Defining ‘Unhealthy’: A Systematic Analysis of Alignment between the Australian Dietary Guidelines and the Health Star Rating System

Abstract

The Australian Dietary Guidelines (ADGs) and Health Star Rating (HSR) front-of-pack labelling system are two national interventions to promote healthier diets. Our aim was to assess the degree of alignment between the two policies. Methods: Nutrition information was extracted for 65,660 packaged foods available in The George Institute’s Australian FoodSwitch database. Products were classified ‘core’ or ‘discretionary’ based on the ADGs, and a HSR generated irrespective of whether currently displayed on pack. Apparent outliers were identified as those products classified ‘core’ that received HSR ≤ 2.0; and those classified ‘discretionary’ that received HSR ≥ 3.5. Nutrient cut-offs were applied to determine whether apparent outliers were ‘high in’ salt, total sugar or saturated fat, and outlier status thereby attributed to a failure of the ADGs or HSR algorithm. Results: 47,116 products (23,460 core; 23,656 discretionary) were included. Median (Q1, Q3) HSRs were 4.0 (3.0 to 4.5) for core and 2.0 (1.0 to 3.0) for discretionary products. Overall alignment was good: 86.6% of products received a HSR aligned with their ADG classification. Among 6324 products identified as apparent outliers, 5246 (83.0%) were ultimately determined to be ADG failures, largely caused by challenges in defining foods as ‘core’ or ‘discretionary’. In total, 1078 (17.0%) were determined to be true failures of the HSR algorithm. Conclusion: The scope of genuine misalignment between the ADGs and HSR algorithm is very small. We provide evidence-informed recommendations for strengthening both policies to more effectively guide Australians towards healthier choices.

1. Introduction

Unhealthy diets—high in salt, harmful fats, added sugar and energy—are a leading cause of death and disability in Australia [1]. Australia has some of the highest obesity rates in the world: nearly two-thirds of Australian adults and one in four children are now overweight or obese. Unprecedented availability and aggressive marketing of processed and pre-packaged foods and beverages are a key driver of obesity and diet-related conditions including high blood pressure, heart disease, stroke, type 2 diabetes, some forms of cancer, dementia and dental caries [2]. Obesity alone is estimated to cost Australia more than $8.6 billion annually [3].

The World Health Organization recommends a comprehensive suite of population health approaches to promote healthier diets. These include laws and regulations, tax and price interventions, community-based measures in facilities such as schools and hospitals, and public education through social marketing campaigns [4]. Despite the increasing impact of poor diet on Australia’s health, few of these preventive strategies have been taken up at a federal level.

Two policy areas where Australia has been benchmarked as performing well against international best-practice are in adoption of food-based dietary guidelines and front-of-pack nutrition labels [5]. The current Australian Dietary Guidelines (ADGs) were introduced in 2013 to promote health and wellbeing while reducing the risk of chronic disease [6]. In 2014, Australia adopted the Health Star Rating System (HSR), an interpretive nutrition labelling scheme that rates foods from 0.5 to 5.0 stars on the front-of-pack [7]. An example of the HSR graphic and the Australian Guide to Healthy Eating, which provides a practical visual representation the food groups recommended by the ADGs and their proportions, are included at Appendix A.

While inherently related in their intent to guide Australians towards healthier choices, the two measures differ in aspects of their purpose and design (Table 1). For example, the ADGs provide information on dietary patterns, amounts and food groups that support health through detailed guidance documents that are for use by health professionals, policy makers, educators, food manufacturers and researchers [6]. By contrast, HSR uses an algorithm to quantify selected aspects of individual foods, generating a summary score displayed in a simple symbol intended to both target consumers at the point-of-sale, and offer incentives for manufacturers to improve recipes to receive a higher rating [8,9]. The relative strengths of each measure suggest potential opportunities for the ADGs and HSR to operate synergistically for maximum public health impact. At the same time, tension between the two approaches is apparent in academic and media critique of HSR, particularly where high profile products have been highlighted for displaying labels allegedly inconsistent with ADG recommendations [10,11].

