Hostname: page-component-586b7cd67f-vdxz6 Total loading time: 0 Render date: 2024-11-22T11:43:25.210Z Has data issue: false hasContentIssue false
  • English
  • Français

Nova diet quality scores and risk of weight gain in the NutriNet-Brasil cohort study

Published online by Cambridge University Press:  31 July 2023

Francine Silva dos Santos ,
Eurídice Martinez Steele ,
Caroline dos Santos Costa Open the ORCID record for Caroline dos Santos Costa [Opens in a new window] ,
Kamila Tiemman Gabe Open the ORCID record for Kamila Tiemman Gabe [Opens in a new window] ,
Maria Alvim Leite Open the ORCID record for Maria Alvim Leite [Opens in a new window] ,
Rafael Moreira Claro ,
Mathilde Touvier ,
Bernard Srour Open the ORCID record for Bernard Srour [Opens in a new window] ,
Maria Laura da Costa Louzada  and
Renata Bertazzi Levy
...Show all authors
Francine Silva dos Santos
Affiliation:
Department of Nutrition, School of Public Health, University of São Paulo, São Paulo 01246-904, Brazil Center for Epidemiological Research in Nutrition and Health, Department of Nutrition, School of Public Health, University of São Paulo, São Paulo, Brazil
Eurídice Martinez Steele
Affiliation:
Department of Nutrition, School of Public Health, University of São Paulo, São Paulo 01246-904, Brazil Center for Epidemiological Research in Nutrition and Health, Department of Nutrition, School of Public Health, University of São Paulo, São Paulo, Brazil
Caroline dos Santos Costa
Affiliation:
Department of Nutrition, School of Public Health, University of São Paulo, São Paulo 01246-904, Brazil Center for Epidemiological Research in Nutrition and Health, Department of Nutrition, School of Public Health, University of São Paulo, São Paulo, Brazil
Kamila Tiemman Gabe
Affiliation:
Center for Epidemiological Research in Nutrition and Health, Department of Nutrition, School of Public Health, University of São Paulo, São Paulo, Brazil Postgraduate Program in Nutrition in Public Health, School of Public Health, University of São Paulo, São Paulo, Brazil
Maria Alvim Leite
Affiliation:
Center for Epidemiological Research in Nutrition and Health, Department of Nutrition, School of Public Health, University of São Paulo, São Paulo, Brazil Department of Nutrition, Federal University of Minas Gerais, Belo Horizonte, Minas Gerais, Brazil
Rafael Moreira Claro
Affiliation:
Department of Nutrition, Federal University of Minas Gerais, Belo Horizonte, Minas Gerais, Brazil
Mathilde Touvier
Affiliation:
Sorbonne Paris Nord University, INSERM U1153, INRAE U1125, CNAM, Nutritional Epidemiology Research Team (EREN), Villetaneuse, France
Bernard Srour
Affiliation:
Sorbonne Paris Nord University, INSERM U1153, INRAE U1125, CNAM, Nutritional Epidemiology Research Team (EREN), Villetaneuse, France
Maria Laura da Costa Louzada
Affiliation:
Department of Nutrition, School of Public Health, University of São Paulo, São Paulo 01246-904, Brazil Center for Epidemiological Research in Nutrition and Health, Department of Nutrition, School of Public Health, University of São Paulo, São Paulo, Brazil
Renata Bertazzi Levy
Affiliation:
Center for Epidemiological Research in Nutrition and Health, Department of Nutrition, School of Public Health, University of São Paulo, São Paulo, Brazil Department of Preventive Medicine, School of Medicine, University of São Paulo, São Paulo, Brazil
Carlos Augusto Monteiro*
Affiliation:
Department of Nutrition, School of Public Health, University of São Paulo, São Paulo 01246-904, Brazil Center for Epidemiological Research in Nutrition and Health, Department of Nutrition, School of Public Health, University of São Paulo, São Paulo, Brazil
*
*Corresponding author: Email [email protected]
Rights & Permissions [Opens in a new window]

Abstract

Objective:

To assess the prospective association of two diet quality scores based on the Nova food classification with BMI gain.

Design:

The NutriNet-Brasil cohort is an ongoing web-based prospective study with continuous recruitment of participants aged ≥ 18 years since January 2020. A short 24-h dietary recall screener including ‘yes/no’ questions about the consumption of whole plant foods (WPF) and ultra-processed foods (UPF) was completed by participants at baseline. The Nova-WPF and the Nova-UPF scores were computed by adding up positive responses regarding the consumption of thirty-three varieties of WPF and twenty-three varieties of UPF, respectively. Participants reported their height at baseline and their weight at both baseline and after approximately 15 months of follow-up. A 15-month BMI (kg/m2) increase of ≥5 % was coded as BMI gain.

