A National Survey of Obesity Prevention Practices in Head Start

Objective  To describe obesity prevention practices and environments in Head Start, the largest federally funded early childhood education program in the United States.

Design  Self-administered survey as part of the Study of Healthy Activity and Eating Practices and Environments in Head Start (SHAPES).

Setting  Head Start, 2008.

Participants  Directors of all 1810 Head Start programs, excluding those in US territories.

Outcome Measures  Descriptive measures of reported practices and environments related to healthy eating and gross motor activity.

Results  The 1583 (87%) programs responding to the survey enrolled 828 707 preschool children. Of these programs, 70% reported serving only nonfat or 1% fat milk. Ninety-four percent of programs reported that each day they served children some fruit other than 100% fruit juice; 97% reported serving some vegetable other than fried potatoes; and 91% reported both of these daily practices. Sixty-six percent of programs said they celebrated special events with healthy foods or nonfood treats, and 54% did not allow vending machines for staff. Having an on-site outdoor play area at every center was reported by 89% of programs. Seventy-four percent of programs reported that children were given structured (adult-led or -guided) gross motor activity for at least 30 minutes each day; 73% reported that children were given unstructured gross motor activity for at least 30 minutes each day, and 56% reported both of these daily practices.

Conclusion  Most Head Start programs report doing more to support healthy eating and gross motor activity than required by federal performance standards in these areas.

Head Start is the largest federally funded early childhood education program in the United States, serving almost 1 million children from socioeconomically disadvantaged families.¹ Head Start uses a holistic approach to children's school readiness that goes beyond the domains of cognitive and socioemotional development to achieve outcomes in physical fitness, healthy nutrition, and gross motor skills.² Therefore, Head Start has great potential to help address the problem of childhood obesity, which affects between 15% and 25% of the children it serves.^3-6

Head Start programs must abide by federal program performance standards.⁷ These standards are intentionally broad to allow for the varied contexts in which programs are run. For example, while the standards do not use the term physical activity, they require programs to provide “sufficient time, indoor and outdoor space, equipment, materials and adult guidance for active play and movement that support the development of gross motor skills.”⁷ Regarding nutrition, the standards require that each child in a full-day program must receive meals and snacks that provide “one half to two-thirds of the child's daily nutritional needs” through foods “high in nutrients and low in fat, sugar and salt.”⁷ To cover the costs of the meals and snacks offered to children, Head Start programs must use funds from the US Department of Agriculture (USDA) by participating in either the Child and Adult Care Food Program⁸ or the National School Lunch and Breakfast Programs,⁹ adhering to the nutritional requirements of these USDA programs.

The Head Start Act of 2007 permits changes in the federal program performance standards including those that support “children's motor development and overall health and nutrition.”¹⁰ Meanwhile, efforts to prevent obesity are under way in Head Start.^11-13 Despite these favorable circumstances for programmatic change to address obesity in Head Start, there are no national data describing what Head Start programs are already doing in this area that might exceed what is required by the existing program performance standards. Therefore, we undertook a study to describe obesity prevention practices and environments in Head Start programs and to identify program characteristics associated with variability in these practices and environments.

Between February and April 2008 we administered a survey to all Head Start programs as part of the Study of Healthy Activity and Eating Practices and Environments in Head Start (SHAPES). The average program has approximately 6 centers, each with 50 to 60 children aged 3 and 4 years. The targeted survey respondents were program directors, who were encouraged to obtain assistance with the survey from their program's specialists in health and/or nutrition.

The survey was developed and administered in partnership with the US Department of Health and Human Services and the USDA. Administrative data and contact information for all 1890 programs were obtained from the Office of Head Start's 2007 Program Information Report.¹⁴ We excluded 50 programs in US territories, 27 that did not provide direct services to children, and 3 that provided all services outside of centers, leaving a final sample of 1810 programs. The study was approved by the Institutional Review Board of Temple University.

Some of the SHAPES survey items were adapted from existing instruments^15-18; others were based on guidelines or policies regarding nutrition and physical activity in childcare or early childhood education settings outside of Head Start.^19,20 Drafts of the instrument were reviewed by federal staff and by several nonfederal content experts. The instrument was further refined based on cognitive interviews and pretesting with 7 Head Start program directors, each from different states. The final survey contained 90 items, could be completed in less than 30 minutes, and did not require program staff to conduct menu analysis or record review.

