Testing front-of-package warnings to discourage red meat consumption: a randomized experiment with US meat consumers.

BACKGROUND Reducing red meat is a strategy to improve public health and mitigate climate change in the United States and other high-income countries. Policies requiring warnings on the front of red meat packages are a promising intervention to shift consumers towards healthier and more sustainable food choices. We aimed to explore participants' reactions to health and environmental warning messages about red meat. METHODS Between June and July 2020, we recruited a national convenience sample of US red meat consumers (n = 1,235; mean age 44 years) for an online survey. Participants were randomly assigned to one of four label conditions: no-label control, health warning, environment warning, and combined health and environment warning (both warnings shown side-by-side). Participants viewed three types of burritos (red meat [steak], chicken, and vegetarian) and selected their preferred item (primary outcome), the item they perceived to be most damaging to health, and the item they perceived to be most damaging to the environment (secondary outcomes). Participants then viewed their assigned warning on a series of other red meat products (no-label control participants were randomly re-assigned to one of the warning conditions) and rated the warnings on perceived message effectiveness, believability, negative emotions, perceived risk, attention, and learning something new. Finally, participants reported their intentions to reduce red meat consumption. RESULTS There were no significant differences in selection of the steak burrito between label conditions or in selection of the item most damaging to the environment. Those exposed to the health warning were more likely to select the steak burrito as most damaging to health compared to those exposed to other label conditions (health 73 %, combined 64 %, environment 60 %, no-label control 63 %, p < 0.05). The combined and health warnings elicited higher perceived message effectiveness ratings than the environment warning (combined mean 2.91, health 2.84, environment 2.61, p < 0.05). CONCLUSIONS Warnings did not have a significant effect on item preference in the choice experiment. However, combined and health warnings performed better than the environment warning across a variety of warning label reaction measures. More research will be needed to understand whether warnings elicit behavioral change in real-world environments. TRIAL REGISTRATION Analyses and hypotheses were preregistered on https://aspredicted.org/ph7mb.pdf on 23 June 2020.