Dietary indices underpinning front-of-pack nutrition labels and health outcomes: a systematic review and meta-analysis of prospective cohort studies

BACKGROUND

Nutrient profiling systems are increasingly used to characterize the healthfulness of foods for front-of-package (FOP) labeling, which have been proposed as an effective public health strategy to help people make healthier food choices.

OBJECTIVE

This study aimed to review available evidence from cohort studies that evaluated the association of dietary indices underpinning FOP nutrition labels with all-cause mortality and incidence of cardiovascular diseases (CVDs) or cancer.

METHODS

PubMed, Web of Science, and Scopus were systematically searched up to October 2023. We included articles if they were prospective cohort studies, if the exposure was any dietary index underpinning FOP nutrition labels [e.g., the modified Food Standard Agency-Nutrient Profiling System (FSAm-NPS) and the Health Star Rating System], and if outcomes were all-cause mortality or incidence of or mortality due to CVD and cancer. Random-effects models were used to calculate the pooled hazard ratios (HRs) and 95% CIs.

RESULTS

We identified 11 records (7 unique prospective studies), which were included in the systematic review. The meta-analysis comprised 8 studies analyzing the FSAm-NPS dietary index (DI) as exposure. The pooled HRs associated with a 2-unit increase in the FSAm-NPS DI of all-cause mortality, CVD, and cancer risk were 1.06 (95% confidence interval [CI]: 0.99, 1.13; I2: 80%), 1.08 (95% CI: 1.00, 1.18; I2: 70%), and 1.09 (95% CI: 1.00, 1.19; I2: 77%), respectively. The Chilean Warning Label score and the Health Star Rating systems were examined by 1 study each and were significantly associated with the outcomes.

CONCLUSIONS

DIs underpinning most common FOP nutrition labels and reflecting nutrient-poor diets show a tendency toward an increased incidence of CVD and cancer, but the observed effects are quite modest in magnitude. Further studies at the population level are needed to support the widely shared hypothesis that FOP labels, possibly in conjunction with other interventions, may contribute to reduce noncommunicable disease risk. This meta-analysis was registered at PROSPERO as CRD42021292625.

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.

Comparison of Nutri-Score and Health Star Rating Nutrient Profiling Models Using Large Branded Foods Composition Database and Sales Data

Front-of-package nutrition labelling (FOPNL) is known as an effective tool that can encourage healthier food choices and food reformulation. A very interesting type of FOPNL is grading schemes. Our objective was to compare two market-implemented grading schemes-European Nutri-Score (NS) and Australian Health Star Rating (HSR), using large Slovenian branded foods database. NS and HSR were used for profiling 17,226 pre-packed foods and drinks, available in Slovenian food supply dataset (2020). Alignment between models was evaluated with agreement (% of agreement and Cohen's Kappa) and correlation (Spearman rho). The 12-month nationwide sales-data were used for sale-weighing, to address market-share differences. Study results indicated that both models have good discriminatory ability between products based on their nutritional composition. NS and HSR ranked 22% and 33% of Slovenian food supply as healthy, respectively. Agreement between NS and HSR was strong (70%, κ = 0.62) with a very strong correlation (rho = 0.87). Observed profiling models were most aligned within food categories Beverages and Bread and bakery products, while less aligned for Dairy and imitates and Edible oils and emulsions. Notable disagreements were particularly observed in subcategories of Cheese and processed cheeses (8%, κ = 0.01, rho = 0.38) and Cooking oils (27%, κ = 0.11, rho = 0.40). Further analysis showed that the main differences in Cooking oils were due to olive oil and walnut oil, which are favoured by NS and grapeseed, flaxseed and sunflower oil that are favoured by HSR. For Cheeses and cheese products, we observed that HSR graded products across the whole scale, with majority (63%) being classified as healthy (≥3.5 *), while NS mostly graded lower scores. Sale-weighting analyses showed that offer in the food supply does not always reflect the sales. Sale-weighting increased overall agreement between profiles from 70% to 81%, with notable differences between food categories. In conclusion, NS and HSR were shown as highly compliant FOPNLs with few divergences in some subcategories. Even these models do not always grade products equally high, very similar ranking trends were observed. However, the observed differences highlight the challenges of FOPNL ranking schemes, which are tailored to address somewhat different public health priorities in different countries. International harmonization can support further development of grading type nutrient profiling models for the use in FOPNL, and make those acceptable for more stake-holders, which will be crucial for their successful regulatory implementation.

The ability of five different front-of-pack labels to assist Australian consumers to identify healthy versus unhealthy foods

OBJECTIVE

The aim of this study was to assess the relative ability of different interpretive front-of-pack food labels to alert consumers to both healthier and unhealthier options to inform their food choices.

