Growing evidence on the potentially harmful health effects of consuming ultra-processed foods (UPF) has drawn attention to the public health importance of industrial food processing.1-8
Definition Ultra-processed foods, as defined by the NOVA food classification system, are industrial formulations of ingredients that undergo a series of physical, chemical and biological processes. 9 They typically do not have intact healthy food components and include various additives.9 |
Ultra-processed foods tend to be more energy-dense and nutritionally poorer (i.e., high in levels of free sugar, salt and saturated fat, but low in levels of protein, dietary fiber and micronutrients) compared to less processed alternatives and They are designed to be cheap, palatable, durable, convenient and attractive.9
These products are aggressively marketed by the food industry to promote purchase and shape dietary preferences, and children are the primary consumers of UPF.9,10
The rapid expansion of global and industrialized food systems has gradually displaced traditional dietary patterns based on fresh and minimally processed foods, in favor of ready-to-eat UPFs.9,10
Currently, UPFs account for 65.4% and 66.2% of daily calorie intake among school-aged children in the UK and US, respectively.11,12 Growing consumption around the world, including low- and middle-income countries, has reflected a parallel increase in the prevalence of childhood and adult obesity globally, 9,10,13 suggesting that the consumption of UPFs may be a key underlying driver of the obesity epidemic and diseases. non-communicable diseases related to diet.9,10,14,15
A recent clinical trial found that UPF consumption leads to excessive calorie intake and weight gain in adults, 1 and cohort studies have reported associations between increased consumption and elevated risk of obesity, 2,3 type 2 diabetes. , 4.5 cardiovascular diseases, 6 cancer, 7 and mortality in adults.8
Associations of UPF consumption with adiposity in children and adolescents remain scarce, with only a few previous small-scale studies available.16-20 This study investigated prospective associations between UPF consumption and objectively evaluated measures of adiposity from childhood to young adulthood in a large cohort of British children.
| Methods |
> Data source
The Avon Longitudinal Study of Parents and Children (ALSPAC) is a prospective birth cohort study that initially enrolled 14,541 pregnant women residing in Avon, England, with an expected date of delivery between April 1, 1991 and April 31. December 1, 1992.21,22 Further enrollment after 1998 resulted in a sample of 14,888 children from singleton or multiple pregnancies.23 In this study, children were followed from ages 7 to 24 during the study period beginning September 1, 1992. 1998 to October 31, 2017.
Data were analyzed from March 1, 2020 to January 31, 2021. ALSPAC participants provided written informed consent, and ethical approval for the study was obtained from the ALSPAC Law and Ethics Committee and local research committees. research ethics. This study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidance.
The ALSPAC study website contains details of all available data via a searchable data dictionary and variable search tool (http://www.bristol.ac.uk/alspac/researchers/our-data /). From 2014, study data were collected and managed using REDCap electronic data capture tools hosted at the University of Bristol, Bristol, United Kingdom.24,25
> Outcome measures
Children were invited to a total of 10 clinical evaluations almost annually from 7 to 17 years of age and then at 24 years of age. Adiposity outcomes were measured following standardized procedures.26 Primary outcomes included body mass index (BMI), fat mass index (FMI), lean mass index (WMI), and percentage of total body fat.
Secondary outcomes were BMI z score, weight, waist circumference, fat mass, and lean mass. Height was measured with a commercially available stadiometer (Harpenden; Holtain); weight, using a body fat analyzer (Tanita); and the waist circumference, using a tape on the minimum circumference of the abdomen between the iliac crests and the lower ribs.26
Total body fat and lean mass were assessed using a dual-energy X-ray absorptiometry scanner (Lunar Prodigy; GE Medical Systems).26 They calculated BMI as weight in kilograms divided by height in square meters. BMI and BMI were calculated using dual-energy X-ray absorptiometry measurements of fat mass and lean mass, respectively, divided by height in meters squared. Total body fat was calculated as the percentage of fat mass divided by body mass.
