A Three-Arm Randomized Controlled Trial Using Ecological Momentary Intervention, Community Health Workers, and Video Feedback at Family Meals to Improve Child Cardiovascular Health: The Family Matters Study Design

Background: Numerous observational studies show associations between family meal frequency and markers of child cardiovascular health including healthful diet quality and lower weight status. Some studies also show the “quality” of family meals, including dietary quality of the food served and the interpersonal atmosphere during meals, is associated with markers of child cardiovascular health. Additionally, prior intervention research indicates that immediate feedback on health behaviors (e.g., ecological momentary intervention (EMI), video feedback) increases the likelihood of behavior change. However, limited studies have tested the combination of these components in a rigorous clinical trial. The main aim of this paper is to describe the Family Matters study design, data collection protocols, measures, intervention components, process evaluation, and analysis plan. Methods/design: The Family Matters intervention utilizes state-of-the-art intervention methods including EMI, video feedback, and home visiting by Community Health Workers (CHWs) to examine whether increasing the quantity (i.e., frequency) and quality of family meals (i.e., diet quality, interpersonal atmosphere) improves child cardiovascular health. Family Matters is an individual randomized controlled trial that tests combinations of the above factors across three study Arms: (1) EMI; (2) EMI+Virtual Home Visiting with CHW+Video Feedback; and (3) EMI+Hybrid Home Visiting with CHW+Video Feedback. The intervention will be carried out across 6 months with children ages 5-10 (n=525) with increased risk for cardiovascular disease (i.e., BMI ≥75%ile) from low income and racially/ethnically diverse households and their families. Data collection will occur at baseline, post-intervention, and 6 months post-intervention. Primary outcomes include child weight, diet quality, and neck circumference. Discussion: This study will be the first to our knowledge to use multiple innovative methods simultaneously including ecological momentary assessment, intervention, video feedback and home visiting with CHWs within the novel intervention context of family meals to evaluate which combination of intervention components are most effective in improving child cardiovascular health. The Family Matters intervention has high potential public health impact as it aims to change clinical practice by creating a new model of care for child cardiovascular health in primary care. Trial Registration: This trial is registered in clinicaltrials.gov (Trial ID: NCT02669797). Date recorded 5/02/22.

Ecological momentary intervention (EMI), or mobile health (mHealth), uses smartphones to send text messages to participants to intervene on behaviors in real-time as they unfold, moment-by-moment, over time and across contexts [28]. For example, a participant responds to a text earlier in the day regarding their stress level and source(s) of stress (e.g., too many things to do) then, an EMI message is sent providing suggestions to support them in making a healthful choice for family meals in the face of stress (e.g., tip for making a quick pasta meal more healthful by adding a vegetable stir in) [29,30]. EMI studies from other elds have shown signi cant improvement in targeted behaviors (e.g., medication compliance, smoking cessation) [31,32], high feasibility [32], validity and reliability [33,34], few logistical problems [26], and low burden [35].
Fourth, interventions utilizing community health workers (CHWs) who can meet participants "where they are at," both with regard to readiness for change and in their own environment (i.e., home visiting) are associated with better outcomes [36]. CHWs link care across clinic and home contexts and have high success with addressing obesity [36], diabetes [36] and other chronic conditions [36]. In addition, given the rise of virtual technology during the COVID-19 pandemic, the need to test virtual versus in-person measurement and delivery of home visiting interventions in a rigorous RCT is key to con rm the bene ts of these approaches [37].
The main aim of the Family Matters Intervention is to target a well-documented family context associated with child CVH (i.e., family meals) using innovative real-time methods (i.e., EMI, video feedback) with CHWs in both virtual and in-person delivery modes to increase child CVH using a three-arm RCT (see Figure 1). The three Arms include: EMI (Arm 1); EMI+Virtual Home Visiting (HV) with CHW+Video Feedback (Arm 2); and EMI+Hybrid HV with CHW+Video Feedback (Arm 3). Our overall hypothesis is that increasing both the quantity and quality of family meals will improve child CVH. Our main study hypotheses include (see Figure 1): Hypothesis 1: BMI percentile (%ile) and neck circumference will decrease and diet quality will increase in children in Arm 3 compared to children in Arms 1 or 2.
Hypothesis 2: Family meal quantity and quality will increase, controlling food parenting practices (e.g., restriction) will decrease, and parent coping skills will increase in parents in Arm 3 compared to parents in Arms 1 or 2.
Hypothesis 3: BMI %ile will decrease in siblings in Arm 3 compared to siblings in Arms 1 or 2.

