Evaluation of a Letter Intervention Promoting a Plant-based Diet

The researchers' mission is to promote healthier eating behavior and to reduce costs associated with healthcare. The purpose of this study is to evaluate the effects of mailing randomly selected participants a letter promoting a plant-based diet. Depending on the experimental condition, participants may additionally receive a free documentary, Forks Over Knives, and they may also get letters which use commitment- or prevention-focused messages to encourage watching the documentary and changing their eating behavior. The researchers hypothesize that receiving the documentary will be associated with lower insurance claims and improved health outcomes one and two years later. The researchers also hypothesize that using either commitment- or prevention-focused messages will also contribute to lower insurance claims and improved health outcomes compared to experimental conditions where materials did not include these messages. This study will help the researchers design evidence-supported programs that can improve people's health.

Note that participant selection and assignment will be done at the level of the household, but analysis of outcomes will include individual-level analysis. This analytic decision was made because the researchers assume that the intervention may potentially affect other members of the household. While one person in a household will be selected for matching the eligibility criteria, the mailed materials will be addressed to all adult members of the household (defined as adults 18+ who are the spouse or child of the selected member) who are members of the Geisinger Health Plan. Outcome data will be pulled from all adult members of the household who have data through the Geisinger Health Plan. The researchers will analyze the data using standard survey research analyses methods, including computing bivariate correlations, using general linear models, using non-parametric models for non-normally distributed insurance data, and entering variables as independent predictors in regression models to attempt to predict desired outcomes. The researchers will also use multilevel models to account for household- and individual-level data. The analysis of primary outcomes will focus on data one year after the intervention begins. Follow-up analysis on secondary outcomes will be conducted two years after the intervention begins. The researchers will conduct separate analyses of the selected participant (i.e., the individual who met the eligibility criteria) and other members of that participant's household. This will test whether there are any spillover effects of the intervention to other people in the household. For clarity in this record, any discussion of sample size and participants refers to households, as this is the primary level for participant selection and assignment.