Table 1. Key features of Australian Dietary Guidelines and the Health Star Rating System.

	Objective	Mechanism	Target Audience	Classification of Foods	Developed by	Governed by
Australian Dietary Guidelines	Provide information on food groups, amounts and dietary patterns that support health.	Guideline Documents, Summary and Educator Guide. Australian Guide to Healthy Eating graphic	Health professionals, policy makers, educators, food manufacturers, food retailers and researchers.	Classification of foods into Five Food Groups that form the basis of a healthy diet, and ‘discretionary’ foods defined by the presence of saturated fat, added sugars, salt and/or alcohol, whose intake is to be limited.	National Health and Medical Research Council (NHMRC) via standardised guideline process. Working Committee incl. public health and industry representation.	NHMRC NHMRC considers whether to update after 5 years. Maximum interval prior to update is 10 years.
Health Star Rating	Simplify nutrition information available on back-of-pack to differentiate between individual foods more likely to be part of a healthy diet from those that are less healthy.	Front-of-pack label to be applied voluntarily by food retailers and manufacturers using relevant policy documents.	Consumers at point of purchase. Food retailers and manufacturers.	A nutrient profile model is used to score individual products from 0.5 to 5.0 stars. The algorithm considers energy, negative nutrients the ADGs recommend eating less of (saturated fat, sugars and sodium), and foods the ADGs recommend eating more of (fruits, vegetables, nuts and legumes) as well as in some instances, allowing points for protein and dietary fibre content.	Australian Fed., State and territory governments in partnership with food industry, consumer and public health groups.	Health Star Rating Advisory Committee. Representation from Australian Federal, State and Territory governments as well as food industry, consumer and public health groups. 2 year monitoring report and 5 year review process set down at adoption.

With a five year review of the HSR system currently underway, our objective was to assess one of the review’s key elements, namely the degree of alignment between the two policies and specifically in their mechanism for defining food as healthy or unhealthy. Policy coherence is important not only because of the need to provide consistent dietary messaging to Australians, but also because inappropriate HSR scores or ADG recommendations threaten the credibility and sustainability of both policies. By identifying areas and causes of misalignment, our aim was to make evidence-informed recommendations for refining both policies, thereby strengthening Australia’s efforts to promote healthier diets.

2. Methods

This was a cross-sectional examination of packaged foods and beverages (hereafter referred to as foods) available in Australia.

2.1. Data Source

We analysed items included in The George Institute for Global Health’s Australian FoodSwitch Database [12]. The database contains nutrition label information from packaged foods systematically collected by The George Institute through large-scale annual surveys, as well as provided directly by manufacturers and consumers on a rolling basis. In total, this data represents more than 90% of products available in the Australian market. For this study, we used information extracted directly from the back-of-pack nutrition information panel. Energy (kJ/100 g), protein (g/100 g), saturated fat (g/100 g), total sugar (g/100 g), and sodium (mg/100 g) are mandatory on the Australian nutrient declaration but details on fruit, vegetable, nut and legume (FVNL) (%), concentrated FVNL (%), and fibre (g/100 g) are optional. Where such details were absent, appropriate levels were estimated using information drawn from the back-of-pack ingredients list, generic food composition databases, or by analogy with similar products using methods described previously [12]. The estimation process provides a proxy value for each nutritional indicator at the finest category level for more than 1000 individual food subcategories. Proxy values are then substituted for each product in that category for which data are missing.

2.2. Product Classification

Classification of products was based on the system developed by the Global Food Monitoring Group and incorporated into FoodSwitch [13]. This hierarchical system is designed to monitor the nutrient composition of processed foods around the world. It classifies foods into major categories (e.g., bread and bakery products), categories (e.g., bread), and subcategories (e.g., pita bread). Our analysis included only packaged food items. We excluded infant foods and formula, vitamins and supplements, formulated supplementary sports foods, foods for special medical purposes and alcoholic beverages because these foods have been specifically deemed outside the scope of HSR [14]. This left 15 major categories for analysis. Within these, we also excluded subcategories of plain tea and coffee, herbs and spices, baking powders, yeasts and gelatines, as these foods do not contribute significantly to nutrient intake, are not required to display a nutrition information panel [15], and are also therefore not required to display a HSR.

