Overweight and Obese Prevalence Rates in African American and Hispanic Children: An Analysis of Data from the 2003–2004 National Survey of Children's Health

Methods

This study is an analysis of secondary data from the NSCH, collected in 2003 and 2004. School-aged children (ages 5 to 18 years) were the population of interest. NSCH is a random digit dial survey conducted by the Centers for Disease Control and Prevention's National Center for Health Statistics and is a part of the State and Local Area Integrated Telephone Survey program. For data collection, one child from each identified household was randomly selected and data were generated through a phone interview with the parent or guardian in the household who was the most familiar with the targeted child. The NSCH collects data in 8 domains, including demographics, physical and mental health status, health insurance, health care use and access to health care, medical home, family functioning, parents' health, and neighborhood characteristics. A full description of the NSCH and the development of sampling frames have been extensively described in other publications.³⁴ Consent for participation in the study was obtained from interview respondents as soon as it was determined that their household contained a child of eligible age. At that time, respondents were informed about the voluntary nature of the survey, the authorizing legislation, and confidentiality of data collected.³⁴

For our study data for the year 2003 to 2004 were analyzed (unweighted, n = 62,976; 21,715 of this sample were overweight or obese). All analyses were performed on weighted data (only weighted numbers appear in the tables). The data were weighted to be representative of the child population in the United States while maintaining complete subject anonymity. The weighting variable is calculated by the Centers for Disease Control and Prevention using the most recently available census data to provide a stratified representation of the nation's children.

A number of the variables were recoded for our analysis. Recoding consisted of collapsing response categories to gain a more meaningful analysis. The variables, original survey questions used for data generation and the recoded categories used in the analyses, are displayed in Table 1.

Sex- and age-specific BMI percentiles were used to measure overweight or obesity among children. For this measure the BMI number was calculated the same way for children as adults (BMI = [weight in lbs / height in inches²] × 703); however, the criteria used to interpret the meaning of the BMI number for children is different. Once the BMI is calculated, its percentile must be determined using a sex- and age-specific BMI percentile chart. Results are interpreted as follows: underweight, <5th percentile; normal weight, [mteq]5th percentile, <85th percentile; overweight, [mteq]85th, <95th percentile; obese, [mteq]95th percentile. Underweight children were not included in this analysis but the proportion of children in our population sample who were underweight is reported in Table 2.

SPSS software version 15.0 (SPSS, Chicago, IL) was used to complete all statistical analyses performed for this study. Univariate descriptions of and bivariate comparisons were made using an unadjusted odds ratio (OR) to test for differences between and within groups. Multivariate logistic regression was performed to ascertain adjusted ORs. Logistic regression refers to statistical models where the dependent, or outcome variable, is categorical. In this study the logistic regression model was performed to predict characteristics of overweight and obese school-aged children compared with healthy weight school-aged children (or with not being overweight or obese). For all analyses, α was set at <0.05. This study was approved by the University of Illinois Chicago College of Medicine at Rockford's Institutional Review Board.

Results

Using BMI, Table 2 displays the proportion of children by race and ethnicity who are underweight and who reported having a healthy weight compared with those who are overweight or obese. Although 32.2% of white children were overweight or obese, 49.2% of the African-American children and 44.0% Hispanic children were overweight or obese. Table 3 presents the findings from the bivariate analysis by race and ethnicity. A higher proportion of Hispanic children were overweight for most of the categories of the predictor or independent variables compared with African-American and white children. For example, a larger proportion (36.9%) of overweight or obese Hispanic children did not get recommended levels of physical activity when compared with overweight or obese white children. Additionally 70.0% of obese or overweight Hispanic children lived in households with annual incomes <150% of the Federal poverty level compared with overweight or obese white children. This pattern continued for the level of parental physical activity, with higher proportions of both the mothers and fathers of overweight or obese Hispanic school-aged children not engaging in recommended levels of moderate physical activity. When assessing hours of television viewing and nonschool computer use, a larger proportion (36.4%) of overweight African-American children reported 3 or more hours of daily television viewing when compared with overweight Hispanic and white children (25.1% and 19.9%, respectively). Higher proportions of overweight African-American children also reported nonschool computer use 3 or more hours daily (11.8% vs 8.4% for white and 9.9% for Hispanic overweight school-aged children).

