An Integrated Nutrition Intervention to Promote Healthy Eating Habits for Children With ASD

The aim of the study is to evaluate the feasibility and preliminary efficacy of Autism Eats in improving dietary intake and mealtime behaviors of children with ASD. We will enroll 30 early intervention (EI) providers who work with dyads of parent and child with ASD (n=48) and randomly assign them into either the Autism Eats or Enhanced Usual Care (EUC) group. We will test the feasibility of enrollment, implementation, and evaluation of Autism Eats. We expect that Autism Eats will be feasible and well-received by EI providers and parent-child dyads. Feasibility will be determined by benchmarks set in each of the following categories: (1) reach/participation; (2) attrition; (3) completion; (4) fidelity; (5) compatibility; and (6) positive participant feedback. To assess the preliminary efficacy of Autism Eats for improving child dietary intake and mealtime behaviors, we will use a validated tool to assess dietary intake of children and problematic mealtime behaviors. While evaluation of outcomes is more appropriate for a fully-powered clinical trial, we will examine variance and effect sizes of key outcomes at baseline, post-intervention, and 5-month follow-up. We will examine differences in children's food intakes (fruit and vegetables), food variety, diet quality, and problematic mealtime behaviors between children in Autism Eats and those in the EUC groups at post-intervention and 5-month follow-up from baseline. As an exploratory aim, we will explore whether child weight status differs between Autism Eats and EUC groups at 5-month follow-up assessment. Height and weight of children will be measured, and weight-for-length based on the CDC growth chart (birth to 3 years)17 will be calculated and weight status will be determined by the percentile chart by sex. The long-term goal is to identify and implement interventions in early childhood that improve health trajectories in populations with ASD.

Random assignment will be made utilizing a random number generator after baseline data collection is completed. The Autism Eats intervention consists of ten 25-minute weekly lessons, two monthly booster sessions, social media components, and provider training and on-going support. Lessons integrate ASD-specific feeding strategies such as repeated exposures, food chaining, and making regular mealtime routines and behaviorally-focused nutrition content and activities utilizing goal setting, healthy meal planning, monitoring progress, strategies to overcome barriers, and creating healthy home food environment. The EI providers are well-trained to use personalized intervention and coaching approach in their Part C EI services, which will be applied to Autism Eats activities as well (e.g., using the child's favorite foods for lesson activities). Both in-person and telehealth versions of the intervention manual will be available (in case of the COVID-19 pandemic service restriction). There is no content difference between in-person and telehealth versions of the program, except that instructions for providers are tailored to each setting. To maintain scientific control for attention threats and increase rigor, EUC families will also receive a one-time 25-minute session on nutrition and publicly available web-based information on healthy eating (e.g., We Can! materials) weekly over 10-weeks to correspond with the number of Autism Eats sessions. These strategies have been successfully used as comparison offerings in previous RCTs.

The primary outcome measure is dietary intake: food groups, food variety, and diet quality. Data collection times are at baseline, post-intervention (about 10-11 weeks from baseline), and 5-month follow-up from baseline. Three-day food records will be completed by parents and data will be entered into the NCI Automated Self-administered 24-hour Dietary Recall (ASA24) by research assistants. We focus on average daily fruit and vegetables and daily food variety. For all children aged 2 years or older, diet quality will be determined by the Healthy Eating Index's total and sub-component scores (adequacy vs. moderation food categories). There are two secondary outcome measures. The first is problematic mealtime behaviors. The time frame is baseline, post-intervention, and 5-month follow-up from baseline. The Brief Autism Mealtime Behavior Inventory (BAMBI), which is validated and used in many previous research studies will be used to assess children with ASD's problematic mealtime behaviors. Parents will complete the survey via REDCap. The BAMBI contains 18 questions and sues a 5-point scale, with response options ranging from 1 (Never) to 5 (Almost every meal), for reporting the frequency of a behavior. Each question also has a yes or no option for parents to indicate if they perceive the behaviors as problematic. The BAMBI includes the subscales of Limited Variety, Food Refusal, and Features of Autism. The survey takes about 5 minutes to complete. The second is anthropometric- height and weight. The time frame is baseline and 5-month follow-up from baseline. Parents will be given options for their child's and their own anthropometric assessments. They can come to the USF office and get children's and their own height and weight measured by a trained research assistant (RA) or a RA can visit participants' house to use the same portable stadiometer and weight scale. If only virtual height and weight measurement is accessible (due to the COVID-19 pandemic), a metal ruler and a bathroom weight scale will be sent to the participant's house and a RA instructed assessment will be virtually completed. As a standard weight status assessment for birth to 36 months, weight-for-length based on the CDC growth chart will be calculated and weight status will be determined by the percentile chart by sex. Parent BMI will be calculated and used as a covariate in data analyses. If height and weight information is available for the spouse of the participating parent or the other biological father/mother, it will be recorded on the form.

