Estimating added sugars in US consumer packaged goods: An application to beverages in 2007-08

Food Additives in Ultra-Processed Packaged Foods: An Examination of US Household Grocery Store Purchases

BACKGROUND

Food additives have been used mainly in the past century to perform specific functions in foods. Some types of food additives have been linked to adverse health outcomes, yet there is little research examining food additives in the US food supply.

OBJECTIVE

To examine the proportion of products purchased by US households containing four common technical food additives using time-specific food composition data and examine whether purchases have changed over time.

PARTICIPANTS/SETTING

Nielsen Homescan Consumer Panels, 2001 and 2019.

MAIN OUTCOME MEASURES

The proportion of packaged food products containing common types of food additives purchased by US households was determined overall and by food category.

STATISTICAL ANALYSIS PERFORMED

Differences were examined using Student t test; P value < 0.001 was considered significant.

RESULTS

Between 2001 and 2019, the proportion of food products purchased by US households that contained additives increased from 49.6% to 59.5% (P < 0.001). The proportion of carbonated soft drinks purchased containing flavors decreased, with a subsequent increase in purchases containing nonnutritive sweeteners. Baby foods showed a 20% increase in the proportion of purchases containing additives and >15% increase in the proportion of purchases containing three or more additives.

CONCLUSIONS

There is convincing evidence that US household purchases of common types of technical food additives are increasing. Despite some positive changes such as a decrease in the use of added flavors in carbonated soft drinks, across most food categories an increase in purchases of all types of products containing additives was observed. In particular the finding that purchases of baby food products containing additives have increased substantially is crucial in informing future research in this area and warrants further investigation.

A Machine Learning Approach to Predict the Added-Sugar Content of Packaged Foods

BACKGROUND

Dietary guidelines recommend limiting the intake of added sugars. However, despite the public health importance, most countries have not mandated the labeling of added-sugar content on packaged foods and beverages, making it difficult for consumers to avoid products with added sugar, and limiting the ability of policymakers to identify priority products for intervention.

OBJECTIVE

The aim was to develop a machine learning approach for the prediction of added-sugar content in packaged products using available nutrient, ingredient, and food category information.

METHODS

The added-sugar prediction algorithm was developed using k-nearest neighbors (KNN) and packaged food information from the US Label Insight dataset (n = 70,522). A synthetic dataset of Australian packaged products (n = 500) was used to assess validity and generalization. Performance metrics included the coefficient of determination (R2), mean absolute error (MAE), and Spearman rank correlation (ρ). To benchmark the KNN approach, the KNN approach was compared with an existing added-sugar prediction approach that relies on a series of manual steps.

RESULTS

Compared with the existing added-sugar prediction approach, the KNN approach was similarly apt at explaining variation in added-sugar content (R2 = 0.96 vs. 0.97, respectively) and ranking products from highest to lowest in added-sugar content (ρ = 0.91 vs. 0.93, respectively), while less apt at minimizing absolute deviations between predicted and true values (MAE = 1.68 g vs. 1.26 g per 100 g or 100 mL, respectively).

CONCLUSIONS

KNN can be used to predict added-sugar content in packaged products with a high degree of validity. Being automated, KNN can easily be applied to large datasets. Such predicted added-sugar levels can be used to monitor the food supply and inform interventions aimed at reducing added-sugar intake.

An Innovative Machine Learning Approach to Predict the Dietary Fiber Content of Packaged Foods

Underconsumption of dietary fiber is prevalent worldwide and is associated with multiple adverse health conditions. Despite the importance of fiber, the labeling of fiber content on packaged foods and beverages is voluntary in most countries, making it challenging for consumers and policy makers to monitor fiber consumption. Here, we developed a machine learning approach for automated and systematic prediction of fiber content using nutrient information commonly available on packaged products. An Australian packaged food dataset with known fiber content information was divided into training (n = 8986) and test datasets (n = 2455). Utilization of a k-nearest neighbors machine learning algorithm explained a greater proportion of variance in fiber content than an existing manual fiber prediction approach (R² = 0.84 vs. R² = 0.68). Our findings highlight the opportunity to use machine learning to efficiently predict the fiber content of packaged products on a large scale.

