Nutrient profiling systems are designed to assess the nutritional quality of packaged foods and beverages, and guide public health policy initiatives(1). In Australia, the Nutrient Profiling Scoring Criterion (NPSC) is used to determine whether foods or beverages are eligible to carry health claims. The NPSC assesses nutrient content using a per 100g or 100mL reference amount. While this approach allows for standardisation of nutritional content across foods, it does not reflect the typical amount consumed. Ready meals, for example, are often sold and consumed as a single serve portion exceeding 300g. As a result, some products may be deemed healthy enough to carry a health claim despite providing a substantial amount of energy, saturated fat, sugar or sodium per serve. This study aimed to examine the impact of applying ‘per serve’ nutrient profiling criteria on the classification of ready meals and compare the proportion of products meeting proposed per serve thresholds with those assessed using per 100g nutrient values.
Data on the nutritional composition and serve sizes of ready meals available in the Australian market was obtained from the 2023 FoodSwitch database, a repository of Australian food packaging images and label data for over 17,000 foods developed by The George Institute for Global Health(2). Four nutrient profiling models were applied to assess the relative healthfulness of ready meals: Model 100g (comparison model using per 100g values; the status quo), Model A (standardised 350g serve), Model B (serve size declared on-pack ), and Model B Adjusted (declared serve with thresholds scaled to 350g). Products were assessed against per serve nutrient thresholds for energy, saturated fat, total sugars, and sodium. Cut off-points were set at 1/3 of daily recommended intakes based on nutrient reference values and other labelling guidance. Products meeting all four nutrient thresholds were classified as meeting model criteria. Cochran’s Q test and McNemar’s post hoc tests were used to compare classification outcomes across models.
Data were available for 775 ready meal products. Of these, 97% met model criteria under Model 100g. Under the per serve models, 16% of products met model criteria for Model A, 20% for Model B, and 16% for Model B Adjusted. Model 100g resulted in a significantly higher proportion of ready meal products meeting nutrient thresholds than all per serve models (p<0.001).
Applying per serve nutrient profiling models substantially reduced the proportion of ready meals meeting model criteria. This study revealed limitations of classifying foods according to nutrient content per 100g, particularly for single serve foods like ready meals that are likely exceed recommended levels of nutrients to limit when consumed as intended. Incorporation of per serve disqualifying criterion into nutrient profiling systems should be considered to better reflect consumption and strengthen food labelling policies.