METHODS

One thousand Australians completed an online experiment where they rated the nutritional quality of sets of fictional products pre- and post-randomisation to one of five front-of-pack labels: Health Star Rating, Multiple Traffic Lights, Nutri-Score, Reference Intakes and Warning Label. Two sample z-tests were used to assess the ability of each label to facilitate the correct identification of the least and most healthy product options.

RESULTS

The Nutri-Score was superior in assisting respondents to identify both the healthiest and unhealthiest options. The Health Star Rating ranked second for both outcomes, followed by the Multiple Traffic Lights.

CONCLUSIONS

Results reinforce the role of interpretive front-of-pack labels in assisting consumers to understand the nutritional quality of food products and suggest spectrum labels may provide superior utility in assisting consumers to identify both the most and least nutritious products from among available product options.

IMPLICATIONS FOR PUBLIC HEALTH

The strongest performance of a highly interpretive front-of-pack label (Nutri-Score) featuring colour in a summary indicator suggests potential strategies for enhancing the performance of the Health Star Rating.

Oh So Sweet: A Comparative Investigation of Retail Market Composition of Sweetened and Flavoured Beverages in Singapore and Australia

The consumption of sugar and non-nutritive sweeteners has been associated with poor health outcomes. The aim of this paper was to provide a comparison of the range of sweetened or flavoured beverages between two high-income countries in the Asia-Pacific region: Australia and Singapore. Following the FoodTrackTM methodology, nutrition, labelling, and price data were collected from major Australian and Singaporean supermarket chains and convenience stores. The nutrient profiles of products were tested for differences using Kruskal−Wallis and Mann−Whitney U tests. The greatest number of products collected in Australia were from the ‘carbonated beverages’ category (n = 215, 40%), and in Singapore the greatest number of products were from the ‘tea and coffee ready-to-drink’ category (n = 182, 35%). There were more calorically sweetened beverages in Singapore compared with Australia (n = 462/517 vs. n = 374/531, p < 0.001). For calorically sweetened products, the median energy of Singaporean products was significantly higher than Australian products (134 kJ vs. 120 kJ per 100 mL, p = 0.009). In Australia, 52% of sweetened or flavoured beverages displayed a front-of-pack nutrient signposting logo, compared with 34% of sweetened or flavoured beverages in Singapore. These findings also indicate that the consumption of just one serving of calorically sweetened carbonated beverages or energy drinks would exceed the WHO maximum daily free sugar recommendations.

Are Front-of-Pack Nutrition Labels Influencing Food Choices and Purchases, Diet Quality, and Modeled Health Outcomes? A Narrative Review of Four Systems

Front-of-Pack Nutrition labels (FOPNLs) aim to improve consumers' food purchases and prompt product reformulation by the food and beverage industry. Despite their widespread use, the effectiveness of FOPNL in achieving these goals is still a matter of debate. This review has gathered 65 original studies exploring the performances of four widely used FOPNLs (Multiple Traffic Light, Warning signs, Nutri-Score and Health Star Rating). Although FOPNLs have been associated with healthier food purchases, the magnitude of improvements was small and dependent on study settings. Any associated health effects were modeled rather than observed. None of the four FOPNLs clearly outperformed the other ones on any outcome. Few studies dealt with the impact of FOPNL on product reformulation. Some of those studies, but not all, found small reductions in energy, sodium, sugar and saturated fat content of foods in some food categories. Although global trends point to a small favorable effect of FOPNL, this conclusion is subject to caution since the evidence is inconsistent and comes from a wide variety of contexts and study designs. There remain numerous research gaps, notably with regard to the optimal characteristics of FOPNLs, the durability of FOPNL effects on consumer behaviors, and any possible unexpected consequences.

Front-of-Package Labels on Unhealthy Packaged Foods in India: Evidence from a Randomized Field Experiment

Policies to require front-of-package labels (FOPLs) on packaged foods may help Indian consumers to better identify foods high in nutrients of concern, including sugar, saturated fat, and sodium, and discourage their consumption, which are outcomes that are critical for preventing rises in diet-related non-communicable disease. The objective was to test whether FOPLs helped Indian consumers identify “high-in” packaged foods and reduce intentions to purchase them. We conducted an in-person randomized experiment (n = 2869 adults between ages 18 and 60 years old) in six states of India in 2022. Participants were randomized to one of five FOPLs: a control label (barcode), warning label (octagon with “High in [nutrient]”), Health Star Rating (HSR), Guideline Daily Amount (GDA), or traffic light label. Participants then viewed a series of packaged foods high in sugar, saturated fat, or sodium with the assigned FOPL, and rated product perceptions and label reactions. Fewer than half of participants in the control group (39.1%) correctly identified all products high in nutrient(s) of concern. All FOPLs led to an increase in this outcome, with the biggest differences observed for the warning label (60.8%, p < 0.001), followed by the traffic light label (54.8%, p < 0.001), GDA (55.0%, p < 0.001), and HSR (45.0%, p < 0.01). While no FOPLs led to a reduction in intentions to purchase the packaged foods, the overall pattern of results suggested that warning labels are the most effective FOPL to help Indian consumers identify unhealthy foods.