For our RCT, we will enroll a total of 48 parent-child dyads with a randomization between Autism Eats intervention and EUC comparison group. The sample size and power calculation was not formally performed because we do not have our own reference effect size values for the study outcomes and because it is not cost-effective to conduct a full-scale RCT without preliminary data. However, based on our previous studies and literature on calculating preliminary RCT sample sizes, our target sample size of 48 parent-child dyads can be justified. We are using the size of the parent-child dyads because children's dietary behavioral outcomes are the primary RCT outcomes. According to Whitehead et al. (2015) in Statistical Methods in Medical Research journal, pilot RCTs do not have the same objectives as a main trial and using formal power considerations is usually not necessary. We justify our target sample size based on their pilot sample size calculations and recommendations. With an estimated small effect size (0.2), an n=25 pilot RCT sample size (per arm) is recommended. For a medium standardized effect size (0.5), a sample size of 15 per arm is recommended. Based on previous nutrition education literature and from our own experiences, we anticipate seeing a small to medium preliminary effect size (approximately 0.3-0.4) conservatively, and therefore, our sample size of 48 parent-child dyads satisfies their pilot RCT sample size recommendations.

Specific Analyses for the Individual Study Aims: Aim 2a. Test the feasibility of enrollment, implementation, and evaluation of Autism Eats. We hypothesize that Autism Eats will be feasible and well received by EI providers and parent-child dyads, which will be assessed with (1) reach/participation; (2) attrition; (3) completion; (4) fidelity; (5) compatibility; and (6) qualitative feedback from participants. The rates of reach, participation, attrition, and completion will be computed. Descriptive statistics will be used to examine data distributions and frequencies. If there is substantial variability in attrition, completion, fidelity, and compatibility, we will examine the relations of such variables with any other variables we collect during the study such as demographic variables of EI providers and parent-child dyads.

Aim 2b.To assess the preliminary efficacy of Autism Eats on child dietary intake and mealtime behaviors. While evaluation of outcomes is more appropriate for a fully-powered RCT, we will examine variance and effect sizes of key outcomes from pre-to post-intervention, as well as post-intervention to 5-month follow-up using non-parametric tests and confidence intervals. We expect differences in children's food group intakes, variety, diet quality, and mealtime behaviors between children who participate in Autism Eats and those in the EUC groups at post-intervention and 5-month follow-up from baseline. We will first assess whether Autism Eats and EUC groups are balanced through randomization in terms of demographic variables at baseline. For each outcome variable, we will use boxplots and scatterplots with confidence intervals to visually inspect the changes across pre-intervention, post-intervention, and 5-month follow-up by group along with descriptive statistics. Also, nonparametric longitudinal analysis will be conducted for each outcome variable with time as a within-subject factor and treatment condition as a between-subject factor to examine the difference in changes between Autism Eats and EUC groups.

Exploratory Aim. We will explore whether weight status of children are different between Autism Eats and EUC groups at 5-month follow-up assessment. Descriptive statistics and visual inspection will be used to summarize weight status by group and effect sizes will be computed. Similar to the analysis of the primary outcomes, we will conduct nonparametric analysis to compare the changes in weight status at 5-month follow-up between Autism Eats and EUC groups. Note that parent BMI will be included as a covariate in the analysis.

Biological variables. Sex: Given the sex difference in ASD prevalence, in addition to the primary data analyses as described above, data distributions will be analyzed by sex. We will then explore how male and female students respond to our intervention differently through qualitative interviews with selected sub-samples. Recommendations will be created to address any sex differences. Genetic factors: Family history of ASD and parent BMI will be measured and considered as covariates. Potential confounding factors to be considered in the analytic models. Parentage, sex, race/ethnicity, baseline child's weight status, parent education, and family income. Examine demographic variations in response to the Autism Eats intervention. Correlation will be used to examine associations between demographic variations (age, sex, race/ethnicity, SES, and baseline weight status) outcomes of the Autism Eats(food intakes, food variety, diet quality, mealtime behaviors, and weight percentile).

Attrition and Missing Data. We aim <20% dropout due to study withdrawal. Every effort will be made to prevent dropouts/missing data, and to complete relevant assessments for participants who drop out or are withdrawn, including reasons why the intervention was ended. To handle the dropout problem in this trial, we will first analyze data to see whether dropout was related to certain baseline measures and demographic variables. We will report the dropout rate and compare the distributions of baseline measures and demographic variables.