The efficacy of 'high in' warning labels, health star and traffic light front-of-package labelling: an online randomised control trial

Abstract

Objective: To examine the impact of front-of-package (FOP) labels on perceived healthfulness, purchasing intentions and understanding of common FOP systems.

Design: A parallel, open-label design randomised participants to different FOP labelling conditions: 'high in' warning labels (WL), multiple traffic light labelling (TLL), health star ratings (HSR) (all displayed per serving) or control with no interpretive FOP labelling. Participants completed a brief educational session via a smartphone application and two experimental tasks. In Task 1, participants viewed healthy or unhealthy versions of four products and rated healthiness and purchasing intention on a seven-point Likert-type scale. In Task 2, participants ranked three sets of five products from healthiest to least healthy.

Setting: Online commercial panel.

Participants: Canadian residents ≥ 18 years who were involved in household grocery shopping, owned a smartphone and met minimum screen requirements.

Results: Data from 1997 participants (n 500/condition) were analysed. Task 1: across most product categories, the TLL and HSR increased perceived healthiness of healthier products. All FOP systems decreased perceived healthiness of less healthy products. Similar, albeit dampened, effects were seen regarding purchasing intentions. Task 2: participants performed best in the HSR, followed by the TLL, WL and control conditions. Lower health literacy was associated with higher perceived healthiness and purchasing intentions and poorer ranking task performance across all conditions.

Conclusions: All FOP labelling systems, after a brief educational session, improved task performance across a wide spectrum of foods. This effect differed depending on the nutritional quality of the products and the information communicated on labels.Trial Registration: NCT03290118.

Keywords: Food labelling; Front-of-package labelling; Health star ratings; Nutrition labelling; Randomised controlled trial; Traffic light labelling; Warning symbols.

Figures

Fig. 1

Examples of the experimental conditions used in all tasks* *In all experimental conditions, participants could access the Nutrition Facts table (NFt) for all products by clicking on a link to ‘See Nutrition Information’, which included ingredient information.

Fig. 2

Task 2 – Ranking products according to healthiness. Examples of cereal, yogurt and pasta image sets to which front-of-package (FOP) labels were applied

Fig. 3

Task 1 – Perceived healthiness of granola bars, pizza, frozen vegetables and soup when they have a healthier or less healthy nutritional profile* (n 1983) *Superscripts indicate results from linear regression models within product category stratified by product healthfulness (healthier or less healthy product), adjusted for gender, income, education, BMI, ethnicity, health literacy and nutrition label use. **Note that in the labels for granola bars, they were erroneously labelled as 500 kcal per serving in the FOP information for the traffic light condition in both the healthier and less healthy scenarios.

Fig. 4

Task 1 – Intention to purchase granola bars, pizza, frozen vegetables and soup when they have a healthier or less healthy nutritional profile* (n 1983) *Superscripts indicate results from linear regression models within product category stratified by product healthfulness (healthier or less healthy product), adjusted for gender, income, education, BMI, ethnicity, health literacy and nutrition label use. **Note that in the labels for granola bars, they were erroneously labelled as 500 Kcal per serving in the FOP information for the traffic light condition in both the healthier and less healthy scenarios.

Fig. 5

Task 2 – The proportion of the sample that (A) correctly selected the healthiest product, (B) correctly selected the least healthy product and (C) the mean number of correct responses in the ranking task.*† *Values with different letters indicate significant differences (P < 0.05) in repeated measures analysis using logistic regression (A & B) or ordinal logistic regression (C), adjusted for gender, income, education, BMI, ethnicity, healthtpfmt 0literacy and nutrition label use †Note that figures 5A and 5B represent a proportion, whereas Figure 5C represents the mean number of responses in each of the FOP conditions.