Studying the impact of COVID-19 mitigation policies on childhood obesity, health behaviors, and disparities in an observational cohort: Protocol for the COVID-19 Family Study

Abstract

Background:

COVID-19 pandemic control policies, including school closures, suspended extra-curricular activities, and social distancing, were introduced to prevent viral transmission, and disrupted children’s daily routines, health behaviors, and wellness. This observational cohort study among 697 families with children or adolescents, based on the Family Stress Model, aims to: 1) evaluate pre- to during-pandemic changes in child health behaviors (diet, physical activity, sleep) and weight gain, 2) identify mechanisms explaining the changes, and 3) determine projected healthcare costs on weight gain and obesity. Each aim includes an examination by racial and ethnic, socioeconomic, and geographic disparities.

Methods:

The study employs a mixed methods design, recruiting children and their caregivers from two obesity prevention trials halted in 2020. Enrolled participants complete annual surveys to assess child health behaviors, family resources, routines, and demographics, and home environment in 2020–2022. Height and weight are measured annually in 2021–2022. Annual semi-structured interviews are conducted within a subsample to understand mechanisms of observed changes. Multilevel mixed models and mediation analyses are used to examine changes in child health behaviors and weight gain and mechanisms underlying the changes. Qualitative data are analyzed within and across time points and integrated with quantitative findings to further explain mechanisms. Markov models are used to determine healthcare costs for unhealthy child behaviors and weight gain.

Conclusion:

Findings from this study will aid in understanding pandemic-related changes in child health behaviors and weight status and will provide insights for the implementation of future programs and policies to improve child and family wellness.

BACKGROUND

In the absence of COVID-19 vaccines or effective treatments, national policies were implemented to prevent the spread of the novel coronavirus beginning in March 2020. Restrictions and social distancing led to closures of childcare centers, schools, work sites, and other public facilities. Families were often confined to their homes with limited external contact.

Pandemic restrictions were frequently accompanied by economic threats, overriding fear of illness, and changes in family lifestyles and routines. Many families experienced socioeconomic hardships, employment instability, food insecurity, limited access to assistance programs, and increased COVID-19 related healthcare expenditures (1–3). Socioeconomic disparities increased, particularly among racial and ethnic minority groups (4). Children experienced increased risks of stress and anxiety (5, 6), unhealthy diets (7), physical inactivity (8), abnormal sleep patterns (9, 10) and, long-term, obesity and co-morbidities (11, 12).

Childhood obesity, a global critical public health problem prior to the onset of the COVID-19 pandemic, increased during the pandemic, as noted by population-level studies using electronic medical records (13–15). Overweight and obesity during childhood increase the risks of adult health complications including type 2 diabetes, hypertension, cardiovascular diseases, and cancer (16, 17), highlighting the urgency of addressing pandemic-related increases in childhood obesity.

In addition to harming health, obesity increases healthcare expenditures. Pre-pandemic trends predicted an increase in annual medical costs from obesity-related medical issues (arthritis, coronary heart disease, diabetes, etc.) of $66 billion/year by 2030 (18). The rising pandemic-related obesity prevalence among children (13–15), may further increase such medical costs in all population (19). However, gaps in how pandemic-related increases in obesity relate to healthcare costs among children hinder predictions.

The Family Stress Model (FSM) posits that declines in socioeconomic status affect child wellness via impacts on caregivers (20, 21). During the pandemic, COVID-19 related socioeconomic challenges altered caregiver wellbeing and family processes that led to changes in child wellness (22). In this study, the FSM is applied to understand the economic, policy, and resource changes during the pandemic and their impact on family routines and child wellness (See Figure 1). Specifically, the FSM provides a framework to study the association between COVID-19 control policies and the family resource pathway (loss of school, income, and services, and food insecurity) and psychosocial pathway (caregiver stress, anxiety, and depression), and how changes in resource and psychosocial pathways impact family routines, subsequently influencing child health related behaviors and health outcomes (22). The FSM also enables us to explore protective factors, such as safety net programs (e.g., federal stimulus funding, access to pandemic relief services, etc.) or interpersonal factors (e.g., social support, family cohesion, etc.) that may have mitigated adverse health outcomes.