2.3. Calculation of the Health Star Rating

The HSR was calculated in alignment with the methods described in the ‘Guide for industry to the Health Star Rating Calculator’ for all products sold, regardless of whether a HSR was reported on the pack [16]. In short, foods were categorised into one of six HSR categories (i.e., non-dairy beverages; dairy beverages; oils and spreads; cheese and processed cheese; all other dairy foods; all other non-dairy foods). Baseline points were calculated based on the energy, saturated fat, total sugar, and sodium content per 100 g. Modifying points for FVNL%, concentrated FVNL%, protein, and fibre were calculated, where applicable. A HSR ‘score’ was calculated by subtracting the modifying points from baseline points. This score is then converted to a HSR based upon a defined scoring matrix for each of the six categories [16]. The HSR ranges from 0.5 to 5.0 stars in ten half-star increments. A higher HSR reflects a healthier product.

2.4. Classification of Foods under the Australian Dietary Guidelines

We classified foods as ‘core’ or ‘discretionary’ according to ADG guidance. In short, core foods were defined as those from the ADGs Five Food Groups: grain (cereal) foods, mostly wholegrain and/or high cereal fibre varieties; vegetables and legumes/beans; fruit; milk, yoghurt, cheese and/or alternatives, mostly reduced fat; and lean meats and poultry, fish, eggs, tofu, nuts and seeds and legumes/beans. Together, these core foods form the basis of a healthy diet. Discretionary foods are those not necessary to provide nutrients the body needs, and are defined for ADG purposes as those ‘high in’ saturated fats, added sugars, and/or salt or alcohol [6]. As the ADG documents themselves provide only limited examples of discretionary choices and no objective measure of ‘high in’, we relied upon the Australian Bureau of Statistics (ABS) ‘Discretionary Food List’ [17] as the best-available reference for classifying each product for the purposes of this analysis. The main principle used by the ABS to classify foods as discretionary is that they were specified or inferred in the ADGs and supporting documents as discretionary [17]. ABS classifications determined at a detailed food category level were matched to FoodSwitch categories to classify each category as ‘core’ or ‘discretionary’.

In some categories, such as those involving mixed foods, the ABS applies additional nutrient criteria to define core and discretionary. The nutrient cut-offs specified are those used in the modelling that supported the original ADG development [18]. Where provided, e.g., pizza with saturated fat content ≤5 g/100 g is ‘core’ while pizza with saturated fat content >5 g/100 g is ‘discretionary’, these were also applied.

2.5. Statistical Analysis

Cross-tabulations of ADG status and HSR were prepared for the twenty cells comprising core, discretionary and the ten possible HSR values (0.5, 1.0, 1.5, 2.0, 2.5, 3.0, 3.5, 4.0, 4.5 or 5.0). In the absence of endorsed HSR cut-offs for healthy or unhealthy, we identified products as ‘apparent outliers’ when the product was categorised ‘core’ by the ADGs but received a HSR ≤ 2.0, suggesting an unhealthy nutritional profile, or the product was categorised as ‘discretionary’ by the ADG but received a HSR ≥ 3.5, suggesting a healthy nutritional profile. The number and proportion of products deemed apparent outliers was determined overall, for each of the 15 major food categories included and by category and sub-category where helpful.