To reduce bias, we assured programs that their individual responses would not be shared with federal agencies, and we introduced the survey with the following statement: “This study is not an assessment of whether your program is meeting certain Head Start program performance standards. We expect that most programs have not adopted many of the practices described in this survey. This is because these practices are not currently an explicit part of Head Start's program performance standards.”

Program directors were mailed a paper survey. After sending reminders by electronic and postal-service mail, we reached nonresponding programs by telephone and allowed them to complete the survey over the telephone. Data from all completed surveys were linked to data on program characteristics contained in the Program Information Report.

This article is focused on 30 yes/no items regarding the presence of practices and environments related to healthy eating (15 items) and physical (gross motor) activity (15 items). From these 30 items, we created 2 summary scores. From the 15 items describing practices and environments related to healthy eating, we developed a healthy eating score. Similarly, we developed a gross motor score from the 15 items related to children's gross motor activity. Similar summary scores have been applied previously to descriptions of school food environments.^21,22 Using a score addresses the inherent limitation of making multiple comparisons between several program characteristics and each of the 30 items. Each score had a value between 0 and 15 (1 point given for each “yes” response), with a higher score indicating a more favorable environment for obesity prevention. We did not compute scores for the 27 programs that were missing more than 3 items from either score. For the remaining 1556 programs, any missing items were assigned a value of 1 because programs were more likely than not to have any given practice or environment. Only 21 programs had missing data for more than 2 of the 30 items.

From the Program Information Report data, we created 8 categorical variables to describe program characteristics (Table 1) that we hypothesized might be associated with variation in the 2 scores. These variables were selected based on discussions with federal staff and prior research assessing sources of variation in the educational environments of Head Start programs^23-25 and in the physical activity levels of preschoolers in child care settings.^26-28 Geographic region and rural-urban location were derived from the program address. Because there were too few programs in some states to make meaningful comparisons between programs at the state level, we grouped programs by 9 geographic regions based on the US Census Bureau geographic divisions.²⁹ We also grouped programs along a rural-urban continuum, linking the county of each program to its USDA rural-urban continuum code.³⁰

We first described the program characteristics using the 8 variables derived from the Program Information Report. We also described the program context for providing meals and snacks using survey items about (1) the source (provider) of meals, (2) the location where meals were prepared, (3) the amount of perceived control over the food and beverages served, and (4) the percentage of food costs reimbursed by the USDA. We then described the 30 obesity prevention practices and environments in Head Start.

Using 1-way analysis of variance, we compared the mean healthy eating and gross motor scores across the levels of each program characteristic. Using multivariable linear regression models, with the healthy eating and gross motor scores as dependent variables, we derived the mean score at each level of a program characteristic after adjusting for the other program characteristics. We also used these regression models to identify the program characteristics that had a significant independent association with the score.

For program characteristics that were independently associated with a given score, we examined the association between that characteristic and each of the 15 binary variables that comprised the score. The purpose of these analyses was to explore whether some items in the score were more strongly associated with the program characteristic than others.

Surveys were completed by 1583 (87%) programs, with 188 programs completing the survey by telephone. In each stratum of the program characteristics shown in Table 1, the response rate was 81% or higher (data not shown). In 27% of programs, the program director completed the survey without assistance from other staff. Of the remaining programs, for whom this task was shared, the primary respondent was the following person: program director (41%), health and/or nutrition specialist (47%), or education specialist or other staff member (12%). Of the 1583 responding programs, the median number of students in each program was 314 and the median number of Head Start centers was 6. These programs enrolled 828 707 children across 13 607 centers, 89% and 90% of all Head Start children and centers, respectively.

Thirty-one percent of programs reported that the primary source of their meals was the food service program of a school or school district, while 55% hired cooks directly (Table 2). Approximately two-thirds of programs reported that they had “a great deal of control” over the types of foods and beverages served to children. More than 75% of programs reported that the USDA reimbursed them less than 100% “of the total costs for all meals and snacks served.” In those programs in which USDA reimbursement did not cover the full costs, 93% used money from their Head Start program budget to make up for some or all of the remaining costs.

The prevalence of the 30 practices and environments that comprised the healthy eating and gross motor scores are shown in Table 3 and Table 4, respectively. Ninety-one percent of programs reported serving some fruit each day other than 100% juice and some vegetable other than fried potatoes. Fifty-six percent of programs provided children with at least 30 minutes per day of structured (adult-led or -guided) gross motor activity and at least 60 minutes per day of unstructured gross motor activity. Eighty-nine percent of programs reported that every center had an on-site play area. Of the programs without an on-site outdoor play area at every center, 77% reported that all of their centers had access to an off-site area within walking distance of the center.