Figure 1:

Conceptual Framework based on the Family Stress Model (FSM)

The impacts of family resources and caregiver psychosocial pathways on children’s wellness may differ by racial and ethnic, socioeconomic, and geographic factors. The pandemic exacerbated socioeconomic challenges, with higher unemployment, income loss, and food insecurity observed in Black and Hispanic communities compared to White communities (23). These hardships negatively affected health and healthcare outcomes, contributing to increased mental distress and obesity (24). Additionally, residents in rural areas faced higher risks of physical and mental health issues and inability to pay bills or afford groceries or other necessities (25). While disparities in outcomes are acknowledged, understanding the underlying mechanisms is essential to advocate for targeted policies and interventions. Research on changes in children’s wellness should consider these factors to tailor health promotion strategies to diverse vulnerable populations.

While many global studies have explored COVID-19 lockdown effects on childhood weight changes, there’s limited longitudinal evidence for U.S. children (14, 26, 27). Previous studies lacked contextual data on family resources, caregiver psychosocial well-being, and protective factors, hindering our understanding of mechanisms linking COVID-19 policies to child health behaviors and obesity. This unique prospective cohort study, spanning pre-pandemic to three years post-pandemic, enables understanding of not only the impact of COVID-19 policies on changes in child health behaviors and well-being, but also the underlying mechanisms driving these changes across a diverse sample, as well as the implications for healthcare costs.

Study Aims

The COVID-19 Family Study seeks to explore the following objectives: 1) to identify the long-term impact of pandemic policies on disparities related to child wellness and health behaviors over two years following the start of the pandemic, 2) to examine the underlying mechanisms linking COVID-19 policies and child health like obesity/excess weight gain, and 3) to investigate the impact of child obesity/excess weight gain on healthcare costs. The Specific Aims and Hypotheses are outlined in Table 1 and shown in the context of the FSM in Figure 1.

Table 1:

Specific Aims and Hypotheses

Aims	Hypotheses
Aim 1: Examine pre-pandemic (< March 2020) to during-pandemic changes [2020, 2021, 2022] in child health behaviors (diet, physical activity, and sleep), and obesity/excess weight gain	Hyp 1a: Families with risks in psychosocial/resource pathways (i.e., high caregiver stress or loss of employment) experience greater declines in child health behaviors (Hyp 1a-i) and greater increases in obesity/ excess weight gain (Hyp 1a-ii), during the pandemic onset and less improvement during the 2 years following the onset compared to families with few risks.
	Hyp 1b: Worsened child health behaviors are related to greater excess weight gain/risk for obesity.
	Hyp 1c: Protective factors moderate the association between the FSM psychosocial and resource pathways and declines in child health behaviors and increases in obesity/excess weight gain over time.
	Hyp 1d: Over time, disparities in child health behaviors and obesity/excess weight gain widen within families with low socio-economic status (SES) (vs. middle or high SES), non-White or Hispanic (vs. non-Hispanic White), or in rural or urban (vs. suburban) areas, resulting in worse outcomes during the pandemic onset and less improvement during the 2 years following the onset.
Aim 2: Examine the mechanisms through which changes in parental psychosocial factors and/or family resources during the COVID-19 pandemic may influence family routines, which are considered as key factors explaining observed changes in child health behaviors and obesity/excess weight gain.	Hyp 2a: Patterns of change in child health behaviors and obesity/excess weight gain are mediated by disruptions in family routines.
	Hyp 2b: Fewer disruptions in family routines are related to better outcomes in child emotional health, predicting better child health behaviors.
Aim 3: Determine projected health care costs related to changes in child health behaviors and obesity/excess weight gain due to COVID-19 policies overall and for specific populations, by race/ethnicity, SES, and geographic locale.

METHODS

Study Design and Settings

The COVID-19 Family Study is a prospective observational cohort that includes families previously recruited from childcare centers or elementary and middle schools across Maryland. This study includes pre-pandemic measures in addition to qualitative and quantitative data collected annually from 2020–2022.