Age- and sex-standardized BMI z-scores were calculated for ages 7 to 17 years because the 1990 British Growth Reference is only available for ages 23 years.27 The completeness of adiposity results ranged from 89.5% and 99.9% in the study cohort. The mean number of repeated measurements was 6.5 for BMI, BMI z score, and weight; 5.3 for waist circumference; and 3.9 for BMI, BMI, fat mass and lean mass.
> Food exhibition and industrial food processing grade
A 3-day food diary was sent to parents prior to the child’s clinical evaluation for the parent to complete at age 7 years and for the child to complete at ages 10 and 13.26 Parents were instructed respondents to record all foods and drinks the child consumed during 2 weekdays and 1 weekend day (not necessarily consecutive).26
Dietary data were reviewed by a nutritionist, and intakes were coded using DIDO (Diet In, Data Out) software and linked to McCance and Widdowson’s British food composition tables.26,28
The authors applied the NOVA food classification and categorized each food and beverage into 1 of 4 food groups based on their scope and purpose of industrial food processing9:
(1) Unprocessed/minimally processed foods are fresh, frozen, ground, pasteurized or fermented (without alcohol) after separation from nature (for example, fruit, vegetables, milk, meat, legumes).
(2) Processed culinary ingredients are substances extracted from foods and used in common culinary preparation, cooking, and seasoning of Group 1 foods (e.g., table salt, sugar, vegetable oils, and butter).
(3) Processed foods are made by adding salt, sugar, or other Group 2 ingredients to Group 1 foods (for example, canned vegetables in brine, canned fish, fresh breads, and cheeses).
(4) UPFs are multi-substance food and beverage formulations, mostly of products for industrial use only (e.g., high fructose corn syrup), and are manufactured through a series of complex industrial processes (e.g., hydrogenation) and often contain cosmetic food additives (e.g., colors, flavors, emulsifiers) that disguise undesirable sensory properties of the final product.9
Examples include carbonated or dairy-based beverages, industrially processed packaged breads with added preservatives or emulsifiers, and frozen or shelf-stable ready meals made with modified starches, stabilizers, or flavor enhancers.
> Covariate study
Covariates included children’s age at clinical evaluation, sex (male or female), race (white or non-white), birth weight (<2500, 2500-3999, or ≥4000 g), baseline physical activity (activity moderate to vigorous physical activity ≥60 minutes per day or otherwise), average daily calorie intake (continuous), and the quintiles of the 2004 Multiple Deprivation Index. The Multiple Deprivation Index is the most common measure of deprivation for each small area of England based on 7 domains.29
Physical activity was based on the earliest recording of accelerometry data (collected at ages 11, 13, and 15 years) where children were instructed to wear a uniaxial accelerometer (model 7164; Actigraph) for 7 days. They categorized the accelerometry data into 2 groups according to the UK government recommendation for children to accumulate at least 60 minutes of moderate to vigorous physical activity per day.26,30,31
Mothers’ self-reported data at baseline included pre-pregnancy BMI (<18.5, 18.5-24.9, 25-29.9, and ≥30), marital status (single or married/living with partner), highest educational attainment (Secondary School Certificate or none, vocational, O-level, A-level, or a degree or above) and socio-economic position based on UK national statistics socio-economic classification (senior managerial, administrative or professional; intermediate; or routine or manual occupation).32
> Statistical analysis
Data were analyzed from March 1, 2020 to January 31, 2021. A total of 9,025 children were included in the study after excluding 4,581 children who did not participate in any clinical evaluation, 1,271 children without dietary data, and 11 children without outcome measurement at or before the collection of their dietary data. Those included were more likely to be female, white, and of a higher socioeconomic status.
The age of each individual at the completion of their first dietary data collection was considered the reference; therefore 7264 (80.5%) were followed up from 7 years of age; 1519 (16.8%), from 10 years of age; and 242 (2.7%), from 13 years of age. Additionally, their dietary data were based on a 1-day food diary for 727 children (8.0%), a 2-day food diary for 1171 children (13.0%), and a 3-day food diary for 7127 children (79%).