Theoretical Framework
Family Systems Theory (FST) [38] guides the current study. According to FST, the family environment is the most proximal in uence on child CVH [39,40]. FST suggests that intervening on individual-level behavior (e.g., dietary intake) has limited success unless the family-level behavior sustaining or overriding the individual-level behavior (e.g., fruits/vegetables served at family meals, food parenting practices) changes too [39,41]. FST also suggests that healthful behaviors learned in one family context (e.g., family dinner meal) will generalize to other family contexts (e.g., breakfast, lunch, snacks) [41,42]. Thus, in the current study it is expected that positive parenting practices learned in the family meal context will generalize to other eating occasions and contribute to child CVH overall. Also, including multiple family members (e.g., parents, grandparents, siblings) in the intervention increases the likelihood of sustainable family-level change [7,20].

Methods
This protocol has been written following the guidelines of the Standard Protocol Items Recommendations for Interventional Trials (SPIRIT) checklist (Additional le 1). A SPIRIT gure is also provided below to demonstrate the ow of the study (see Figure 2).

Study Design
The Family Matters intervention is a single site RCT with child as the unit of randomization and analysis (see Figure 3). The study is funded by the National Institutes of Health (HL151978) and is registered at clinicaltrial.gov (Trial ID: NCT02669797; May 2, 2022). This RCT lasts 12 months for each family, with a four month active intervention phase, a two month maintenance phase, and data collection at baseline, 6 months (i.e., post-intervention), and 12 months (i.e., 6 months post-intervention). All study materials are created in both English and Spanish.

Study Recruitment
Children (n=525) and their families are recruited via family medicine and pediatric primary care clinics in Minneapolis and St. Paul, MN. Recruitment is ongoing for 42 months. Eligible children receive a letter inviting participation. Parents then ll out a REDCap survey assessing eligibility criteria.
Children at high risk for CVD, de ned as BMI ≥75th percentile [43].
Children from African American/Black, Asian, Hispanic, Native American, or White households who speak Spanish or English.
Exclusion Criteria: Children with medically necessary dietary restrictions (e.g., feeding tubes) or who are developmentally unable to participate (e.g., non-verbal).
Non-custodial parent who lives with the child <50% of the time.
Children participating in a weight management study.

Study Arms and Randomization
Families are randomized into one of three intervention Arms: (1) EMI; (2) EMI+Virtual HV with a CHW+Video Feedback; and (3) EMI+Hybrid HV with a CHW+Video Feedback. All Arms receive 16 weeks (4 months) of EMI stress reduction and family meal tip messages via smartphones. Arms 2 and 3 additionally receive eight home visits by CHWs focused on family meal quantity and quality, a meal preparation activity, and video feedback on their family meal behaviors/patterns every-other-week for 16 weeks. Arm 2 receives all of these components virtually and Arm 3 receives these components half in person and half virtual (hybrid). In between weeks, families in Arms 2 and 3 complete a Try-it-Yourself activity to apply the new skills/behaviors they have been taught. All Arms receive an 8-week (2 months) maintenance phase allowing for progressively less support so they can increase self-e cacy and sustainability of behavior change.
Once participants complete their baseline data collection visit, they are randomized into one of three study Arms (n=175 per Arm). If households have multiple eligible children, one child is randomly selected to minimize bias that could affect generalizability due to parent selection. Randomization is strati ed by ve racial/ethnic groups (African American, Hispanic, Native American, Asian American, White; n≈105 per race group). Block randomization schedules were produced in PASS 2021 (Kaysville, Utah) to account for the racial/ethnic strati cation. Schedules are maintained by the biostatistician to keep team members blinded.