Study Population

In this study, we re-enroll families from “Creating Healthy Habits Among Maryland Preschoolers” (CHAMP) and “Wellness Champions for Change” (WCC), pre-pandemic intervention trials in Maryland childcare centers and elementary and middle schools, respectively. Details for these projects have been described previously (28, 29). Briefly, CHAMP was an NIH-funded cluster randomized trial implemented in 56 childcare centers in 10 Maryland counties that aimed to build healthy diet and physical activity habits into childcare activities and extend them to caregivers and families (28). WCC was a USDA-funded cluster randomized control trial implemented in 33 elementary and middle schools in 5 Maryland counties that aimed to examine how increased implementation of local wellness policy in schools impacts children’s health behaviors including dietary intake and physical activity (29). The data collected in CHAMP and WCC (from 2017 – January 2020) are used as pre-pandemic data. Protocols for CHAMP and WCC were approved by the University of Maryland School of Medicine Institutional Review Board, and the current COVID-19 Family Study has been approved by the University of Maryland School of Medicine and Johns Hopkins Bloomberg School of Public Health Institutional Review Boards.

Participant Recruitment Procedures

Caregivers and children are eligible if 1) the caregiver completed a pre-pandemic baseline survey from WCC or CHAMP, and 2) the family lives in Maryland during the pandemic. Of 1,835 children and caregivers enrolled in the CHAMP and WCC projects, a sub-sample of 1,048 (57.1%) families meet criteria for re-enrollment in the COVID-19 Family Study. Eligible caregivers are contacted by e-mail, mail, or phone in May 2020 to assess their interest. Caregivers provide consent for their participation and their child through an electronic informed consent process. WCC children who are invited to complete a child survey, also provide electronic assent. For qualitative interviews, we recruit a maximum variation sub-sample from the enrolled families in Fall 2020 and conduct longitudinal interviews with caregivers and WCC children at each time point. Families are recruited in equal numbers from strata using five criteria to ensure a balanced selection: study (WCC or CHAMP), locale (urban or rural), child race (white or non-white), and socioeconomic status, including household food security (high or low) and income (high or low). We select 2–3 caregiver-child dyads using a randomization procedure within each stratum (~50 total), aiming to retain ~40 (80%) over the course of the study.

Data Collection Procedures

Surveys are administered electronically (e-mail) or mail (caregiver preference) and available only in English. Among the interviewed subsample, interviews are scheduled approximately six months after surveys to enable time for survey analysis to inform interview guides. For caregivers only, interviews are ~40 minutes. For caregiver-child dyads (from WCC), interviews are ~60 minutes (40 minutes with the caregiver and ~20 with the child). Interviews are administered and recorded by trained research assistants via Zoom or phone. A subset of child participants is invited to wear an accelerometer to measure physical activity and sleep for one week (as done in WCC and CHAMP). After confirming willingness to participate in accelerometer data collection, devices are mailed to participants’ homes with instructions, and a pre-paid return envelope. Beginning in 2021, all participants are invited annually to meet the study team in community sites for anthropometric data collection. Participants receive an electronic gift card after each contact: $25 for caregiver survey, interview, or child anthropometric measures, and $15 for child survey, interview, or accelerometer data.

Power Calculation and Sample Size

In 2020, we recruited families from CHAMP and WCC projects to enroll in the COVID-19 Family Study. Given a type I error of 0.05, a sample of 600 child-caregiver pairs provides enough power (>0.80) to detect: 1) a small change in child behavior, excess weight gain, or obesity over time (Cohen’s d>=0.13), considering 20% attrition, 2) a modest difference in changes for moderators (Cohen’s d>=0.25), 3) a modest relation between psychosocial and resource pathways and child health behaviors or obesity/weight gain, or between child health behaviors, weight gain, and obesity (Cohen’s d>=0.20–0.22, separately), 4) a small moderating effect (Cohen’s d>=0.10–0.11) of the protective factors on the relations based on G-power software, and 5) power (p> 0.80) to examine mediation through structural equation modeling, assessed with a simulation study using Mplus.

Study Measures

We employ 6 methods of data collection in this study, outlined below.

