Our Choice/Nuestra Opción: The Imperial County, California, Childhood Obesity Research Demonstration Study (CA-CORD)

Introduction

Childhood obesity remains high in the United States at 18% among 6- to 11-year-old children, despite recent declines among 2- to 5-year-olds (currently 8%).¹ Among 6- to 11-year-olds, rates are highest among Hispanic children, with 27% of boys and 23% of girls in the obese range.¹ More concerning are rates of overweight and obesity among Hispanic children 2–5 years of age (boys, 49%; girls, 43%).¹ Children living in rural communities and on the US-Mexico border are at even greater risk. For example, Imperial County, California, has higher overweight and obese rates among fifth, seventh, and ninth graders than overall in California (47% vs. 38%).² These rates are concerning given the consequences of childhood obesity, including adult obesity,³ associated comorbidities (e.g., diabetes⁴), compromises in quality of life,^5,6 and early mortality.⁷ The “Our Choice…is to be healthy/Nuestra Opción…es ser saludables” (OCNO) project is designed to prevent and control obesity in a vulnerable young population living in a rural, border community.

Preventing and Controlling Childhood Obesity

Successful approaches need to target multiple sectors and levels of influence simultaneously,^8–11 as indicated by the whole-child approach,¹² “Total Worker Health,”¹³ the Social Ecological Framework (SEF),^14,15 and several reviews and commentaries.^16–21 According to SEF, including its application to US Latinos,^22,23 levels of influence range from most to least proximal to the individual (see Fig. 1). Factors within these levels exert influence over an individual's health behaviors and outcomes; these become the targets of our interventions.²⁴ The influence exerted at various levels is not always most proximal to the individual (e.g., restaurant policies on cigarette smoking). Within the many social and structural environments in which children's behaviors are shaped and enacted, specific sectors can be identified.²⁵ Of interest to OCNO are the sectors of home, healthcare, early care and education (ECE), school, community recreation, and restaurants.

Present Study

This article describes the “Our Choice” project, herein referred to as the Imperial County, California, Childhood Obesity Research Demonstration study (CA-CORD). It is one of three CORD studies funded by the CDC in 2011 to test multisector, multilevel approaches to prevent and control childhood obesity.²⁶ CA-CORD is translating evidence-based approaches for modifying behaviors, policies, systems, and environments to promote fruit, vegetable, and water consumption, physical activity (PA), and quality sleep. The ultimate goal is to assist Children's Health Insurance Program–eligible children between 2 and 12 years of age to attain a healthy weight. In addition to sector-specific intervention and evaluation activities, CA-CORD is working with the University of Houston Evaluation Center to conduct a cross-site evaluation on a set of shared measures.

Methods

Study Design

CA-CORD is a 2×2 factorial study (see Table 1) designed to assess BMI z-score change in 1200 children, ages 2–11 years, assigned to a Health Care (HC) plus Public Health (PH) intervention, a HC intervention only, a PH intervention only, or a control condition. The HC intervention involves the implementation of an obesity care model within a federally qualified health center (FQHC) and includes a family wellness program (FWP) delivered by community health workers (CHWs). The PH intervention involves working with ECE centers, schools, community recreation organizations, and restaurants to promote four health behaviors: fruit, vegetable, and water consumption; PA; and quality sleep.

Setting

Imperial County, California, is located on the US-Mexico border, and in 2010 had an estimated 174,528 residents, 77% of whom were of Mexican origin, including 32% who were foreign born.²⁷ Three quarters of all residents reported speaking a language other than English at home. The median household income was $39,402, compared to $61,632 in the state; and income disparities are reflected further in the differential poverty rates (county=23% vs. state=14%). Despite these data, the county has numerous strengths upon which CA-CORD was built, including pre-existing strong partnerships.