To further understand the reasons for outlier status of products, and in particular the potential impact of the undefined ‘high in’ terminology used by the ADGs, we applied additional nutrient cut-off criteria. In the absence of any existing international standard or guidance, we drew from the United Kingdom (UK) multiple traffic-light nutrient profile model, which was validated during development in the UK context, and has been subsequently used to model dietary outcomes elsewhere [19,20,21]. Specifically, the cut points used to apply red traffic lights for salt, total sugar and saturated fat were used to provide a quantitative measure of the ADGs ‘high in’ terminology. These were applied to apparent outliers, and products were removed or retained on the basis of qualifying for none, one or multiple red traffic lights:

• Apparent outliers that scored a low HSR but were assigned ‘core’ status by the ADGs were not considered ‘genuine outliers’ if they were sufficiently ‘high in’ salt, saturated fat and/or sugar to warrant at least one red traffic light.
• Apparent outliers that scored a high HSR but were assigned ‘discretionary’ status by the ADGs were not considered ‘genuine outliers’ if nutrient values for salt, saturated fat and/or sugar were not sufficiently high to warrant at least one red traffic light.

Apparent outliers that were deemed not to be genuine outliers after application of these ‘high in’ cut points were deemed ‘ADG failures’. All others were deemed ‘HSR failures’. Reasons for failures and potential solutions were systematically recorded (Appendix B).

3. Results

In total, 65,660 packaged products available in Australian supermarkets between 1 January 2013 and 30 June 2017 were identified in the FoodSwitch database. Of these, 11,431 had insufficient product information to enable categorization and generate a HSR, and a further 7113 were in categories excluded from HSR.

This left 47,116 products for analysis. Of these, only 3524 (7.5%) were displaying HSR on pack at the date of data extraction (30 June 2017).

3.1. HSR Distribution by Core and Discretionary

There were 23,460 (49.8%) core products and 23,656 (50.2%) discretionary products. Distribution of HSR by core and discretionary is shown in Figure 1.

Figure 1. Distribution of HSR by core and discretionary with apparent outliers.

In total, the median (IQR) of calculated HSR scores was 3 (1.5 to 4). Core products had a median (IQR) calculated HSR score of 4.0 (3.0 to 4.5) and discretionary products had a median (IQR) calculated HSR score of 2.0 (1.0 to 3.0).

Of products displaying HSR on pack, the overall median (IQR) value was 4.0 (3.0 to 4.5); out of these, 2131 (60.5%) were core products and 1393 (39.5%) were discretionary.

3.2. Apparent Outliers

There were 6324 (13.4%) apparent outliers. In total, 2219 were apparent core outliers, representing 9.5% of all core products and 4.7% of the total sample. In total, 4105 apparent discretionary outliers were identified, representing 17.4% of all discretionary products and 8.7% of the total sample (Figure 1 and Table 2).

Table 2. Cross tabulation of apparent outliers, and attribution of reason for outlier status (i.e., ADG failure or HSR failure) after application of traffic light cut-offs.

	Apparent Outliers	ADG Failure	HSR Failure
HSR ≤ 2.0	2219	2116	103
HSR ≥ 3.5	4105	3130	795

As seen in Figure 2, the distribution of products and number of apparent outliers varied greatly across the 15 major food categories and by core and discretionary classification. The major categories with the largest proportion of apparent outliers were sauces, dressings, spreads and dips (19.9%); dairy (18.4%); and snack foods (10.3%).

Figure 2. Apparent core and discretionary outliers—numbers of products and numbers of outliers by major food category.

3.3. Application of Traffic Light Cut-Offs

Application of cut-points to identify foods ‘high in’ salt, total sugar and saturated fat greatly reduced the number of outliers (Table 2).

Among the apparent core outliers, 2116/2219 (95.4%) had at least one red traffic light, signifying high levels of salt (1159), saturated fat (1136), and/or sugar (538). These foods were deemed ADG failures on the basis that high levels of these negative nutrients form the basis of the ADG definition of discretionary foods. This left 103 (4.6%) core food outliers that received a low HSR despite not being ‘high in’ any negative nutrients. These results were genuinely misaligned with the ADGs and deemed HSR failures. Figure 3 and Table 3 detail these results by major food category. The three categories with HSR failures were fruit and flavoured yoghurts; and flavoured teas. The yoghurts had amber lights for saturated fat and sugar, and the flavoured teas had amber lights for sugar. Both categories likely had a mix of naturally occurring and added sugar.