1. Caregiver Survey

Caregiver survey constructs are described in Table 2. For child health and behavioral outcomes, caregivers from CHAMP families respond to additional questions due to the age of the children (child diet and sleep behaviors). Survey items related to other key constructs were completed by both CHAMP and WCC caregivers, including disparity variables (demographics such as race and ethnicity, income, receipt of services, etc.), resource pathway variables (reflecting changes in daily necessities, schools, social interactions, and food security, etc. during pandemic) and psychosocial pathway variables are asked to. For mechanisms, CHAMP and WCC caregivers respond to survey items on family routines, and CHAMP caregivers also respond to additional information on comprehensive feeding practices and child mental health.

Table 2:

Constructs Measured in the Caregiver Survey

SECTION	Construct	Measure/Reference	Data Collection
			Pre-pandemic (2017-Jan 2020)	Early pandemic (2020)	Mid-pandemic (2021)	Late pandemic (2022)
			T0	T1	T2	T3
Outcomes	Child Diet *	Brief FFQ(30)	X	X	X	X
	Child Sleep *	Child Sleep Habits Questionnaire(73)	X	X	X	X
Disparity Variables	Demographics	Includes race and ethnicity, income, dependents, education, receipt of services, etc.	X	X	X	X
Resource Pathway	COVID-19 Resources (Lost/Gained)	Newly developed/adapted from scales used in surveys administered in Tennessee, Pittsburgh, and Vermont		X	X	X
	Food Insecurity	2-item screen(35) and Household food security short form(36)	X	X	X	X
Psychosocial Pathway	CG Social Support	Functional Social Support Scale(37)-Appraisal		X	X	X
	HH Chaos	CHAOS (Confusion, Hubbub And Order Scale)(39)		X	X	X
	Stress	Perceived Stress Scale(40)	X	X	X	X
	Anxiety	State-Trait Anxiety Inventory(44)	X	X	X	X
	Depression	Center for Epidemiologic Studies Depression Scale-Revised(74)	X	X	X	X
Mechanisms	Family Routines-Diet, Physical Activity	Comprehensive Home Environment Survey (CHES)(47)	X	X	X	X
		Comprehensive Feeding Practices Questionnaire*(48)	X	X	X	X
	Child mental health	Child Strengths and Difficulties Questionnaire * (51)	X		X	X

^*CHAMP only

Child health and behavioral outcomes

Child Diet:

CHAMP caregivers report their child’s usual eating habits using the 17-item validated Food Frequency Questionnaire (FFQ) (30) assessing: 1) daily servings of fruits, vegetables, and cups of beverages; 2) frequency of eating lean meats, processed meats, take-out foods, snack foods, and desserts; and 3) behavioral habits including the regularity of consuming breakfast and eating in front of the television (31).

Child Sleep:

CHAMP uses the abbreviated 18-item validated Children’s Sleep Questionnaire to collect information on children’s usual sleep quantity, perceived quality, and adequacy over the previous week (32). For each item, caregivers report the frequency of sleep problems (33).

Disparity variables (Demographics)

Both CHAMP and WCC caregivers provide demographic information on age, gender, race, and ethnicity on behalf of themselves and their participating child. Information is also collected on caregivers’ marital status, highest level of education, employment status, income status, number of children and adults in the household, and receipt of federal assistance services for the family. Information on reception of community services, public assistance/safety net, and federal stimulus support, etc. is considered as protective factors underlying the mechanisms of observed changes.

Resource pathway variables

COVID-19 Resources:

Caregiver surveys address: 1) access to food, essential items, medicine, babysitters, and parks and playgrounds specific to the pandemic, 2) pandemic impact on school learning mode and meals (in-person, hybrid, and remote), social interactions (caregiver and child), and employment; and 3) COVID-19 exposure (34).

Household Food Insecurity:

In the caregiver survey, households at risk of food insecurity are identified using a validated 2-item screener (97% sensitivity) (35). Household food insecurity is further determined using the 6-item USDA-developed Household Food Security Short Form (36).

Psychosocial pathway variables

Family Social Support:

Social support is measured using the “appraisal” subscale from the 12-item Interpersonal Support Evaluation List (37). This 4-item subscale measures perceived availability of someone to talk to about personal problems on a four-point scale (38).

Family Chaos:

Household chaos is assessed using the Confusion, Hubbub, and Order Scale questionnaire, a 15 true/false items to capture daily disruption or environmental confusion within the home setting regardless of child age (example: “we almost always seem to be rushed”) (39).

Caregiver Perceived Stress:

The 4-item Perceived Stress Scale measures the perceived stress of generalized situations (not specific events or experiences) over the previous month using a 5-point Likert scale (40). The Perceived Stress Scale, designed for use in community samples, has been used to assess parenting stress (41).

Caregiver Anxiety:

The validated 6-item short-form State-Trait Anxiety Inventory measures anxiety among caregivers (42, 43). The short-form produces scores similar to those obtained using the 20-item measure across subject groups manifesting normal and raised levels of anxiety (44).

Caregiver Depressive Symptoms:

Caregiver’s depressive symptoms over the past week are assessed by the 20-item Center for Epidemiologic Studies Depression Scale (45). Participants report the frequency of each depressive symptom on a 4-point Likert scale (46).

Mechanisms (Family Routines on Diet, Physical Activity, and Child Mental Health)

Family Routines:

Both Champ and WCC caregivers are asked about family routines using scales adapted from the Comprehensive Home Environment Survey, a measure of the home food, physical activity, and media environment including 18 subscales and a total score. Caregivers report on a 30-item home food and physical activity availability inventory and a 10-item screen access survey (47). In addition, using four Comprehensive Home Environment Survey subscales (32 items), they report their perceived support for: 1) parental policies for physical activity, 2) parental policies for healthy eating, 3) parent rules for healthy eating, and 4) parent healthy role modeling using a 5-item Likert scale (47).

Comprehensive Feeding Practices Questionnaire (CFPQ):

CHAMP caregivers complete a Comprehensive Feeding Practices Questionnaire to assess the feeding practices of young children (48). The adapted questionnaire includes 8 subscales (one to three items/subscale), including emotion regulation, child control, monitoring, encourage balance and variety, food as reward, involvement, pressure, and restriction for health (49).

Child Mental Health:

For CHAMP children, caregivers respond to the 25-item validated Child Strengths and Difficulties Questionnaire (50) which assesses child attributes across five 5-item scales: Hyperactivity, Emotional Symptoms, Conduct Problems, Peer Problems, and Prosocial (51). The questionnaire uses a 3-point Likert scale (“Not true”, “Somewhat true”, to “Certainly true”).

2. Child Survey

Child survey constructs (Table 3) are administered to WCC children only (enrolled elementary and middle schools, pre-pandemic).

Table 3.

Constructs Measured in the Child Survey (WCC only)

SECTION	Construct	Measure/Reference	Data Collection
			Pre-pandemic	Early pandemic	Mid-pandemic	Late pandemic
			T0	T1	T2	T3
Outcomes	Child Diet 1	Patterns of Diet at School (PODS, adapted for weekdays at home)(75)	X	X	X	X
	Child Sleep	PROMIS sleep disturbance short form(54) Pediatric Daytime Sleepiness Scale (PDSS)(55)		X	X	X
Resource Pathway	COVID-19 Resources (Lost/Gained)	Newly developed/adapted from scales used in surveys administered in Tennessee, Pittsburgh, and Vermont		X	X	X
Mechanisms	Family Routines-Diet, Physical Activity	Comprehensive Home Environment Survey (CHES)(47)	X	X	X	X
	Child Mental Health	NIH Toolbox: Child Emotional Battery-Surveys (loneliness, emotional support, positive affect, fear)(57)		X	X	X

¹Because of school closures during COVID-19, the survey was modified to emphasize weekdays (Monday to Friday) over the past two weeks and removing the location information, reflecting a typical “school” week.

Child health and behavioral outcomes

Child Diet:

WCC children report their eating patterns during the school week using the validated Patterns of Diet at School (PODS) dietary screener (52), yielding type and frequency of foods and beverages consumed within MyPlate categories (53).