Partnerships

CA-CORD is a partnership between San Diego State University and the Institute for Behavioral and Community Health (IBACH), Clínicas de Salud Del Pueblo, Inc. (CDSDP), and the Imperial County Public Health Department (ICPHD). IBACH is an academically affiliated research institute.²⁸ Since 1987, IBACH's research has focused on the top contributors to early mortality, including diet, PA, and tobacco use, with a focus on the Latino community. For CA-CORD, IBACH leads on cohort recruitment and retention and on the school and restaurant interventions. CDSDP has had a 7-year partnership with IBACH and is a private, nonprofit, FQHC with the mission of providing direct access to comprehensive, quality, primary, and preventive healthcare for underserved residents. In 2013, CDSDP served 52,511 patients, including 37% who were under 18 years of age.^29,30 For CA-CORD, CDSDP leads the HC intervention. ICPHD monitors local health problems, needs and resources, policy development, and leadership that foster local involvement and equitable distribution of resources. ICPHD was instrumental in forming the Childhood Obesity Prevention Alliance (COPA), a key strategy of the 2008 Imperial County Children and Families First Commission Prop 10 funded program. For CA-CORD, ICPHD leads the Community Advisory Committee (CAC) and works with IBACH on the ECE and recreation interventions. The partners developed the media intervention collaboratively.

Development of the CA-CORD Intervention Approach

The CA-CORD intervention is based on the study team's previous research,^29,31–36 several systematic reviews,^{8–10,21,37–41} CA-CORD CAC input, and formative research. CA-CORD CAC members were identified during the grant-writing phase by responding to a request for participation in a community-wide approach to prevent and control childhood obesity. Once the grant was funded, these individuals and COPA members were invited to an informational meeting. Their participation was formalized by completing a commitment form and a baseline survey. Over the first year of funding, the committee met 10 times to plan CA-CORD, and they established three working groups (healthcare, ECE, and school), ensuring that members represented a variety of agencies that serve Imperial County residents.

Focus groups of parents, healthcare providers, CHWs (promotores), ECE providers, and Special Supplemental Nutrition Program for Women, Infants, and Children providers revealed the importance of a multilevel, multisector approach to preventing and controlling overweight/obesity. It was also not uncommon for individuals from the various sectors to place responsibility on other sector members. In-depth interviews with school, restaurant, and community recreation representatives stressed the importance of feasibility and sustainability in selecting intervention activities. See Table 2 for key findings.

The CA-CORD Intervention

The CA-CORD intervention is multisector and multilevel. As depicted in Figure 1 and reflected in Table 3, it involves working with families, an FQHC, ECE centers, schools, community recreation organizations, and restaurants to create socially and physically supportive environments for the four targeted health behaviors of fruit, vegetable, and water consumption, PA, and quality sleep (e.g., sufficient duration of sleep). Three hallmarks of the CA-CORD intervention are that it: (1) is contextually and behaviorally focused; (2) targets multiple levels and systems simultaneously; and (3) is informed by theory.

CA-CORD seeks to modify behaviors within the contexts in which they occur. This contextualized approach to behavior change is further supported by changes to the social and structural environments, a strategy for sustained behavior change supported by several theoretical models.^14,15 The focus on increasing the frequency with which the four healthy behaviors occur is supported by previous research.^{29,31,33,34,42} Nevertheless, demonstrated through the concept of behavioral substitution,⁴³ the intervention also targets decreases in the consumption of high-fat, high-sugar snacks and beverages, as well as sedentary behaviors. For example, the family intervention promotes substituting television (TV) watching by engaging family participation in active play. An additional expected benefit of decreasing TV watching is improved sleep quality.⁴⁴ Based on previous research, behavioral substitution is an effective strategy for improving the diets of Mexican-origin children through a family-based intervention.³³ It was also successful in a tailored nutrition communication intervention with Mexican-origin adult women.^45,46 Finally, evidence-based and theory-informed change strategies are prioritized over those with insufficient evidence; where necessary, however, new strategies and associated materials were developed in response to our formative research findings. Described next are the HC and PH interventions.