The adaptation, validation, and application of a methodology for estimating the added sugar content of packaged food products when total and added sugar labels are not mandatory

Nutrition policies recommend limiting the intake of added sugars. Information about added sugar content is not provided on packaged foods in Brazil, and even total sugar content information is often absent. This study aimed to (i) adapt a systematic methodology for estimating added sugar content in packaged foods when information on total and added sugar contents is not mandatory on labels, (ii) apply the adapted methodology to a Brazilian food composition database to estimate the extent of added sugar content in the national food supply, and (iii) assess the validity of the adapted methodology. We developed an 8-step protocol to estimate added sugar content using information provided on food labels. These steps included objective and subjective estimation procedures. Mean, median, and quartiles of the added sugar content of 4,805 Brazilian foods were determined and presented by food categories. Validity was assessed using a US database containing values of added sugar as displayed on the product labels. Objective estimation of added sugar content could be conducted for 3,119 products (64.9%), with the remainder 1,686 (35.1%) being assessed using subjective estimation. We found that 3,093 (64.4%) foods contained added sugar ingredients and the overall estimated median added sugar content was 4.7 g (interquartile range 0-29.3) per 100 g or 100 ml. The validity testing on US data for products with known added sugar values showed excellent agreement between estimated and reported added sugar values (ICC = 0.98). This new methodology is a useful approach for estimating the added sugar content of products in countries where both added and total sugar information are not mandated on food labels. The method can be used to monitor added sugar levels and support interventions aimed at limiting added sugar intake.

Use of added sugars in packaged foods sold in Brazil

Objective

Studies have shown that the consumption of added sugars may be associated with harmful health effects. The present study investigated the presence and types of added sugars in packaged foods.

Design

Cross-sectional analysis of the presence and types of added sugars in the ingredients lists of packaged foods sold in a major Brazilian supermarket. The nomenclature of ingredients classified as added sugars and their frequency were identified. Data were organised and analysed through descriptive statistics: absolute and relative frequencies of the presence of added sugars categorised by food groups were calculated, and ingredients were analysed by text mining using R statistical environment.

Setting

A supermarket in Florianópolis, a metropolis in southern Brazil.

Subjects

Packaged food products (n 4539) classified into eight groups.

Results

Of the 4539 products, 71 % had some type of added sugar. The group with the highest frequency of added sugars was ‘products in which carbohydrates and fats are the main energy source’ (93 %). Food groups containing predominantly salty foods had a high frequency of added sugars, such as ‘meats and eggs’ (61 %). In total, 179 different terms for added sugars were identified, of which sugar, maltodextrin and glucose syrup were the most frequent.

Conclusions

Most of the packaged foods sold in Brazil contain added sugars, which may hamper adherence to the recommendation of limiting added sugars intake. The data may be useful for monitoring tendencies in the use of added sugars in packaged foods and as supplementary information to support the improvement of food label regulations.

A Disaggregation Methodology to Estimate Intake of Added Sugars and Free Sugars: An Illustration from the UK National Diet and Nutrition Survey

Various and inconsistent definitions for free and added sugars are used in the consideration and assessment of dietary intakes across public health, presenting challenges for nutritional surveillance, research, and policy. Furthermore, analytical methods to identify those sugars which are not naturally incorporated into the cellular structure of foods are lacking, thus free and added sugars are difficult to estimate in an efficient and accurate way. We aimed to establish a feasible and accurate method that can be applied flexibly to different definitions. Based on recipe disaggregation, our method involved five steps and showed good repeatability and validity. The resulting Free Sugars Database provided data for seven components of sugars; (1) table sugar; (2) other sugars; (3) honey; (4) fruit juice; (5) fruit puree; (6) dried fruit; and (7) stewed fruit, for ~9000 foods. Our approach facilitates a standardized and efficient assessment of added and free sugars, offering benefit and potential for nutrition research and surveillance, and for the food industry, for example to support sugar reduction and reformulation agendas.