Child Sleep:

WCC children report on the general time they go to bed at night and wake up in the morning in addition to sleep disturbances (quality, trouble falling asleep/staying asleep, etc.) using the validated PROMIS sleep disturbance short form (54) and daytime sleepiness (daytime sleepiness/grumpiness, etc.) using the validated Pediatric Daytime Sleepiness Scale (55).

Resource pathway variables

COVID-19 Resources:

Children in WCC are asked additional questions regarding the social interactions, school mode and meals, COVID-19 exposure, and feelings regarding lost access and closures related to COVID-19.

Mechanisms (Family Routines on Diet, Physical Activity, and Child Mental Health)

Family Routines:

WCC children are asked about perceived parental support on physical activity and healthy eating via an adapted and validated version of the Comprehensive Home Environment Survey (CHES), which consists of 32 items with 5-item Likert response sets in 4 subscales: 1) parental policies for physical activity, 2) parental policies for healthy eating, 3) parent rules for healthy eating, and 4) parent healthy role modeling (47, 56).

Child Mental Health:

For WCC children, the NIH Toolbox Child Emotional Battery is used to assess symptoms of depression, anxiety, and associated emotional and interpersonal problems (57). The Toolbox includes measures of negative affect (anger, fear, and sadness), psychological wellbeing (positive affect, life satisfaction, and meaning and purpose), stress and self-efficacy (perceived stress and self-efficacy), and social relationships (social support, companionship, and social distress) (57).

3. Obesity/Excess weight gain

Height (cm) and weight (kg) are measured in triplicate using a portable stadiometer (Shorr Productions, Olney MD) and standard scale (300 GS, Tanita Corp, Tokyo Japan) for both CHAMP and WCC pre-pandemic, using the same protocols and training. During the pandemic, trained data collectors with tents and mobile equipment are dispatched to sites that are convenient to study participants, thus ensuring the rigor and reproducibility that is necessary to study changes in weight and obesity. If families are unable to schedule an in-person data collection, caregivers are given an option to provide details from a physician visit within the prior 12 months (date of visit, recorded height and weight, physician contact information). To examine incidence of overweight or obesity over time, BMI-for-age z-scores and percentiles are calculated according to CDC growth charts, with thresholds applied for (58, 59): underweight (<5th %tile), healthy weight (5th ≤ BMI-for-age <85th %tile), overweight (85^th ≤ BMI-for-age < 95th %tile), obesity (BMI-for-age ≥ 95th Percentile), and severe obesity (BMI-for-age ≥ 120% of the 95th %tile or ≥35 kg/m²). Excess weight gain is assessed by a gain of >3% in BMI-for-age z-scores compared with the expected maintenance value from the growth curve (59).

4. Accelerometry

A random sub-sample is selected for assessments on physical activity and sleep using the Actical accelerometer (Philips Respironics, Bend, OR) annually, beginning in 2020. Subsample stratification is used to ensure data are captured across age ranges (pre-pandemic pre-school, elementary school, middle school) and gender, given a limited number of accelerometers. Data is not collected in summer months. Children with prior accelerometry data from CHAMP and WCC are eligible to receive an accelerometer (~80% of sample) and asked to wear the accelerometer using ankle placement for seven consecutive days. Validated thresholds are applied to identify time spent in 24-hour movement: sedentary, light, and moderate-to-vigorous activity (MVPA) (60, 61) and sleep (62).

5. Tracking COVID-19 Policies over Time

This study gathers state- and county-level policies on pandemic mitigation strategies and school meal policies and practices (34). We track policies related to sports, youth activities, childcare centers, restaurants, and other public facilities by county. In addition, local and state government websites and media outlets are searched for other policies relevant for children and families. These data are matched to the child’s school or family’s neighborhood to determine school- and community-level resources.

6. Caregiver and Child Interviews

As is standard with an explanatory design, interview guides are developed iteratively based on survey findings each year. To maintain a trajectory analytical approach with comparability over time, we have similar topics of interest across all three time points. Specifically, the guides align with quantitative instruments to further investigate family-level resource and psychosocial pathways and seeks to identify and investigate family-level protective factors that are not well-known and thus not captured via survey. Additionally, interview guides explore the role of changing family routines (e.g., physical activity, meals, sleep patterns) as a mediator. We anticipate additional unique questions as the pandemic landscape evolves.

