Effectiveness of Differing Levels of Support for Family Mealtimes on Obesity Prevention Among Head Start Preschools

Introduction Despite slight decreases in obesity prevalence in low-income children, nearly
25% of children under age 5 years are overweight or obese. Socioeconomic disparities in early
childhood place low-income children at 1.5 to 2 times higher risk for obesity compared to
middle- to upper-income children making obesity prevention targeting low-income children a
top public health priority. Recently, obesity interventions have turned toward the promotion
of family mealtimes as an obesity prevention strategy. Even among families with economic
resources, planning and implementing family mealtimes is challenging. Barriers to family
mealtimes among low-income families, however, are intensified. Further, low-income parents
facing the chronic stressors of poverty are more likely to experience compromised executive
functioning skills, impeding their abilities to plan and organize events such as mealtimes.
They are also more likely to encounter decision fatigue, the deteriorating quality of
decisions made after multiple decision making demands1 (e.g., continual daily decisions about
how to maximize limited financial, social, and emotional resources). In the context of
poverty, a parent's meal planning capacity is diminished and mealtime decisions may be made
impulsively. Most family mealtime intervention approaches have targeted meal planning skills
and knowledge and parental modeling of healthy eating, without significant regard to the
practical resources that would enable meal planning and implementation. There is a striking
lack of evidence about what level of resources low-income parents need to implement family
mealtimes. The investigators' approach is built on the premises that: 1) enhancing the
ability of parents to implement healthy family mealtimes is critical to effective obesity
prevention; and 2) when parents do not implement regular healthy family mealtimes, it is not
solely due to a lack of nutrition education or knowledge, but to the combined adverse effects
of poverty on the capability of parents to plan and execute healthy family mealtimes.
Identifying what resources are needed to aid low-income families in planning and implementing
family meals is crucial but unknown information. Without such information, effective
interventions cannot be planned.

In this integrated Coordinated Agricultural Project (CAP) application, the investigators
propose to test the main, additive and interactive effects of 6 intervention components
reflecting differing levels of practical resources to increase the frequency of healthy
family mealtimes, improve children's dietary quality and ultimately reduce childhood obesity
prevalence. Using the innovative Multiphase Optimization Strategy (MOST; a cutting-edge
approach to maximizing resources in behavioral interventions by identifying the most
efficient intervention model possible), the investigators will test the 6 intervention
components in a Screening Phase (N = 520), resulting in the implementation and evaluation of
a "final" intervention model in the Confirming Phase (N = 250) that bundles the components
yielding the most robust effects. The 6 intervention components include: (1) Meal Delivery:
the home delivery of pre-made healthy family meals including recipes weekly for 8 weeks; (2)
Ingredient Delivery: the home delivery of ingredients and recipes to make healthy family
meals weekly for 8 weeks; (3) Community Kitchen: community kitchen sessions in which families
prepare healthy meals with recipes to take home weekly for 8 weeks; (4) Didactics: nutrition
education classes using the Preschool Obesity Prevention Series (POPS) curriculum, developed
and evaluated by the team with more 400 low-income Head Start parents 2 weekly for 8 weeks
(5) Cooking Lessons: cooking lessons/demonstrations with recipes weekly for 8 weeks; and (6)
Cookware/Flatware: delivery of flatware/ cookware to utilize for family meals delivered at
the beginning of the 8 week intervention.

MOST is a comprehensive, principled, engineering-inspired framework for optimizing and
evaluating multicomponent behavioral interventions. MOST includes a randomized, controlled
trial (RCT) for intervention evaluation, but unlike the standard approach to intervention
development, also includes key steps before the RCT. These steps are aimed at intervention
optimization by identifying what intervention components yield robust effects. MOST is
designed to be practical, and holds the possibility of achieving more rapid long-run
improvement of interventions without a dramatic increase in intervention research resources.
By testing 6 different intervention components simultaneously using the factorial design that
forms the basis of MOST, we will be able to identify the most cost-effective strategies for
achieving childhood obesity prevention through the Extension framework. Given the substantial
amount of resources dedicated to obesity prevention through Extension, the results will
inform the most appropriate resource allocation.