Setting:

Brazil.

Participants:

9551 participants from the NutriNet-Brasil cohort.

Results:

Increasing quintiles of the Nova-UPF score were linearly associated with higher risk of BMI gain (relative risk Q5/Q1 = 1·34; 95 % CI 1·15, 1·56), whereas increasing quintiles of the Nova-WPF score were linearly associated with lower risk (relative risk Q5/Q1 = 0·80; 95 % CI 0·69, 0·94). We identified a moderate inverse correlation between the two scores (–0·33) and a partial mediating effect of the alternative score: 15 % for the total effect of the Nova-UPF score and 25 % for the total effect of the Nova-WPF score.

Conclusions:

The Nova-UPF and Nova-WPF scores are independently associated with mid-term BMI gain further justifying their use in diet quality monitoring systems.

Keywords

Food processingDiet quality metricsBMICohort studiesBrazil
Type
Research Paper
Information
Public Health Nutrition , Volume 26 , Issue 11 , November 2023 , pp. 2366 - 2373
Check for updates
Creative Commons
Creative Common License - CCCreative Common License - BY
This is an Open Access article, distributed under the terms of the Creative Commons Attribution licence (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted re-use, distribution and reproduction, provided the original article is properly cited.
Copyright
© The Author(s), 2023. Published by Cambridge University Press on behalf of The Nutrition Society

The Nova system classifies all foods and food products according to the extent and purpose of the industrial processing to which they were subjected into four groups: unprocessed or minimally processed foods, processed culinary ingredients, processed foods, and ultra-processed foods (UPF)(Reference Monteiro, Cannon and Levy1,Reference Monteiro, Cannon and Levy2) .

The metric most frequently used in studies that have employed Nova to assess the quality of diets and their impact on health is the dietary share, in calories or grams, of UPF3–5. These are formulations of processed food substances (e.g. oils, fats, sugars, starch and protein isolates) and cosmetic additives (e.g. flavourings, colourings and emulsifiers) created by the industry as profitable and attractive alternatives to unprocessed or minimally processed foods and their culinary preparations(Reference Monteiro, Cannon and Levy2). Various UPF attributes of acting through known, plausible, or suggested physiologic and behavioural mechanisms relate them to several chronic diseases including the risk of overweight and obesity(Reference Monteiro and Astrup3,Reference Juul, Vaidean and Parekh4) . Furthermore, a 4-week crossover randomised clinical trial, with twenty weight-stable adults, showed that, compared to a diet with no UPF, a diet with 83 % of energy from UPF caused a substantial increase in ad libitum calorie intake and consequent weight gain(Reference Hall, Ayuketah and Brychta5).

Although less often, the dietary share of Nova unprocessed/minimally processed foods has also been used to assess the quality of diets and their impact on health. These foods include unprocessed plant- and animal-sourced foods and those that were submitted to removal of inedible or undesired parts, chilling, freezing, pasteurisation, fermentation, boiling, drying, grinding and other processes that do not add salt, sugar, fat, or any other food substance to the original foods(Reference Monteiro, Cannon and Levy1). One cohort study observed that the dietary contribution of Nova unprocessed/minimally processed foods was inversely associated with overweight risk(Reference Beslay, Srour and Méjean6). Furthermore, it is well established that a diet based on a variety of unprocessed or minimally processed foods, predominantly plants, has a protective effect on health including the prevention of overweight/obesity and other obesity-related diseases(Reference Katz and Meller7).

However, the estimation of the dietary contribution of Nova groups requires detailed knowledge of total food intake which demands complex, high-cost and time-consuming data collection instruments such as 24-h dietary recalls. Aiming for simpler metrics that could be used in surveillance systems, authors of the Nova classification designed a short 24-h dietary recall screener that enables the calculation of two scores to capture the consumption of unprocessed or minimally processed whole plant foods (WPF) and UPF. The current study aims to assess the association of these two Nova diet quality scores with weight gain in the NutriNet-Brasil cohort study.

