An Analysis of Socioeconomic Differences in Diet Composition and How This Influences Variations in Obesity Among Women

Obesity is a multi-factorial entity which tends to be resistant to intervention, pointing to the need for prevention. Obesity is more prevalent in lower socioeconomic status (SES) groups than in higher SES groups. This inverse relationship has been well-documented in women, and to a lesser degree, in men. Fewer studies have investigated the behavioural factors which may contribute to the different prevalence in obesity. This study has three purposes: 1) to investigate SES differences in intake of total energy, fat, carbohydrate, total, insoluble and soluble non-starch polysaccharide (NSP), and fruit and vegetables; 2) to examine SES and intake of dietary variables above while considering rural/urban differences, cigarette smoking status, obesity, under-reporting of dietary intake and physical activity 3) to assess whether SES differences in diet contribute to the differential prevalence of obesity by SES. Data from the Heart Health Intervention Evaluation baseline study (1992-93) was used to provide data on women between 18-74 years old. Three-day food diaries were used to gather nutrient information, which were analyzed using, for nutrients, the NUTS database and, for NSP, CarbCount. A supplementary questionnaire, given to those who completed the 3-day diary, was used to assess physical activity. All analyses were age-stratified. For the 18-34 year group, there were socioeconomic differences in the percent energy from fat, percent energy from carbohydrate, grams of total, insoluble and soluble NSP, and total, insoluble and soluble NSP densities (g/1000kcal). SES groups did not differ in intake of total energy or in grams of fruit and vegetables. In the 35-54 year group, there were socioeconomic differences in grams of total, insoluble and soluble NSP, and total, insoluble and soluble NSP densities, and grams of fruit and vegetables, but not in percent energy from fat or carbohydrate, or in total energy. In the 55-74 year group, there were no SES differences in intake of dietary variables evaluated. There was no evidence of effect modification of the SES-diet relationships by rural/urban status, cigarette smoking status, under-reporting, obesity or physical activity, although the power to detect these differences was low. Multi-variate logistic regression analysis revealed that the best independent predictors of obesity varied as a function of age. In the 18-34 year group, the following variables were significant unconfounded predictors of the presence of obesity: having a diet with > 35% energy from fat, being an ex-smoker, under-reporting and low educational attainment. Education differences in fat intake may predict the presence of obesity. In the 35-54 year group, the following conditions were significant predictors of being obese: under-reporting, and the joint combination of being an ex-smoker and consuming a diet of 35% energy from fat significantly predicted the presence of obesity. There were no education differences in diet, and thus they could not predict the presence of obesity. SES had the strongest relationship with dietary variables in the 18-34 year group, to a lesser extent in the 35-54 year group, and no effect in the 55-74 year group. Since the magnitude of dietary differences with SES were small, the issue of statistical versus clinical significance, in terms of their relationship to obesity, is raised. Education may not be a sensitive indicator of SES in the 55-74 year group. These findings have implications for health promotion by identifying an area of focus for nutrition education and community programs. Research should move forward into understanding why there are SES differences in diet. From this understanding, health promotion efforts may be targeted more appropriately at specific groups