Figure 3. Apparent and genuine outliers by major food category (areas of circles are proportional to numbers of products).

Table 3. Core products with HSR ≤ 2.0.

Major Food Category	Number of Apparent Outliers	Number of Products with Any Red TLL	Outliers Remaining	Illustrative Examples of Outliers Remaining	Characteristics of Remaining Outliers
1 Bread and bakery products	193	187	6	Pancake mixes, tortillas	All have amber traffic lights for saturated fat and salt, and some also for sugar
2 Cereal and grain products	174	172	2	Rice puff cereal, noodles	Breakfast cereal has amber lights for salt and sugar Noodles have amber lights for saturated fat and salt
3 Confectionery	0	-	-		-
4 Convenience foods	57	56	1	Antipasto product	Has amber traffic lights for salt, saturated fat and sugar
5 Dairy	1056	997	58	Fruit and flavoured yoghurts, natural yoghurt	All are in yoghurt category. More than 90% are yoghurts with fruit or flavourings that have amber traffic lights for saturated fat and sugar Two products are natural yoghurts with amber traffic lights for saturated fat and salt
6 Edible oils and oil emulsions	73	73	0		-
7 Eggs	0	-	-		-
8 Fish and fish products	163	160	3	Salmon pate, garlic prawns	All have amber lights for salt and at least one other of sugar and saturated fat
9 Fruit and vegetables	181	181	0		-
10 Meat and meat products	168	168	0		-
11 Non-alcoholic beverages	46	13	33	Fruit flavoured teas and iced teas, matcha	Some teas, unlike most, carried a nutrient information panel and therefore had a HSR generated despite being low in nutrients overall Iced teas with added sugar are discretionary but these teas contained fruit and in the absence of added sugar labelling it was not possible to definitively categorise these drinks as core or discretionary Growth in popularity of new beverages categories (e.g., matcha, chai) suggest more classification guidance needed
12 Sauces, dressings, spreads and dips	11	11	0		-
13 Snack foods	0	-	-		-
14 Special foods	75	75	0		-
15 Sugars, honey and related products	23	23	0		-
All	2219	2116	103

TLL = Traffic Light Label, referring to the threshold set for a red traffic light under the UK nutrient profiling model.

In the discretionary outlier group, 3130/4105 (76.2%) of apparent outliers had no red traffic lights, signifying they were not ‘high in’ salt, sugar or saturated fat and were therefore deemed ADG failures. This left 975 (23.8%) apparent discretionary outliers receiving a high HSR despite being ‘high in’ salt (510), sugar (296) and/or saturated fat (235). These foods were deemed HSR failures. Figure 3 and Table 4 outline these findings by major food category. The largest number of HSR failures occurred in sauces, dressings, spreads and dips; savoury snacks; meat and meat products; and, convenience foods. Products in these categories predominantly had red traffic lights for salt and to a lesser degree, saturated fat despite receiving HSR ≥ 3.5.

Table 4. Discretionary products with HSR ≥ 3.5.