Procedures and Data Collection
Virtual Data Collection. Once child eligibility is con rmed, baseline data collection occurs via a virtual zoom visit including: guided anthropometry [43] and neck circumference measurements, a child 24hr. dietary recall, registration for one-week of Ecological Momentary Assessment (EMA) on their phone [44], and training on video recording of family meals. Following the virtual visit, a 14-day observational period ensues including a parent online survey, two additional 24-hr dietary recalls, ten days of EMA measuring parent stress and parenting practices, and a 2-day video-recorded family meal observation period (1 weeknight, 1 weekend night) measuring family meal quality (i.e., dietary, interpersonal) [45]. Virtual data collection occurs at baseline, 6 months (post-intervention), and 12 months (6-month post-intervention). Primary and secondary outcome measures are described in Table 1 and are collected at all three data collection time points in all Arms. Virtual protocols are based on ours [37] and other's [7,37] prior studies. Data collection tools and databases (i.e., REDCap) include features to support HIPAA compliance and allows for data checks to ensure data quality during data entry. Access to data collection tools and databases including REDCap and Box will be strictly limited and regulated through personal user pro les. Both of these platforms are password protected and all data will be regularly backed up into a passwordprotected database.

Child Neck Circumference
Neck circumference (NC) has been shown to be highly associated with cardiovascular disease in children [47] and adults [47], and is strongly correlated with waist circumference [47]. NC is often chosen over other measures (e.g., waist circumference) as it is not impacted by respiratory movement or postprandial abdominal distension, is non-invasive, does not require clothing removal, and is easy to collect [47] Procedure: Families are sent a 24" measuring tape prior to their virtual visit to collect NC. Families watch a short training video on how to take NC. After the video, trained and certi ed research staff guide parents/guardians on taking the child's NC using standardized protocols.
Measure: Locate the prominence on the neck (i.e, Adam's apple), and wrapping the measuring tape around the child's neck directly under this prominence. The research staff con rm the child is looking straight ahead and that the measuring tape is straight. To ensure reliability, NC measurements are taken twice and measurements need to be within 0.5 cm of each other.
Child Diet Quality Child diet quality was used as a primary outcome given its link to child risk for heart disease and obesity in adulthood.

Family Meal Quality
Family meal quality consists of both dietary quality and interpersonal quality. These two factors will be measured via video-recorded (via Zoom) family meals as well as ecological momentary assessment (EMA). Procedures: · Video recorded family meals: Following their virtual visit, families are asked to record two family dinner meals (1 weekday, 1 weekend day) over the next 2 weeks where the majority of the family is present. Our prior research showed that video-recording two family meals, including one week day and one weekend day, was adequate for measuring dietary healthfulness and interpersonal atmosphere at family meals [7,45]. Families are sent a HIPAAcompliant Zoom link where they are able to log on when their meal begins; the link is set up so that recording starts immediately and the video is automatically uploaded to the cloud. At the start of the video, families are instructed to introduce everyone present at the meal and to say what is being served at the meal.

·
Ecological Momentary Assessment: At the virtual visit, staff members register parents for EMA [44], which includes asking about wake and dinner times on weekdays and weekend days. Parents are randomly texted three signalcontingent surveys during the day (determined by parent's wake time) as well as a meal survey, which is scheduled to be sent a minimum of an hour after the parent reports usually having dinner. Parents use their own smartphone in order to receive texts and access surveys. Parents have 1 hour to nish the rst three signal-contingent surveys and 4 hours to nish the meal survey sent at the end of the day. Following the virtual visit, parents have 14 days to nish 10 "complete" days of EMA; a complete day includes at least 2/3 signal-contingent surveys as well as the meal survey. If parents are able to consecutively nish 7 complete days, they earn a $25 bonus.

·
During the meal survey, parents are asked whether they had a family dinner meal. If they did, they are asked to complete a meal screener. This includes: 1) Listing all the food served at the dinner (e.g., tacos, tortilla chips, refried beans, fruit salad); 2) For each food served, asking about the components of each food, including: Fruits, Vegetables, Dairy, Meat Protein, Other Non-Meat Protein (e.g., tofu), Whole grains, Re ned grains, Snack Foods, and Desserts. If a participant selected Vegetables, Dairy, Meat Protein, Snack Foods, Desserts, and/or Sauces, they were asked to categorize the food even further. For example, if Vegetables was selected, the participant was asked if the food contained Dark Green Vegetables (e.g., kale, romaine lettuce, broccoli), Other Vegetables (e.g., corn, tomatoes, peas, potatoes), Fried Vegetables (e.g., French fries), and/or Pickles or olives.

