Table 2.

Issues for ECE-Based Interventions and Policy Research

Intervention approaches

• Few studies have assessed weight outcomes of child care–based interventions, and less than half (2 of 5) demonstrated positive outcomes. This failure to measure an impact on child weight may be the result of a variety of issues related to intervention approach, study design, and/or measurement.

• Successful interventions to affect children's behavior may require more complex strategies, such as multilevel interventions guided by the Social Ecologic Model that target child (individual level), staff (inter-personal level), and policy and environment (organizational level).

• New theories or multiple theories need to be applied to address the unique structure of child care settings and the multiple targets of interest (e.g., child, staff, environment, policies, parent, community). This may require that researchers look beyond their immediate fields for potential frameworks and behavior change theories to employ.

• The most appropriate intervention length has not been determined. Existing interventions with a child-focused component have provided anywhere between 21 and 72 hours of intervention time.

• Child care workers are important in changing children's behavior, but it is unclear what level of training is needed to effectively change staff behavior. These individuals have their own health challenges. Most are low-wage earners without insurance who are at high risk for health disparities. It may be critical to address staff's own health issues before they take on new health promotion efforts.

• Parent engagement is another critical component because they can be important reinforcers and/or barriers for children's behavior. However, few studies (child care or school) provide effective models for reaching parents that take into consideration the multiple demands on parents with young children (e.g., work schedules, limited resources, cultural intrusions, child demands).

• Type and structure of child care setting should be considered when planning interventions. Family child care homes are smaller operations run out of the provider's own home. Centers and faith-based programs can vary largely in size, from just a few children to more than 100. Child care programs also vary in structure. Some programs offer year-round care while others run only 9 months; some offer full-time care and others offer only part-time care; some serve food and others do not.

• Although most child care programs are regulated at the state level, policies vary greatly from state to state. It may be impossible to create a universally accommodating intervention, but it is important for researchers to think through these issues so that they can make informed choices when designing their intervention and selecting inclusion/exclusion criteria.

Intervention study designs

• Children are naturally clustered within classrooms and centers. Depending on the intervention, children may need to be randomized by one of these groups.

• Traditional cohort designs are challenging because of the high turnover rates for children enrolled in child care. Unlike schools, enrollment in child care is optional (in the United States). It is very common for parents to move children in and out of programs often due to employment changes. A more appropriate strategy may be to intervene with the child care program and to assess repeated cross-sectional samples of children over time.

• When selecting an outcome, researchers should remember that children 2–5 years old are going through a period of rapid growth and development. Traditional indicators of weight for height (BMI, BMI percentage, BMI z-score) are crude measures for these children. Multiple measures of height and weight would allow modeling of weight gain trajectories, but this method has high participant burden and is more costly. Waist circumference and/or sum of skinfolds may be useful alternatives; however, both are technically challenging to collect, and no norms exist for comparisons.

• Other issues may be choice of outcome measure, faulty design and reporting, and lack of consideration for early life determinants (e.g., birth weight, rapid infant growth rate, sleep duration) (see below).

Considerations when working with minority populations

• Minority populations, particularly African Americans, American Indians, and Hispanics, suffer disproportionately high rates of obesity, thus making them important targets for public health intervention. Nearly half (47%) of the nation's children younger than 5 years old are from a minority group, making child care–based interventions an important avenue to reach these populations.

• Segregation levels for African-American and Hispanic children are higher than for their adult counterparts, despite a general reduction in segregation over the last 10 years.

• African-American, American Indian, and Hispanic children also have disproportionately high poverty rates (between 31% and 35%).

• Despite the great need, many of the ECE studies regarding policies and practices have not even reported race/ethnicity of their participants. Among those that have, it seems that policies and practices vary depending on the race/ethnicity of the provider. For example, fewer Hispanic providers report eating meals together with children (24% compared to 86% of white and Asian providers); and Hispanic providers were more likely to report making children eat foods they think are good for them (85% compared to 69% of Asians and 44% of whites).

• While most ECE-based obesity prevention intervention studies report race/ethnicity, many find different outcomes depending on sample characteristics (e.g., Hip-Hop to Health, Jr).

Policy research

• When assessing the impact of local, state, and federal policies, there is a wide spectrum of outcomes possible, including environment, knowledge/attitudes/beliefs, behaviors, health indicators, and disease.

• Structural and environmental variables include such aspects as examining the legislation enacted, funds appropriated, institutional changes (e.g., tax credits), and environmental changes.

• Policies can also impact knowledge, attitudes, and social norms or may change individuals' behaviors such as diet, physical activity, sedentariness, and breastfeeding; or behaviors and practices of the organization, such as food offerings.

• Policy makers and researchers also need to understand how the policy impacts health indicators like child BMI, or disease prevalence such as diabetes, stroke, and cancer.

• Randomized controlled trials are not generally practical in “real world” policy evaluation; therefore, other designs of varying degrees of strength must be employed. These include, at the lower end, single group and posttest-only designs, whereas higher-end designs include multiple time series data collection.

• Even in natural experiments, it is important to capture key demographic (or other) variables, to recruit a sample large enough to provide sufficient power, and to include adequate sampling of important subgroups.

• Policy research studies should consider what other possible groups, sites, or situations they care to generalize to.

• Process evaluation is also critical to insure internal validity. It is important to capture things like the extent to which the intervention was implemented, degree to which other events or experiences outside of the policy being evaluated may have affected behavior, whether enough time elapsed between implementation and the measurement of the intended effects, and any unintended effects.