Background: Front-of-pack nutrition labelling may support healthier packaged food purchases. Australia has adopted a novel Health Star Rating (HSR) system, but the legitimacy of this choice is unknown.
Objective: To define the effects of different formats of front-of-pack labelling on the healthiness of food purchases and consumer perceptions.
Design: Individuals were assigned at random to access one of four different formats of nutrition labelling-HSR, multiple traffic light labels (MTL), daily intake guides (DIG), recommendations/warnings (WARN)-or control (the nutrition information panel, NIP). Participants accessed nutrition information by using a smartphone application to scan the bar-codes of packaged foods, while shopping. The primary outcome was healthiness defined by the mean transformed nutrient profile score of packaged foods that were purchased over four weeks.
Results: The 1578 participants, mean age 38 years, 84% female recorded purchases of 148,727 evaluable food items. The mean healthiness of the purchases in the HSR group was non-inferior to MTL, DIG, or WARN (all p < 0.001 at 2% non-inferiority margin). When compared to the NIP control, there was no difference in the mean healthiness of purchases for HSR, MTL, or DIG (all p > 0.07), but WARN resulted in healthier packaged food purchases (mean difference 0.87; 95% confidence interval 0.03 to 1.72; p = 0.04). HSR was perceived by participants as more useful than DIG, and easier to understand than MTL or DIG (all p < 0.05). Participants also reported the HSR to be easier to understand, and the HSR and MTL to be more useful, than NIP (all p < 0.03).
Conclusions: These real-world data align with experimental findings and provide support for the policy choice of HSR. Recommendation/warning labels warrant further exploration, as they may be a stronger driver of healthy food purchases.
Keywords: food industry; food labelling; food purchases; policy; randomised trial.