Obesity in the U.S. is a major public health concern and U.S. Hispanics are disproportionately affected. The purpose of this study was to explore the associated factors that predict variation in body mass index (BMI) among four U.S. Hispanic populations including Puerto Ricans, Dominicans, Cubans and Mexicans. This study is significant as there is a gap in the literature using time discounting and a health economic theoretical approach to understanding obesity among adult U.S. Hispanics. In addition, the prevalence of obesity has increased in the U.S. across demographics, ethnicities and regions, and its largest increase is among the Hispanic community. Health economic theory (HET) was used to develop research questions and hypotheses.
HET contains elements from human capital theory, household consumption theory of consumer behavior and characteristic theory. Human capital theory states that health is an investment that is made over time and is a function of time discounting. Time discounting looks at how the future is devalued for immediate gratification of present day choices. Household consumption theory proposes that health investment is made possible with resource inputs of market goods, services and time. Thus, market resources aid in accessibility, affordability and availability of health-based mediums. Characteristic theory proposes that a healthy lifestyle carries different meanings dependent upon social background, culture and status. The constructs measured from these three components of HET are time discounting (human capital), market (household consumption) and sociocultural (characteristic). HET provided the framework in which the research questions and hypotheses were developed.
This study had two research questions. Research question one asked if the variation in BMI among Dominican Americans, Mexican Americans, Cuban Americans and Puerto Ricans was more explainable via time discounting, market or sociocultural factors. Research question two asked if time discounting was differentially important compared to sociocultural and market factors in predicting changes in BMI.
Time discounting was measured by three variables including frequency of alcohol consumption in the past year, current cigarette intake and frequency of 10 minutes of light to moderate physical activity per week. Market factors were measured by family income at/above or below poverty ($20,000) and education from never attended/kindergarten to college degree. Sociocultural factors were measured by gender (categorized as 0=female, 1=male) and age which was continuous from 18–85 years and over. BMI was a continuous variable ranging from .01–99.94 and was categorized for descriptive analyses using the World Health Organization cutoffs of 30 and over=obese, 25 and over=overweight and 24.99 and below=normal weight.
Data from the National Health Interview Survey (2002) was used to explore body mass index and associated factors among 4, 244 U.S. Hispanics including Dominicans, Mexicans, Cubans and Puerto Ricans. NHIS used multistage probability sampling to identify U.S. households to conduct one-on-one interviews. A descriptive analysis was conducted as well as stepwise multiple regression. The stepwise regression model was developed by an inclusion criterion of .05 to enter the independents into the final outcomes model. This step was iterated three times.
Descriptive analysis findings for this study showed women were more obese than men, a lack of physical activity among all four groups and increases in levels of education and decreases in smoking were associated with obesity. Stepwise multiple regression statistics showed that little variance in BMI was explained by combining education, gender, age, poverty, smoking, alcohol and physical activity. Age, poverty and alcohol were the key indicators selected into the model equation. Age (.094, p<.01), alcohol consumption (.055, p<.01) and poverty (-.072, p<.01) were important predictors of body mass index. Thus, BMI regressed significantly on alcohol (time discounting), poverty (market) and age (sociocultural) only. High time discounters (alcohol: r=.039*), increased age (r=.086**) and increased market resources (poverty: r=-.067**) were associated with increases in BMI. Alcohol (B=.055) was the only time discounting indicator included in the model and was not as important as age (B=.094) and poverty (B=-.072) in predicting BMI. The directionality of poverty (r=-.067), education (.024) and smoking (r=-.015) did not support the theoretical model.
Using the NHIS (2002) has some limitations in terms of the questionnaire items available to explore sociocultural factors. Additional variables of generation, migration, urban/rural residence, and acculturation may have tapped into more aspects of this factor. A food intake variable could have also gauged the sociocultural factor as well. Poverty level cutoff points lacked variation with a family income of $20,000 at/above or below and individual versus family income should be considered as well. In addition, the sample sizes were not comparable. This affected the ability to effectively do multivariate analyses across ethnic groups.
Culturally based information on the importance of physical activity (appropriate and accessible) would be important in terms of public health interventions. Assessing underlying beliefs, concepts and behaviors of healthy living (time discounting; hot/cold humors) and approaching U.S. Hispanics as heterogeneous group are significant as well. Future research should use a mixed methods approach for triangulation of data, as well as longitudinal methods, more measures and variation in questionnaire items and address obesity in relation to co-morbidities (such as metabolic syndrome).