A novel food processing-based nutrition classification scheme for guiding policy actions applied to the Australian food supply

Extent of alignment between the Australian Dietary Guidelines and the NOVA classification system across the Australian packaged food supply

AIMS

The Australian Dietary Guidelines are currently being revised and ultra-processed foods have been identified as a high priority action area. To better understand how well the current Dietary Guidelines align with level of processing classifications, the aim of this study was to assess the alignment between the Australian Dietary Guidelines and the NOVA classification system for classifying the healthiness of packaged foods.

METHODS

Data were sourced from the Australian FoodSwitch dataset, which included 28 071 packaged food and beverage products available in major Australian supermarkets in 2022. Products were classified as (i) core or discretionary (Australian Dietary Guidelines) and (ii) non-ultra-processed or ultra-processed (NOVA). Agreement between the two systems (core vs. non-ultra-processed and discretionary vs. ultra-processed) was evaluated using the kappa statistic.

RESULTS

There was 'moderate' agreement (κ = 0.41, 95% CI: 0.40-0.42) between the Australian Dietary Guidelines and the NOVA system, with 69.8% of products aligned across the two systems. Alignment was more common for discretionary foods (80.6% were ultra-processed) than core foods (59.9% aligned were not-ultra-processed). Food categories exhibiting the strongest levels of alignment included confectionary, foods for specific dietary use, and egg and egg products. Discordance was most common for convenience foods, sugars, honey and related products, and cereal and grain products.

CONCLUSIONS

Despite moderate alignment between the Australian Dietary Guidelines and NOVA, the discordance observed for almost one-third of products highlights the opportunity to develop recommendations for ultra-processed foods within the guidelines to advise Australians how these foods should be considered as part of a healthy diet.

Modifying the Health Star Rating nutrient profiling algorithm to account for ultra-processing

AIM

To modify the Australian and New Zealand Health Star Rating to account for ultra-processing and compare the alignment of the modified ratings with NOVA classifications and the current Australian Dietary Guidelines classifications of core (recommended foods) and discretionary (foods to limit).

METHODS

Data was cross-sectionally analysed for 25 486 products. Four approaches were compared to the original Health Star Rating: (1) five 'negative' points added to ultra-processed products (modification 1; inclusion approach); (2) ultra-processed products restricted to a maximum of 3.0 Health Stars (modification 2; capping approach); (3 and 4) same approach used for modifications 1 and 2 but only applied to products that already exceeded 10 'negative' points from existing Health Star Rating attributes (modifications 3 and 4, respectively; hybrid approaches). Alignment occurred when products (i) received <3.5 Health Stars and were NOVA group 4 (for NOVA comparison) or discretionary (for Dietary Guidelines comparison), or (ii) received ≥3.5 Health Stars and were NOVA groups 1-3 or core.

RESULTS

All Health Star Rating modifications resulted in greater alignment with NOVA (ranging from 69% to 88%) compared to the original Health Star Rating (66%). None of the modifications resulted in greater alignment to the Dietary Guidelines classifications overall (69% to 76%, compared with 77% for the original Health Star Rating), but alignment varied considerably by food category.

CONCLUSIONS

If ultra-processing were incorporated into the Australian and New Zealand Health Star Rating, consideration of ultra-processing within the broader dietary guidance framework would be essential to ensure coherent dietary messaging to Australians.

The Development of a Food Frequency Questionnaire for the Assessment of Ultra-Processed Food Consumption in the Italian Adult Population: Protocol for a Validity and Reproducibility Study

BACKGROUND

Over the last decade, while studies on the detrimental effects of ultra-processed food (UPF) consumption have increased, methodological limitations on the quality of available evidence have emerged. Starting from a critical reassessment of the NOVA classification, this project will aim to develop and validate a food frequency questionnaire (FFQ), which is based on the processing of consumed foods and specifically designed to estimate the UPF consumption and total dietary intake of macro- and micronutrients in the Italian adult population.

METHODS

This study will take place in selected workplaces and include healthy males and females aged ≥18 years, residing in Italy and with Italian citizenship. The FFQ will be online, voluntary, self-administered, semi-quantitative, and designed to assess food intake over the past year and distinguish between industrial, artisanal, and home-made products. This project will consist of two phases. First, a pilot study will be conducted to obtain the final version of the FFQ. The current food consumption of the target population will be investigated, through a 24 h dietary recall, and the face validity of the new tool will be tested. The second phase will involve at least 436 participants. To assess reproducibility, the FFQ will be administered twice (at an interval of 3-10 months), and the test-retest method will be used. A 7-day weighed dietary record (WDR) will also be completed after each FFQ administration. To evaluate criterion validity, data from the two WDRs will be compared against those from the first FFQ administration.