Major Food Category	Number of Apparent Outliers	Number of Products No Red TLL	Outliers Remaining	Illustrative Examples of Outliers Remaining	Characteristics of Remaining Outliers
1 Bread and bakery products	166	141	25	Sweet biscuits, savoury breads and pastries	Products have red TLL for sugar (sweet breads and biscuits), salt (savoury breads and biscuits) or saturated fat (puff pastries, quiche)
2 Cereal and grain products	254	190	64	Breakfast cereals, cereal and nut-based bars	Most have red TLL for sugar and a few salt or saturated fat
3 Confectionery	173	161	12	Jellies, cocoa powder, chocolate strawberries	Products have red TLL for sugar (jellies) or saturated fat (chocolate based items)
4 Convenience foods	508	372	136	Ready meals, meal kits	Most products have red TLL for saturated fat and/or salt A smaller number have red TLL for sugar
5 Dairy	108	76	32	Rice puddings	All products have red TLL for sugar
6 Edible oils and oil emulsions	2	0	2	Almond oil, lemon butter	All products have red TLL for saturated fat or sugar.
7 Eggs	0	-	-		-
8 Fish and fish products	219	206	13	Salt and pepper products, fish cakes	All products have red TLL for salt.
9 Fruit and vegetables	347	261	86	Fruit bars and bites, pickled vegetables	Fruit products have red TLL for sugar (fruit bars, bites) and sometimes saturated fat (fruit bites with coconut) Vegetable products have red TLL for salt (pickled vegetables, olives)
10 Meat and meat products	384	213	171	Sliced meats, frozen and chilled meats	Most products have red TLL for salt or saturated fat and a few for both
11 Non-alcoholic beverages	30	22	8	Milk flavourings, beverages mixes	Products have red TLL for sugar. Are able to take advantage of ‘as prepared’ rules
12 Sauces, dressings, spreads and dips	1245	1015	230	Salty dips, relishes and chutneys	Most products have red TLL for salt, some for sugar and a few for saturated fat
13 Snack foods	652	461	191	Potato chips, vegetable and legume-based snacks, corn chips	Most products have red TLL for salt, some for saturated fat, a few for sugar and a few for several
14 Special foods	2	2	-		-
15 Sugars, honey and related products	15	10	5	Syrups	Products all have red TLL for sugar
All	4105	3130	975

TLL = Traffic Light Label, referring to the threshold set for a red traffic light under the UK nutrient profiling model.

Taking core and discretionary together, 5246 outliers (83%) were attributable to ADG failure. This left 1078 outliers (17%) attributable to a failure of the algorithm.

A detailed list of core and discretionary outliers is included at Appendix B.

4. Discussion

In contrast to intense media attention on occasional anomalies, this large quantitative analysis suggests that the scope of genuine misalignment between the ADGs and the HSR algorithm across the Australian food supply is very small.

The degree of policy coherence demonstrated by our results is encouraging, though perhaps not surprising given the well-recognised relationship between nutrients, foods and dietary patterns [22]. Our results are consistent with existing research demonstrating that the HSR algorithm is aligned well overall with the ADGs [23,24,25], and that the median HSR of core foods is significantly higher than that of discretionary foods [10,23,26].

While these results are promising, it is reasonable to seek better alignment between the HSR and the ADGs if this increases their public health impact. The results of this work suggest directions for improvement.

Specific recommendations for improving alignment are set out in Table 5.

Table 5. Priority recommendations for reviewing the HSR algorithm and ADG definitions to improve alignment.

Review the weighting given to salt in HSR algorithm, given the large number of sauces, dips, savoury snacks, sliced meats and convenience foods that receive a high HSR despite being high in salt. This would be supported by the 2017 update of Australia’s Nutrient Reference Value for sodium.

Review the eligibility of fried or pickled vegetables and dried fruits for FVNL points given their receipt of high HSR scores despite being high in negative nutrients. This would be supported by references in ADG text (but not the ABS Table) that such products should be only consumed occasionally and in small amounts.

Review the weighting given to sugar, and/or incorporate added sugars given the large number of outliers in categories likely to contain a mix of naturally occurring and added sugars. These appeared at both ends, i.e., core yoghurts with fruit or flavours, fruit flavoured teas, as well as discretionary chutneys, breakfast cereals, muesli and fruit bars, dairy desserts and table sauces.

Review the ADG definition of discretionary, including additional guidance on ‘high in’ criteria for saturated fat, added sugars and salt to elucidate, for example, at what point a flavoured yoghurt can more properly be considered a dairy dessert.

Review a wide range of ABS table classifications (see Appendix B) including: ‘core’ status of breakfast cereals with sugar up to 30 g/100 g, cheese regardless of salt and saturated fat content, yoghurt and flavoured milks regardless of sugar or saturated fat content, most meat and fish regardless of salt content; ‘discretionary’ status of vegetable and legume based dips, ‘potato products’ and crumbed fish intended for home baking; and, appropriate treatment of growing product categories such as breakfast beverages and coconut products.