·
Parents were also asked to report all of the drinks served at dinner, including: Water, Skim/low-fat milk, 2% or whole milk, Non-dairy drinks (e.g., soy milk), Tea, Coffee, Sports drinks, Fruit drinks (e.g., SunnyD), Regular pop/soda, Diet pop/soda, 100% fruit juice, No drinks were served, or None of the above. Measures: Quality. An adapted Healthfulness of Meal (HOM) index [21] is used to measure the dietary quality of foods served at family meals via the EMA meal screener. The HOM index is a quantitative coding system used in prior studies [21]. It was developed based on the Healthy Eating Index (HEI)-2015, which includes 13 components that are either adequacy components (food groups/subgroups that are encouraged) or moderation components (food groups where lower intake is encouraged.) Like the HEI-2015, points are given for the presence of some food components (e.g., Dark Green Vegetables), and the absence of other food components (e.g., Desserts). Higher scores represent a meal with higher dietary quality.

·
Meal Interpersonal Quality. The Iowa Family Interaction Rating Scales (IFIRS) [50] will be used to measure the interpersonal interactions that occur at the two video-recorded family meals. The IFIRS is a quantitative direct observational coding system that measures behavioral and interpersonal interactions (e.g., dyadic, family level) [50]. IFIRS scales have been used with diverse families and have high validity (r=.77-.86) [50]. Example measures include: parent feeding practices (i.e., restriction, pressure-to-eat), parenting style (i.e., authoritative, authoritarian, permissive, neglectful), and family functioning (e.g., communication, con ict). Higher scores (range 1-9) represent healthy interpersonal dynamics.

Parent BMI
Procedure: Parent height and weight are taken during the virtual visit using the same procedures as child height and weight.
Measure: BMI values are computed according to the following formula: weight (kg)/height (meters) 2 and cut offs for weight status are calculated using CDC guidelines [51].

Parent Neck Circumference
Procedure: Parent NC is taken during the virtual visit using the same procedures as child NC.
Measure: The same measure as child NC is used.

Food Parenting Practices
Procedure: Food parenting practices are assessed via an on-line survey completed by the parent after the virtual visit as well as through the EMA meal survey.
Measure: Parents are asked about a broad range of food parenting practices [52], which measure structure, autonomy support, coercive control, and indulgent y pp g feeding practices [52]. In the on-line survey, parents are asked about the frequency in which they engage in these food parenting practices; in EMA, they are asked (Yes/No) if they engaged in the food parenting practice at the meal that evening.

Parent Coping Skills
Procedure: Self-report on all four EMA surveys (three signal-contingent and the meal survey).
Measure: Parents are asked to report on their current stress level (How would you rate your level of stress right now?), coping ability (How would you rate your ability to manage stress right now?), and current sources of stress (Childcare, Job dissatisfaction, Body image/weight concerns, Too many things to do).

Sibling BMI%ile
Procedure: Sibling height and weight are taken during the virtual visit using the same procedures as child height and weight.
Measure: Sibling BMI%iles are calculated using the same procedures as child BMI%ile.

Other Secondary Outcomes and Sub-analysis Variables
Procedure: Other sociodemographic and home environment variables will be measured via validated self-report online measures. Staff only code families in which they did not participate in the measurement visit [7,20,21]. Practice videos are used until coders reach 95% inter-rater reliability and 100% after consensus meetings; 25% of videos are double coded and checked at a 1:5 ratio to ensure high inter-rater reliability and delity to protocols. For NDS-R, quality assurance is conducted on 100% of recalls [64].

Ongoing Supervision
After training, practice role-plays, and certi cation, the rst 10 virtual data collection visits are reviewed in detail by a supervisor (i.e., measurement team director or co-investigator) with the data collector who conducted the visit. Measurement team members continue to receive feedback on a randomly selected recorded virtual data collection visit by a supervisor on a weekly basis at either an individual supervision session or a group measurement team meeting.

