Overview
Our research plan is to conduct a series of seven studies over a 5-year period to better understand and manage SO in children. The studies are diverse in setting (e.g., community, primary- and tertiary-level care) and methods (e.g., observational studies, clinical trials, and qualitative). Studies one through five, which address issues such as health risks and drivers of unhealthy weight gain, aim to better understand issues unique to children with SO, while studies six and seven aim to examine different intervention models for managing SO.
Study 1: Prevalence and health care utilization of SO in children accessing primary health care in Ontario
Rationale and objectives
No national or provincial estimates of excess weight in Canadian children are available. This knowledge gap led to provincial recommendations for a system to monitor prevalence of childhood obesity through existing mechanisms, including primary care electronic medical records [18]. Children in Ontario attend ~19 primary health care visits in the first two years of life [41]. As it is standard care to measure height and weight at these visits [42], an excellent source of weight-related data should be available but to date has not been accessed. It has been suggested that health care utilization is higher among children with obesity versus their leaner peers [43,44,45,46,47]. Knowing the prevalence of SO and health care utilization of children with SO is essential to develop interventions, evaluate the impact of these interventions, and monitor trends over time. Thus, Team ABC3 objectives for this study are to determine (i) the prevalence of SO in three cohorts of 0 to 18 year olds in Ontario and (ii) whether SO is associated with increased all-cause health care utilization in 0–18 year olds.
Research plan
Three Ontario-based data sources will be accessed: (1) The Applied Research Group for Kids (TARGet Kids!) [48], a practice-based research network in primary care that includes >7500 children, recruited between 0 to 5 years of age, with measured height and weight data, and obesity-related health behaviors and outcomes, (2) the Electronic Medical Record Administrative Data Linked Database (EMERALD) [49] which includes data collected during primary care visits of >30,000 children 0 to 18 years of age and (3) the Better Outcomes Registry and Network (BORN Ontario) [50] which collects, interprets, shares and protects health-related data about pregnancy, birth and childhood from primary care practices in Ontario. For objective 1, we will complete cross-sectional analyses to determine the prevalence of SO in the three data sources. Height/length, weight, age, sex, family income and postal code data will be extracted. Although BMI ≥99th percentile is our working definition of SO in children [17], we will also calculate prevalence by additional criteria (e.g., ≥120% of the 95th percentile) to create a comparative approach as done previously by our team members to determine obesity prevalence [51, 52]. Additionally, longitudinal analyses will be performed over different time periods based on available data from these datasets. Longitudinal data is available for children in TARGet Kids! from 2008 to present, and from EMERALD from 1997 to 2016. Two analytic approaches will be conducted: (1) following a subset of the same children over time and (2) using serial-cross-sectional prevalence estimates. For objective 2, we will undertake analyses to assess health care utilization of children with SO. We will link our three databases with Ontario’s administrative health services data housed at the Institute of Clinical Evaluative Sciences (ICES), through children’s Ontario Health Insurance Plan (OHIP) numbers. All-cause health care utilization will be defined as the number of hospitalizations, emergency department (ED) visits, and physician visits, including primary care and specialist visits. Using ICES databases, we will access the Canadian Institute of Health Information—Discharge Abstract Database, the National Ambulatory Care Reporting System (ED visits) and the OHIP billing claims database (physician visits). Descriptive statistics will be calculated for all variables to determine distributions. A multivariable Poisson regression will be performed to determine the association between SO and health care utilization adjusted for age, sex, family income, and geographic region (postal code). Sensitivity analyses will be used to determine how changes in the definition of SO affect the association with health care utilization. The primary analyses will be using the Canadian definition, prosed by the Dietitians of Canada, suggesting zBMI > 3 or ≥99.9th BMI percentile. We will also use the proposed definition of ≥120% of 95th percentile [53].
Outcomes
This study will identify the prevalence of SO in children and the impact of SO on health care utilization in Ontario and demonstrate the feasibility of using primary care data to monitor SO prevalence and health care utilization. These findings have the potential to be extended to other provinces through our team of researchers and decision-makers.