Four logistic regression models were tested and the results are displayed in Table 4.

For the first model tested, the dependent variable was all overweight or obese children; the second logistic regression model included only overweight or obese African-American children as the dependent variable whereas the dependent variable for the third model tested was overweight or obese Hispanic children. The fourth and last logistic regression model included overweight or obese white children as the dependent variable. The results of the first multivariate model yielded that overweight or obese children aged ≥5 years were more likely to be African-American (OR 1.568; 95% CI, 1.562–1.574) and Hispanic (OR 1.298; 95% CI, 1.293–1.303) than white. Male children rather than female children were more likely to be overweight or obese in all 4 tested models. Overweight or obese children from all 4 models tested were more likely to live in households with incomes below 150% of Federal poverty level compared with households with incomes >150% of the Federal poverty level and to have not received preventive care in the past 12 months. With respect to physical activity, all of the models tested found that overweight and obese children were less likely to get the minimum levels of moderate physical activity and were less likely to have participated on a sports team. Overall, overweight or obese children were also more likely to have a mother who did not get at least recommended levels of moderate activity. This was also true for school-aged overweight or obese Hispanic and white children; however, the results for mothers of African-American children were inconclusive. All overweight or obese children across the 4 models tested were less likely to have a father who met recommended levels of moderate physical activity. In all logistic regression models tested, overweight or obese children were more likely to watch ≥3 hours of television daily than they were to watch <1 hour of television. Finally, with regard to nonschool computer use, overweight or obese Hispanic and white children were more likely to use the computer for ≤3 hours daily than they were to use it for <1 hour.

Discussion

Similar to other recently published studies,^1,2,11 our findings indicate that, when compared with white school-aged children, African-American and Hispanic children were more likely to be overweight and/or obese. For African-American children the increased risk was slightly more than 50% compared with white children, whereas it was approximately 30% for Hispanic children. Our finding that children of poorer families are more likely to be overweight or obese is also consistent with previous studies that demonstrated that children in lower income groups are at higher risk of being overweight or obese.^2,35–37 Because African-American and Hispanic families are 3 times more likely than whites to live in poverty,^2,38 socioeconomic status clearly places these groups at risk.

Unraveling the relative impact of socioeconomic status, race, and ethnicity to health disparities is difficult. Income impacts childhood BMI in at least 2 specific ways that cross racial and ethnic boundaries: (1) unsafe neighborhoods and (2) the cost and accessibility of healthy foods in low-income communities. Children living in low income neighborhoods are more likely to face safety concerns, which may make physical activity through outside play less likely, and more likely to have poorer access to stores that carry healthier food choices such as fresh fruits and vegetables and whole grain and low-fat dairy products.³⁹ Even if available, such foods are often less obtainable to families with limited financial means.

However, our results did differ from earlier research³⁸ by finding that the association of living in a poorer household and being overweight or obese was strongest for Hispanic children than for white or African-Americans. The reason why these children may be more vulnerable to living in poverty is unclear and could be related factors such as language or an enhanced susceptibility to the marketing of calorie-dense fast food.⁴⁰ Further study to corroborate this finding and to explore possible explanations would likely be helpful for developing strategies to address obesity in Hispanic children.

Among school aged children of the 3 races/ethnicities examined here, boys were more likely to be overweight or obese than girls. Another finding was that overweight or obese school-aged children of all 3 racial/ethnic groups had a slightly greater likelihood of not receiving preventive care in the past 12 months. Although the association between childhood overweight and obesity and receiving preventive care makes intuitive sense, this study is the first to provide evidence of this relationship. Hispanic children who did not have a preventive health visit were 22% more likely to be overweight or obese, a much higher chance compared with African-American or white children and a finding that may be helpful for developing interventions targeted at Hispanic children. For example increasing the number of well child check-ups that are mandated by government agencies and/or required by school systems and during which family physicians, pediatricians, or nurse practitioners could provide counseling and education about nutrition and exercise might be a policy of benefit. Furthermore, policymakers concerned with issues of childhood obesity might pursue the creation of school-based health clinics in schools where at least 50% of the student body live in households with incomes <150% of the Federal poverty level. These clinics could be located in schools, contract for physician and nurse practitioner time, and provide medical services for both students and their families. In addition, if children are not visiting physician offices for preventive care, alternative points of intervention such as Parent Teacher Association meetings or other school-initiated events should be considered. These events could also be viewed as opportunities to address healthy nutrition in the context of family because unhealthy parental lifestyles increase the risk of childhood obesity. Interventions in which both parents and children interactively learn about nutrition and strategies for a healthy diet could be used to teach parents how to shop for healthy food within the context of limited resources and potentially restricted shopping options.

Not surprisingly, physical activity levels and participation on sports teams emerged as significant factors related to childhood overweight and obesity for all 3 racial/ethnic groups of school-aged children. Although all groups of overweight or obese children were more likely to not get recommended amounts of at least moderate physical activity, the likelihood was greater for overweight and obese African-American school-aged children. One possible factor contributing to lower levels of exercise is that both overweight and obese African-American and Hispanic school-aged children were less likely to participate on sports teams, particularly Hispanic school-aged children, who had higher odds of not participating than the African-American children. Sports team participation may very well be connected to a family's financial circumstances; fees for participation and uniforms or other equipment might very well exclude a child from participating. That Hispanic children were less likely to participate in organized sports suggests that cultural differences might also play an important role.

Previous work by Dennison et al found that the mean television/video viewing time was higher for African-American and Hispanic children than White children and concluded that this increased the risk for overweight in those populations.⁴¹ Our findings partially support Dennison et al's findings. Overweight or obese Hispanic children were also more likely than their healthy weighted counterparts to engage in ≥3 hours of nonschool computer use daily; overweight or obese African-American children were more likely to watch ≥3 hours of television daily. Ultimately, spending time in these passive activities may decrease a child's time spent engaged in physical activity of some form. Future studies should examine what these types of activities (watching television and computer use) mean for these different groups of children to devise interventions that might redirect their attention to physical activity or incorporate television viewing and/or computer use into physically active avenues.

There are several limitations to this study. First, our attempt to parse out the effect of race/ethnicity from socioeconomic status was limited by the data collected. Perhaps with more parameters to more specifically define socioeconomic status, ethnicity/race effect might be attenuated. Second, the data are self-reported and subject to error. However, the bias would most likely be to under-report overweight and/or obesity, suggesting that the prevalence of such might be even higher. Third, although several confounding variables associated with obesity were accounted for, it is possible other unidentified factors may account for the increased risk of obesity. For instance, the survey did not ask parents to describe the neighborhood or environments they lived in nor was parental BMI, one of the strongest predictors of childhood obesity, assessed.^19,39,42 Future research to further unravel factors contributing to obesity in the African-American and Hispanic child population may aid in developing effective strategies geared to preventing obesity. The NCHS survey did not query parents about either their children's eating habits or more broad family nutrition patterns. To fully understand issues related to overweight and obesity, eating habits and nutritional patterns are important considerations.^43,44 In addition, although participation on a sports team emerged as a significant factor related to childhood overweight and obesity, there is no way of knowing from the collected data what barriers, if any, African-American and Hispanic children may have encountered that prevented such participation. For example, some of the children included in the study may have attended schools that did not have sports teams or some of the children in the study may not have been able to participate on sports teams because they needed transportation that was not available to them or they may not have been able to pay necessary fees (eg, for uniforms or other equipment). Future efforts should consider collecting data about these issues. Finally, although excessive time watching television was predictive of childhood risk for being overweight or obese, data were not collected about whether or not the child had a television in his/her bedroom. Again such data could prove informative to the development of public health interventions.

Conclusion

The prevalence of overweight and obese children is higher among African-American and Hispanic populations than among their white peers. Effectively addressing this disparity will require a greater understanding of its causes and what factors might magnify the risk among minority populations. Despite the identified limitations, the research findings presented here offer some important insights into the issues surrounding overweight and obesity in African-American and Hispanic children in the United States. Further study is still needed to more completely understand how the various aspects of social, cultural, and economic factors place these children at risk and to develop effective strategies for combating the epidemic of childhood obesity.