Recerti cation
Recerti cation of measurement team members on all study data collection protocols and measures are conducted every quarter throughout the intervention.

Intervention Community Health Workers (CHWs) Supervision
Initial Training and Certi cation When onboarding, CHWs have 6-8 weeks of training focused on program session content (8 total sessions), foundation training (e.g., motivational interviewing (MI), SMART goal setting, family systems theory, trauma-informed care, food parenting practices, nutrition 101, video feedback) and additional training components (e.g., REDCap, zoom, scheduling visits). They also have the following practice and supervision sessions in preparation for certi cation and conducting in-home visits: provided with detailed information about the study by our research team via consent and assent forms including study aims and detailed procedures. Participants are informed that their participation is voluntary and that they have the right to withdraw from the study without any consequences at any point.
They will be assured of anonymity in participation and con dentially of any data they provide throughout the study, through the use of study IDs and the storage of sensitive information in secure online platforms (i.e., REDCap and Box). Participants can be enrolled into the study only after they have provided written consent and assent forms to our research team.
Regulatory Oversight/Monitoring. All study modi cations will be communicated with and regulated by the IRB. Even though the study is expected to pose minimal risk, the Data Safety Monitoring Board (DSMB), in collaboration with the study investigators will closely monitor recruitment, process evaluation, and retention activities. The DSMB will meet yearly with the study investigators and staff, or more often as needed, for oversight of the study. Any adverse events will be reported to the NHLBI and the IRB at the University of Minnesota. This trial is also registered in the OnCore clinical trial management system and is audited by the Medical School at the University of Minnesota.

INTERVENTION
The Family Matters three-arm intervention, known as the Family Matters Program to our study families, components and dose are described below.

Study Arm #1: EMI
Parents randomized to study Arm 1 receive EMI text messages twice a day for 16 weeks via their smartphone. A study smartphone is provided for use if needed.
EMI. Our prior research showed parental stress early in the day was associated with more controlling food parenting practices and serving more unhealthful foods (e.g., fast food) at dinner the same night [66]. Therefore, in all Arms, parents receive EMI text messages to their phones that include two parts. First, the parent is sent a text message with a survey link, between 11am-2pm to report their stress level (i.e., scale of 0-10) and sources of stress (e.g., child demands, busy at home/work, social media). Second, a text message is sent back to the participant from a bank of tips (approximately 50 tips per source of stress) for the particular source of stress they reported. This tailored tip is intended to help them cope with the reported stress and increase the likelihood that they will still carry out a family meal the same evening in the face of stress [66]. After the tip is sent, parents are also asked to report whether or not the tip was helpful, which then adjusts their individual EMI algorithm so there is an increased likelihood of them receiving more or less of these types of tips. If parents report no stress, they receive a tip to facilitate having a family meal (e.g., recipe ideas, meal prep tips, mealtime conversation starters).

Study Arm #2: EMI+Virtual Home Visiting (HV) with CHW+Video Feedback
Parents randomized to study Arm 2 receive all elements of study Arm 1, in addition to home visiting by a CHW. Visits by CHWs are virtual via zoom and occur every-other-week (8 total) simultaneously with the 16 weeks of EMI. In between CHW home visits, families complete "Try-it-Yourself" activities (8 total) to reinforce new behaviors and meal preparation skills (e.g., batch cooking recipe, shopping scavenger hunt, stress reduction coping skills).
Home Visiting. CHWs carry out the 60-90-minute HVs using Motivational Interviewing (MI) [67,68] and psychoeducation [35]. The visits focus on family meal quantity and quality factors [7,20,21] known to be associated with child CVH. Family members are taught speci c skills through didactic and interactive session activities (e.g., AHA Slides, Figma games) to improve family meal processes and behaviors. Session content and activities are described in Table 3  Google slides c "You're the Expert" scenarios: 'What would you do if…?' scenarios game: focus on using 4-step process for stress reduction to handle common stressful family situations/ challenges · SMART goal setting focused on ingredient #2: have good communication and interactions: focus on using 4-step process for stress reduction; using apps/exercise to reduce stress (coping tools) You are equipped with the tools and knowledge you need to do this on your own · Barriers happen; your family barriers can be reduced (i.e., identify them and make a plan to mitigate) to ensure future success · Celebrate and Try-It-Forever!! · High 5 bingo board game: video feedback (for family recorded meal) · Family Matters Storybook a demonstrating concepts taught during session 8 · Aha slides f : 'Ask the Expert' scenarios game: discussion of common scenarios and potential solutions from concepts taught throughout the program · SMART goal setting focused on 3 ingredients of a successful family meal: "try-it-forever" "Try-it-Yourself" Activities. Families are given food-related (e.g., recipes, meal planning strategies) and interpersonal (e.g., food prep with kids, family meal communication game, stress reduction) activities to try out in between visits to increase their self-e cacy in preparing family meals on their own and reinforce messages they are taught during HVs. The study child and all siblings in the home are also given an activity book with games that reinforce session content.

