Socioeconomic inequality of obesity in the United States: do gender, age, and ethnicity matter?
Introduction
Obesity has become a global epidemic, and the prevalence of obesity continues to increase in both developed and developing countries (Wang, Monteiro, & Popkin, 2002; WHO, 1998). Obesity increases the risk of a number of diseases and health conditions, including cardiovascular disease, hypertension, Type II diabetes, and certain types of cancer (Bray, Bouchard, & James, 1998; WHO, 1998). In the United States, over 60% of the adult population is overweight or obese, and obesity is currently the second leading cause of preventable disease and death, next to smoking (US HHSD, 2001). The magnitude of obesity as a major public burden is reflected in its total direct and indirect cost, which was estimated at 120 billion dollars in 2000 (US HHSD, 2001).
Effective interventions to control or manage obesity will necessarily hinge on understanding the complex processes that determine body composition, including excess adiposity, in the population. These processes involve the interactions of numerous factors, including biological characteristics such as genetic predisposition, social, cultural, environmental, and behavioral factors (Bray et al., 1998; WHO, 1998). For example, socioeconomic status (SES) has been demonstrated to influence individual's energy intake and energy expenditure, and as a result, to affect body fat storage (Sundquist & Johansson, 1998). Further evidence can be found in the repeated findings of disproportionally high rates of obesity among minority and low-income groups (Dreeben, 2001; Flegal, Carroll, Kuczmarski, & Johnson, 1998).
In the biomedical field, linear and logistic regression analyses are the classical approaches to studying the association between SES and obesity. Usually, odds ratios (OR) or beta coefficients are reported to indicate the magnitude and direction of the association (Sobal & Stunkard, 1989; Wang, 2001). These methods are straightforward, but suffer from several limitations. First, although linear regression analysis can help examine whether there is an association between SES and obesity, it is not powerful enough to measure the disparity quantitatively, i.e., to tell how severe the inequality is. Second, comparing inequality across studies or over time using traditional regression analysis is difficult, since the validity of regression analysis is based on the assumption of multi-normality and independence between study variables over time (Zeger, Liang, & Albert, 1988). Third, from a statistical perspective, linear regression analysis assesses the relationship between the outcome and explanatory variables on average but ignores the possibility that the effect of explanatory variables may vary across the distribution. To illustrate, suppose SES is positively related to BMI at one tail of the distribution of SES, while SES is negatively related to BMI at the other tail of the distribution. Simply regressing BMI on SES may result in a finding of zero mean effect of SES on BMI.
To solve similar problems, economists have developed summary indices such as the Gini coefficient and the concentration index to quantitatively measure the degree of income-related inequality. A variety of techniques based on these indices have since been developed to address the limitations of linear regression (Lambert, 1993). These indices and methods are used extensively in economics and have proven useful in studying issues like tax progressivity (Kakwani, 1977; Plotnick, 1981; Jenkins, 1988).
More recently, economists have exploited the intuitive analogy between taxes and poor health outcomes as burdens that may be unevenly apportioned to members of different SES groups, and have applied techniques of analyzing income inequality to analyze inequality in general health status (Kunst, Geurts, & van den Berg, 1995; Wilkinson, 1996; van Doorslaer et al., 1997). However, only a few studies have focused on SES inequality of specific diseases (see Pamuk, 1988; Sturm & Gresenz, 2002). To our knowledge, no published study has quantitatively measured the degree of SES inequality of obesity.
The primary goal of this study is to apply a summary index, the concentration index, to assess the degree of inequality in the distribution of obesity across SES using a national representative survey data set. We also investigate whether there are differences in the inequality across sex, age, and ethnic groups. Our study yields potentially important implications for medical scientists engaged in studying the relationship between SES and obesity, social scientists interested in the effects of health inequality with respect to disease, and policy makers committed to reduce health disparities across socioeconomic groups.
Section snippets
The concentration index as a measure of health inequality
A few summary indices of health inequality have been reported in the literature, including the Gini coefficient, the index of dissimilarity, the index of inequality, the relative index of inequality and the concentration index (see Wagstaff, Paci, & van Doorslaer, 1991 for a review). Of these summary indices, Wagstaff et al. (1991) argued that the concentration index (CI) is the most appropriate measure of health inequality, since it meets the three basic requirements of a health inequality
Data
We used data collected from the National Health and Nutrition Examination Survey III (NHANES III, 1988–1994), which is a cross-sectional representative sample of the US civilian, non-institutionalized population aged 2 months and older. NHANES III contains data for a sample of 33,994 individuals. Data on weight and height were collected for each individual in the full mobile examination center through direct physical examinations. Based on self-reported race and ethnicity, subjects were
Overall socioeconomic inequality of obesity and overweight
The CIs of obesity and overweight are presented in Table 2. CIs in row 1 are measures of the socioeconomic inequality in obesity and overweight for the whole study population. The CI of obesity (BMI⩾30) was –0.055 and was statistically significant (P<0.05), indicating that socioeconomic inequality favors higher SES groups. In other words, SES was negatively related to obesity. The CI of overweight (BMI⩾25) was –0.007, which was not significantly different from 0. This suggested that there was
Discussion
To our knowledge, no previous studies have quantitatively measured socioeconomic inequality in the distribution of obesity in the US. using the concentration index. Our analyses are among the first to examine the socioeconomic inequalities across gender, age, and ethnic groups by using a national sample. We found that the SES inequality of obesity varied dramatically across demographic groups and there were several interesting patterns. First, large ethnic differences exist in the association
Conclusions
We introduced the concentration curve and concentration index as an alternative approach to study the socioeconomic disparity in obesity. The concentration index provides a quantitative summary measure of socioeconomic inequality in obesity. It not only tells the direction of the association between SES and obesity, but also indicates how severe the inequality is. This method could be a useful tool in future research that compares the socioeconomic inequality in obesity over time and across
Acknowledgements
We would like to thank Dr. Jeanette W. Chung, Ms. Lisa Johns for her comments and assistance on editing to improve the manuscript. We appreciate Dr. John Formby, Dr. John Bishop, and two reviewers’ invaluable comments.
References (46)
- et al.
Socioeconomic inequalities in healthMeasurement, computation, and statistical inference
Journal of Econometrics
(1997) Understanding ethnic differences in energy balanceCan we get there from here?
American Journal of Clinical Nutrition
(1999)- et al.
Risk factors for obesity in young adultsHispanic, African American and Whites in the transition years, age 16–28 years
Biomed & Pharmacother
(1994) Income-related inequalities in healthSome international comparisons
Journal of Health Economics
(1997)- et al.
On the measurement of inequalities in health
Social Science & Medicine
(1991) - et al.
Trends of obesity and underweight in older children and adolescents in the United States, Brazil, China and Russia
American Journal of Clinical Nutrition
(2002) - et al.
Energy metabolism in African AmericansPotential risk factors for obesity
American Journal of Clinical Nutrition
(1999) Accelerating cardiovascular risk factor change in ethnic minority and low socioeconomic groups
Annals of Epidemiology
(1997)- et al.
Handbook of obesity
(1998) - et al.
Correlates of obesity in young blank and white womenThe Cardia study
American Journal of Public Health
(1992)