Does the prevalence of promotions on foods and beverages vary by product healthiness? A population-based study of household food and drink purchases in New Zealand

OBJECTIVE

To assess the prevalence of promotions on foods and non-alcoholic drinks purchased by New Zealand households and to determine if they vary according to healthiness of products.

DESIGN

We undertook a cross-sectional analysis of Nielsen New Zealand Homescan® 2018/19 panel data. We conducted multivariate analyses to examine the variability in quantities of healthy v. unhealthy food and beverage products purchased on promotion. Promotion was self-reported by the panellist. Healthiness of products was measured by the Health Star Rating (HSR) system. We also carried out a subgroup analysis for beverages according to the threshold of < 5 g v. ≥ 5 g sugar per 100 ml content of products.

SETTING

The Nielsen New Zealand Homescan® data were linked with two New Zealand Food Composition Databases (Nutritrack and the FOODfiles).

PARTICIPANTS

Food and beverage purchases data by 1800 panel households were used.

RESULTS

Overall, 46 % (1 803 601/3 940 458) of all purchases made were on promotion. Compared with purchases of food and beverage products with HSR < 3·5 (unhealthy), food and beverage products with HSR ≥ 3·5 (healthy) were significantly less likely to be on promotion (OR = 0·78, 95 % CI 0·77, 0·79). The subgroup analysis for beverages shows that products with < 5 g sugar per 100 ml were significantly less likely to be on promotion than those with ≥ 5 g sugar per 100 ml (OR = 0·77, 95 % CI 0·75, 0·79).

CONCLUSIONS

Policies to improve healthy food retailing should focus on increasing the promotion of healthier food and drink options in stores and supermarkets.

Measuring the Healthiness of Ready-to-Eat Child-Targeted Cereals: Evaluation of the FoodSwitch Platform in Sweden

BACKGROUND

Childhood obesity is a major public health issue. The increase in the consumption of foods with poor nutritional value, such as processed foods, contributes to this. Breakfast cereals are often advertised as a healthy way to start the day, but the healthiness of these products varies greatly.

OBJECTIVE

Our main objective was to gather information about the nutritional characteristics of ready-to-eat breakfast cereals in Sweden and to investigate the healthiness of products targeted at children compared to other cereals by use of the FoodSwitch platform. A secondary objective was to evaluate the alignment between the Keyhole symbol and the Health Star Rating.

METHODS

The FoodSwitch app is a mobile health (mHealth) tool used to present nutrition data and healthier alternative products to consumers. Ready-to-eat breakfast cereals from the largest Swedish grocery retailers were collected using the FoodSwitch platform. Products were defined as targeting children if they presented features addressing children on the package.

RESULTS

Overall, information on 261 ready-to-eat cereals was examined. Of this total, 8% (n=21) were targeted at children. Child-targeted cereals were higher in sugar (22.3 g/100 g vs 12.8 g/100 g, P<.001) and lower in fiber (6.2 g/100 g vs 9.8 g/100 g, P<.001) and protein (8.1 g/100 g vs 10.5 g/100 g, P<.001). Total fat (3 g/100 g vs 10.5 g/100 g, P<.001) and saturated fat (0.8 g/100 g vs 2.6 g/100 g, P<.001) were also lower. No difference was found in salt content (P=.61). Fewer child-targeted breakfast cereals displayed an on-pack Keyhole label (n=1, 5% vs n=53, 22%; P=.06), and the mean Health Star Rating value was 3.5 for child-targeted cereals compared to others (mean 3.8, P=.07). A correlation was found between the Keyhole symbol and the Health Star Rating.

CONCLUSIONS

Ready-to-eat breakfast cereals targeted at children were less healthy in terms of sugar and fiber content compared to products not targeted at children. There is a need to improve the nutritional quality of child-targeted cereals.

Growth of Ready Meals in Australian Supermarkets: Nutrient Composition, Price and Serving Size

Pre-prepared, or ready meals (frozen, chilled and shelf-stable) are increasingly available in supermarkets in developed countries. This study aimed to investigate how the range of ready meals in Australian supermarkets has changed from 2014 to 2020, and how products vary by price, serving size, nutrient composition and Health Star Rating. Product information was obtained from the FoodTrack™ packaged food database for the years 2014 to 2019 and from an instore audit of products available in Adelaide, Australia for 2020. There was a 13% annual average increase in the number of ready meals available in supermarkets. Serving size did not change (median 350 g, p-trend = 0.100) and price increased modestly from 2014 to 2020 (median $1.67 to $1.79/100 g, p-trend < 0.001), with chilled ready meals being the most expensive. A modest decrease in sodium density from 2014 to 2020 (median 275 to 240 mg/100 g, p-trend < 0.001) was seen. However, the category has a wide range in Health Star Ratings and nutrient composition, highlighting the importance of appropriate consumer choice to optimise health benefits. With the increasing availability of ready meals, global improvements within this category should be encouraged and consumers guided to choose healthier products.

