Comparing Front-of-package Nutrient Labels

The goal of this study is to compare front-of-package labels to a no-label control and to one another: 1.) FDA's Nutrition Info with %DV, 2.) Nutrition Info with red "high", 3.) FDA's High In, 4.) multiple High In labels-one for each nutrient and 5.) a no-label control. Primary outcomes include 1.) correct identification of foods and beverages with the healthiest and least healthy overall nutrient profiles, 2.) perceived healthfulness of products low in two nutrients of concern, and 3.) correct identification of foods and beverages as high in saturated fat, sodium, and added sugars.

The overall goal of this online randomized controlled trial is to compare the performance of FDA's Nutrition Info %DV front-of-package label (FOPL) to a no-label control and other FOPL designs, including the Nutrition Info label without the %DV and with red color for "high in" nutrients, the FDA's High In label, and multiple black box High In labels-one for each nutrient.

The primary outcomes are: 1.) correct identification of foods and beverages with the healthiest and least healthy overall nutrient profiles, 2.) perceived healthfulness of products low in two nutrients of concern, and 3.) correct identification of foods and beverages as high in saturated fat, sodium, and added sugars. Secondary outcomes include: 1.) noticing the nutrition label, 2.) recalling the label's content, 3.) reported use of label when selecting food items, 4.) perceived healthfulness of i.) products not high in any nutrients of concern and ii.) products with different nutrient levels for each nutrient of concern, 5.) selection of a food item high in at least one nutrient of concern for themselves, 6.) selection of a food item high in at least one nutrient of concern for their youngest child (at least 2 years old), 7.) perceived message effectiveness of label for discouraging consumption of foods high in saturated fat, sodium, and added sugars, 8.) time spent on identifying healthiest and least healthy overall nutrient profiles, and 9.) time spent on identifying foods and beverages as high in saturated fat, sodium, and added sugars.

Repeated dichotomous outcomes (e.g., correct selection of healthiest profile, correct selection of least healthy profile, correct identification of product high in nutrients of concern) will be analyzed in mixed effects Poisson regression with robust standard errors. Single dichotomous outcomes (e.g., noticing, use of label, selection of at least one item high in one nutrient of concern) will be analyzed in Poisson regression with robust standard errors. Results will report proportions in each condition and absolute (i.e., percentage point) and relative differences (i.e., probability ratios). Continuous outcomes (e.g., perceived healthfulness, time spent) will be used in linear regression. All regression models will regress the outcome on an indicator for experimental condition. results will report averages by condition and average differential effects between conditions. Analyses will use a two-sided critical alpha of 0.05 to conduct all statistical tests. All confidence intervals presented will be 95% and two-sided. Given the preliminary nature of this study, which is comparing novel FOPLs for the first time, no p-value adjustments for multiple comparisons are planned.

In moderation analyses using interaction terms and stratified models, effect modification of label design on the primary outcomes will be examined by participant nutrition literacy, education, grocery shopping frequency, having children, and type of device used to take the survey.

Additionally, differences in all outcomes between each of the 5 unique conditions will be examined. The modeling approach may be modified if assumptions are not met.

A soft launch of approximately 500 completed surveys will be conducted to verify that the online survey is functioning as intended and to detect any programming errors. There may be another soft launch if any corrections are needed. If the target sample size is not achieved by April 30th, 2025, the data collected thus far will be used in preliminary analysis to provide public comments. The analytic sample will include participants with complete data on the primary outcomes.