Table 2.
Overweight, obesity and cardio-metabolic risks as outcomes in studies in adults *.
| Study Details | UPF Exposure | Outcomes | Results | |||||
|---|---|---|---|---|---|---|---|---|
| Publication Author(s) Year | Study Type (Year) Setting | Population (Number) | Extraction Level | Relative exposure [UPF Reference Year] | Data Collection Method | Health Outcome(s) (Study Definition) | Data Collection Method | Key Findings |
| Overweight and obesity | ||||||||
| Juul 2015 [36] |
Ecological (1960–2010) Sweden |
Adults ≥18 years (n = −4000 household) |
National + household sampling |
National: per capita UPF consumption + Household: UPF % share food purchase (kg or litre per capita per annum) [NOVA.2014] [29] |
National: Swedish BOA net food ** available Household: 2-week purchase record by interview |
BMI classified in prevalence overweight (BMI ≥ 25) and obesity (BMI ≥ 30) | National population statistics | From 1980 to 2008: rise in overweight prevalence for men from 35% to 54–56% and women from 26% to 39%; and obesity prevalence for men rose from 4.5% to 11% and for women from 5% to 10%. From 1960 to 2010 rise in UPF consumption of 142% tracks increase in overweight and obesity prevalence. |
| Monteiro 2017 [52] |
Ecological (1991–2008) Europe | Adults ≥ 18 years except Belgium ≥ 15 years (n = 19 countries) | Household (National Sample) | UPF % total E purchases (continuous) [NOVA.2018] [30] | Belgium, Sweden, Germany = one month food ** purchase record; all others = 14 day record g/mL. |
BMI classified in prevalence obesity (BMI ≥ 30) | National reports | UPF ranged 10.2–50.7% (median 26.4) of household total E in food purchases. Each 1% increase in UPF E availability was associated with 0.25% increase in obesity prevalence. |
| Vandevijvere 2019 [37] |
Ecological (Repeated cross-sectional) (2002–2014) Global |
Adults ≥18 years (n = 80 countries) |
National | UPF total sales (volume/capita) [NOVA.2018] [30] |
Volume sales of UPF (137 items from 212 food ** subgroups) | Mean population BMI | National reports | Increases in UPF volume sales/capita were directly associated with mean BMI trajectories. Every standard deviation increase in volume sales of UPF, mean BMI increased by 0.195 kg/m2 for men and 0.072 kg/m2 for women (drinks only), and 0.316 kg/m2 for men (foods only). |
| Canella 2014 [53] |
Cross-sectional (2008–2009) Brazil |
All ages (n = 55,970 households; 190,159) individuals) | Household (National Sample) |
UPF % total E purchases (quartiles) [NOVA.2012] [54] |
7-day food ** purchase record | BMI classified in excess weight (BMI > 25), obesity (BMI > 30) WHO BMI for age Z scores [children] | Trained personnel | UPF contributed 25.5% of total E purchased. Participants living in household strata belonging to the upper quartile of UPF consumption had higher mean BMI (Z score) (β = 0.19; 95% CI 0.14, 0.25) prevalence of obesity (β = 3.72; 95% CI 2.50, 4.94) and prevalence of excess weight (β = 6.27; 95% CI 4.15, 8.39), compared with those in the lowest quartile. As UPF consumption rose from Quartile 1 to Quartile 4, the prevalence of excess weight rose from 34.1% to 43.9%, and prevalence of obesity rose from 9.8% to 13.1%. |
| Adams 2015 [55] |
Cross-sectional (2008–2012) UK |
Adults > 18 years (n = 2174) |
Individual (National Sample) |
UPF % total E intake (continuous) [NOVA.2010] [25] |
4-day food ** intake diary | BMI classified in overweight (BMI ≥ 25); obesity (BMI ≥ 30) | Trained personnel | UPF contributed 53% of total E intake. UPF consumption was not significantly associated with BMI, overweight and obesity, and obesity. |