Healthcare Intervention

The HC intervention is based on the obesity care model^41,47 and several systematic reviews.^10,48 Within the three largest of the 12 CDSDP clinics, new policies were adopted to promote the assessment and treatment of childhood obesity, including authorization to modify the electronic health records (EHRs) to accommodate practice changes. Using these policies and accompanying trainings developed by the study team, pediatric providers and staff were oriented to the new system, new EHR alerts, and a treatment plan for children in the overweight or obese range. To facilitate the adoption of system changes, a full-time patient care coordinator was hired to work across the three clinics to facilitate model implementation.

In addition to changes in healthcare delivery, an adapted FWP^{29,32–34,36} was integrated into CDSDP's Programs Department and is being administered by a full-time CHW coordinator and implemented by six full-time CHWs. A key feature of the FWP is its delivery by CHWs within a healthcare system and the documentation of CHW support in the child's EHRs. CHWs have been shown to be able to deliver an intervention with fidelity and achieve behavior change,^29,45,49 but their involvement as a team member to prevent and control childhood obesity is an innovation of CA-CORD. A second key feature of the FWP is that it simultaneously targets improvements in several health behaviors and parenting practices associated with these health behaviors, an evidence-based strategy.^34,36,37 This provides the opportunity for parents and their children to select a focus (e.g., reduce sugary beverage consumption) or behavioral strategy (e.g., do not have sugary drinks in the home) from several presented consistent with the Our Choice theme.

CHWs received 114 hours of training, were evaluated by direct observation preceding engaging study families, and were supervised by the CHW coordinator. The FWP is based on social cognitive theory (SCT),⁵⁰ family systems theory,⁵¹ and health behavior change research.^52,53 The program includes separate wellness workshops for parents and children, which are followed by PA workshops for families. Groups of families participate in these workshops together over the course of 12 weeks. The 1.5-hour wellness workshops provide opportunities for understanding and practicing the behavioral goals in age-appropriate ways. Parents were provided opportunities to discuss potential solutions to challenges encountered in adopting new behaviors when attempting to achieve goals and were empowered to try new behaviors and generalize behavior change to other settings. The 45-minute PA workshops provide family members opportunities to engage in fun PAs together and learn how to encourage others to be active. From the CHWs, families also receive 12 monthly newsletters, quarterly motivational interviewing phone calls, and an invitation to an annual PA community event.

Public Health Intervention

Early care and education centers and schools

The ECE and school interventions use similar change strategies, with the ultimate goal of providing opportunities for children to engage in healthy eating, PA, and sleep behaviors through policy, system, and environmental changes. The development of the interventions in both locations was informed by SCT⁵⁰ and organizational change theory⁵⁴ and recommendations from numerous professional organizations to affect changes in these sectors.^25,55 As an example, ECE providers and teachers were trained to use SPARK©⁵⁶ to promote PA throughout the day and as part of physical education (PE). ECE centers are encouraged to provide opportunities for active play throughout the day. Schools are mandated by the State of California to provide a minimum of 200 hours every 10 days for elementary schools.⁵⁷ This recognizes theory-informed recommendations to intervene in the environment to support healthier behaviors,⁵⁸ the importance of building organizational capacity to address identified needs,⁵⁹ and to respond to policy demands.^11,58,60 Despite their similarities, implementation of the ECE and school interventions are somewhat different (detailed further in Table 3).

Community: Recreation Organizations and Restaurants

The recreation intervention involves trainings and technical assistance to build capacity to promote the four behaviors, work toward policy changes in needed areas, and make environmental changes. The recreation intervention is notable for the development of gardens in two locations.

The restaurant intervention involves testing several strategies to promote healthy menu options for children in local independent restaurants. A systematic enumeration of all restaurants using three sources identified 132 restaurants in the PH intervention communities. Fifty percent (n=66) were randomly sampled and audits conducted to describe the restaurant environment. The pilot restaurant intervention in three restaurants, informed by the work of others on how to market to children and parents,^61–63 has involved the implementation of healthy children menu options and their promotion through new menus, table tents, and wait-staff trainings.

Plans for sustaining the interventions in all sectors are ongoing,⁶⁴ with strategies including recognizing achievements of sector representatives and stakeholders, identifying organizational champions to facilitate the ongoing change process (e.g., the superintendent, the assistant superintendent, an academic coach, an elementary school principal, and a teacher), making print resources available on the Web, and providing ongoing technical assistance through participation in COPA.

Evaluation Overview

The CA-CORD evaluation protocol includes process, impact, and outcome evaluation components that involve children, parents, and organizations. It complements and supports the University of Houston Evaluation Center activities through a set of shared consensus measures. To assess the primary aim of determining whether the HC plus PH intervention is more effective than the other three conditions at helping children to achieve a healthy weight, 1200 children and parents are being recruited to participate in data collection at baseline and again 12 and 18 months later. As depicted in Figure 2, in this quasi-experimental study, condition assignment is based on whether the child is a patient of CDSDP and if primary-school aged, a student in a participating school (or for non-school-age children, the family's residence is in an intervention city). For example, 3-year-old children who are patients of CDSDP and residents of El Centro are placed in the HC plus PH intervention condition, whereas resident children who are not CDSDP patients are assigned to the PH intervention condition.

Initial inclusion criteria for families was 1 child per household between 2 and 10 years of age, the child's BMI at or above the 75th percentile, and the child not currently taking medications that affect weight. The family had to have plans to remain in the area for the duration of the study, and the participating parent had to understand English or Spanish. Owing to recruitment challenges, inclusion criteria were expanded to include up to 2 children per household, children up to 11 years of age, those with a BMI at or above the 5th percentile, and those on stimulant medications for attention deficit hyperactivity disorder.

The outcome and impact evaluation protocols include measured and observed data collection from children and parents, interviews with parents reporting on themselves, their children, and their family, and interviews and observations with agency representatives from all sectors. To collect additional PA and sleep data, a subsample of cohort children and parents are also asked to wear a triaxial accelerometer (ActiGraph GT3X+; ActiGraph, Pensacola, FL) for 7 days. Cost data are also being collected from the clinic and the school perspectives, including the extent to which the various organizations incurred costs associated with implementation.^65,66 A cost-effectiveness analysis will combine these costs (excluding research costs) and the health outcomes achieved.

Process evaluation protocols are comprehensive, including measures of context, reach, dose delivered, dose received, and intervention fidelity.^67,68 Various methods are being used, including abstracting information from the EHR, CHWs using iPADs to log family contact, and interviews with parents and agency representatives.

Power Calculations

Several initial assumptions were made for the power calculations. The mean standardized BMI z-score in the control is assumed to be 0.8. The standard deviation of the mean standardized BMI z-score is assumed to be 0.2. The largest reduction (0.1) of the standardized BMI z-score is assumed to occur in the HC plus PH intervention condition versus the evaluation-only condition. The reduction from the HC intervention is assumed to be at 0.05 and that from the PH intervention at 0.025. The alpha level is set at 0.05. With N=1007, the study has 80% power to detect significant changes for different comparisons, except the PH intervention condition versus control condition. Inflated by 20% to account for attrition, the required sample size is 300 per condition or 1200 total, including a 20% attrition that is expected over the 18-month evaluation period. A second power calculation was performed when multiple children from the same household were included owing to recruitment challenges. The intraclass correlation on BMI between 2 children from the same household was found to be 0.246. Simulating the inflated sample size needed in order to maintain the previously estimated statistical power if 30% of the children were from the same household for example, then 1400 are needed to maintain the same statistical power of the previous study design.

Analysis of Primary Aim

Univariate statistics will be conducted on primary and secondary endpoints as well as baseline covariates to describe the sample characteristics. Cross-group comparisons will be conducted on all of the baseline covariates to identify potential sample differences and adjusted in the subsequent multivariate models. Propensity scores for belonging to each site on baseline covariates will be estimated using polychomotous logistic regression and used as an additional covariate in the subsequent multilevel modeling. Correlation matrices of repeated measures of different endpoints will be calculated to suggest the correct correlation structures in the subsequent multilevel modeling.

Mixed-effects models will be used for analyzing the primary endpoint—BMI z-score. The main independent variable will be the CONDITION and a TIME variable to represent the repeated measure time points. The main effects of CONDITION, TIME, and CONDITION×TIME will be tested. Other unbalanced covariates and the estimated propensity score will be added to the mixed-effects model to control for their confounding effects. Whereas the primary endpoint is BMI z-score, path analytic partial least squares will explore the potential underlying mediation and moderation effects using demographics, cultural, behavioral, and environmental data. Finally, when evaluating multicomponent interventions, it is usually of interest to explore which component produced the largest effect on the outcome. The variance importance of projection in the partial least squares method will help determine which components contributed more influence on BMI z-score.

Discussion

CA-CORD is one of three field sites funded by the CDC to identify multisector, multilevel approaches to prevent and control childhood obesity. Sites in Massachusetts and Texas are implementing similar strategies, providing the Evaluation Center the opportunity to examine strategies across diverse communities and its members. There remains an urgent need to identify cost-effective strategies that target multiple sectors and levels of influence simultaneously. The CA-CORD study is working with families, an FQHC, ECE centers, schools, community recreation organizations, and restaurants to support healthier lifestyle behaviors.

Limitations

A number of limitations should be noted. This article describes the CA-CORD intervention as intended to be delivered. Differences, however, are emerging between what was planned and what is being delivered. For example, in the HC intervention, a registered dietitian was going to be hired per the obesity care model;^41,47 but no qualified personnel applied. Intervention fidelity has been challenging given staff changes, time constraints, and competing demands. A second limitation is that only three cities are involved, and city is nested within study design; although the cities are comparable on some dimensions, such as being rural, they differ on others, such as proximity to the border. Analyses will attempt to control for sources of variance associated with city. Despite this concern, in our previous studies in these three communities, we found no differences in similar health outcomes and behaviors among children and their parents participating in a healthy eating randomized, controlled trial.³³ A third limitation is the inclusion of children below the 75th percentile. This study design element was modified during cohort recruitment to reach the goal of 1200 children. Such a wide variance in BMI percentile will make it more challenging to find a statistically significant effect.

Conclusions

These studies and studies, funded by the NIH, the Robert Wood Johnson Foundation's Active Living Research, Healthy Eating Research, and SaludAmerica! programs, are among the leaders in identifying the best approaches for maintaining the health of children. This research has the potential to inform best practices across sectors.

Acknowledgments

This research was supported by cooperative agreement U18DP003377-01 from the CDC. The content is solely the responsibility of the authors and does not necessarily represent the official views of the CDC. The authors thank the wonderful staff and administration of San Diego State University Research Foundation, Clinicas de Salud del Pueblo, Inc., and the ICPHD, who are making these efforts happen. The authors also thank the members of the Our Choice Community Advisory Committee presented in alphabetical order by organization: Brawley Elementary School District; Campesinos Unidos, Inc. Head Start; City of El Centro Community Services Department; City of Brawley Parks and Recreation Department; Clinicas de Salud del Pueblo, Inc.–Women, Infants, and Childrens' Program; Cooperative Extension, Imperial County; El Centro Elementary School District; Imperial County Office of Education; Imperial County Office of Education–Early Care and Education Program; Imperial Valley Boys and Girls Club; and Imperial Valley Food Bank.

Author Disclosure Statement

No competing financial interests exist.