For each child, the proportion of UPFs consumed in total daily food intake was calculated and expressed as a percentage. This was considered the primary exposure because it better captures zero-calorie UPFs, such as artificially sweetened beverages. However, they also derived for sensitivity analysis a secondary exposure defined as the percentage caloric contribution of UPF relative to total daily energy intake.
They categorized individuals’ baseline UPF intake into quintiles based on cutoff points based on dietary data at age 7 because most children were followed from age 7 onwards. They then compared this to quintiles derived from dietary data at ages 10 and 13. The quintiles were similar and no sex-specific differences were identified.
Time-varying exposure was not considered because although a total of 7072 children (78.4%) provided follow-up dietary data, an absolute change in UPF consumption of 20% or more was observed in only 1288 children. (14.2%) between 7 and 10 years old and in 1831 children (20.2%) between 10 and 13 years old.
Differences in baseline characteristics by UPF quintiles were compared using χ2 tests and analysis of variance where appropriate. Linear growth curve models were used to investigate longitudinal associations between baseline UPF quintile and trajectories of adiposity outcomes. These 2-level linear regression models allow for random intercept and slope modeled with age as the underlying time scale.
Models included 3 key variables: age, UPF quintile, and an interaction term between age and UPF quintile examining the difference in the mean growth trajectories of those in the highest UPF quintiles compared to the reference of the lowest quintile. They assessed nonlinearity by fitting a quadratic age term in the fixed and random parts of the growth models. These terms were retained if there was evidence of improved model fit.
They used multiple imputation by chained equation to impute missing covariate data (range, 1.8%-27.7%) under the assumption that they were missing at random. Five imputed data sets were generated where analytical models were performed on each, and the results were combined using Rubin’s rule.33
Analyzes based on complete data were performed for comparison. Study covariates were included in a stepwise manner. Model 1 did not adjust for any covariates; Model 2 was adjusted for sex, race, birth weight, physical activity level, and quintile of the child’s Multiple Deprivation Index; Model 3 additionally adjusted for mother’s prepregnancy BMI, marital status, highest educational level, and socioeconomic position; and model 4 was additionally adjusted for the child’s baseline daily energy intake.
> Sensitivity analysis
They performed a series of sensitivity analyses, including additional adjustment for baseline fruit and vegetable intake; intakes of saturated fat, sugar, fiber and sodium; restrictive analyzes to people with follow-up data; excluding twin children from the study cohort; stratifying by boys and girls; and recategorizing the reference UPF consumption into 5 groups by 20% absolute increase in their percentage of weight contribution towards daily food intake. All statistical analyzes were performed using Stata SE, version 12.1 (StataCorp LLC). All statistical tests were two-sided and p <0.05 was considered significant.
| Results |
A total of 9025 children (4481 [49.7%] females and 4544 [50.3%] males) were followed for a median of 10.2 (interquartile range, 5.2-16.4) years. Mean (SD) UPF consumption at baseline by quintile (Q1-Q5) was 23.2% (5%) of total daily food intake in Q1 (lowest), 34.7% ( 2.5%) in Q2, 43.4% (2.5%) in Q·, 52.7% (2.8%) in Q4 and 67.8% (8.1%) in Q5 (the most high).
Children assigned to different UPF quintiles were not significantly different by sex, race, or birth weight. However, children with higher UPF consumption were more likely to have lower maternal socioeconomic status profiles compared to those in lower UPF quintiles (e.g., 600 of 1858 [32.3%] for routine occupation or manual in Q5 vs. 418 of 1708 [24.5%] in Q1).
The main sources of UPF among children in Q5 included fruit-based drinks (22.2%), carbonated drinks (11.5%), ready-to-eat/heat foods (8.6%) and industrially produced breads and rolls (5.9%). In contrast, children’s diets in Q1 were largely based on minimally processed foods, including water and tea (22.2%), milk and plain yogurt (20.2%), and fruit (6%).
The results of the growth models were kept constant when adjusting for covariates in several steps. BMI at baseline (7 years of age) did not differ significantly between baseline UPF quintiles (e.g., β, 0.08 [95% CI, −0.09 to 0.24] for Q5 vs Q1). The mean BMI among children in Q1 increased by 0.55 (95% CI, 0.53-0.56) per year. However, increases in BMI were significantly greater among the 3 highest UPF quintiles with a dose-response association (e.g., BMI increased by an additional 0.06 [95% CI, 0.04-0 .08] per year in Q5 compared to Q1).
