Research Roundup
Modifying the Health Star Rating nutrient profiling algorithm to account for ultra-processing
2 August 2024
Some public health advocates and academics have raised concerns that the Health Star Rating (HSR) system is too lenient on unhealthy products that might be ultra-processed yet still receive high ratings due to its nutrient-centric focus and lack of consideration for ultra-processing.
A recent Australian study (1) has investigated the impact of combining ultra-processing with the HSR, to determine how modified ratings compare with NOVA and the current Australian dietary guidelines of core and discretionary foods.
The George Institute for Global Health’s Australian 2022 FoodSwitch data was cross-sectionally analysed for 25 486 products. Ultra-processed foods were defined by the presence of industrial food substances and “cosmetic” additives in the product’s ingredient list. These included additives such as anti-caking, firming or glazing agents, colour, flavour, emulsifier, extractor thickener; any sweeteners other than sugar, maple syrup and honey (e.g. dextrose, fructose and maltitol); whole foods isolates (e.g. lactose, wheat gluten and triglycerides); and protein powders.
Four modifications were used to determine the potential impacts of ultra-processing-related adjustments to the HSR algorithm. These were based on 1. inclusion in which ultra-processing was assessed by a negative component (negative points) added to the HSR algorithm; 2. capping in which UPF products are restricted to a maximum HSR; and 3. hybrid approach in which UPF products were penalised if meeting a points threshold ‘tipping point’. This last approach had two variations.
All modifications resulted in greater alignment with the NOVA classification system when compared to the current HSR. This is not surprising as modifications were designed to include ultra-processing. Cereal and grain products were the food categories with the greatest gain in alignment with the NOVA classification system. However, none of the modifications resulted in overall greater alignment with the Australian Dietary Guidelines classifications, although there was considerable variation among major food categories.
In conclusion, the authors state that incorporating ultra-processing into Australia’s current HSR system through several mechanisms is technically feasible. Additionally, they added that before this could occur, consideration of the concept of ultra-processing within the broader framework of dietary guidance would be essential.
A strength of the study was the use of data from over 25, 000 products representative of the current Australian packaged food supply and the systematic process of categorising individual products according to an adapted NOVA classification system. There are limitations such as the continual refining of the definition of ‘ultra-processed’ that will influence how it could be incorporated into food policies; and data gaps in the labelling of fruit, vegetable, nut and legume content or fibre content of products to determine the HSR. As a result, missing values were estimated using ingredient lists, food composition databases, and other available data. Finally, the HSR cut-off of ≥3.5 used across analyses is not universally accepted as appropriate to distinguish healthy and unhealthy ratings. Other researchers in Australia have suggested a more stringent cut-off (≥2.5). The issue of a different cut-off could result in a different degree of alignment than found in the study for each modification.
Note – The AFGC does not support the NOVA classification nor the use of the term ultra-processing to describe foods. The AFCG has developed resources on food processing here.
Reference
- Barrett E M, Pettigrew S, Neal B, Rayner M, Coyle D H, Jones A, Maganja D, Gaines A, Mozaffarian D, Taylor F, Ghammachi N, Wu J H. Modifying the Health Star Rating nutrient profiling algorithm to account for ultra-processing. Nutrition & Dietetics. 2024 https://doi.org/10.1111/1747-0080.12892
