Effects of interpretive nutrition labels on consumer food purchases: the Starlight randomized controlled trial

Background: Nutrition labeling is a prominent policy to promote healthy eating. Objective: We aimed to evaluate the effects of 2 interpretive nutrition labels compared with a noninterpretive label on consumer food purchases. Design: In this parallel-group randomized controlled trial, we enrolled household shoppers across New Zealand who owned smartphones and were aged ≥18 y. Eligible participants were randomly assigned (1:1:1) to receive either traffic light labels (TLLs), Health Star Rating labels (HSRs), or a control [nutrition information panel (NIP)]. Smartphone technology allowed participants to scan barcodes of packaged foods and to receive allocated labels on their smartphone screens. The primary outcome was the mean healthiness of all packaged food purchases over the 4-wk intervention period, which was measured by using the Food Standards Australia New Zealand Nutrient Profiling Scoring Criterion (NPSC). Results: Between October 2014 and November 2015, 1357 eligible shoppers were randomly assigned to TLL (n = 459), HSR (n = 443), or NIP (n = 455) labels. Overall difference in the mean transformed NPSC score for the TLL group compared with the NIP group was −0.20 (95% CI: −0.94, 0.54; P = 0.60). The corresponding difference for HSR compared with NIP was −0.60 (95% CI: −1.35, 0.15; P = 0.12). In an exploratory per-protocol analysis of participants who used the labeling intervention more often than average (n = 423, 31%), those who were assigned to TLL and HSR had significantly better NPSC scores [TLL compared with NIP: −1.33 (95% CI: −2.63, −0.04; P = 0.04); HSR compared with NIP: −1.70 (95% CI: −2.97, −0.43; P = 0.01)]. Shoppers who were randomly assigned to HSR and TLL also found the labels significantly more useful and easy to understand than the NIP (all P values