Mean GMI at baseline (9 years of age) was significantly higher at Q5 by 0.27 (95% CI, 0.09-0.45) compared to Q1. The mean GMI increased by 0.22 (95% CI, 0.20-0.23) per year in Q1, and this growth trajectory was found to be significantly greater in Q5 than in Q1 by an additional 0.03 (95% CI %, 0.01- 0.05) per year.
The mean percentage of body fat at baseline (9 years of age) was significantly higher among children in the 3 highest UPF quintiles (e.g., 1.47% [95% CI, 0.81% -2, 13%] higher in Q5 than in Q1).
However, increasing trajectories of body fat percentage were not significantly different between UPF quintiles. The mean BMI was estimated to grow at an annual rate of 0.55 - (2 × 0.02 × years of follow-up) starting at age 9, but neither the BMI at age 9 nor its growth trajectory was found significantly different between children from different UPF quintiles.
Mean levels of BMI z-score, weight, and waist circumference were not significantly different at baseline (age 7) across UPF quintiles, except for children’s weight at Q2 (β = 0 .35 [95% CI, 0.007-0.69]).
However, compared with children in Q1, weight gain and waist circumference trajectories were significantly greater in the highest quintiles 2 and 3 of UPF, respectively, with a dose-response association (p For example, mean weight increased an additional 0.10 [95% CI, 0.01-0.18] kg per year in Q4 compared to Q1 and an additional 0.20 [95% CI, 0.01-0.18] kg per year in Q4 compared to Q1. 11-0.28] kg per year in Q5 compared to Q1).
BMI z-score trajectories were only significantly higher at Q5 (β=0.01 [95% CI, 0.003-0.01]). Fat mass and lean mass results were similar to IMG and IMM findings, respectively.
At age 24, significantly higher mean levels of BMI were observed at 1.18 (95% CI, 0.78-1.57), GMI at 0.78 (95% CI, 0.46-1.08), body fat percentage in 1.53% (95% CI, 0.81% - 2.25%), weight in 3.66 (95% CI, 2.18-5.12) kg, and waist circumference in 3.08 (95% CI, 2.08-4.06) cm in Q5 compared to Q1.
The results of the sensitivity analyzes were largely consistent with the main findings. Girls were observed to have a steeper trajectory of body fat measurements than boys, although their BMI trajectories were similar.
Analyzes using secondary exposure showed that the mean UPF consumption in the study cohort was 61.4% of daily energy intake, and the main contributors to energy intake were ready-to-eat/heat UPF and breads and industrially processed buns.
| Discussion |
In this large prospective follow-up study of British children aged 7 to 24 years , growth trajectories among children with the highest (vs. lowest) UPF consumption increased by an additional 0.06 (95% CI, 0.04 -0.08) per year for BMI, 0.03 (95% CI, 0.01-0.05) per year for IMG, 0.20 (95% CI, 0.11-0.28) kg per year for weight, and 0.17 (95% CI, 0.11-0.22) cm per year for waist circumference trajectories among those in the 2 highest UPF quintiles. At 24 years of age, children with the highest (vs. lowest) UPF intake were found to have a higher BMI of 1.18 (95% CI, 0.78-1.57), 0.78 (95% CI, 0.46-1.08) and a higher percentage of body fat by 1.53% (95% CI, 0.81%-2.25%).
Previous cohort studies of children/adolescents (sample size, 307-3454 participants) 16-20 had shorter follow-up and yielded inconsistent results. Two studies16,17 found no significant associations between UPF consumption at 4 years of age and BMI measurements 3 to 4 years later, while 1 study20 reported no differences in BMI growth between 16 and 18 years of age. old.
However, a Portuguese study19 reported a 0.028 increase in BMI z-score at 10 years for every 100 kcal/d of increased UPF consumption at 4 years, and a Brazilian study18 reported a 0.20 increase in BMI and 0.14 increase in BMI, from 6 to 11 years of age per 100 g/d increase in UPF consumption.
