An Online RCT Testing Added-Sugars Warning Label Public Service Announcements Among U.S. Hispanic Adults

This online randomized controlled trial (RCT) will assess if language-matched English-Spanish public service announcements (PSAs) can increase the understanding of and response to added-sugars menu warning labels among U.S. Hispanic adults. The investigators will recruit a sample of 2,000 U.S. Hispanic/Latino adults, oversampling such that approximately 50% prefer Spanish. Participants will be assigned to either an experimental PSA arm or a control PSA arm using a 2:1 ratio. Within the experimental PSA arm, participants will be randomized using a 1:1:1 ratio to one of three interventions, each with different PSA themes: (1) Labels are coming (i.e., informing consumers that added-sugars menu warning labels are coming soon), (2) Hidden sugars (i.e., informing consumers that added sugars are hidden in many common restaurant foods), and (3), Health outcomes (i.e., informing consumers that added sugars increase the risk of getting type 2 diabetes, cardiovascular disease, and tooth decay). Within the control PSA arm, all participants will be exposed to the same PSA intervention theme: Safe driving (i.e., informing consumers that using a phone while driving is dangerous). For each PSA intervention theme, participants will view two PSAs (e.g., two PSAs about hidden sugars).

This RCT will be administered using an online survey for which participants will select a language (i.e., English, Spanish) for survey navigation. Then, after randomization to a study arm and intervention, they will view their two assigned (and language-matched) PSAs in random order within the context of four mock advertisements, presented in identical order in each arm and condition. These mock advertisements will be the same for each intervention and will promote car insurance, a food delivery service, a music streaming platform, and a credit card to simulate a realistic advertising environment. After exposure to the experimental or control PSAs and advertisements, participants will place hypothetical meal orders from two realistic chain-restaurant menus, one fast-food menu and one full-service menu, shown in random order. The menus will display an added-sugars warning label next to items containing ≥100% Daily Value for added sugars and a disclosure statement explaining the label's meaning. After completion of the ordering tasks, participants will answer additional questions.

In both study arms, if the participant's language preference is Spanish, they will be shown the PSAs and mock advertisement graphics in Spanish. Otherwise, they will view the PSAs and mock advertisement graphics in English. Explanations of the added-sugars warning label on the restaurant menus and the menus for the ordering task will always be in English to simulate a realistic U.S. environment.

Primary outcomes for this study include: (1) noticing, recall, understanding of the added-sugars menu label; and (2) grams of added sugars ordered. Secondary outcomes include: (1) noticing, recall, understanding the PSAs; (2) reported use of the added-sugars label when selecting food items; (3) selection of a food item high in added sugars for themselves; (4) perceived message effectiveness of the label; (5) perceived message effectiveness of a PSA; (6) Perceived trust of a PSA; (7) correct identification of menu items high in added sugars; (8) anticipated social interaction about a PSA; (9) selection of 2 PSAs perceived as most effective in discouraging the consumption of added sugars; (10) perceived PSA weight stigmatization of people with obesity; and (11) support for the use of the added-sugars label in the community.

The primary analysis will compare label understanding between the intervention and control arms on 1) noticing and 2) added sugars ordered. All statistical analyses will use two-tailed tests with a significance level of alpha < 0.05. For continuous outcomes, linear regression models will be used, but if there is evidence of non-normality, necessarily transformations or alternative modeling approaches will be used. For dichotomous outcomes, risk ratios (or probability ratios) will be estimated using modified Poisson regression with robust standard errors. Results will report means and proportions in each condition and differences between groups. All models comparing labels will be bivariate, regressing outcomes on an indicator variable for assignment to the experimental arm. Sensitivity analyses will examine primary outcomes by chain-restaurant menu (i.e., fast-food and full-service). Exploratory analyses will compare the efficacy of each experimental PSA theme to the control and to one another. These models will regress outcomes on indicators for each experimental PSA and use post-regression pairwise comparisons. Effect modification of the PSAs on primary outcomes will be explored for sociodemographic variables using interaction terms and stratified models.