The mean (SD) healthy eating and gross motor scores were 11.8 (2.0) and 11.2 (2.1), respectively. The healthy eating scores ranged from 4 to 15, and the gross motor scores ranged from 2 to 15. We considered differences between scores of more than 0.5 points (approximately 0.25 SD) to be meaningful. Geographic region was the program characteristic with the strongest independent relationship to both scores (Table 5). Healthy eating scores also had a strong independent association with the type of entity administering the program, indicating that those programs administered by a school system had lower (less healthy) scores. Programs administered by tribal governments had somewhat lower gross motor scores. We found no meaningful independent associations between either of the 2 scores and the characteristics of programs' teachers or students.

Across the 9 geographic regions, the prevalence of the individual practice or environmental factor differed by 20% or more for 8 of the 15 healthy eating items (eTable 1) and for 7 of the 15 gross motor items (eTable 2). We divided the 9 regions into tertiles (3 groups of 3) based on the adjusted mean healthy eating scores in each region, and we did the same based on the adjusted mean gross motor scores (Table 5). Only 1 region (Pacific) was in the highest tertile for both the healthy eating and gross motor scores, and no regions were in the lowest tertile for both scores.

Because of the association between programs being administered by a school system and the healthy eating score, we examined whether being administered by a school system was related to whether a program obtained its meals from the food service program of a school or school district, and how this, in turn, was related to the program's healthy eating score. Of the programs administered by a school system, 88% used a school food service as the source of their meals, compared with 17% of programs not administered by a school system (P < .001). Programs that used a school food service as their source of meals had significantly lower healthy eating scores than programs that did not use a school food service (10.9 vs 12.2; P < .001). This difference was also observed in programs administered by a school system (10.7 vs 12.0; P < .001) and those not administered by a school system (11.3 vs 12.2; P < .001). Across all programs, those administered by a school system had mean (SD) healthy eating scores that were approximately 1 (0.5) point lower than those not administered by a school system (10.8 vs 12.0; P < .001). However, this difference was reduced to approximately half of a point when controlling for the source of meals (11.4 vs 11.9; P = .002). This finding suggests that the relationship between programs being administered by a school system and having a lower healthy eating score is mediated, in part, by the program's use of a school food service as the source of its meals.³¹

Programs that used a school food service as the source of their meals differed in several ways from those that did not. For 5 of the 15 practices and environments on the healthy eating score, the prevalence was at least 10% lower in programs that used a school food service (Table 6). In addition, programs that used a school food service were less likely to obtain 100% reimbursement of their food costs from the USDA (18% vs 25%; P = .001), less likely to perceive a “great deal of control” over the food and beverages served (22% vs 88%; P < .001), and less likely to have meals prepared at or adjacent to their Head Start centers (55% vs 70%; P < .001).

In this national survey of all Head Start programs, we found that most programs reported practices and environments related to healthy eating and gross motor activity that went beyond the existing federal program performance standards in these areas. For example, while the performance standards have no quantitative guidelines for the amount of children's daily physical activity, more than half of programs reported that they provided at least 30 minutes of daily, structured gross motor activity and at least 60 minutes of unstructured gross motor activity. In the standards for meals served in Head Start, as outlined by the Child and Adult Care Food Program,⁸ 100% fruit juice and fried potatoes are classified as a fruit and a vegetable, respectively, and whole milk can be served. However, most Head Start programs are serving a daily fruit other than 100% fruit juice, a daily vegetable other than fried potatoes, and lowfat milk (skim or 1% fat).

The variation we observed in obesity prevention practices by geographic region was not explained by any of the characteristics of the programs' staff or children. The variation might be due to geographic differences reflected in policies, broader sociocultural norms, or economic conditions. Such place-based factors, ranging from land-use mix³² to regional food preferences³³ to income inequality,³⁴ have been associated with variation in the prevalence of obesity. Similar factors may also shape the contexts in which individual Head Start programs are nested,³⁵ making it more or less difficult for Head Start programs in some regions to support obesity prevention efforts.