The authors’ findings were based on multiple measurements of adiposity from ages 7 to 24 years and detailed 3-day food diaries, whereas previous studies relied heavily on food frequency questionnaires that may have a limited ability to accurately capture UPFs.
Notably, British children have a high consumption of UPF compared to previous studies based in Brazil, 16,18 Portugal, 19 or Spain17 (range, 27.3%-42% of daily caloric intake).
The positive longitudinal association between childhood consumption of sugary drinks and adiposity has been widely documented34; The authors’ results reflect this because sugar-sweetened and artificially sweetened beverages constituted a large proportion of UPF consumption, especially in those with the highest quintile of consumption (33.7%).
The increasing availability and variety of UPFs has reshaped global food systems by displacing dietary patterns that were previously based on fresh and minimally processed foods. Of particular concern is the increasing consumption of UPFs among children and adolescents, who are the main consumers, even in middle-income countries.11,12,35,36
These findings have important implications for public health, with increased consumption of UPFs associated with excessive calorie intake1 and elevated risk of obesity, 2,3 type 2 diabetes, 4,5 hypertension, 37 cardiovascular disease, 6 cancer, 7 and mortality.8
The authors’ findings add positive associations between UPF consumption and adiposity outcomes during childhood, which is critically important given that lifelong dietary patterns develop from childhood and can have widespread health consequences. and well-being throughout life.38
The UPF industry is highly profitable through the use of low-cost supply chains and aggressive marketing strategies to promote overconsumption.14,15
Global economic policies and trade agreements that favor the interests of transnational food corporations have further reinforced their central role in the global transformation of food systems and have undermined the implementation of effective policies to curb the consumption of UPFs.10 15 However, policies are emerging that explicitly target UPFs.10
Public health authorities in Brazil, Uruguay, Ecuador, Peru, France, Canada and Israel have modified their national dietary guidelines with recommendations to limit consumption of UPFs.10,39,40 France has set an ambitious goal to reduce consumption of UPFs by 20% by 2022. Action on UPFs in the UK and elsewhere remains limited, emphasizing instead the reduction of certain nutrients.14,41
Voluntary product reformulations have proven to be ineffective, 10,41 and even bolder regulations may not address health harms because they may miss several UPFs (e.g., artificially sweetened beverages) that contain industrial trans fatty acids. , 42 food additives or toxic contaminants, 43, 44 even when their calories, salt, and sugar content are reduced.
Only mandatory policies that target UPFs comprehensively, with globally cooperative strengthening of regulations and trade agreements to reduce the supply and consumption of UPFs, will offset the substantial burden of UPF consumption on the environment and health. healthcare systems around the world.14,41,45
| Limitations |
The authors’ study has several limitations.
1. First, in some people fewer adiposity measurements were collected and no data were collected between 17 and 24 years of age. However, completeness of outcome data was high in the study cohort (89.5%-99.9%) and a mean of 3.9 to 6.5 repeated measurements were available in the study results.
2. Second, misclassification of foods/drinks based on NOVA classification can occur, but this is probably minimal given the detailed food diaries that were used.
3. Third, large changes in UPF intake may contribute to a change in adiposity trajectories, but we did not use time-varying exposure due to modest changes in UPF intake between ages 7 and 13. year old.
4. Fourth, the availability of multiple food diaries reduces measurement bias, with only 727 (8%) of the cohort completing on a single occasion, while the majority of participants completed 2 or more days.
5. Fifth, they examined potential dietary misstatements based on the relationship between energy intake and estimated energy expenditure46. The results remained consistent after exclusion of 1314 underreporters (14.6%) and 715 overreporters (7.9%).
6. Sixth, missing data may introduce bias, but they used multiple imputation, while auxiliary variables were included as appropriate. A comparison of the main findings with those from complete case analyzes yielded similar results.
Finally, although the authors took into account a wide range of factors, the observational nature of the study means that residual confounding may have affected the results.
Conclusions
|
