The NOVA system categorises foods based on the degree of industrial processing, with category 4 “ultra-processed foods” (UPFs) which typically comprise of products high in fat, sugar, and salt (HFSS), associated with poor nutritional value and adverse health outcomes (1). However, this classification does not distinguish non-HFSS UPFs, such as fortified or nutritionally enhanced products (2). To address this, we developed the ‘Processed with Purpose’ (PwP) framework, which differentiates UPFs based on functionality: PwP UPFs provide nutritional benefits and are non-HFSS, whereas non-PwP UPFs are generally HFSS with poor nutritional profiles. This study explored how demographic characteristics, and psychological traits relate to UPF consumption, using both the traditional NOVA system and the PwP framework. Data from the OMNIPLaNT study (3) (N = 296) were analysed via mediation models in JASP, examining demographic predictors including age, employment, income, gender, education, ethnicity, and dietary pattern (e. g. , vegan) on UPF intake, and testing whether psychological traits, specifically the Three-Factor Eating Questionnaire (TFEQ) (4) and Food Neophobia Scale (FNS) (5), mediated these relationships. This study used data from the OMNIPLaNT cohort, approved by Swansea University (Approval Number: JP₂4-06-21b). The dataset included 609 UK participants aged 18–82 years who provided informed consent. Data were collected via two online questionnaires. Questionnaire 1 captured demographics, medical history, and self-reported dietary patterns; Questionnaire 2 assessed food intake using a 112-item Food Frequency Questionnaire (FFQ) validated in the EPIC study. Questionnaires were separated by two weeks to reduce participant fatigue. Analyses focused on demographics, dietary patterns, FFQ, Food Neophobia Scale (FNS), and Three-Factor Eating Questionnaire (TFEQ). The study was preregistered on OSF. Participants missing either questionnaire or with incomplete responses were excluded. Of the 609 original participants, 296 remained after exclusions for invalid age/date of birth (n=4), missing income (n=6), incomplete FFQ (n=271), and inconsistent dietary reporting (n=32). Sample size was adequate to detect medium-sized mediation effects; power=0. 8). MeasuresFFQ: Assessed frequency of 112 foods and beverages on a 9-point Likert scale. FNS: Ten items on a 7-point scale measured reluctance to try novel foods; five items were reverse-scored. Higher scores indicated greater neophobia (α=0. 77). TFEQ-R18: Eighteen items on a 4-point scale assessed cognitive restraint, uncontrolled eating, and emotional eating (α=0. 76–0. 87). Dietary Patterns: Self-reported as omnivore, vegetarian, vegan, pescatarian, semi-vegetarian, or whole-food plant-based. Vegetarian subtypes were combined; semi-vegetarians were included in the omnivore group based on FFQ data. FFQ items were classified using NOVA (processing level) and the newly developed Processed with Purpose (PwP) framework. PwP was designed to address limitations in NOVA by distinguishing ultra-processed foods (UPFs) based on nutritional or functional purpose. NOVA Groups 1–3 were condensed into PwP Categories 1–2 (unprocessed/minimally processed and traditionally processed). NOVA Group 4 (UPFs) was divided into PwP Category 3 (UPFs with nutritive/functional benefit) and PwP Category 4 (UPFs without functional benefit) using HFSS/Nutrient Profiling Model scores. PwP considers fortification, enrichment, or other industrial modifications that deliver physiological or health-related benefits, while non-functional UPFs are primarily energy-dense with limited nutritional value. This framework aligns with UK nutrition policy, supports more nuanced dietary analyses, and helps distinguish foods that may contribute positively versus negatively to health. Two models examined predictors (age, sex, employment, education, dietary pattern, income, ethnicity) via FNS and TFEQ mediators on UPF intake outcomes: NOVA UPFs (model 1) and PwP/non-PwP UPFs (model 2). Analyses used JASP v0. 18. 3 In Model 1, the outcome was the frequency of consumption of NOVA-classified UPFs (Figure 1). In Model 2, the outcome was the frequency of UPFs further categorized as PwP UPFs or non-PwP UPFs (Figure 2). Both mediation models were bootstrapped with 1, 000 replications for bias-corrected percentile estimation. Model fit was evaluated using R 2 to quantify the proportion of variance in UPF intake explained by the predictors. Age emerged as a significant negative predictor of NOVA UPF intake (Estimate = –1. 146, SE = 0. 500, z = –2. 293, p=. 022, 95% CI –2. 112, –0. 161) ; as age increased intakes of NOVA UPF decreased. All other predictors from direct and indirect pathways (including sex, employment, household income, education, ethnicity, and diet quality, were non-significant). The model fit is R 2 = 0. 08 (Figure 4). PwP UPF; Vegetarian dietary patterns were a significant positive predictor (Estimate = 4. 330, SE = 1. 423, z = 3. 044, p =. 00295% CI 1. 376, 7. 282). Full-time employment status was positively associated with PwP UPF intake (Estimate = 4. 778, SE = 1. 625, z = 2. 940, p =. 003, 95% CI 1. 292, 8. 849). No other predictors were statistically significant. Model fit is R 2 = 0. 07. Non-PwP UPF: An increased in age showed a significant negative association with non-PwP UPF intakes (Estimate = –1. 018, SE = 0. 519, z = –1. 962, p =. 005, 95% CI –2. 086, 0. 172). All other predictors were non-significant. Model fit: R 2 = 0. 075All other predictors from direct and indirect pathways were non-significant. The use of modelling allowed for the conceptualisation of some variables as control variables that reflect the complexity of consumption behaviour such as emotional eating or cognitive control. The models were structured to test both direct and indirect pathways- diagram available Future research should build on these strengths by incorporating finer measurement: portion sizes, specific product information (including additive content, brand and manufacturing details) and longitudinal designs to unpack causal pathways. Expanding the PwP framework to include more granular food group classifications (for example ready to drink vs snack bar formats) could enhance its utility in both dietary research and public health monitoring. Further studies should also examine additional mediators of UPF consumption such as consumer perceptions, individual metabolic differences and lifestyle goals. In particular, research should explore how discourse around ultra processing influences intake for example, whether overly negative framing may inadvertently discourage consumption of nutritionally beneficial UPFs among populations with limited cooking skills or fresh food access. Consumer based qualitative work, including findings from this study on nutritionally complete meal replacements, indicates that familiarity, perceived health benefit and convenience are key drivers of choice and these should be explored further. In summary, this study reinforces the importance of avoiding a binary view of UPF versus non UPF foods. It emphasises that certain ultra processed foods may contribute to public health nutrition in specific contexts, while others remain associated with elevated health risk when eaten in frequently and in excess. This study highlights differential patterns of consumption for UPFs offering functional benefits (PwP) compared to those that do not (non PwP), as a function of age, employment and diet type. This knowledge is essential for developing pragmatic dietary advice targeted at key demographic groups especially those who rely on UPFs due to cost, shelf life or convenience and who may benefit from shifting toward nutritionally purposeful options. Overall, these findings indicate that demographic groups with consistently high UPF intake, such as younger adults or time-constrained individuals, could particularly benefit from substituting non-PwP HFSS UPFs with PwP alternatives, improving nutritional quality without requiring substantial dietary changes.
Rouse et al. (Fri,) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: