Indian children and youth, although not reported separately in the NHANES data, are also particularly affected by obesity (Caballero et al., 2003). For example, the prevalence of obesity in 7-year-old American-Indian children has been estimated recently at nearly 30 percent, representing twice the current estimated prevalence among all U.S. children of that age (Caballero et al., 2003).
Moreover, ethnicity-specific plots of the cross-sectional NHES and NHANES data for children aged 6 to 19 years suggest accelerated rates of increase in obesity prevalence for non-Hispanic black and Mexican-American children of both sexes (Figure 2-4), creating a disparity in obesity prevalence between non-Hispanic white and black children (particularly among girls) (CDC, 2003).
Additional evidence that some ethnic disparities for obesity are increasing over time is drawn from the National Longitudinal Survey of Youth (NLSY). Between 1986 and 1998, the prevalence of obesity increased 120 percent among African Americans and Hispanics while it increased 50 percent among non-Hispanic whites (Strauss and Pollack, 2001).
Evidence also suggests significant variation in BMI as a function of both socioeconomic status and ethnicity based on NHANES III in girls aged 6 to 9 years (Winkelby et al., 1999). An increase in obesity prevalence among African Americans appears greatest for those at the lowest income (Strauss and Pollack, 2001). But uncertainties remain. These disparities are not the same across ethnic groups and they do not emerge at comparable times during childhood. Also, there is almost no consensus, despite many theories, about the mechanisms by which they occur. For instance, analysis of the data from the 1988-1994 NHANES shows that the prevalence of obesity in white adolescents is higher among those in low-income families but there is no clear relationship between family income and obesity in other age or ethnic subgroups (Troiano and Flegal, 1998; Ogden et al., 2003).
Nonetheless, two analyses of nationally representative longitudinal data—the NLSY (Strauss and Knight, 1999; Strauss and Pollack, 2001) and the National Longitudinal Study of Adolescent Health (Goodman, 1999; Goodman et al., 2003)—have suggested that family socioeconomic status is inversely related to obesity prevalence in children and that the effects of socioeconomic status and race or ethnicity were independent of other variables.
One explanation is insurance status, which is related to socioeconomic status; the uninsured may face barriers to accessing health care (Haas et al., 2003). Insurance coverage has been associated with the prevalence of obe-