| Louzada ‡ 2015 [56] |
Cross-sectional (2008–2009) Brazil |
Adults > 20 years; children > 10 years (n = 30,243) |
Individual (National Sample) | UPF % total E intake (quintiles) [NOVA.2012] [54] |
2 × 24-h food ** intake record | BMI classified in excess weight (BMI ≥ 25), obesity (BMI ≥ 30) [adults]; WHO BMI for age Z scores [children] | Trained personnel | UPF contributed to 29.6% of total E intake. Individuals in the upper quintile of UPF intake had significantly higher BMI (0.94 kg/m2; 95% CI = 0.42, 1.47) and higher odds of being obese (OR = 1.98; 95% CI = 1.26, 3.12) compared with the lowest quintile. No significant association with excess weight was found. |
| Nardocci 2018 [57] |
Cross-sectional (2004–2005) Canada |
Adults > 18 years (19,363) |
Individual (National Sample) | UPF % total E intake (quintiles, and continuous) [NOVA2016.2018] [30,58] | 1 × 24-h recall | BMI classified in overweight (25.0 ≤ BMI < 30.0); obesity (BMI ≥ 30) | Trained personnel | UPF contributed 45.1% of total E intake. Individuals in highest quintile UPF intake significantly had higher odds of being obese (OR = 1.32, 95% CI 1.05, 1.57, and overweight (OR = 1.03; 95% CI 1.01, 1.07), compared with individuals in lowest quintile. |
| Juul 2018 [59] |
Cross-sectional (2005–2014) USA |
Adults 20–64 years (15,977) |
Individual (National sample) |
UPF % total E intake (quintiles) [NOVA.2014] [29] |
2 available 24-h recall or 1 day otherwise | BMI classified in overweight and obesity (BMI ≥ 25), obesity (BMI ≥ 30); WC classified in abdominal obesity (AO) [men ≥ 102 cm, women ≥ 88 cm) |
Trained personnel | UPF contributed 56.1% of total E intake. Individuals in the highest quintile of UPF intake had significantly higher BMI (1.61 kg/m²; 95% CI 1.11, 2.10), and WC (4.07 cm, 95% CI 2.94, 5.19), and higher odds of having excess weight (OR = 1.48; 95% CI 1.25 to 1.76), obesity (OR = 1.53, 95% CI 1.29, 1.81), and abdominal obesity (OR = 1.62; 95% CI 1.39 to 1.89) compared with those in the lowest quintile. |
| Rauber 202 [60] | Cross-sectional (2008–2016) UK |
Adults 19−96 years (n = 6143) |
Individual (National sample) | UPF % total E intake (quartiles) [NOVA.2019] [21] |
4-day food ** intake diary | BMI classified in obesity (BMI ≥ 30). WC classified in AO | Trained personnel | UPF contributed 54.3% of total E intake. Individuals in the highest quartile of UPF intake had higher BMI (1.66 kg/m2; 95%CI 0.96, 2.36) and WC (3.56cm, 95% CI 1.79, 5.33), and higher odds of obesity (OR = 1.90, 95% CI 1.39, 2.61) compared with the lowest quartile. |
| Julia 2018 [61] |
Cross-sectional (2014) France |
Adults Mean 43.8 years (n = 74,470) | Individual | UPF % total grams (quartiles) [NOVA.2016] [22,33] |
3 × 24 h records | BMI classified in overweight (25–29.9), obesity (≥30) |
Self-report # | UPF contributed 18.4% of total weight intake, and 35.9% of total E intake. Higher consumption of UPF by % E intake was independently associated with overweight (p < 0.0001); and higher intake by energy-weighted UPF was independently associated with overweight, and obesity (both p < 0.0001). |
| Silva 2018 [62] |
Cross-sectional (2008–2010) Brazil | Active and retired civil servants 35–64 years (n = 8977) | Individual | UPF % total E intake (quartiles) [NOVA.2016] [22] |
114 item-FFQ | BMI classified in overweight (25.0-29.9); obesity (≥30); WC classified in increased WC (men ≥ 94; women ≥ 80); significantly increased WC (men ≥ 102; women ≥ 88) |