Perspective: Why Whole Grains Should Be Incorporated into Nutrient-Profile Models to Better Capture Nutrient Density

Healthy eating patterns, as described by dietary guidelines, typically favor whole grains, low-fat dairy, vegetables, fruit, legumes, and nuts and seeds. Nutrient-profiling (NP) models capture nutrient density of individual foods and can inform healthier food choices. Although whole grains are prominently featured in most dietary guidelines, they are not included in most NP models. Healthy foods, as identified by most NP models, are those that contain limited amounts of energy, saturated fat, total or added sugar, and sodium. As global dietary guidance turns to foods and food groups as opposed to individual nutrients, future nutrient-density metrics may need to do the same. Potential methods to incorporate whole grains into the overall concept of nutrient density and into selected NP models are outlined in this review. Incorporating whole grains into the Nutri-Score, Health Star Rating, or the Nutrient Rich Food index will require further analyses of dietary nutrient density in relation to health outcomes across diverse population subgroups. We present the rationale for how the inclusion of whole grains in NP models can assist in the implementation of dietary guidance.

Seventeen-Year Associations between Diet Quality Defined by the Health Star Rating and Mortality in Australians: The Australian Diabetes, Obesity and Lifestyle Study (AusDiab)

BACKGROUND

The Health Star Rating (HSR) is the government-endorsed front-of-pack labeling system in Australia and New Zealand.

OBJECTIVES

We aimed to examine prospective associations of a dietary index (DI) based on the HSR, as an indicator of overall diet quality, with all-cause and cardiovascular disease (CVD) mortality.

METHODS

We utilized data from the national population-based Australian Diabetes, Obesity and Lifestyle Study. The HSR-DI at baseline (1999-2000) was constructed by 1) calculation of the HSR points for individual foods in the baseline FFQ, and 2) calculation of the HSR-DI for each participant based on pooled HSR points across foods, weighted by the proportion of energy contributed by each food. Vital status was ascertained by linkage to the Australian National Death Index. Associations of HSR-DI with mortality risk were assessed by Cox proportional hazards regression.

RESULTS

Among 10,025 eligible participants [baseline age: 51.6 ± 14.3 y (mean ± standard deviation)] at entry, higher HSR-DI (healthier) was associated with higher consumption of healthy foods such as fruits, vegetables, and nuts, and lower consumption of discretionary foods such as processed meats and confectionery (P-trend < 0.001 for each). During a median follow-up of 16.9 y, 1682 deaths occurred with 507 CVD deaths. In multivariable models adjusted for demographic characteristics, lifestyle factors, and medical conditions, higher HSR-DI was associated with lower risk of all-cause mortality, with a hazard ratio (95% confidence interval) of 0.80 (0.69, 0.94; P-trend < 0.001) comparing the fifth with the first HSR-DI quintile. A corresponding inverse association was observed for CVD mortality (0.71; 0.54, 0.94; P-trend = 0.008).

CONCLUSIONS

Better diet quality as defined by the HSR-DI was associated with lower risk of all-cause and CVD mortality among Australian adults. Our findings support the use of the HSR nutrient profiling algorithm as a valid tool for guiding consumer food choices.

The Effects of a Supermarket-Based Intervention on the Nutritional Quality of Private-Label Foods: A Prospective Study

Private-label products, products owned by supermarkets, are a growing area of the food supply. The aim of this study was to assess the effect of an intervention that provided an Australian supermarket (‘intervention supermarket’) with comparative nutrition data to improve the healthiness of their private-label range. Between 2015 and 2016, the intervention supermarket received reports that ranked the nutritional quality of their products against competitors. Changes in the nutrient content (sodium, sugar, saturated fat, energy and Health Star Rating) of products from the intervention supermarket between 2015 and 2018 were compared against changes achieved for three comparators (private-label products from two other supermarkets and branded products). The intervention supermarket achieved a significantly greater reduction in the sodium content of their products relative to all three comparators, which ranged between −104 and −52 mg/100 g (all p < 0.05). Conversely, the three comparators each achieved a greater relative reduction in the sugar content of their products by between −3.5 and −1.6 g/100 g (all p < 0.05). One of the comparators also had a greater relative reduction in the saturated fat and energy content of their products compared to the intervention supermarket (both p < 0.05). There were negligible differences in the Health Star Rating of products between the intervention supermarket and comparators (all p > 0.05). Providing comparative nutrition information to a supermarket may be ineffective in improving the healthiness of their private-label products, likely due to competing factors that play a role in the decision-making process behind product reformulation and product discontinuation/innovation.