Methods

Study participants

Participants are enrolled at NutriNet-Brasil, an ongoing web-based cohort study with continuous recruitment of residents aged ≥18 years from all Brazilian regions since 26 January 2020. The NutriNet-Brasil primary aim is to investigate the prospective association between characteristics of the diet and health outcomes related to chronic non-communicable diseases in Brazil(8). The study, approved by the ethics committee of the School of Public Health from São Paulo University (process No. 88455417.8.0000.5421), recruits volunteers mainly through campaigns using traditional media (e.g. television, radio, newspapers and institutional websites), as well as social medias of digital influencers and the ones created for the study. More than 100 000 volunteers have already registered at the study digital platform (https://nutrinetbrasil.fsp.usp.br/). In a web interface, volunteers answer baseline and follow-up questionnaires. These include web-based 24-h dietary recalls, alternating the short screener and the complete recall, as well as questionnaires about health status (including weight and height) and aspects potentially associated with diet and health, such as sociodemographic characteristics, physical activity, smoking, and other health-related behaviours.

Dietary assessment

The short Nova 24-h dietary recall screener (hereinafter Nova 24-h screener) used in the current study is a low-burden survey questionnaire including checkbox format questions about the consumption of selected food items on the previous day(Reference Steele, Rauber and dos Santos Costa9). These food items include thirty-three varieties of unprocessed or minimally processed WPF (vegetables excluding potatoes and cassava, fruits excluding fruit juice, wholegrain cereals, pulses, and unsalted nuts and seeds) and twenty-three varieties of UPF (drink products such as soft drinks and flavoured dairy drinks, snack products such as packaged chips and sweetened cereal bars, and ready meals such as instant noodles and reconstituted meat products, and food products that commonly accompany main meals such as salad dressings) (Table 1). Varieties of WPF and UPF were those with the greatest energy contributions to the Brazilian diet as informed by a national population-based dietary survey(10).

Table 1 Food items considered in NOVA diet quality scores

The current study used the Nova 24-h screener completed by participants at baseline, on three non-consecutive days over the period of 2 weeks (2 weekdays and 1 weekend day), starting on the same day they register at the study platform. Based on the responses to each screener, two diet quality scores were calculated: the Nova score of WPF (Nova-WPF score) and the Nova score of UPF (Nova-UPF score).

The Nova-WPF score was computed by adding up positive responses regarding the day before consumption of each of the thirty-three varieties of unprocessed or minimally processed WPF (therefore ranging from zero to thirty-three), whereas the Nova-UPF score was computed by adding up positive responses regarding the day before consumption of the twenty-three varieties of UPF (therefore ranging from zero to twenty-three). The complete list of the food items included in each score is described in Table 1. For this study, we averaged the scores from each of the three completed screeners, to obtain the baseline Nova-WPF and Nova-UPF participant scores.

The Nova-UPF score had been previously validated in a clientele of primary health services from the city of São Paulo, with a similar agreement(Reference dos Santos Costa, de Faria and Gabe11). A validation study of both scores was conducted in a random sample of 812 NutriNet-Brasil cohort participants using a full 24-h recall as the reference dietary intake method. The prevalence-adjusted and bias-adjusted kappa (Pabak) index(Reference Byrt, Bishop and Carlin12) was used to assess the degree of agreement in the classification of participants according to the quintiles of the dietary energy share of WPF and UPF in the diet (provided from the 24-h recall) and quintiles of the Nova diet quality scores. The Pakab index with values greater than 0·80 indicates an almost perfect agreement: 0·61 to 0·80, a substantial agreement; 0·41 to 0·60, moderate; 0·21 to 0·40, fair; and equal to or less than 0·20, slight agreement(Reference Landis and Koch13). The validation study available in preprint version(Reference Costa, dos Santos and Gabe14) showed substantial agreement between participants classified, alternatively, across quintiles of the Nova-WPF scores and quintiles of the dietary energy share of all WPF (Pabak index of 0·72, 95 % CI 0·64, 0·81) as well as between participants classified across quintiles of the Nova-UPF score and quintiles of the energy share of all UPF (Pabak index of 0·79, 95 % CI 0·69, 0·88)(Reference Costa, dos Santos and Gabe14).

BMI change assessment

Participants self-reported their height in metres with two decimal points at baseline. At both the baseline and approximately 15 months after baseline (mean 14·9 (±1·2) months; min. 11·0 to max. 18·0 months), participants also self-reported the date they measured their weight for the last time and their weight measured in kg. The BMI in each date was calculated as the weight in kilograms divided by the height in squared metres (kg/m²)(15). The relative BMI change in the period between the two dates, expressed as a percentage of the initial BMI, was then standardised for the exact period of 15 months.