Trained personnel | UPF contributed 22.7% of total E intake. Individuals in highest quartile UPF intake had significantly higher BMI (0.80 kg/m2; 95% CI 0.53, 1.07), WC (1.71 cm; 95% CI 1.02, 2.40), and higher odds of being overweight (OR = 1.31; 95% CI 1.13, 1.51), obese (OR = 1.41, 95% CI 1.18, 1.69), increased WC (OR = 1.31, 95% CI 0.96, 1.32), and significantly increased WC (OR = 1.41; 95% CI 1.20, 1.66), compared with individuals in the lowest quartile. |
| Da Silveira 2017 [63] |
Cross-sectional (2015) Brazil |
Vegetarians > 16 years (n = 503) |
Individual | UPF intake frequency (≥3 times per day) [DGB.2014] [28] |
FFQ (number of items not specified) | BMI classified in overweight BMI ≥ 25 (16–59 years), BMI ≥ 27 (≥60 years) |
Self-report # | Higher intake of UPF (≥3 times/day) was independently associated with overweight (OR = 2.33; 95% CI 1.36, 4.03). |
| Ali 2020 [64] |
Cross-sectional (2018) Malaysia |
Adults 18–59 years (n = 167) University personnel | Individual | UPF % total E intake (+continuous) [NOVA. 2018] [30] |
2-day 24 h recall | BMI % Body fat | Trained personnel |
UPF contributed 23 % of total E intake. No significant findings between ultra-processed food consumption BMI, body fat percent (p = 0.954). |
| Mendonca 2016 [65] |
Prospective Cohort (1999–2012) 8.9 years median follow-up Spain |
Adults Mean 37.6 years (n = 8451) |
Individual | UPF intake servings/day (quartiles) [NOVA.2016] [22] |
136-item FFQ | BMI classified in overweight/obesity (BMI ≥ 25), obesity (BMI ≥ 30). |
Self-report # | Participants in the highest quartile of UPF consumption were at a higher risk of developing overweight/obesity (HR = 1.26; 95% CI 1.10, 1.45) compared with those in the lowest quartile of consumption. |
| Canhada 2020 [66] | Prospective Cohort (2008–2010) 3.8 years median follow-up Brazil |
Adults 35–74 years (n = 11,827) |
Individual | UPF % total E intake (quartiles) [NOVA 2016] [22] | 114-item FFQ | Large weight gain (≥1·68 kg/year) Large WC gain (≥2·42 cm/year) Overweight/obesity (BMI ≥ 25 kg/m2) Obesity (BMI ≥ 30) |
Trained personnel | UPF contributed 24.6% of total E intake. Participants in the highest quartile of UPF intake had greater risk of large weight (RR = 1.27; 95% CI 1.07, 1.50) and waist gains (RR = 1.33; 95% CI 1.12, 1.58), and of developing overweight/obesity (RR = 1.20; 95% CI 1.03, 1.40) compared with individuals in the lowest quartile. |
| Hall et al. 2019 [67] |
Randomised Controlled Trial (2018, 4 weeks) USA |
Weight stable adults Mean 31.2 years (n = 20) |
Individual | Whole diet UPF vs. MPF diet (ad libitum) [NOVA.2018] [30] | Diets designed and analysed using ProNutra software | Energy Intake (kcal) Change in body weight (kg) |
Trained personnel | Energy intake was greater during exposure to the UPF diet (508 ± 106 kcal/day; p = 0.0001). Participants gained 0.9 ± 0.3 kg (p = 0.009) during the UPF diet, and lost 0.9 ± 0.3 kg (p = 0.007) during the MPF diet. |
| Cardio-metabolic risks | ||||||||
| Lavigne- Robichaud 2017 [68] |
Cross-sectional (2005–2009) Canada |
Adults ≥ 18 years (n = 811) |
Individual | UPF total E % intake (quintiles) [NOVA.2010] [25] | 1 × 24-h food ** recall | Metabolic syndrome (MetS) (≥3 factor: high WC, HT TAG, BG; low HDL-C) | Trained personnel | UPF contributed 51.9% of total E intake. Those in highest quintile of UPF intake significantly associated with higher prevalence of MetS (OR = 1.90; 95% CI 1.14), higher prevalence of reduced HDL-C (OR = 2.05; 95% CI 1.25, 3.38), elevated fasting plasma glucose (OR = 1.76, 95% CI 1.04, 2.97) compared with those in the lowest quintile. |
