.3, NCI Epidemiology and Genomics Research Program) | Two researchers independently categorized all items using the NOVA system. Discordantly assigned items were placed in the less-processed group. Sensitivity analysis performed with items assigned to more-processed group. | Not described | Tofu and honey were grouped into UPF categories because they could not be disaggregated from mixed foods |
| Wolfson, 2024 [80] Refers to Martinez Steele, 2016 [98] and 2023 [99] |
Two 24 h dietary recalls, 3‒10 days apart on different days of the week | Food coded using FNDDS and NNDSR | Food items were classified according to the NOVA system using a unique 8-digit food code. Foods likely to be homemade or artisanal were linked to scratch ingredients while foods likely purchased ready-to-eat were not disaggregated. | Not described | Several uncertain breads, such as sourdough, Italian, and naan, excluding from fast-food restaurants, categorized as NOVA 3 Some uncertain breakfast cereals such as corn flakes, frosted corn flakes, puffed rice, and raisin bran categorized as NOVA 3 Some uncertain salty snacks such as chips, crackers, and popcorn categorized as NOVA 3 |
| Zancheta Ricardo, 2023 [92] | 24 h dietary recall | SER-24 (CIAPEC) | Three different methods used to identify UPFs based on the NOVA system: (1) using the usual NOVA categories, (2) if they contained at least one ingredient not commonly used in home cooking, and/or (3) cosmetic additives. Food was classified by one dietitian and reviewed by a second dietitian. A third person classified a small random subset of records to verify. Homemade recipes were disaggregated into their components and classified. | Disagreements were discussed and resolved by consensus | Unbranded traditional Chilean bread assigned NOVA 3, while industrially produced, packaged, and branded bread assigned to NOVA 4 |
Abbreviations: ASA24 = National Cancer Institute’s web-based Automated Self-Administered 24-h Dietary Assessment Tool; UPF = ultra-processed food; FFQ = food frequency questionnaire; NHANES = National Health and Nutrition Examination Survey; EPIC = European Prospective Investigation into Cancer and Nutrition; UK = United Kingdom; NNPAS = Australian National Nutrition and Physical Activity Survey; FNDDS = Food and Nutrient Database for Dietary Studies; NNDSR = National Nutrient Database for Standard Reference; NCI = National Cancer Institute; CIAPEC = Center for Research in Food Environments and Prevention of Nutrition-Associated Diseases; N/A = not applicable.