Covariate assessment

Data on demographics, socio-economic characteristics and health behaviours were self-reported by the participants at the baseline. The baseline covariates included sex (male/female), age (years, continuous), schooling level (0–11/≥ 12 years of schooling), region of residence in Brazil (North/Northeast/Midwest/Southeast/South), smoking status (never/former/current), diet for weight loss (no/yes) and physical activity (insufficiently active/active). Physical activity was assessed with the Global Physical Activity Questionnaire (GPAQ). According to GPAQ’s data processing guidelines, we summed the time spent on work, transport, and leisure-time physical activity and classified using the cut-off points of 150 min of moderate-intensity physical activity, 75 min of vigorous-intensity physical activity, or combination of moderate- and vigorous-intensity physical activity achieving at least 600 MET-minutes/week. We consider insufficiently active participants below these cut-off points and active those with physical activity values equal to or greater than those of the cut-off points(16).

Data analysis

Data were extracted for analysis on January 2022 and included 12 074 participants who had completed the Nova 24-h screener at baseline in three non-consecutive days and had self-reported their height at baseline and their weight and date they measured their weight for the last time both at baseline and at the 15th month of follow-up. We excluded participants who reported weight measured more than 2 months before baseline or more than 6 months before the 15-month follow-up questionnaire (n 1690), women who were pregnant at baseline or became pregnant during the follow-up period (n 355), participants with implausible initial or 15-month BMI values (<15 or ≥60 kg/m²) (n 20), outliers for 15-month BMI change (< 0·1 centile (–10·19 kg) or > 99·9 centile (+7·63 kg), (n 15)) and participants with incomplete data for covariates (n 443).

We first described the distribution of the covariates across quintiles of each Nova dietary quality score and tested the differences using Pearson’s χ2 test.

The association between each Nova dietary quality score and the 15-month change in BMI (kg/m2) was first assessed using restricted cubic spline linear regression with knots at the 10th, 50th and 90th centiles of the Nova score distribution(Reference Desquilbet and Mariotti17) with adjustment for all the previously mentioned study covariates. The statistical significance of linear and non-linear terms was evaluated using Wald tests.

Then, Poisson crude and adjusted regression models with robust variance were used to assess the association between quintiles of each Nova diet quality score and the risk of 15-month BMI gain of 5 % or more of the initial BMI.

Two adjusted models were used to test the associations between the quintiles of each score and BMI gain: one including all the study covariates and an additional model also adjusting for the quintiles of the alternative score. Mediation analysis conducted with the medsem command in Stata was used to investigate the proportion of the total effect of each Nova diet quality score on the risk of 15-month BMI gain that could be explained by the alternative Nova score(Reference Mehmetoglu18).

All statistical analyses were conducted using Stata (Stata Corp, College Station Texas) version 14.0; a P-value of <0·05 was considered statistically significant.

Results

Table 2 presents the baseline characteristics of participants included in the analysis (n 9551), overall and according to quintiles of the Nova dietary quality scores. Higher Nova-WPF scores were significantly associated with older age, higher education (≥12 years of schooling), not currently smoking, not being on weight loss diet, physically active status and having lower BMI at the baseline. Higher Nova-UPF scores were significantly associated with being male and younger, having lower education (<12 years of schooling), being insufficiently active, having higher BMI at the baseline, and living in the more economically developed South and Southeast Brazilian regions. An inverse correlation was observed between the Nova-WPF score and the Nova-UPF score at the baseline (Spearman’s coefficient = –0·33; P-value < 0·0001) (data not shown).

Table 2 Distribution (%) according to baseline sociodemographic and behavioural characteristics, and BMI intervals, overall and across quintiles of Nova dietary quality scores. Participants of the NutriNet-Brasil cohort study (2020–2022) (n 9551)

WPF, whole plant food; UPF, ultra-processed food.

P-values were estimated through Pearson’s χ2.

* Nova-WPF score (mean (min.–max.)): Q1 (2·4 (0·0–3·3)); Q2 (4·2 (3·7–4·7)); Q3 (5·7 (5·0–6·3)); Q4 (7·3 (6·7–8·0)); Q5 (10·3 (8·3–22·0)).

† Nova-UPF score (mean (min.–max.)): Q1 (0·4 (0·0–0·7)); Q2 (1·2 (1·0–1·3)); Q3 (1·8 (1·7–2·0)); Q4 (2·6 (2·3–3·0)); Q5 (4·3 (3·3–13·7)).

Restricted cubic splines linear regression models showed an inverse association between the Nova-WPF score and 15-month BMI change (kg/m2) (with slight departure from linearity) and a direct dose–response association in the case of the Nova-UPF score (with no departure from linearity) (Fig. 1). Supplementary Table 1 presents the results of the restricted cubic splines for all analysis models.