Study 2: Predictors of treatment initiation for children with SO referred for tertiary-level management of obesity
Rationale and objective
To benefit from lifestyle and behavioral interventions for managing SO, families must initiate treatment. Data are limited, but one recent study found that only 10–15% of children referred for additional care actually initiated obesity management services [54]. To optimize the impact of health services for managing SO in children, clinicians and health care administrators must understand the factors that influence treatment initiation in families. With that understanding, clinicians can make appropriate referrals and incorporate strategies to increase the likelihood that families engage in treatment. Therefore, we will examine demographic (children’s sex and age), anthropometric (children’s BMI z-score), procedural (type of referral provider, length of the enrollment process, and treatment clinic), and contextual (distance between families’ home and treatment venues and seasonality) variables possibly associated with initiation of multidisciplinary management for SO in children in the province of Alberta.
Research plan
This retrospective study will examine data from all children (n ≈ 2500) referred by Alberta-based physicians and nurse practitioners to tertiary-level, multidisciplinary obesity management clinics (2005–2013). Since 2005, Alberta Health Services (AHS) has dedicated administrative and informatics infrastructure to receive and process referrals for pediatric obesity management, including a standard referral form and data management resources to convert referral data to electronic format. Team members helped to design, implement and refine the referral form and associated procedures. We will work with our AHS decision-maker partners to obtain institutional approval and access to demographic, geographic, anthropometric and clinical data, as well as history of obesity management and potential barriers to treatment, from children’s referral documents. Along with defining SO in children as BMI ≥99th percentile we will also calculate prevalence by additional criteria (e.g., ≥120% of the 95th percentile) to understand a more contemporary view of SO in children. Research by team members [55, 56] indicates that 50–90% of children referred for obesity management will have SO. We will perform descriptive statistics (e.g., means, 95% CIs, proportions) and multivariable logistic regression analyses, with treatment initiation as the dependent variable, to determine the proportion of children who initiate treatment and characteristics of families participating/not participating in care. Independent variables, both continuous and categorical, will be considered for inclusion in our models based on preliminary analyses, an approach used previously by our team members [57, 58]. Similar to previous studies [59], data will be managed using LabKey®, an open-source, password-protected data repository available to our team through the University of Alberta Women and Children’s Health Research Institute. We will not examine reasons, barriers, and facilitators of initiation, as our research team has previously investigated these factors [60, 61].
Outcomes
This project will determine whether treatment initiation is related to SO and whether demographic, anthropocentric, procedural, and contextual factors are associated with family initiation of treatment, informing clinical or administrative strategies to help children and families overcome barriers to care.
Study 3: Pathways to eating: determining eating behavior phenotypes in children with SO
Rationale and objective
Pediatric obesity is a complex condition with heterogeneous phenotypes, yet recommended treatments fail to target these multifaceted etiological pathways. Overeating is a major contributor to obesity; recent evidence supports notable differences in eating behaviors between children with and without obesity [62,63,64,65,66]. Various eating behaviors, including loss of control eating, emotional eating, excessive hunger, impulsivity/delay of gratification, and responsivity to external cues mark important and distinct triggers. They often co-occur [67] but are infrequently investigated, particularly in children with SO [67, 68]. We will determine eating behavior phenotypes in children with SO by examining the clustering of eating triggers, relating identified phenotypes to demographic, physical, and environmental characteristics of children with SO, and investigate whether treatment outcomes vary according to eating phenotypes and changes in eating pathways.