The nutritional quality of supermarket own brand chilled convenience foods: an Australian cross-sectional study reveals limitations of the Health Star Rating

OBJECTIVE

To assess the nutritional quality of Australian supermarket own brand chilled convenience foods (SOBCCF), for example, ready meals, pizza, pies and desserts.

DESIGN

Cross-sectional.

SETTING

Two large supermarkets (Coles and Woolworths) in Perth, Western Australia were audited in February 2017.

PARTICIPANTS

Data were extracted from photographic images of 291 SOBCCF, including front-of-pack information (i.e. product name, description and nutrition labels including Health Star Rating (HSR)) and back-of-pack information (i.e. nutrition information panel and ingredients list). SOBCCF were classified as healthy or unhealthy consistent with principles of the Australian Guide to Healthy Eating (AGTHE), NOVA classification of level of food processing and HSR score.

RESULTS

Fifty-four percentage of SOBCCF were classified as unhealthy according to AGTHE principles, 94 % were ultra-processed foods using NOVA and 81 % scored a HSR of ≥2·5, implying that they were a healthy choice. Some convenience food groups comprised more healthy choices overall including prepared vegetables, salad kits and bowls, soups and vegetarian food. A significantly larger proportion of SOBCCF from Coles were classified as unhealthy compared with Woolworths (70 v. 44 %, P < 0·05) using the AGTHE.

CONCLUSIONS

The findings suggest there is potential for Australian supermarkets to improve the nutritional quality of their SOBCCF and highlights the differences between supermarkets in applying their corporate social responsibility policies. Policies to assist consumers to select healthier foods should address difficulties in identifying healthy convenience foods. The findings reveal misclassification of unhealthy SOBCCF as healthy by the HSR suggesting that its algorithm should be reformed to align with recommendations of the AGTHE.

Stars versus warnings: Comparison of the Australasian Health Star Rating nutrition labelling system with Chilean Warning Labels

OBJECTIVE

The Health Star Rating (HSR) is a voluntary front-of-pack nutrition labelling system that rates products from ½ to 5 stars (five being healthiest). The Chilean Warning Label system displays warnings on foods high in sugar, saturated fat, sodium, or energy. We aimed to evaluate alignment between the systems.

METHODS

New Zealand packaged products (n=13,868) were classified according to the two systems. Alignment was assessed by cross-checking the number of products meeting the criteria for warnings against star ratings. Products with no warnings but an HSR <2, or with >1 warning but an HSR of ≥3.5 were considered outliers.

RESULTS

Two-thirds of products met the criteria for at least one warning. There was a significant positive relationship between the number of warnings and mean HSR: 0 warnings = HSR 3.77±.0166 (p<0.001), 1 warning = HSR 2.70±.0206 (p<0.001) and >1 warning = HSR 2.00±.0160 (p<0.001). The systems were non-aligned for 1,117 products (8%).

CONCLUSION

HSR and the Chilean Warning Label systems are broadly aligned. Non-alignment is due to the Chilean system restricting warnings to foods containing added ingredients and HSR awarding points for positive components. Implications for public health: These results could be helpful in informing improvements to the HSR system.

The Healthfulness of the US Packaged Food and Beverage Supply: A Cross-Sectional Study

The US food supply is dominated by highly-processed packaged food and beverage products that are high in energy, saturated fat, sugar, and salt. We report results of a cross-sectional assessment of the 2018 US packaged food and beverage supply by nutritional composition and indicators of healthfulness and level of processing. Data were obtained through Label Insight’s Open Data database, which represents >80% of all food and beverage products sold in the US over the past three years. Healthfulness and the level of processing, measured by the Health Star Rating (HSR) system and the NOVA classification framework, respectively, were compared across product categories and leading manufacturers. Among 230,156 food and beverage products, the mean HSR was 2.7 (standard deviation (SD) 1.4) from a possible maximum rating of 5.0, and 71% of products were classified as ultra-processed. Healthfulness and level of processing varied substantially by category (range: HSR 1.1–3.9; 0–100% ultra-processed) and manufacturer (range: HSR 0.9–4.6; 26–100% ultra-processed). The US packaged food and beverage supply is large, heterogeneous, highly processed, and generally unhealthy. The wide variability in healthfulness and level of processing demonstrates that opportunities exist, through reformulation or replacement, for large-scale improvements to the healthfulness of the US packaged food and beverage supply.

Nutrient profiling and food prices: what is the cost of choosing healthier products?

BACKGROUND

The Health Star Rating (HSR) is a front-of-pack label designed to help Australian consumers identify healthier packaged foods. Price is an important determinant of food choice and yet no previous studies have examined the relationship between HSR and price. In the present study, we investigated whether (i) healthier packaged food products, as determined by HSR, are more expensive than less healthy alternatives and (ii) products displaying the HSR are more expensive than similar products that do not.