Fig. 1 The 15-month BMI change (kg/m2) regressed on the baseline Nova-WPF and Nova-UPF scores using adjusted restricted cubic spline linear regression. Participants of the NutriNet-Brasil Cohort study, 2020–2022, (n 9551)a. WPF, whole plant food; UPF, ultra-processed food. aNova-WPF score: Wald test for linear term P < 0·01; Wald test for all non-linear terms P = 0·03; values corresponding to the 10th, 50th and 90th centiles knots were 2·7, 5·7, and 10·0, respectively. Nova-UPF score: Wald test for linear term P < 0·001; Wald test for all non-linear terms P = 0·12; values corresponding to the 10th, 50th and 90th centiles knots were 0·33, 1·67 and 4·0, respectively. Adjustment for sex, age (continuous), macro-region of residence, educational level, baseline smoking status, physical activity, diet for weight loss and BMI (continuous)

A 15-month BMI gain of ≥5 % of the initial BMI was observed in 16·7 % of participants (CI95 % 16·0; 17·5). Table 3 shows results from crude and adjusted regression analysis between quintiles of each Nova diet quality score and the risk of BMI gain. Increases in quintiles of the Nova-UPF score and decreases in quintiles of the Nova-WPF score were both directly and linearly associated with increased risk of BMI gain in all models including the models with adjustment for the alternative score. The protective effect of the Nova-WPF score on the risk of BMI gain was more strongly attenuated with the adjustment for the Nova-UPF score than the opposite effect of the Nova-UPF score when adjusting for Nova-WPF score. Mediation analysis showed that 15·0 % of the total effect of the Nova-UPF score on the risk of 15-month BMI gain could be explained by the Nova-WPF score, and 25·7 % of the total effect of the Nova-WPF score could be explained by the NOVA-UPF score (data not shown).

Table 3 Frequency (%) and relative risk (RR) of 15-month BMI gain (≥5 %) according to quintiles (Q) of baseline Nova diet quality scores. Participants of the NutriNet-Brasil cohort study (2020–2022) (n 9551)

WPF, whole plant food; UPF, ultra-processed food.

* Nova-WPF score (min.–max.): Q1 (0·0–3·3); Q2 (3·7–4·7); Q3 (5·0–6·3); Q4 (6·7–8·0); Q5 (8·3–22·0).

† Nova-UPF score (min.–max.): Q1 (0·0–0·7); Q2 (1·0–1·3); Q3 (1·7–2·0); Q4 (2·3–3·0); Q5 (3·3–13·7).

‡ Adjusted for sex, age (continuous), macro-region of residence, educational level, smoking status, physical activity, diet for weight loss and BMI at baseline (continuous).

§ Additionally adjusted for quintiles of the other dietary score.

Discussion

In this large prospective study with participants from all Brazilian regions, we have shown that quintiles of new diet quality scores based on the day before intake of unprocessed or minimally processed WPF and UPF, as defined by Nova food classification system, were associated, respectively, with lower and higher risk of 15-month BMI gain of ≥ 5 %. We also identified a moderate inverse correlation between the two scores (–0·33) and the partial mediating effect of the alternative score: 25 % of the protective effect of Nova-WPF score on BMI gain could be explained by its inverse correlation with the Nova-UPF score, while 15 % of the risk effect of the Nova-UPF score could be explained by its inverse correlation with the Nova-WPF score.

Direct comparisons of these findings with other studies are not possible because this is the first study that employed the two new Nova scores to study their association with health outcomes. However, in consistency with our results, previous studies have observed a direct association between the overall dietary contribution of UPF with both short- and long-term risk for weight gain and the risk of overweight/obesity(Reference Pagliai, Dinu and Madarena19). Some of these studies also have shown that these risks remained largely unchanged after adjusting for the dietary contribution of unprocessed/minimally processed fruit and vegetables(Reference Beslay, Srour and Méjean6,Reference Mendonca, Pimenta and Gea20,Reference Canhada, Luft and Giatti21) . Likewise, an inverse association between the dietary contribution of Nova unprocessed/ minimally processed foods and risk of overweight has been previously described(Reference Beslay, Srour and Méjean6).