Research plan
The Canadian Pediatric Weight Management Registry (CANPWR) [55] is the largest obesity research study in Canada, encompassing ten multidisciplinary clinical centers. Through the CANPWR data collection process, we will recruit a sample of youth 10–18 years old (n = 500) for a longitudinal (2-year) sub-study. They will complete validated surveys (duration: ~15 min) during their annual CANPWR visits. Surveys will supplement existing demographic, anthropometric and lifestyle measures already being collected [55] and provide detailed information on five proposed key pathways: loss of control eating, emotional eating, hunger, impulsivity, and eating in response to external cues (Fig. 1). Data will be collected at three time points (0-, 12-, and 24-months follow-up). Eating phenotypes will be assigned membership in a latent cluster from patterns of interrelationships among indicator variables, using Latent Profile Analysis (LPA). LPA uses categorical and continuous indicators from cross-sectional data to identify latent subgroups of individuals (e.g., two to five mutually exclusive groups); a statistical approach our team members have applied previously [69, 70]. Second, we will compare demographic (e.g., age, sex) anthropometric (e.g., BMI), and environmental (e.g. family structure) variables across the identified LPA phenotypes using one-way analysis of variance and post-hoc comparisons. Third, we will examine treatment outcomes (e.g., longitudinal changes in demographic, physical, and environmental variables at 12- and 24-months follow-up) in relation to the LPA phenotypes and changes in individual eating components (e.g., improvements in impulse control), using mixed effects models.
Outcomes
This research will reveal the degree to which triggers of eating are present in children with SO enrolled in CANPWR and offer insights into which specific triggers influence overeating, informing tailored interventions to improve eating behaviors.
Study 4: Family recommendations for improving health services to manage SO in children with disabilities
Rationale and objectives
Low treatment initiation, high program attrition, and poor adherence to lifestyle and behavioral recommendations limit the successful management of pediatric obesity [54, 71,72,73]. Team members have explored family preferences for care which suggested that families desire better help from health care professionals, family-centered treatment, a desire for increased social support, and need for policy/program-level changes to assist their weight management efforts [74]. Extending this research, team members are currently completing a qualitative study exploring families’ reasons and decisions for initiating, terminating or continuing health care to manage pediatric obesity [59]. To date, we have interviewed four families who have a child with SO and a disability and identified several unique issues associated with SO in children with disabilities. Recent reviews note that children with disabilities are at a heightened risk of developing obesity [75, 76], but little information exists on managing obesity in children with any disabilities. Thus, we aim to explore families’ experiences in managing SO, and identify families’ recommendations for improving health services to manage SO in children with disabilities and their parents.
Research plan
Parents (primary caregivers or guardians) and children with disabilities will be recruited through CANPWR sites in six of the largest cities across Canada (Vancouver, Edmonton, Hamilton, Toronto, Ottawa and Montreal) and through the pediatric outpatient clinic at a children’s rehabilitation centre in Toronto. We will enroll 35 parent–child dyads (n = 5 dyads per site) for a total of 70 interviews. English-speaking families will be eligible if children are 10–17 years old, have SO (BMI ≥99th percentile), and experience participation restrictions or activity limitations associated with a neurological, musculoskeletal or developmental disorder [77]. If a child’s cognitive challenges limit the quality of interview data, we will rely on the parent interview as the primary data source. Data will be gathered with semi-structured interviews. Questions asked of children and parents will be similar, however questions will be modified in the case of children due to age and cognitive ability. Families will be asked about their experiences with and recommendations for improving health services to manage SO in children with disabilities. Interview data will be digitally recorded, transcribed verbatim, and subsequently managed using NVivo 10 (QSR International). Demographic and clinical data will be collected for descriptive purposes. Guided by an ecological perspective [74], we will use thematic data analysis [78] to identify family experiences and recommendations at the family, social and health care services levels.
Outcomes
This qualitative study will reveal families’ experiences and recommendations for improving health services to manage SO in children with disabilities, inform modifications to health services delivery for managing obesity in children with disabilities, and identify intervention approaches to best meet the needs of this population.