Our data suggest that programs administered by a school system are very likely to use the food service program of that school system as the source of the meals they serve. Using the school food program, compared with not doing so, appears to be associated with at least 3 disadvantages: a less healthy eating environment, lower perceived control over the foods and beverages served, and a lower proportion of food costs being reimbursed. More needs to be known before making any programmatic recommendation based on these findings. There may be potential advantages that we did not identify associated with using the school food service. More importantly, there may be cost and quality advantages, unrelated to nutrition, for programs that are administered by a school system. It is not clear whether these programs have other options for obtaining their meals outside the school food service, such as by hiring cooks directly.

Despite the high response rate of the survey, which attempted to reach all Head Start programs, this study had several limitations. Programs may have reported practices that were expected in centers and classrooms but which were not always occurring there. We did not attempt to validate program reports of practices or environments with on-site observations or record reviews such as analysis of a program's meal menus or written staff guidelines. In completing the survey, most program directors received assistance from other management staff. However, in large programs with many centers, the respondents may still have lacked knowledge of specific practices related to obesity prevention. In addition, the survey required programs to characterize their average practices across centers. This made it more likely that programs with large between-center variability would misclassify their program practices. The healthy eating and gross motor scores combined variables that described different dimensions within these 2 broad domains (eg, type of milk served combined with teacher access to vending machines and minutes of structured physical activity combined with access to outdoor play areas). It is possible, therefore, that aggregating 30 items on practices and environments into 2 scores obscured some meaningful differences in these 30 individual items by program characteristics.

In the last several years, there has been great interest in the opportunities afforded by schools to prevent obesity,^36-38 and there have been several comprehensive assessments of eating and physical activity-related practices and environments in US public schools.^21,39-41 However, there are very few data describing obesity prevention practices and environments in either childcare or early childhood education settings.^42,43 This is the first national report on practices and environments related to healthy eating and physical activity in Head Start. This report comes at a time when there is increasing consensus to make greater public investments in early childhood education^44,45 and to begin childhood obesity prevention efforts early in life.^37,46 As Head Start and other early childhood programs try to take advantage of their unique position to prevent childhood obesity, the results of this survey provide programs with a list of practices and environments that are potential targets for change and with a baseline against which these changes can be assessed.

Correspondence: Robert C. Whitaker, MD, MPH, Temple University, Center for Obesity Research and Education, 3223 N Broad St, Ste 175, Philadelphia, PA 19140 (rwhitaker@temple.edu).

Submitted for Publication: May 11, 2009; final revision received June 14, 2009; accepted June 15, 2009.

Author Contributions: Dr Whitaker had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Whitaker. Acquisition of data: Whitaker. Analysis and interpretation of data: Whitaker, Gooze, Hughes, and Finkelstein. Drafting of the manuscript: Whitaker, Gooze, and Hughes. Critical revision of the manuscript for important intellectual content: Whitaker, Gooze, Hughes, and Finkelstein. Statistical analysis: Whitaker. Obtained funding: Whitaker. Administrative, technical, and material support: Whitaker, Gooze, Hughes, and Finkelstein. Study supervision: Whitaker.

Financial Disclosure: None reported.

Funding/Support: This study was supported by Healthy Eating Research Program grant 63042 and Active Living Research Program grant 64114 from the Robert Wood Johnson Foundation, and it was carried out in partnership with the US Department of Health and Human Services and the US Department of Agriculture.

Disclaimer: The opinions expressed do not reflect those of either the Robert Wood Johnson Foundation or the US Government.

Additional Contributions: We thank the following individuals at Mathematica Policy Research, Inc, for their assistance in survey development and data collection: Linda Mendenko, Alison Guy, Elaine Hill, and Anna Comerford. We thank Sara Benjamin, MD, PhD, Marsha Dowda, DrPH, Kerry McIver, PhD, and Catherine Polanski for their helpful suggestions during survey development, and Prabhu Ponkshe, Amy Requa, MSN, CRNP, James Sallis, PhD, and Mary Story, PhD, for feedback on an earlier draft of this manuscript. The following federal staff provided critical review on drafts of the survey instrument and/or this manuscript: Julie Brewer, MPA, Keith Churchill, John Endhal, PhD, Joanne Guthrie, PhD, MPH, and Jay Hirschman, MPH, CNS, United States Department of Agriculture; Robin Brocato, MHS, Amanda Bryans, Laura Hoard, PhD, and Lauren Supplee, PhD, Department of Health and Human Services. Finally, we would like to thank the National Head Start Association for announcing the Study of Healthy Activity and Eating Practices and Environments in Head Start and the programs for completing the survey.