Several pathways may explain the independent and opposite association between each of the two Nova diet quality scores and the risk of BMI gain. One possible route is the well-known positive effects of WPF on dietary parameters critical to weight gain(Reference Canhada, Luft and Giatti21) and the more recently demonstrated negative effects of UPF on energy density, sugar, saturated fat, Na, fibre, and protein contents(Reference Martinez Steele, Marrón Ponce and Cediel22,Reference Martini, Godos and Bonaccio23) , non-nutrient bioactive compound intakes(Reference Steele and Monteiro24) and total water intake(Reference Baraldi, Steele and Louzada25). Lower or greater exposure to a modified food matrix(Reference Wahlqvist26), cosmetic food additives(Reference Roca-Saavedra, Mendez-Vilabrille and Miranda27), trans fatty acids(Reference Thompson, Minihane and Williams28), plastic packaging contaminants(Reference Steele, Khandpur and da Costa Louzada29), or neo-formed processing contaminants(Reference Lee and Pyo30), and a disturbed gut microbiota profile and integrity(Reference Zinöcker and Lindseth31) may be additional pathways explaining the opposite associations of BMI gain with the Nova-WPF and the Nova-UPF scores.

The persistent effect of each score on the risk of BMI gain after adjusting for the alternative score indicates that strategies to prevent weight gain should promote diets that are simultaneously rich in whole plant foods and reduced in UPF. These results highlight the importance of including both scores in surveillance systems that monitor population diet quality.

The strengths of this study relate to its prospective design and large sample size, as well as the baseline assessment in three non-consecutive days of the intake of WPF and UPF using a data collection tool specifically developed to capture the consumption of these foods in Brazil.

There are also some limitations that should be considered in the interpretation of our findings. This study was undertaken on volunteer participants of a cohort study on diet and health that may have a more health-conscious behaviour than the general population including more homogenous and heathier diets. This could have determined the lower contrast between extreme quintiles of each score observed in this study as compared with the overall Brazilian population(Reference Costa, Sattamini and Steele32,Reference Costa, Steele and de Faria33) . This lower contrast would tend to reduce the magnitude of the association of each score with BMI gain.

The self-reporting of anthropometric data may have introduced classification bias regarding BMI gain. However, a previous similar study with French participants observed a high concordance between web-based self-reported and objectively measured weight and height(Reference Lassale, Péneau and Touvier34). Among the Brazilian population aged 18 years or older, self-reported measurements of weight and height were shown to be a valid method to estimate anthropometric data(Reference Moreira, Luz and Moreira35).

Furthermore, non-differential social desirability bias may increase the chances of overestimating Nova-WPF or underestimating Nova-UPF scores, as well as underestimating the self-reported weight gain, biasing the studied associations towards the null.

Finally, given the observational design of the study, residual confounding arising from unadjusted factors or imprecise measurement of self-reported covariates cannot be ruled out, limiting our capacity to make causal inferences.

Conclusions

The current study shows that Nova-UPF and Nova-WPF scores are independently associated with mid-term BMI gain further justifying the use of both scores in diet quality monitoring systems.

Acknowledgements

The authors are grateful to the participants of the NutriNet-Brasil cohort.

Financial support

The NutriNet-Brasil cohort study is funded by the Brazilian Ministry of Health through CNPq, the Brazilian National Council for Scientific and Technological Development (grant number 404211/2020-8), FAPESP, the São Paulo state Research Foundation (grant number 2022/05636-0) and from UMANE. FSS was awarded a postdoctoral fellowship from Provost of Inclusion and Belonging, University of São Paulo (USP). MAL was awarded a postdoctoral fellowship from Harvard University Lemann Brazil Research Fund. The funders had no role in the design, analysis or writing of this manuscript.

Conflict of interest

The authors declare no competing interests.

Authorship

All the authors contributed to the study conception and design. FSS performed the data analysis. F.S.S. and E.M.S. wrote the first draft of the manuscript. All authors critically reviewed and approved the final manuscript.

Ethics of human subject participation

This study was conducted according to the guidelines laid down in the Declaration of Helsinki, and all procedures involving research participants were approved by the School of Public Health from São Paulo University Ethics Committee (process No. 88455417.8.0000.5421) and informed consent was obtained from all participants.

The datasets used and analysed during the current study are available from the corresponding author on reasonable request.