Planned Analytical Approaches

For aim 1, descriptive analyses will be conducted. Data will be checked for normality and transformation will be conducted if necessary. To examine changes over time, multilevel models including linear regression models for continuous variables (e.g. diet quality, minutes in MVPA, BMI-for-age z-score) and generalized linear regression models for categorical variables (e.g. overweight and obesity) will be conducted with time as an independent categorical variable, accounting for the clustering of repeated measures over time within each child. Changes between each pair of adjacent assessments will be estimated using a post-estimation method. Time-lagged models will be used to examine how antecedent risks in the psychosocial/resource pathways (i.e.: caregiver stress, loss of employment, etc.) impact child health behaviors and how antecedent child health behaviors impact excess weight gain/obesity. Protective factors, race and ethnicity, socioeconomic status, and locale, will be integrated to examine potential moderating effects between psychosocial/resource pathways and child health outcomes. Detailed analytical methods for each hypothesis in Aim 1 are included in Appendix 1.

For aim 2, mixed methods analyses including quantitative and qualitative analysis will be applied. Longitudinal sequential mediation analyses will be conducted using the product of regression coefficient method to understand how family routines and child emotional health may function as mediators between child health behaviors and weight outcomes. Autoregressive mediation models over time will also be conducted to assess contemporaneous (mediation effect at the same assessment) and longitudinal (mediation effect across time with predictors ahead of mediators and mediators ahead of outcomes) mediation in the same model. Sequential mediation via family routines and further via the children’s emotion health will be assessed to understand the complex relations among psychosocial/resource factors, family routines, emotional health and children’s behaviors or obesity/excess weight gain. In all analyses, maximum likelihood will be used to account for missing values due to loss to follow up.

The interview data will further explore the quantitative mechanisms through integrating in the design phase (via the interview guide), analysis, and reporting phases (63). Interviews will be transcribed verbatim and analyzed using a hybrid deductive/inductive coding approach, whereby initial analysis is performed to develop a codebook in alignment with the FSM and related to the relationships or mechanisms of interest from the quantitative phase. Multiple coders will continue to apply these codes related to the model until a stable thematic structure is developed, which will be applied to the full dataset. The quantitative findings and qualitive themes in our study will be compared to identify areas of congruence and discrepancy, which will further examine our hypotheses and potentially contribute to development of a stronger pathway and offer alternative findings Qualitative analysis and visualization of mixed methods findings will be conducted in MAXQDA 2020 (Verbi Software, 2019). We will report findings for which both qualitative and quantitative methods were used via joint display (64).

For aim 3, Markov models will be developed to project 20-year healthcare costs related to changes in weight status among children as a whole population group and by subgroups such as race and ethnicity, SES, and geographic locale. Markov models will be built with five health states including healthy weight (BMI<85th %tile), overweight (85th≤ BMI<95th %tile), class 1 obesity (95th≤ BMI<99th %tile), class 2 obesity (BMI ≥ 99th %tile) (65), and age-dependent all-cause mortality. The time horizon of our Markov model will be 20 years and the cycle length will be one year. The models will be designed under two scenarios for 20 years: 1) counterfactual arm in the absence of COVID-19, and 2) status quo with COVID-19 prevention policies. The excess healthcare costs of obesity among children associated with the COVID-19 pandemic will be estimated by performing a probabilistic analysis using a Monte Carlo simulation with 10,000 random samplings from parameters’ statistical distributions based on literature (66). Specifically, in the base case analysis, modeling over 20 years will be run. The future costs will be discounted at 3% in concordance with the current guidelines (67). In addition, in a sensitivity analysis, the undiscounted outcomes for 20 years, as well as for other time horizons including 5 years, 10 years, and lifetime, will be reported. For all analyses, a societal perspective and model indirect costs in line with recommendations by the Second Panel on Cost-Effectiveness in Health and Medicine will be adopted (67). All models for aim 3 will be built in R Statistical Programming (68).