Study Arm #3: EMI+Hybrid HV with CHW+Video Feedback
Parents randomized to study Arm 3 receive all elements of study Arm 2, but they are delivered hybrid.
Speci cally, CHWs meet in-person with families every other HV and then virtually via zoom on the other weeks. Families also engage in two cooking demonstration activities with the CHW during in-person HVs to reinforce messages taught, share easy recipes (e.g., batch cooking, one ingredient for multiple meals), and teach food prep skills to increase family's self-e cacy for having family meals. This Arm is important to examine whether relationship building and creating an atmosphere conducive to health behavior change requires an in-person component. This Arm is also critical to examine COVID's impact on moving research to virtual modes.

Maintenance Phase
After completing four months of the intervention, all study Arms transition to a two-month maintenance phase, based on best practice [70]. For all study Arms, EMI family meal tips are reduced to the three days per week that parents reported their highest stress levels during the 16-week active intervention phase.
Stress pro les corresponding to the high risk stress days are created for each participant to maximize intervention uptake and subsequent sustainability [24,25].  [7,27], and HV protocols. The CHW supervision process provides multiple levels of supervision throughout training and intervention delivery (see Table 2). Co-investigators who provide supervision and the intervention director (MA) are trained in MI and are licensed mental health clinicians (JB, TM) or registered dietitians (KL, DNS). All role-plays, certi cation, and family intervention visits in all study Arms are video-recorded. Both the CHW and supervisor watch and give feedback on the video-recordings, which allows for thorough feedback. Just as families receive video feedback on their recorded family meals from the CHWs as part of the intervention, the CHWs are given feedback as well, thus creating a parallel process that models to the CHW how to give feedback that is collaborative, focusing on both strengths and areas of growth, with their own intervention families.

Process Evaluation
A robust feasibility and process evaluation protocol was designed for this intervention (see Table 4), to ensure feasibility, generalizability, and dissemination into primary care and other health care settings [70,71]. Collect and analyze outcome data at post-intervention and 6-month post-intervention