Supplementary material

For supplementary material accompanying this paper visit https://doi.org/10.1017/S1368980023001532

References

Monteiro, CA, Cannon, G, Levy, RB et al. (2016) NOVA. The star shines bright. World Nutr 7, 11.Google Scholar
Monteiro, CA, Cannon, G, Levy, RB et al. (2019) Ultra-processed foods: what they are and how to identify them. Public Health Nutr 22, 936–941.CrossRefGoogle Scholar
Monteiro, CA & Astrup, A (2022) Does the concept of ‘ultra-processed foods’ help inform dietary guidelines, beyond conventional classification systems? YES. Am J Clin Nutr 116, 1476–1481.CrossRefGoogle Scholar
Juul, F, Vaidean, G & Parekh, N (2021) Ultra-processed foods and cardiovascular diseases: potential mechanisms of action. Adv Nutr 12, 1673–1680.CrossRefGoogle ScholarPubMed
Hall, KD, Ayuketah, A, Brychta, R et al. (2019) Erratum: ultra-processed diets cause excess calorie intake and weight gain: an inpatient randomized controlled trial of ad libitum food intake. Cell Metab 30, 67–77.e3.CrossRefGoogle Scholar
Beslay, M, Srour, B, Méjean, C et al. (2020) Ultra-processed food intake in association with BMI change and risk of overweight and obesity: a prospective analysis of the French NutriNet-Santé cohort. PLoS Med 17, 1–19.CrossRefGoogle ScholarPubMed
Katz, DL & Meller, S (2014) Can we say what diet is best for health? Annu Rev Public Health 35, 83–103.CrossRefGoogle ScholarPubMed
NutriNet Brasil (2020) University of São Paulo, School of Public Health, Center for Epidemiological Research in Nutrition and Health. https://nutrinetbrasil.fsp.usp.br/ (accessed November 2022).Google Scholar
Steele, EM, Rauber, F, dos Santos Costa, C et al. (2020) Dietary changes in the NutriNet Brasil cohort during the covid-19 pandemic. Rev Saude Publica 54, 1–8.CrossRefGoogle ScholarPubMed
Brazilian National Institute of Geography and Statistics - IBGE (2011) Household Budget Survey 2008–2009: analysis of individual food consumption in Brazil. Rio de Janeiro: IBGE.Google Scholar
dos Santos Costa, C, de Faria, FR, Gabe, KT et al. (2021) Nova score for the consumption of ultra-processed foods: description and performance evaluation in Brazil. Rev Saude Publica 55, 1–9.Google Scholar
Byrt, T, Bishop, J & Carlin, JB (1993) Bias, prevalence and kappa. J Clin Epidemiol 46, 423–429.CrossRefGoogle ScholarPubMed
Landis, JR & Koch, GG (1977) The measurement of observer agreement for categorical data. Biometrics 33, 159–174.CrossRefGoogle ScholarPubMed
Costa, CD, dos Santos, FS, Gabe, KT et al. (2023) Description and performance evaluation of two diet quality scores based on the Nova classification system. medRxiv doi: 10.1101/2023.05.19.23290255.Google Scholar
World Health Organization Technical Report Series (2000) Report of a WHO Consultation. Obesity: Preventing and Managing the Global Epidemic. Geneva: WHO.Google Scholar
World Health Organization (2002) Global Physical Activity Questionnaire (GPAQ) Analysis Guide. https://www.who.int/ncds/surveillance/steps/resources/GPAQ_Analysis_Guide.pdf (accessed November 2022).Google Scholar
Desquilbet, L & Mariotti, F (2010) Dose-response analyses using restricted cubic spline functions in public health research. Stat Med 29, 1037–1057.Google ScholarPubMed
Mehmetoglu, M (2018) medsem: a Stata package for statistical mediation analysis. Int J Comput Econ 8, 63.Google Scholar
Pagliai, G, Dinu, M, Madarena, MP et al. (2021) Consumption of ultra-processed foods and health status: a systematic review and meta-analysis. Br J Nutr 125, 308–318.CrossRefGoogle ScholarPubMed
Mendonca, RD, Pimenta, AM, Gea, A et al. (2016) Ultraprocessed food consumption and risk of overweight and obesity: the University of Navarra Follow-Up (SUN) cohort study. Am J Clin Nutr 104, 1433–1440.CrossRefGoogle ScholarPubMed
Canhada, SL, Luft, VC, Giatti, L et al. (2020) Ultra-processed foods, incident overweight and obesity, and longitudinal changes in weight and waist circumference: the Brazilian Longitudinal Study of Adult Health (ELSA-Brasil). Public Health Nutr 23, 1076–1086.CrossRefGoogle ScholarPubMed
Martinez Steele, E, Marrón Ponce, JA, Cediel, G et al. (2022) Potential reductions in ultra-processed food consumption substantially improve population cardiometabolic-related dietary nutrient profiles in eight countries. Nutr Metab Cardiovasc Dis 32, 2739–2750.CrossRefGoogle ScholarPubMed
Martini, D, Godos, J, Bonaccio, M et al. (2021) Ultra-processed foods and nutritional dietary profile: a meta-analysis of nationally representative samples. Nutrients 13, 3390.CrossRefGoogle ScholarPubMed
Steele, EM & Monteiro, CA (2017) Association between dietary share of ultra-processed foods and urinary concentrations of phytoestrogens in the US. Nutrients 9, 209.CrossRefGoogle Scholar
Baraldi, LG, Steele, EM, Louzada, MLC et al. (2021) Associations between ultraprocessed food consumption and total water intake in the US population. J Acad Nutr Diet 121, 1695–1703.CrossRefGoogle ScholarPubMed
Wahlqvist, ML (2016) Food structure is critical for optimal health. Food Funct 7, 1245–1250.CrossRefGoogle ScholarPubMed
Roca-Saavedra, P, Mendez-Vilabrille, V, Miranda, JM et al. (2018) Food additives, contaminants and other minor components: effects on human gut microbiota—a review. J Physiol Biochem 74, 69–83.CrossRefGoogle ScholarPubMed
Thompson, AK, Minihane, AM & Williams, CM (2011) Trans fatty acids and weight gain. Int J Obes 35, 315–324.CrossRefGoogle ScholarPubMed
Steele, EM, Khandpur, N, da Costa Louzada, ML et al. (2020) Association between dietary contribution of ultra-processed foods and urinary concentrations of phthalates and bisphenol in a nationally representative sample of the US population aged 6 years and older. PLoS One 15, e0236738.CrossRefGoogle Scholar
Lee, HW & Pyo, S (2019) Acrylamide induces adipocyte differentiation and obesity in mice. Chem Biol Interact 298, 24–34.CrossRefGoogle ScholarPubMed
Zinöcker, MK & Lindseth, IA (2018) The western diet-microbiome-host interaction and its role in metabolic disease. Nutrients 10, 365.CrossRefGoogle ScholarPubMed
Costa, CDS, Sattamini, IF, Steele, EM et al. (2021) Consumption of ultra-processed foods and its association with sociodemographic factors in the adult population of the 27 Brazilian state capitals (2019). Rev Saude Publica 55, 1–9.CrossRefGoogle ScholarPubMed
Costa, CD, Steele, EM, de Faria, FR et al. (2022) Score of ultra-processed food consumption and its association with sociodemographic factors in the Brazilian National Health Survey, 2019. Cad Saude Publica 38, e00119421.CrossRefGoogle Scholar
Lassale, C, Péneau, S, Touvier, M et al. (2013) Validity of web-based self-reported weight and height: results of the Nutrinet-Santé study. J Med Internet Res 15, e2575.CrossRefGoogle ScholarPubMed
Moreira, NF, Luz, VG, Moreira, CC et al. (2018) Self-reported weight and height are valid measures to determine weight status: results from the Brazilian National Health Survey (PNS 2013). Cad Saude Publica 34, e00063917.CrossRefGoogle ScholarPubMed
Figure 0