Once the algorithm is reviewed, make HSR mandatory to enable consumers to receive the full benefit of the system’s performance across the food supply.

While there is considerable current focus on review of the HSR algorithm, these results highlight a parallel urgent need for review of the ADG text and corresponding ABS Table (See Appendix B).

Our findings can be differentiated from a recent analysis of foods carrying HSR labels on pack during voluntary implementation [19]. That analysis of 1269 new products found that 57% were core items and 43% discretionary. The authors concluded that HSR labelling was undermining ADG recommendations by facilitating the marketing of discretionary foods because more than half of those defined as discretionary displayed a HSR ≥ 2.5. Our work suggests caution in this interpretation given the small sample size of the prior study, identified failings of the ADGs in regard to the classification of some foods, and the highly selective subset of products studied. The conclusions may also reflect the simplistic approach ultimately taken by the ADGs, in seeking to dichotomise foods as core or discretionary, when healthiness of products is almost certainly distributed along a continuum.

The present analysis benefits from the large number of foods included, their robust representation of the broader Australian food supply and the comprehensive and systematic approach taken to the evaluation and presentation of the data. Our decision to use the UK traffic light criteria as a quantitative measure of ‘high in’ was an objective approach to quantifying the textual guidance provided by the ADGs. The international food standards agency, The Codex Alimentarius Committee, recently agreed to commence work to develop ‘high in’ criteria for salt, sugar and saturated fat given increasing international interest in this area. This process is likely to take several years [27].

Some limitations also need to be mentioned. FVNL content and fibre are not currently mandatory on back-of-pack nutrition information panels in Australia and missing values were therefore estimated from ingredients lists, food composition databases, and other sources. This analysis does not capture related concerns raised by stakeholders to the Government’s five year review [28] regarding the high HSRs received by fruit juices, breakfast cereals with a sugar content ≤30 g/100 g, and breakfast beverages. These issues likely represent HSR failures but are not identified as ‘outliers’ in this analysis because the ADG text and/or ABS Table also identify these products (arguably incorrectly) as core.

Our ability to measure alignment was limited by the components of foods considered by the HSR algorithm. For example, HSR currently relies on total sugar but the ADGs recommend to limit added sugars specifically. Areas where we believe this distinction may have impacted outlier status include flavoured yoghurts and milks, breakfast cereals, muesli bars, chutneys and table sauces as noted in detail in Appendix B. Previous work has indicated that incorporating added sugar into the HSR algorithm would improve its ability to discriminate between core and discretionary [26,29]. This would be best facilitated by updating mandatory nutrient information panel (NIP) requirements to include transparent information on added sugars. Alternatively, added sugar values could be systematically estimated using available information from current NIPs in combination with the ingredients list, using published methods [30]. A similar approach is currently provided to companies claiming points for fruit, vegetable, nut and legume content that is also not required in the NIP.

Our analysis was also limited in scope to packaged products only. While half our sample were classified core (suggesting that not all packaged foods are unhealthy), some foods recommended by the ADGs (i.e., whole fresh fruit and vegetables) are generally sold without packaging. Current consideration of whether to extend HSR to these products (e.g., through shelf talkers) could further enhance alignment between the HSR and ADGs [31].

5. Conclusions

This work illustrates the complexity of defining foods as ‘healthy’ or ‘unhealthy’ across the huge range of packaged products available in the current Australian food supply.

Like other front-of-pack labelling systems that rely on an underlying profiling model, the HSR algorithm is intended as a tool to quantify selected aspects of individual foods rather than to be a complete source of dietary advice. Nevertheless, our results are consistent with WHO recognition that such tools are a helpful method to use in conjunction with interventions aimed at improving the overall nutritional quality of diets [32].

Rather than undue focus on perfect alignment or determination of the superiority of the HSR or ADGs, a more nuanced understanding of the relative contribution (and inherent limitations) of each suggests areas where the design and implementation of both policies could be strengthened to guide Australian consumers towards healthier choices.