| Nasreddine 2018 [69] |
Cross-sectional (2014) Lebanon | Adults ≥18 years (n = 302) |
Individual | UPF ‘pattern’ vs. MPF and PF ‘pattern’ (quartiles) [NOVA.2012] [54] |
88-item FFQ | Metabolic syndrome (≥3 factors: high WC, HT, TAG, BG; low HDL-C) |
Trained personnel | UPF vs. MPF were 36.5% vs. 27.1% of total E intake. Those in highest quartile MPF/PF significantly lower odds MetS (OR = 0.18, 95% CI 0.04, 0.77); hyperglycaemia (OR = 0.25, 95% CI 0.07, 0.98), low HDL-C (OR = 0.17, 95% CI 0.05, 0.60) compared with those in the lowest quartile. No significant association between MetS and UPF. |
| Lopes 2019 [70] |
Cross-sectional (2008–2010) Brazil |
Adults 35–74 years (n = 8468) |
Individual | UPF % total E intake (terciles) [NOVA 2016] [22] |
114–item FFQ | C-reactive protein (CRP) level (mg/L) | Trained personnel | UPF contributed to 20% total E intake. Women in highest tercile UPF intake had higher levels of CRP (arithmetic mean = 1.14; 95% CI: 1.04–1.24) than lowest tercile of intake, no significance when controlling for BMI. No significant association was observed in men. |
| Martinez Steele 2019 [71] |
Cross-sectional (2009–2014) US |
Adults ≥ 20 years (n = 6385) |
Individual (National sample) |
UPF Total E % intake (quintiles and continuous) [NOVA.2018.2019] [21,30] |
2 available ×24-h recall, or 1 day otherwise. | Metabolic syndrome (≥3 factor of high WC, HT, TAG, BG; low HDL) | Trained personnel | UPF contributed 55.5% of total E intake. The highest quintile of UPF consumption was associated with higher MetS prevalence (PR = 1.28; 95% CI 1.09, 1.50) compared with the lowest quintile of UPF consumption. Each 10% increase in the consumption of UPF was associated with 4% increase in MetS prevalence (PR = 1.04; 95% CI 1.02, 1.07) |
| Mendonca 2017 [72] |
Prospective Cohort (1999–2013) 9.1 years median follow-up Spain |
Adult graduates (n = 14,790) | Individual | UPF E intake servings per day (tertiles) [NOVA.2016] [22] |
136-item FFQ | Hypertension (BP: Systolic ≥ 140 mm Hg and/or Diastolic ≥ 90 mm Hg) |
Self-report ξ | Participants in the highest tertile of UPF intake had higher risk of developing hypertension (HR = 1.21; 95% CI 1.06–1.37) compared with those in the lowest tertile of intake. |
Results are presented for adjusted associations for potential confounders and statistically significant associations. NOVA refers to the food classification system [21] or earlier versions, as referenced; * Includes studies on all ages; ** includes beverages; # anthropometrics; ξ reported medical diagnosis, medication, or BP readings, ‡ results for adolescents are presented in Table 3; UPF: ultra-processed food (includes foods and beverages); BOA: Board of Agriculture; BMI: Body Mass Index [weight (kilograms)/height (metres)2]; E: energy in kilocalories or kilojoules; WHO: World Health Organisation; OR: odds ratio; CI: confidence interval; WC: waist circumference (cm); increased WC: (men ≥ 94; women ≥ 80; significantly increased WC (men ≥ 102; women ≥ 88); AO: abdominal obesity (men ≥ 102 cm; women ≥ 88 cm); FFQ: food frequency questionnaire; DGB: Dietary Guidelines for the Brazilian Population; HR: hazards ratio; RR: relative risk; MPF: unprocessed or minimally processed food; MetS: metabolic syndrome; HT: hypertension; TAG: triacylglycerol; BG: blood glucose; HDL-C: high density lipoprotein cholesterol; MPF and PF ‘pattern’: factor derived ‘pattern’ of mainly MPF and processed food (PF); CRP = C-reactive protein; BP = blood pressure.