Table 1 Food items considered in NOVA diet quality scores

Figure 1

Table 2 Distribution (%) according to baseline sociodemographic and behavioural characteristics, and BMI intervals, overall and across quintiles of Nova dietary quality scores. Participants of the NutriNet-Brasil cohort study (2020–2022) (n 9551)

Figure 2

Fig. 1 The 15-month BMI change (kg/m2) regressed on the baseline Nova-WPF and Nova-UPF scores using adjusted restricted cubic spline linear regression. Participants of the NutriNet-Brasil Cohort study, 2020–2022, (n 9551)a. WPF, whole plant food; UPF, ultra-processed food. aNova-WPF score: Wald test for linear term P < 0·01; Wald test for all non-linear terms P = 0·03; values corresponding to the 10th, 50th and 90th centiles knots were 2·7, 5·7, and 10·0, respectively. Nova-UPF score: Wald test for linear term P < 0·001; Wald test for all non-linear terms P = 0·12; values corresponding to the 10th, 50th and 90th centiles knots were 0·33, 1·67 and 4·0, respectively. Adjustment for sex, age (continuous), macro-region of residence, educational level, baseline smoking status, physical activity, diet for weight loss and BMI (continuous)

Figure 3

Table 3 Frequency (%) and relative risk (RR) of 15-month BMI gain (≥5 %) according to quintiles (Q) of baseline Nova diet quality scores. Participants of the NutriNet-Brasil cohort study (2020–2022) (n 9551)

Supplementary material: File

Santos et al. supplementary material

Santos et al. supplementary material

Download Santos et al. supplementary material(File)
File 14.6 KB