Study 5: Examining obesity-related health outcomes using the 4Ms framework (metabolic, mechanical, mental, milieu) in children with SO
Rationale and objectives
Most studies linking SO in children to adverse health outcomes risk such as type 2 diabetes and cardiovascular disease have focused on cardiometabolic health (e.g., insulin resistance, hypertension, dysglycemia, dyslipidemia) [8, 20, 21]. However, biomechanical, psychological and social health measures have received much less research attention, yet may be most salient for families and clinicians [79]. Further, the lack of universally-accepted criteria for SO in children highlights the need for empirical evidence to inform a definition. We aim to identify the presence of adverse health outcomes, namely metabolic, mechanical, mental health, and social milieu (the 4Ms), and compare and contrast 4Ms in children across the range of obesity and across definitions of SO.
Research plan
Using cross-sectional baseline data collected from children and families enrolled in CANPWR [55], we will examine the burden of illness based on a diverse set of conditions organized under the 4M framework. Health measures under the 4Ms framework are (1) metabolic (fasting levels of glucose, insulin, total cholesterol, HDL-cholesterol, LDL-cholesterol, triglycerides, systolic and diastolic blood pressure, liver enzymes, Acanthosis Nigricans); (2) mechanical (sleep quality and apnea, musculoskeletal problems, gastroesophageal reflux disorder, physical functioning); (3) mental health (depression, anxiety, attention deficit hyperactivity disorder, learning disability, emotional functioning); and (4) social milieu (household income, parent education and health status, school and social functioning, inter-personal interactions [e.g., bullying]). Our secondary data analysis will utilize data already collected within CANPWR using validated questionnaires that are standardized across sites. We will compare and contrast variables across the 4M categories in a sample of CANPWR participants (n = 1600) along a spectrum of increased weight status: overweight (BMI ≥85th percentile), obese class I (BMI ≥95th percentile), obese class II (BMI ≥120% of the 95th percentile), and obese class III (BMI ≥140% of the 95th percentile). This classification system was proposed recently [19, 80], but lacked empirical data. Consideration will also be given to BMI ≥99th percentile. Descriptive statistics will be calculated for participant characteristics and health outcomes (e.g., means, 95% CIs, proportions), for continuous and categorical data. Group differences will be examined by analysis of (co)variance with post-hoc comparisons and Chi squared tests for continuous and categorical data, respectively.
Outcomes
This study will examine a diverse set of health outcomes associated with obesity across a spectrum of excess weight for children enrolled in CANPWR, contribute data to determine criteria for SO based on health risks, and quantify the burden of illness associated with SO in children (as in adults [81, 82]) to inform development of clinical tools and decision-making.
Study 6: Does integration of health coaches improve adherence to an e-health lifestyle and behavioral intervention? A randomized controlled trial (RCT) of LiGHT (Living Green and Healthy for Teens)
Rationale and objective
Geographic or physical barriers prevent access to in person health services for some children with SO, while others prefer self-guided treatment options. E-health strategies offer a cost-effective means to broadly disseminate lifestyle and behavioral interventions through the Internet. However, the extent to which participants adhere to web-based intervention components remains relatively low [83], limiting intervention effects [83,84,85,86,87,88]. LiGHT (Living Green and Healthy for Teens) is an e-health lifestyle and behavioral intervention for managing SO in children. Previously, families rated LiGHT favorably and expressed a desire to incorporate health coaches into LiGHT to enhance engagement, support and motivation [89]. We will examine whether adding health coaches to LiGHT increases intervention adherences and improves anthropometry, lifestyle habits, cardiometabolic risk factors, and family psychosocial health.
Research plan
We will conduct a seven-site comparative effectiveness RCT with two parallel groups and a 1:1 allocation ratio across Canada. Our protocol adheres to the Standard Protocol Items for Randomized Trials guidelines [90] and will be registered publically and prospectively at ClinicalTrials.gov.
Intervention groups
LiGHT is an e-health intervention designed for children and their families, delivered in 12 modules. It combines evidence-based obesity management techniques with environmental and economic information, and is made available to families through a secure, password-protected website via desktop, laptop or tablet. LiGHT is designed to increase intervention adherence and retention by emphasizing the impact on personal health outcomes of individual lifestyle and behavioral habits, the environment (e.g., food packaging materials) and family finances (e.g., commuting costs). Following feedback from the pilot study [91], we are incorporating gamification into LiGHT (v2.0) to enhance visual appeal and interactivity. For this trial, we will compare LiGHT (v2.0) to LiGHT+, which adds personal health coaches (PHCs) to improve intervention adherence instead of providing a virtual coach. PHCs will encourage LiGHT+ participants to complete behavioral techniques (e.g., self-monitor, set goals) that can enable changes in eating and physical activity habits [91]. They will support families during the trial through their preferred mode of contact (text message, email and/or telephone). PHCs will have health professional training (e.g., psychology, nutrition), and be employed locally at each site as research assistants. Team members with expertise in health coaching and communication will train PHCs at study onset and support them throughout the trial. Children are the primary intervention recipients, with parents playing secondary roles. Children and parents in both trial groups will complete a comprehensive assessment at 0-, and 4-months follow-up.
Participants
Children 10–17 years old with SO (BMI ≥99th percentile) [17] and at least one parent (primary caregiver) will be eligible to participate. Families will be recruited through six CANPWR sites across Canada. In our team members’ experience, ~50% of families attending an information session do not initiate care. As an alternative to in-person care, we will recruit families who decided not to initiate care after initial referral to one of the seven sites, a strategy that offers an alternative for families who declined in-person care and avoids co-intervention effects with families currently enrolled in a clinical program.
Outcome measures
Our primary outcome will be adherence. This outcome will be tracked continuously within LiGHT and measured as a latent variable by monitoring the extent to which participants access the intervention (number of weeks accessed), percentage of pages viewed, and adherence to behavioral change techniques (percentage of use of self-monitoring techniques). Secondary outcomes include children’s anthropometry, lifestyle habits, cardiometabolic health measures, family psychosocial health, and factors at individual, social and environmental levels that can influence adherence (e.g., intrinsic motivation, peer support). All outcomes will be measured at 0-, and 4-months follow-up using standardized measures.
Sample size
We will recruit 186 participants in two groups (n = 93/group). The trial can detect a 20% difference in adherence (odds ratio of 2.33) at 4-months follow-up at an alpha of 0.05 and 80% power.
Randomization and blinding
A biostatistician will complete the computer-generated random allocation sequence, in blocks of variable sizes. Allocation concealment will be achieved by a central randomization system. Individuals collecting families’ outcome data and completing data analysis will be blind to group allocation, which will be concealed until data analyses are completed.
Data management and analysis
We will capitalize on technical and research support from the University of Alberta Women and Children’s Health Research Institute to use REDCap, a secure online platform, for data management and storage. Baseline characteristics and outcomes will be calculated with appropriate descriptive statistics (e.g., means, 95% CIs, proportions). LiGHT has built-in capacity to gather metrics on intervention use and adherence. To examine our primary objective, we will compare the difference in adherence between LiGHT and LiGHT+ using a risk difference estimator based on logistic regression [92]. We will also use generalized estimating equations to model change in adherence over time to identify individual, social and environmental variables associated with adherence. Adherence will be modeled using an auto-regressive correlation structure to account for the expectation that adherence will decrease over time. To evaluate our secondary objective, we will compare groups using the same risk difference estimator procedure.
Outcomes
This trial will determine whether health coaching enhances adherence (and other health-related outcomes) to LiGHT and provide further evidence for this novel and accessible treatment option for managing SO in children who are referred for, but decline, in-person care.