Development of a food composition database to monitor changes in packaged foods and beverages

In order to monitor nutritional changes in the US food supply and assess potential impact on individual dietary intake, an approach was developed to enhance existing standard food composition tables with time-varying product- and brand-specific information for barcoded packaged foods. A "Crosswalk" was formed between barcoded products and USDA foodcodes in a time-specific manner, such that sales-weighted average nutritional profiles were generated for each foodcode based on corresponding products (275,000 to 350,000 per 2-year cycle). This Crosswalk-enhanced food composition table was applied to dietary intake data from the National Health and Nutrition Examination Survey (cycles 2007-2008, 2009-2010, and 2011-2012). Total energy density of foods consumed by Americans from stores/vending was stable over time and differed by <5 kcal/100g using the Crosswalk-enhanced vs standard database. However, changes in the energy density of food groups were found utilizing the Crosswalk that were not detected using the standard database. Likewise, significant declines in energy intake from beverages among children (288±7.3 to 258±6.8 kcal/d) were found using the Crosswalk-enhanced database but were non-significant using the standard database. The Crosswalk approach can potentially augment national nutrition surveys by utilizing commercial food purchase and nutrient databases to capture changes in the nutrient content of packaged foods.

Trends in added sugars from packaged beverages available and purchased by US households, 2007-2012

Background: The US Food and Drug Administration's updated nutrition labeling requirements will include added sugars starting in July 2018, but no measure currently exists to identify the added sugar content of products and what it represents among purchases. Beverages are one of the first targets for reducing added sugar consumption, and hence are the focus here.Objective: Our goal was to estimate trends in added sugars in nonalcoholic packaged beverage products available in the United States and to estimate amounts of added sugars obtained from these beverages given the purchases of US households overall and by subpopulations.Design: On the basis of nutrition label data from multiple sources, we used a stepwise approach to derive the added sugar content of 160,713 beverage products recorded as purchased by US households in 2007-2012 (345,193 observations from 110,539 unique households). We estimated the amounts of added sugars obtained from packaged beverages US households reported buying in 2007-2008, 2009-2010, and 2011-2012, overall and by subpopulations based on household composition, race/ethnicity, and income. The key outcomes are added sugars in terms of per capita grams per day and the percentage of calories from packaged beverages.Results: Packaged beverages alone account for per capita consumption of 12 g/d of added sugars purchased by US households in 2007-2012, representing 32-48% of calories from packaged beverages. Whereas the absolute amount of added sugars from beverages has not changed meaningfully over time, the relative contribution of added sugars to calories from beverages has increased. Non-Hispanic black households and low-income households obtain both higher absolute and relative amounts of added sugars from beverages than non-Hispanic white households and high-income households (all P < 0.01).Conclusions: These results provide measures of added sugars from packaged beverages at both the product level and the population level in the United States and can be used for comparisons after the revised nutrition labels are implemented and for future monitoring.

Sweetening of the global diet, particularly beverages: patterns, trends, and policy responses

Evidence suggests that excessive intake of added sugars has adverse effects on cardiometabolic health, which is consistent with many reviews and consensus reports from WHO and other unbiased sources. 74% of products in the US food supply contain caloric or low-calorie sweeteners, or both. Of all packaged foods and beverages purchased by a nationally representative sample of US households in 2013, 68% (by proportion of calories) contain caloric sweeteners and 2% contain low-calorie sweeteners. We believe that in the absence of intervention, the rest of the world will move towards this pervasiveness of added sugars in the food supply. Our analysis of trends in sales of sugar-sweetened beverages around the world, in terms of calories sold per person per day and volume sold per person per day, shows that the four regions with the highest consumption are North America, Latin America, Australasia, and western Europe. The fastest absolute growth in sales of sugar-sweetened beverages by country in 2009-14 was seen in Chile. We believe that action is needed to tackle the high levels and continuing growth in sales of such beverages worldwide. Many governments have initiated actions to reduce consumption of sugar-sweetened beverages in the past few years, including taxation (eg, in Mexico); reduction of their availability in schools; restrictions on marketing of sugary foods to children; public awareness campaigns; and positive and negative front-of-pack labelling. In our opinion, evidence of the effectiveness of these actions shows that they are moving in the right direction, but governments should view them as a learning process and improve their design over time. A key challenge for policy makers and researchers is the absence of a consensus on the relation of beverages containing low-calorie sweeteners and fruit juices with cardiometabolic outcomes, since decisions about whether these are healthy substitutes for sugar-sweetened beverages are an integral part of policy design.