Study Design The first phase of MOST, which the investigators will undertake in Years 1 to 2
of this proposal, is referred to as the "Screening Phase", which refers to the systematic
testing of intervention components to ultimately build the most effective intervention. Using
a factorial design, the objective of implementing the MOST Screening Phase is to answer
questions such as: Which of the set of program components are active and contributing to
positive outcomes, and should be included in the intervention? Which program components are
inactive or counterproductive, and should be discarded? In the Screening Phase, 520
participants will be enrolled over four cohorts. After the indentification of the key
components for the packaged intervention, the critical interventions will be bundled in the
latter years of the study and evaluated as a standard 2-arm RCT, referred to in MOST
terminology as the "Confirming Phase". The Confirming Phase will answer the question: Is this
intervention, as a package, efficacious and is the intervention effect large enough to
justify investment in community implementation? In the Confirming Phase, the final model of
the intervention will be delivered over the school year between October and April, beginning
in Fall 2017 and then in Fall 2018 in a randomized controlled trial (N = 250, half randomly
assigned to the bundled intervention and half receiving usual Head Start exposure). In the
Confirming Phase, the investigators hypothesize that the finalized intervention will lead to
children's improved dietary quality and, most importantly, to improved adiposity indices.

Participants will be recruited from a partnering Head Start program which serves mixed rural
and urban areas. Children and families represent a diverse group (race is 67% white, 28%
black; ethnicity 23% Hispanic; 12.6% English as second language) per year. Data will be
collected in the home prior to and at the conclusion of the interventions. Head Start
teachers will also complete behavioral ratings of children.

Statistical Analysis- Screening Phase In the Screening Phase, the primary aim is to test on
an intent-to-treat basis using ANOVA models to test whether each factor (intervention) has a
significant effect on frequency of healthy, family mealtimes and dietary quality over an 8
week-period pre- to post-intervention. The investigators will also include two- and three-way
interactions between components to identify the effects of interactions between intervention
components on outcomes. The secondary primary aim is to make decisions about intervention
components and component levels based on the factorial experiment results. The investigators
will use a modified version of a decision making approach frequently used in engineering,
which emphasizes main effects, using interactions as additional valuable information.

Statistical Analysis- Confirming Phase The primary objective is to determine whether the
bundled intervention vs. no intervention comparison group is more effective in reducing
children's adiposity indices. Baseline comparability of the 2 groups will be assessed using
t-test for continuous variables and χ2 tests for categorical variables. To account for
clustering of children within a Head Start classroom, Proc Mixed in SAS with a random
intercept for the classroom will be used. The investigators will employ logistic regression
models using General Estimating Equations (GEE) techniques to account for classroom
clustering to compare changes in the prevalence of obesity and overweight across the 2
groups. Analyses will be conducted both with and without adjustment for baseline
characteristics such as age, sex, race, and baseline BMI or BMI z-score.

Evaluation of Educational Objectives, Process and Fidelity Evaluation Maternal mealtime
self-efficacy, mealtime barriers, knowledge of obesity prevention behaviors, resources, and
perceptions of the mealtime climate will be compared pre-post in the intervention groups
using ANOVA models to determine change compared between groups. The investigators will
analyze data on recruitment, retention, participant and educator satisfaction, and also the
implementation fidelity data, using basic descriptive statistics. The investigators will use
qualitative approaches to analyze the interview data obtained from HS staff, Extension and
participating parents (i.e., identifying themes; segmenting responses into meaningful
categories). The process evaluation data will inform continued collaborations with HS and
Extension, and allow the investigators to more effectively disseminate our findings.
Likewise, results from descriptive analyses of the fidelity self-report (Extension) and
observational data will be analyzed.

Inclusion Criteria:

- Child is enrolled in Head Start

Exclusion Criteria:

- Significant feeding/eating disorders that would preclude participation in the
interventions

- Child is a foster child

- Parent is non-English speaking