Effect of voluntary Health Star Rating labels on healthier food purchasing in New Zealand: longitudinal evidence using representative household purchase data

Abstract

Front-of-pack labelling (FoPL) aims to promote healthier diets by altering consumer food purchasing behaviour. We quantify the impact of the voluntary Health Star Rating (HSR) FoPL adopted by New Zealand (NZ) in 2014, on (i) the quantity of foods purchased by HSR scores and food groups and (ii) the quantities of different nutrients purchased. We used Nielsen HomeScan household purchasing panel data over 2013-2019, linked to Nutritrack packaged food composition data. Fixed effects analyses were used to estimate the association of HSR with product and nutrient purchasing. We controlled for NZ-wide purchasing trends and potential confounding at the household and product level. In 2019, HSR-labelled products accounted for 24% (2890) of 12 040 products in the dataset and 32% of purchasing volume. Of HSR-labelled products, 1339 (46%) displayed a rating of 4.0-5.0 stars and 556 (19%) displayed a rating of 0.5-2.0 stars. We found little or no association between HSR labelling and the quantities of different foods purchased. Introduction of HSR was, however, associated with lower sodium (-9%, 95% CI -13% to -5%), lower protein (-3%, 95% CI -5% to 0%) and higher fibre (5%, 95% CI 2% to 7%) purchases when purchased products carrying an HSR were compared with the same products prior to introduction of the programme. Robust evidence of HSR labelling changing consumer purchasing behaviour was not observed. The positive effect on nutrient purchasing of HSR-labelled foods likely arises from reformulation of products to achieve a better HSR label.

Keywords: nutrition assessment.

Figure 1

Examples of Health Star Rating configurations. Clockwise from top left: (i) Health Star Rating icon only, (ii) Health Star Rating and energy icons, (iii) Health Star Rating, energy, three prescribed nutrients, high and low indicators and optional nutrient.

Figure 2

Conceptual diagram of empirical methodology.

Figure 3

Adoption of Health Star Rating (HSR) by rating in the merged HomeScan-Nutritrack dataset 2013–2019, as per cent of unique products; and as per cent overall New Zealand-wide purchasing in g or mL.

Figure 4

Percentage purchasing (using grams and millilitres) over time of products adopting the Health Star Rating (HSR) during the study period, for these products observed in each year over 2013–2019 (ie, including before they adopted the HSR), by the mean HSR rating eventually displayed on product packaging.

Figure 5

Health Star Rating (HSR)-linked per cent change in purchasing versus the mean HSR score displayed by category products in 2019. Markers scaled to 1/SE of the coefficient. The three solid outline circles had coeffficients from the fixed effects regression with 95% CI that excluded the null. The figure shows 42 food groups; 1 group (sugars, 22.5% increase) was removed as an outlier for clarity.

Figure 6

Per cent changes and 95% CI in the product-level log of nutrient purchasing for Health Star Rating (HSR)-labelled products, compared with non-HSR-labelled products. All estimates control for household, product and time fixed effects. Other covariates include household characteristics (number of adults, number of children, main buyer sex, main buyer age, household type, household income) and product characteristics (mean selling price, % purchases made on when the product on promotion, number of products in brand portfolio and number of products in HomeScan product category).