METHODS

Prices of three packaged foods categories (breakfast cereals, cereal-based bars and fruit juices) and nutrient data (to calculate HSR) were obtained from shopping receipts of approximately 1600 Australians between June 2014 and September 2016. Associations between HSR and price [per energy ($/100 kJ) and per unit ($/100 g)] for products of comparable package sizes were assessed by linear regression and the results are presented as differences in average price over the theoretical maximum range of HSR from 0.5 to 5 stars.

RESULTS

The HSR of products was not consistently related to price. Small positive associations were observed for juice ($0.08/100 mL; P = 0.03) and for cereal-based bars ($0.04/100 kJ; P = 0.02). No other associations between HSR and price were observed (P ≥ 0.23). Products that displayed the HSR were no more expensive on average than products that received a similar HSR but did not display the HSR (P ≥ 0.16).

CONCLUSIONS

In summary, the findings of the present study suggest that healthier packaged food products were not consistently more expensive than less healthy products and also that price is unlikely to be a barrier for consumers to use the HSR to select healthier packaged foods.

A comparison of the Health Star Rating and nutrient profiles of branded and generic food products in Sydney supermarkets, Australia

OBJECTIVE

To compare the Health Star Rating (HSR) and the nutritional profile of branded and generic packaged foods in Australia.

DESIGN

In-store audits of packaged food products capturing data on HSR and nutritional content to analyse differences between branded and generic foods across ten food categories.

SETTING

The audit was conducted in four major supermarket chains across various locations within metropolitan Sydney regions, Australia.

RESULTS

A total of 6269 products were analysed with 57 % of generic products and 28 % of branded products displaying an HSR. The median HSR of branded products was significantly greater than for generic products overall (4·0 v. 3·5, P&lt;0·005) and in six out of ten food categories (P&lt;0·005). However, when branded products could be matched to their generic counterparts for paired comparisons (n 146), no statistical difference was observed in all ten food categories. Branded products that chose to display an HSR had significantly lower saturated fat and Na, but higher fibre contents than branded products not displaying an HSR.

CONCLUSIONS

Our data show no difference in the HSR or nutrient profiles of similar branded and generic products that display HSR. Branded products appear to exploit the voluntary nature of the HSR scheme, preferentially displaying an HSR on healthier products compared with their generic counterparts.

Nutritional quality and reformulation of a selection of children's packaged foods available in Australian supermarkets: Has the Health Star Rating had an impact?

AIM

To examine whether the nutritional quality of children's packaged food products available in Australian supermarkets improved between 2013 and 2016, and whether any change could be detected in product reformulation since the introduction of the Health Star Rating (HSR) labelling scheme.

METHODS

Packaged food products marketed towards children were purchased from three Australian supermarkets in July 2013 (for a previous study) and July 2016. Nutritional quality was assessed using the Food Standards Australian New Zealand Nutrient Profiling Scoring Criterion. Comparisons were made between the nutrient composition and formulation of products (a) available in 2013 and 2016; and (b) with and without HSR graphics.

RESULTS

Of the 252 children's packaged products analysed, 53.6% were classified as 'less healthy'. HSR-labelled products had a significantly higher proportion classified as 'healthy' than those without the HSR (χ² = 26.5; P < 0.0001; 73.8% and 59.0%, respectively). Overall, 28.5% displayed the HSR; the majority (81.5%) having a rating of ≥3.0 stars. Cereal-based products had the greatest uptake of the scheme, with HSR-labelled products having significantly lower mean energy and saturated fat content (P < 0.01) and higher mean protein and fibre content (P < 0.001) than non-HSR products. Reformulation of products that were available in 2013 had occurred in 100% of HSR-labelled products in comparison to 61.3% of non-HSR labelled products.

CONCLUSIONS

Despite the introduction of the HSR, more than half of children's packaged foods sampled are 'less healthy'. However, early indications suggest that the HSR may stimulate healthier product reformulation.

Alignment of Supermarket Own Brand Foods' Front-of-Pack Nutrition Labelling with Measures of Nutritional Quality: An Australian Perspective

Two voluntary front-of-pack nutrition labels (FOPNL) are present in Australia: the government-led Health Star Ratings (HSR) and food industry-led Daily Intake Guide (DIG). Australia's two largest supermarkets are key supporters of HSR, pledging uptake on all supermarket own brand foods (SOBF). This study aimed to examine prevalence of FOPNL on SOBF, and alignment with patterns of nutritional quality. Photographic audits of all SOBF present in three large supermarkets were conducted in Perth, Western Australia, in 2017. Foods were classified as nutritious or nutrient-poor based on the Australian Guide to Healthy Eating (AGTHE), NOVA level of food processing, and HSR score. Most (81.5%) SOBF featured FOPNL, with only 55.1% displaying HSR. HSR was present on 69.2% of Coles, 54.0% of Woolworths, and none of IGA SOBF. Half (51.3%) of SOBF were classified as nutritious using the AGTHE, but using NOVA, 56.9% were ultra-processed foods. Nutrient-poor and ultra-processed SOBF were more likely than nutritious foods to include HSR, yet many of these foods achieved HSR scores of 2.5 stars or above, implying they were a healthy choice. Supermarkets have a powerful position in the Australian food system, and they could do more to support healthy food selection through responsible FOPNL.