STATISTICAL ANALYSIS PLAN
Overview. This study is powered for three pairs of tests [72] to evaluate intervention effectiveness: (a) Arm 2 vs. 1, (b) Arm 3 vs. 1, and (c) Arm 2 vs. 3 over three time points. Multi-level, general linear mixed models (MLMM) with a clinic random intercept that nests participants within clinics to address any clinic differences in the recruitment populations, with participant random slopes for time to examine intervention treatment effects, and conditional xed effects regression models (within-child analytic contrast against baseline), are the primary analytical models for all study hypotheses. Participants' randomized condition will be examined irrespective of adherence to the study protocol in accordance with an intent-to-treat (ITT) analysis. After the intervention has been fully administered, data will be assessed for balance across arms, outliers, missingness, and other modeling assumptions. Although randomization is expected to produce balance on measured and unmeasured characteristics, variables will be considered for inclusion as controls in adjusted analyses to reduce test statistic variance [73]. We expect little missing data based on our prior work, but if needed, we will employ methods recommended for clinical trials to minimize analytical assumptions required when missing data are present (e.g., follow up all randomized participants prior to unblinding [74], evaluating if results from the primary analysis differ when sensitivity analyses are performed [75]). Reasons for participant withdrawal and nonadherence will be analyzed and reported in the nal ITT analysis [76].
Sample Size and Power Computations. Study design features were accounted for in powering the study that required increases in sample size to minimize an in ated experiment-wise error rate (EER) due to three pairwise tests between each treatment arm for three primary outcomes (i.e., BMI%ile, neck circumference, and the Healthy Eating Index (HEI)). Accounting for these nine tests, sample size was determined using a conservative two-sided critical value of z=2.77 (P=0.006) to achieve experiment-wise Type I error of 0.05. Our power calculations were based on prior studies showing that a decline of two BMI%ile points was a clinically meaningful difference in children with overweight/obesity [46]. BMI%ile is a continuous outcome with a variance of 18.8. Eighteen-month follow up data with a comparable cohort provided intraclass correlation coe cient estimates to inform sample size determination (BMI%ile ICC ).
At 80% power and multiple-outcome and pairwise testing corrected EER of 0.05, with a sample size of 525, we will be able to detect a minimum average difference in BMI%ile as small as 1.67 (or 0.38 SD) with 15% attrition. This magnitude translates to approximately a 2.8lb difference in a six year-old boy who is 3.8 feet tall and 45 lbs, or approximately a 7.5 ounce per month change in weight by the post-intervention endpoint.
Aim 1 (Primary Outcomes): Examine intervention effects on markers of child CVH including BMI%ile, diet quality, neck circumference. Treatment condition mean differences on the three primary outcomes will be examined at the post-intervention primary endpoint (6 months after baseline). Sample size determination allows for primary outcome standardized effect size assessment of all three outcomes of at least 0.38, which is a small-to-moderate minimum detectable effect.
Aim 2 (Secondary Outcomes): Examine intervention effects on family, parental, and sibling factors. Family meal quantity and quality, food parenting practices and stress, and sibling BMI %ile outcomes are powered at similar levels as in Aim 1 with the ability to detect standardized effect sizes as small as 0.38.
Sub-group Exploratory Analyses: Analyses exploring whether interaction effects depend on participant sex, race/ethnicity, and seasonality will also be conducted. These post-hoc analyses will examine whether the intervention has synergistic effects in speci c populations or during different seasonal contexts. Post-hoc analyses will be conducted to explore the interaction of child/parent sex and baseline weight status on intervention treatment effects to determine whether the intervention is particularly effective in certain subpopulations.
Other Exploratory Hypotheses: A model incorporating an interaction effect of treatment arm crossed with the change in family meal quality and quantity between observation periods will be used to evaluate if increases (or decreases) in the quality and quantity of family meals correspond with synergistically favorable (or unfavorable) child outcomes. This analysis will inform whether intervention effects depend on participants' changes in family meal quality and quantity. This analysis is powered to detect a between-within intervention slope difference over the 6-month intervention period of as little as 9.6 ounces per month, depending on whether participants had high or low change in the moderating variables. Seasonality robustness checks will also be performed to evaluate whether results differ substantively for participants who received the intensive intervention during the summer months.

Discussion
The Family Matters intervention has high potential public health impact as it aims to change clinical practice by creating a new model of care for child CVH in primary care. Research in this eld is needed given the low to moderate success of lifestyle behavior interventions for children at risk for CVD and the persistent high prevalence of disparities across race/ethnicity groups. The state-of-the-art measures being used including EMA, EMI, and video feedback combined with the novel intervention context of family meals and CHWs as interventionists will greatly advance the eld. In addition, the three-arm study design will allow for testing which combinations of intervention components are most effective in improving child CVH by race/ethnicity as well as whether a virtual or hybrid Arm is more effective.
Dependent on study ndings, this intervention will be disseminated to other primary care settings. The study was conducted according to the guidelines of the Declaration of Helsinki and was approved by the Institutional Review Board Human Subjects Committee at the University of Minnesota (IRB ID: STUDY00014632). All participants provide their written informed consent or assent before participating in the study. For minors, a signed assent form will be obtained from their parents/or legal guardians before participating in the study.

Consent for publication
Not applicable.   The SPIRIT diagram