Nutrition Classification Schemes for Informing Nutrition Policy in Australia: Nutrient-Based, Food-Based, or Dietary-Based?

BACKGROUND

Policy makers are increasingly using nutrition classification schemes (NCSs) to assess a food's health potential for informing nutrition policy actions. However, there is wide variability among the NCSs implemented and no standard benchmark against which their contrasting assessments can be validated.

OBJECTIVES

This study aimed to compare the agreement of nutrient-, food-, and dietary-based NCSs in their assessment of a food's health potential within the Australian food supply, and examine the conceptual underpinnings and technical characteristics that explain differences in performance.

METHODS

A dataset combining food compositional data from the Mintel Global New Products Database and the Australian Food Composition Database (AUSNUT 2011-2012) (= 7322) was assembled. Products were classified by 7 prominent NCSs that were selected as representative of one or other of ) nutrient-based NCSs [the Chilean nutrient profile model (NPM), Health Star Rating (HSR), Nutri-Score, the WHO European Region's NPM (WHO-Euro NPM), and the Pan American Health Organization's (PAHO) NPM]; ) food-based NCS (NOVA), and ) dietary-based NCS [Australian Dietary Guidelines (ADGs)].

RESULTS

The PAHO NPM classified the lowest proportion (22%) of products as "healthy", and the HSR the highest (63%). The PAHO NPM, NOVA, WHO-Euro NPM, and the Chilean NPM classified >50% of products as "unhealthy," and the ADGs, HSR, and Nutri-Score classified <50% of products as "unhealthy." The HSR and Nutri-Score had the highest pairwise agreement (κ = 0.7809, 89.70%), and the PAHO NPM and HSR the lowest (κ = 0.1793, 53.22%). Characteristics of NCSs that more effectively identified ultraprocessed and discretionary foods were: category-specific assessment, the classification of categories as always "healthy" or "unhealthy," consideration of level of food processing, thresholds for "risk" nutrients that do not penalize whole foods; and no allowance for the substitution of ingredients.

CONCLUSIONS

Wide variation was observed in agreement of the assessment of a food's health potential among the NCSs analyzed due to differing conceptual underpinnings and technical characteristics.