Evaluation of Alignment between the Health Claims Nutrient Profiling Scoring Criterion (NPSC) and the Health Star Rating (HSR) Nutrient Profiling Models

In Australia, manufacturers can use two government-endorsed approaches to advertise product healthiness: the Health Star Rating (HSR) front-of-pack nutrition labelling system, and health claims. Related, but different, algorithms determine the star rating of a product (the HSR algorithm) and eligibility to display claims (the Nutrient Profiling Scoring Criterion (NPSC) algorithm). The objective of this study was to examine the agreement between the HSR and NPSC algorithms. Food composition information for 41,297 packaged products was extracted from The George Institute’s FoodSwitch database. HSR and the NPSC scores were calculated, and the proportion of products in each HSR category that were eligible to display a health claim under the NPSC was examined. The highest agreement between the HSR scoring algorithm and the NPSC threshold to determine eligibility to display a health claim was at the HSR cut-off of 3.5 stars (k = 0.83). Overall, 97.3% (n = 40,167) of products with star ratings of 3.5 or higher were also eligible to display a health claim, and 94.3% (n = 38,939) of products with star ratings less than 3.5 were ineligible to display a health claim. The food group with greatest divergence was “edible oils”, with 45% products (n = 342) with HSR >3.5, but 64% (n = 495) eligible to display a claim. Categories with large absolute numbers of products with HSR <3.5, but eligible to display a claim, were “yoghurts and yoghurt drinks” (335 products, 25.4%) and “soft drinks” (299 products, 29.7%). Categories with a large number of products with HSR ≥3.5, but ineligible to display a claim, were “milk” (260 products, 21.2%) and “nuts and seeds” (173 products, 19.7%). We conclude that there is good agreement between the HSR and the NPSC systems overall, but divergence in some food groups is likely to result in confusion for consumers, particularly where foods with low HSRs are eligible to display a health claim. The alignment of the NPSC and HSR scoring algorithms should be improved.

Comparative Analysis of the Classification of Food Products in the Mexican Market According to Seven Different Nutrient Profiling Systems

Nutrient profiling systems (NPS) are used around the world. In some countries, the food industry participates in the design of these systems. We aimed to compare the ability of various NPS to identify processed and ultra-processed Mexican products containing excessive amounts of critical nutrients. A sample of 2544 foods and beverages available in the Mexican market were classified as compliant and non-compliant according to seven NPS: the Pan American Health Organization (PAHO) model, which served as our reference, the Nutrient Profiling Scoring Criterion (NPSC), the Mexican Committee of Nutrition Experts (MCNE), the Health Star Rating (HSR), the Mexican Nutritional Seal (MNS), the Chilean Warning Octagons (CWO) 2016, 2018 and 2019 criteria, and Ecuador's Multiple Traffic Light (MTL). Overall, the proportion of foods classified as compliant by the HSR, MTL and MCNE models was similar to the PAHO model. In contrast, the NPSC, the MNS and the CWO-2016 classified a higher amount of foods as compliant. Larger differences between NPS classification were observed across food categories. Results support the notion that models developed with the involvement of food manufacturers are more permissive than those based on scientific evidence. Results highlight the importance of thoroughly evaluating the underlying criteria of a model.

Cost-Effectiveness of Product Reformulation in Response to the Health Star Rating Food Labelling System in Australia

The Health Star Rating (HSR) system is a voluntary front-of-pack labelling (FoPL) initiative endorsed by the Australian government in 2014. This study examines the impact of the HSR system on pre-packaged food reformulation measured by changes in energy density between products with and without HSR. The cost-effectiveness of the HSR system was modelled using a proportional multi-state life table Markov model for the 2010 Australian population. We evaluated scenarios in which the HSR system was implemented on a voluntary and mandatory basis (i.e., HSR uptake across 6.7% and 100% of applicable products, respectively). The main outcomes were health-adjusted life years (HALYs), net costs, and incremental cost-effectiveness ratios (ICERs). These were calculated with accompanying 95% uncertainty intervals (95% UI). The model predicted that HSR-attributable reformulation leads to small changes [corrected] in mean population energy intake (voluntary: -0.98 kJ/day; mandatory: -11.81 kJ/day). [corrected]. These are likely to result in changes in mean body weight (voluntary: -0.01 kg [95% UI: -0.012 to -0.006]; mandatory: -0.11 kg [95% UI: -0.14 to -0.07, and HALYs gained [corrected] (voluntary: 4207 HALYs gained [corrected] [95% UI: 2438 to 6081]; mandatory: 49,949 HALYs gained [95% UI: 29,291 to 72,153]). The HSR system [corrected] could be considered cost-effective relative to a willingness-to-pay threshold of A$50,000 per HALY (incremental cost effectiveness ratio for voluntary: [corrected] A$1728 per HALY [95% UI: dominant to 10,445] and mandatory: A$4752 per HALY [95% UI: dominant to 16,236]).