CONCLUSIONS

The results will provide a new valid tool focused on food processing, potentially useful for future studies.

The Nutritional Quality of Food Donated to a Western Australian Food Bank

Food banks provide an indispensable service to people experiencing severe food insecurity. Food banks source donations from across the food system; however, the food redistributed to clients across the developed world is nutritionally poor. This, together with the increasing prevalence of diet-related diseases and food insecurity, has prompted a focus on nutritional quality. Despite more food being distributed via food banks in Australia, the nutritional quality of donated food remains unreported. This study analyzed all food (84,996 kg (1216 products)) donated to Foodbank WA over a 5-day period using diet-, food-, and nutrient-based nutrition classification schemes (NCSs). A total of 42% (27% of total weight) of donated food products were deemed 'unsuitable' and 19% (23% by weight) were 'suitable' according to all NCSs. There was no agreement on 39% of products (50% by weight). Overall, NOVA and the Healthy Eating Research Nutrition Guidelines (HERNG) (κ = 0.521) had the highest level of agreement and the ADGs and HERNGs the lowest (κ = 0.329). The findings confirm the poor nutritional quality of food donated to food banks and the need to work with donors to improve the food they donate. Fit-for-purpose nutrition guidelines are urgently needed for Australian food banks to support them in providing nutritious food to their vulnerable clients.

Flavour, emulsifiers and colour are the most frequent markers to detect food ultra-processing in a UK food market analysis

OBJECTIVE

To elucidate which markers of ultra-processing (MUP) and their combinations are best suited to detect ultra-processed food (UPF).

DESIGN

The study was based on the 206 food and 32 beverage items of the Oxford WebQ which encompass all major foods consumed in the UK. For each Oxford WebQ question, ingredient lists of up to ten matching different commercial products (n 2146) were researched online using data from the two market leaders of groceries in the UK sorted by relevance (Tesco) and by top sellers (Sainsbury's), respectively. According to the NOVA classification, sixty-five MUP were defined, and if the ingredient list of a food product was positive for at least one MUP, it was regarded as UPF. The percentage of UPF items containing specific MUP was calculated. In addition, all combinations of two to six different MUP were assessed concerning the percentage of identified UPF items.

SETTING

Cross-sectional analysis.

PARTICIPANTS

None.

RESULTS

A total of 990 products contained at least one MUP and were, therefore, regarded as UPF. The most frequent MUP were flavour (578 items, 58·4 % of all UPF), emulsifiers (353 items, 35·7 % of all UPF) and colour (262 items, 26·5 % of all UPF). Combined, these three MUP detected 79·2 % of all UPF products. Detection rate increased to 88·4 % of all UPF if ingredient lists were analysed concerning three additional MUP, that is, fibre, dextrose and firming agent.

CONCLUSIONS

Almost 90 % of all UPF items can be detected by six MUP.

Comparing product healthiness according to the Health Star Rating and the NOVA classification system and implications for food labelling systems: An analysis of 25 486 products in Australia

We investigated the extent of alignment between 'healthiness' defined by a food classification system that classifies foods and beverages primarily by their nutrient composition, the Health Star Rating (HSR) and a system that considers only the degree of processing of the product, the NOVA classification system. We used data for 25 486 products contained within the George Institute for Global Health's Australian 2022 FoodSwitch Dataset. Agreement between the two systems in the proportion of products classified as 'healthier' (HSR ≥3.5 or NOVA group 1-3) or 'less healthy' (HSR <3.5 or NOVA group 4) was assessed using the κ statistic. There was 'fair' agreement (κ = 0.30, 95%CI: 0.29-0.31) between both systems in the proportion of all products classified as healthier or less healthy. Approximately one-third (n = 8729) of all products were defined as 'discordant', including 34.3% (n = 5620) of NOVA group 4 products with HSR ≥3.5 (commonly convenience foods, sports/diet foods, meat alternatives, as well as products containing non-sugar sweeteners) and 34.1% (n = 3109) of NOVA group 1-3 products with HSR <3.5 (commonly single-ingredient foods such as sugars/syrups, full-fat dairy and products specially produced to contain no ultra-processed ingredients). Our analysis strengthens the evidence for the similarities and differences in product healthiness according to a nutrient-based classification system and a processing-based classification system. Although the systems' classifications align for the majority of food and beverage products, the discordance found for some product categories indicates potential for confusion if systems are deployed alongside each other within food policies.