Study 7: STOMP early years: a pilot randomized controlled trial of an intensive, family-centered, home visiting intervention for young children with SO
Rationale and objectives
To date, few reports have been published on obesity management interventions in young children (<5 years old) [30]. Evidence is emerging that community—[93] and home-based [94, 95] interventions can help young children to improve their health behaviours and weight status. Interventions available to families in their local communities and homes can reduce barriers to accessing health services [96]. The SickKids Team Obesity Management Program (STOMP) Early Years Program is a unique and intensive pediatric obesity management program designed for 1 to 5 year olds with SO and their families. This evidence-based program combines lifestyle, behavioral and parenting strategies and is offered in partnership with Toronto-based public health professionals with a mandate to provide both community- and home-based care. In our health services experience to date, the intervention is acceptable to families and health care providers, but the feasibility of scientific issues (e.g., sample size, feasibility of family consent and outcome measurement tools) remains unknown. Overall, we will determine the feasibility of using this intervention to manage SO in young children, obtain a reliable estimate of the variance in the primary outcome (BMI z-score) and use data from this study to calculate a sample size estimate for a definitive, future RCT.
Research plan
We will conduct a single-site (Toronto) internal pilot RCT with two parallel groups and a 1:1 allocation ratio. This internal pilot RCT [97, 98] will be co-led in partnership with Toronto Public Health. Our protocol adheres to the Standard Protocol Items for Randomized Trials guidelines [90] and will be registered publically and prospectively at ClinicalTrials.gov.
Intervention and control groups
Parents randomized to the intervention group will participate in a modified, version of the Chicago Parenting Program [99] (10 weekly sessions over 6 months), which focuses on lifestyle, behavioral and parenting issues and strategies for managing SO in young children. A mental health specialist–dietitian–nurse team will deliver the first four sessions, which focus on health behaviours. The next 6 sessions will be delivered by a public health nurse and focus on parenting skills to help implement behaviour change. This is complemented by four home-based visits by a public health nurse. Home visits help families to incorporate healthy nutrition and physical activity habits into their home environments through effective parenting practices learned in the program. Families randomized to the control group will be offered this program as a delayed treatment following the study period. Children and parents in both groups will complete a comprehensive multidisciplinary assessment at 0- and 6-months follow-up. To minimize the risk of bias, study data will be collected by a trained research assistant who will not participate in intervention delivery.
Participants
Young children (1–5 years old) with SO (BMI ≥99th percentile) and at least one parent (primary caregiver) who are referred for the STOMP program will be eligible to participate. Families will be excluded if they reside beyond the Toronto Public Health catchment area for home visiting.
Outcome measures
Our primary outcome will be BMI z-score. Secondary outcomes will include children’s dietary intake (e.g., NutriSTEP® [100]), physical activity, sedentary behavior (e.g., Nutrition and Health Questionnaire [NHQ]–questions based on the Canadian health Measures Survey [CHMS] [101]), cardiometabolic risk factors (e.g., blood pressure, lipids, insulin resistance), and family psychosocial health (e.g., Parental Stress Index [102]).
Sample size
Recommendations for an internal pilot study are to include half the anticipated sample size for a full-scale trial and at least 10 children/group [97]. We will include a total of 38 children (n = 19/group) for this pilot RCT. A power calculation is not appropriate as the study does not aim to provide a definitive estimate of treatment effect. The aim is to provide robust estimates of the likely rates of recruitment and retention, and to yield estimates of the variability of the primary and secondary outcomes to inform power calculations for a future large-scale trial [103, 104].
Randomization and blinding
A biostatistician will complete the computer-generated random allocation sequence, in blocks of variable sizes. Allocation concealment will be achieved by a central randomization system. Individuals collecting families’ outcome data and completing data analysis will be blind to group allocation, which will be concealed until data analyses are completed.
Data management and analysis
The Applied Health Research Centre (University of Toronto) will provide support for data management and analysis. Baseline characteristics and outcomes will be calculated with appropriate descriptive statistics (e.g., means, 95% CIs). Feasibility will be assessed as our ability to recruit, consent and collect data from families at 0- and 6-months follow-up. We will calculate variance around the primary outcome (BMI z-score) in the control group to inform sample size for a definitive future RCT.
Outcomes
This study will provide essential data and experience to plan a definitive, multi-site RCT for managing SO in young children [90] and enable our team to partner with public health professionals (both front-line staff and decision-makers) to manage SO in the community. This trial will offer invaluable experience to evaluate the STOMP Early Years program in other communities.
