Evaluation of a Beverage Media Campaign and Added-sugar Warning Labels

The primary objective of this study is to examine differences in intentions (measured by hypothetical beverage selection) between the following conditions: (1) Control condition: Viewing beverages with a barcode, (2) viewing beverages that have an icon-plus-text warning label, "WARNING: High in added sugar", (3) viewing a campaign message about warning labels in one's preferred language + viewing beverages with a barcode, and (4) viewing a campaign message about warning labels in one's preferred language + viewing beverages with the warning label above. An additional objective is to compare other message reactions and perceptions and perceptions of SSBs (sugar-sweetened beverages) between the conditions. This study will use an online randomized controlled trial among Latino/a/x adults to assign participants to one of the above conditions.

The primary objective of this study is to examine differences in intentions (measured by hypothetical beverage selection) between the following conditions: (1) Control condition: Viewing beverages with a barcode, (2) viewing beverages that have an icon-plus-text warning label, "WARNING: High in added sugar", (3) viewing a campaign message about warning labels in one's preferred language + viewing beverages with a barcode, and (4) viewing a campaign message about warning labels in one's preferred language + viewing beverages with the warning label above. An additional objective is to compare other message reactions and perceptions and perceptions of SSBs (sugar-sweetened beverages) between the conditions. This study will use an online randomized controlled trial among Latino/a/x adults to assign participants to one of the above conditions. Although this study uses prospective assignment, it is not a clinical trial because the outcomes are intentions, attitudes, and other reactions and perceptions and not behavioral or health outcomes.

Participants can take the survey in English or Spanish, but the warning labels on the beverages will always be in English. In the campaign conditions, if the participant generally speaks and reads in Spanish better than English, they will be shown the campaign message in Spanish, and otherwise, they will be shown the campaign message in English.

Within the campaign conditions, participants will be randomized to view 1 of 3 campaign posters that have different messages. Also, at the end of the questionnaire, all participants will be shown 1 of the 3 campaign message posters and asked to provide ratings of perceived discouragement (one of the 3 PME questions), how much the poster grabs attention, and how trustworthy the poster is. Following that item, all participants will view all 3 posters one after the other on the same page (shown in random order) and asked to select the one that discourages them the most from wanting to drink beverages high in added sugar.

1. Primary outcome: hypothetical selection of a sugar-sweetened beverage (SSB) for oneself.

   Analysis will use Poisson regression with robust standard errors, regressing dichotomous selection of an SSB on indicators for condition, which are described below:

   1. Main effects: Indicators for the warning label condition and campaign poster condition.
   2. To examine interaction effects: Indicators for warning label condition and campaign poster condition as well as a label by campaign interaction term.
2. Secondary outcomes: For binary outcomes, the analysis will entail the same modeling approach as above, and for continuous outcomes, the analysis will use linear regression models with the same indicators in the model. For outcomes about campaign posters, instead of including indicators for condition in the model, the analysis will include indicators for type of poster. For the question that asks participants to select the 1 poster (out of 3) that most discourages consumption, the analysis will use a chi-square goodness of fit test to see if the proportion of participants selecting any one poster differs significantly than what would be expected by chance. That model will include indicators for experimental condition. Within family of outcomes, the Holm-Bonferroni procedure will be used to adjust for multiple comparisons.

Exploratory analyses:

1. For the outcome of hypothetical selection of an SSB for one's child, the same modeling approach as specified for the primary outcome will be used.
2. Exploration of effect modification will be examined in separate models using interaction terms between condition and indicators for levels of the following characteristics: Participant reported speaking and reading better in Spanish than English, cut-points on the Short Acculturation Scale for Hispanics (SASH), household income and poverty level, gender, and educational attainment.