Profiling ultra-processed foods in Thailand: sales trend, consumer expenditure and nutritional quality

BACKGROUND

Ultra-processed foods (UPF) are associated with adverse health outcomes. This study aimed to analyse the national trends in retail sales, consumer expenditure and nutritional quality of UPFs in Thailand.

METHODS

The study used data from the Euromonitor Passport database for analysis of retail sales and consumer expenditure, and from the Mintel Global New Products Database for nutritional analysis using the WHO Southeast Asian Region nutrient profile model.

RESULTS

The study found the highest per capita sales volume and value of UPFs in 2021 were sauces, dressings & condiments (8.4 kg/capita) and carbonated soft drinks (27.1 L/capita), respectively. However, functional & flavoured water, ready-made meals and baked goods had the highest observed (2012-2021) and expected (2021-2026) sales growth. Supermarkets were responsible for most of the UPF sales since 2012, but convenience stores had larger growth in retail values. Growth in consumer expenditure per capita on UPFs from 2012 to 2020, ranged between 12.7% and 34%, and till 2026 is forecast to grow between 26% and 30%. More than half of UPFs exceeded at least one nutrient cutoff, 59.3% for total fats, 24.8% for saturated fats, 68.2% for total sugars and 94.3% for sodium.

CONCLUSIONS

The findings suggest a need for regulatory and non-regulatory measures such as UPF taxation and marketing restrictions, and market incentives for producing non-UPFs. A system for regularly monitoring and evaluating healthiness (both nutritional and processing aspects) of food products, especially UPFs, is required.

Choosing an effective food classification system for promoting healthy diets in Thailand: a comparative evaluation of three nutrient profiling-based food classification systems (government, WHO, and Healthier Choice Logo) and a food-processing-based food classification system (NOVA)

Introduction

This study aimed to assess the nutritional quality of food and beverage products in Thailand by comparing four different food classification systems: the nutrient profiling-based food classification systems by the Department of Health (DOH), the WHO South-East Asia Region (WHO SEA), the Healthier Choice Logo (HCL), and the food-processing-based food classification system, NOVA.

Methods

This study used secondary data from the Mintel Global New Products Database (N = 17,414). Food subgroups were classified differently based on these four systems. The DOH classified food products into three groups: Group A-healthy pass or meeting standard, Group B-not meeting the standard, and Group C-far below standard. The WHO SEA classified food products into two groups: marketing prohibited products and marketing permitted products. The HCL classified food products into two groups: eligible products for the logo; and ineligible products for the logo. The NOVA classified food products into four groups: unprocessed or minimally processed foods (MP), processed culinary ingredients (PCI), processed foods (P), and ultra-processed foods (UPF). Descriptive statistics (percentage and frequency) were used for analysis. Agreement analysis was conducted using Cohen's kappa statistic between each pair of food classification systems.

Results

Of the total sample that could be classified by any of the four classification systems (n = 10,486), the DOH, the WHO SEA and the HCL systems classified products as healthy (Group A, marketing permitted or eligible for HCL logo) at 10.4, 11.1, and 10.9%, respectively. Only 5.6% were classified as minimally processed foods using NOVA and 83.1% were ultra-processed foods (UPFs). Over 50% of products classified as healthy by the nutrient profiling systems were classified as UPF according to the NOVA system. Products that were eligible for the HCL had the highest proportion of UPF products (84.4%), followed by the Group A products (69.2%) and the WHO marketing-permitted products (65.0%).

Conclusion

A hybrid food classification approach taking both nutrients and food processing into account is needed to comprehensively assess the nutritional quality of food and beverage products in Thailand.

Development of the Menu Assessment Scoring Tool (MAST) to Assess the Nutritional Quality of Food Service Menus

Preventing the rise in obesity is a global public health priority. Neighbourhood environments can help or undermine people's efforts to manage their weight, depending on availability of nutritious and nutrient-poor 'discretionary' foods. The proportion of household food budgets spent on eating outside the home is increasing. To inform nutrition policy at a local level, an objective assessment of the nutritional quality of foods and beverages on food service menus that is context-specific is needed. This study describes the development and piloting of the Menu Assessment Scoring Tool (MAST), used to assess the nutritional quality of food service menus in Australia. The MAST is a desk-based tool designed to objectively assess availability of nutrient-poor and absence of nutritious food and beverages on food service menus. A risk assessment approach was applied, using the best available evidence in an iterative way. MAST scores for 30 food service outlets in one Local Government Authority in Perth, Western Australia highlight opportunities for improvements. MAST is the first tool of its kind in Australia to assess the nutritional quality of food service menus. It was practical and feasible to use by public health nutritionists/dietitians and can be adapted to suit other settings or countries.