Use of Added Sugars Instead of Total Sugars May Improve the Capacity of the Health Star Rating System to Discriminate between Core and Discretionary Foods

BACKGROUND

The Australian Government has introduced a voluntary front-of-package labeling system that includes total sugar in the calculation.

OBJECTIVE

Our aim was to determine the effect of substituting added sugars for total sugars when calculating Health Star Ratings (HSR) and identify whether use of added sugars improves the capacity to distinguish between core and discretionary food products.

DESIGN

This study included packaged food and beverage products available in Australian supermarkets (n=3,610). The product categories included in the analyses were breakfast cereals (n=513), fruit (n=571), milk (n=309), non-alcoholic beverages (n=1,040), vegetables (n=787), and yogurt (n=390). Added sugar values were estimated for each product using a validated method. HSRs were then estimated for every product according to the established method using total sugar, and then by substituting added sugar for total sugar. The scoring system was not modified when added sugar was used in place of total sugar in the HSR calculation. Products were classified as core or discretionary based on the Australian Dietary Guidelines. To investigate whether use of added sugar in the HSR algorithm improved the distinction between core and discretionary products as defined by the Australian Dietary Guidelines, the proportion of core products that received an HSR of ≥3.5 stars and the proportion of discretionary products that received an HSR of <3.5 stars, for algorithms based upon total vs added sugars were determined.

RESULTS

There were 2,263 core and 1,347 discretionary foods; 1,684 of 3,610 (47%) products contained added sugar (median 8.4 g/100 g, interquartile range=5.0 to 12.2 g). When the HSR was calculated with added sugar instead of total sugar, an additional 166 (7.3%) core products received an HSR of ≥3.5 stars and 103 (7.6%) discretionary products received a rating of ≥3.5 stars. The odds of correctly identifying a product as core vs discretionary were increased by 61% (odds ratio 1.61, 95% CI 1.26 to 2.06; P<0.001) when the algorithm was based on added compared to total sugars.

CONCLUSIONS

In the six product categories examined, substitution of added sugars for total sugars better aligned the HSR with the Australian Dietary Guidelines. Future work is required to investigate the impact in other product categories.

Incorporating Added Sugar Improves the Performance of the Health Star Rating Front-of-Pack Labelling System in Australia

Background: The Health Star Rating (HSR) is an interpretive front-of-pack labelling system that rates the overall nutritional profile of packaged foods. The algorithm underpinning the HSR includes total sugar content as one of the components. This has been criticised because intrinsic sugars naturally present in dairy, fruits, and vegetables are treated the same as sugars added during food processing. We assessed whether the HSR could better discriminate between core and discretionary foods by including added sugar in the underlying algorithm. Methods: Nutrition information was extracted for 34,135 packaged foods available in The George Institute’s Australian FoodSwitch database. Added sugar levels were imputed from food composition databases. Products were classified as ‘core’ or ‘discretionary’ based on the Australian Dietary Guidelines. The ability of each of the nutrients included in the HSR algorithm, as well as added sugar, to discriminate between core and discretionary foods was estimated using the area under the curve (AUC). Results: 15,965 core and 18,350 discretionary foods were included. Of these, 8230 (52%) core foods and 15,947 (87%) discretionary foods contained added sugar. Median (Q1, Q3) HSRs were 4.0 (3.0, 4.5) for core foods and 2.0 (1.0, 3.0) for discretionary foods. Median added sugar contents (g/100 g) were 3.3 (1.5, 5.5) for core foods and 14.6 (1.8, 37.2) for discretionary foods. Of all the nutrients used in the current HSR algorithm, total sugar had the greatest individual capacity to discriminate between core and discretionary foods; AUC 0.692 (0.686; 0.697). Added sugar alone achieved an AUC of 0.777 (0.772; 0.782). A model with all nutrients in the current HSR algorithm had an AUC of 0.817 (0.812; 0.821), which increased to 0.871 (0.867; 0.874) with inclusion of added sugar. Conclusion: The HSR nutrients discriminate well between core and discretionary packaged foods. However, discrimination was improved when added sugar was also included. These data argue for inclusion of added sugar in an updated HSR algorithm and declaration of added sugar as part of mandatory nutrient declarations.