Study Timeline and Study Status

This study re-recruited 583 families in 2020 to participate in this prospective cohort study. In 2020, interviews were conducted with 38 caregivers and 20 children. In 2021, retention of enrolled families was 84% (n=491), and an additional 115 families were recruited from our participant pool, increasing our sample to n=697. Data collection for the final planned timepoint is underway.

DISCUSSION

The COVID-19 Family Study, based on the FSM, examines family routines as a mechanism for understanding associations between COVID-19 control policies and children’s health behaviors and weight outcomes. Changes in family routines during pandemic may contribute to children’s unhealthy behaviors, including sedentary behaviors, increased screen time, and unhealthy snacking (69, 70). As a result, obesogenic behaviors may compound during periods of minimal structure, such as during school closures implemented to prevent the spread of COVID-19, leading to weight gain. Changes in family routines, for example, an increased child perceived parental policies to support physical activity and healthy eating (56), along with structure in the context of psychosocial well-being can promote healthy behaviors among children (71, 72). Understanding how families have adapted to changes associated with pandemic control policies and have implemented health-promoting family routines, and the impact on children’s health and wellbeing, will inform future health promotion strategies.

This innovative study has several important strengths. First, the longitudinal design creates a unique opportunity to track the relation between COVID-19 control policies and children’s health behaviors, weight gain, and obesity from pre-, during, to 2 years following the COVID-19 pandemic. Second, the diversity of the samples in CHAMP and WCC (race, ethnicity, SES, and locale) across a wide age group allows the study to examine widening disparities associated with COVID-19 policies on health behaviors, weight gain, and obesity. Third, the use of objective measures of anthropometry and accelerometry in a longitudinal design enhances the rigor and reproducibility of the study findings, compared to studies that rely on self-report or recall. Fourth, this study applies a mixed methods approach which allows for the identification of underlying pathways via family routines and emotional health, in-depth explanations, and clarification of pathways and relationships. Finally, the inclusion of an economic model to project the impact of COVID-19 policies on healthcare costs provides information regarding investments in obesity treatment and prevention in the coming decades.

The study has several limitations. Prior to the pandemic, children participated in randomized controlled trials to evaluate childcare and school-promoting environments; though, all analyses in this project will account for intervention exposures in previous trials. This study heavily relies on participants’ self-reported information on child health and health behaviors, family routines, caregiver psychosocial and household resource factors, which may cause biased responses. Allowing caregivers to self-report children’s height and weight measured at doctors’ office during early and mid-pandemic may introduce recall bias and measurement error. Although this study aims to recruit a sample that is diverse in its geographic locations (urbanicity), conducting the study solely within a single state may limit the generalizability of the findings to broader populations. Participation in this study relies on families completing online surveys, which may inadvertently exclude families with lower-incomes without internet access, thereby potentially narrowing the representativeness of the current sample. An English-only online survey excludes Spanish-speaking participants from the original trials, compromising the representativeness of our sample. Future prospective cohort studies should also consider accounting for the emerging growth in young children and adolescents by employing timely and frequent measurements to monitor these changes, which play a pivotal role in determining child wellness and health behavior outcomes over time.

CONCLUSION

COVID-19 control policies have created disruptions to daily life for families with school-age children, threatening family routines and children’s health. Understanding how families’ adaptations to the disruptions and their daily routines relate to children’s health and health-related behaviors provides opportunities to inform future wellness guidelines and policies on timely and targeted actions to promote children’s health.

LIST OF ABBREVIATIONS

BMI

Body Mass Index

CES-D

Center for Epidemiologic Studies Depression Scale

CFPQ

Comprehensive Feeding Practices Questionnaire

CHAMP

Creating Healthy Habits among Maryland Preschoolers

CHAOS

Confusion, Hubbub and Order Scale

CHES

Comprehensive Home Environment Survey

FFQ

Food Frequency Questionnaire

FSM

Family Stress Model

IRB

Institutional Review Board

ISEL

Interpersonal Support Evaluation List

PDSS

Pediatric Daytime Sleepiness Scale

PODS

Patterns of Diet at School

PSS

Perceived Stress Scale

SDQ

Strengths and Difficulties Questionnaire

STAI

State-Trait Anxiety Inventory

WCC

Wellness Champions for Change