Disentangling effects of socioeconomic status on obesity: A cross-sectional study of the Spanish adult population

Highlights

• We provide new evidence about the mechanisms through which SES affects obesity.

• The indirect effects of SES are quite modest (3.3% for males, 2.4% for females).

• Dietary habits do not show a significant mediating effect.

• Mediating effect of sedentarism in leisure time is offset by that related to main activity.

• Rest habits contribution to total effect of SES on obesity is significant but small.

Abstract

This paper complements previous estimations regarding socioeconomic inequalities in obesity for Spanish adults, and provides new evidence about the mechanisms through which socioeconomic status (SES) affects obesity. Microdata from the Spanish National Health Survey (SNHS) 2011–2012 are analysed. Corrected concentration indices (CCI) are calculated to measure inequality. Path analysis is employed to disentangle direct and indirect effects of SES on obesity, where dietary patterns, physical activity and sleep habits act as mediator variables. Multivariate logistic models are used to select those exogenous variables to be included in the path diagram. Men and women are analysed separately. Our results show significant pro-rich inequality in the distribution of obesity (the poorer the more obese), particularly for women (CCI = −0.070 for men, CCI = −0.079 for women). The indirect effects of SES on obesity (those transmitted via mediator variables) are quite modest (3.3% for males, 2.4% for females) due to three reasons. Firstly, dietary habits do not show a significant mediating effect. Secondly, the mediating effect of physical activity in leisure time, although significant (14% for males, 11.1% for females), is offset by that related to main activity. Finally, sleep habits contribution to total effect of SES on obesity is statistically significant but small (roughly 1%). Our results indicate that promoting physical activity in leisure time for those with a low SES, particularly for men, would contribute to prevent obesity and to reduce health inequalities. Promotion of adequate sleep habits for women with a low SES might have a similar effect. However, interventions aimed to reduce sedentarism related to main activity, although useful to prevent obesity, would amplify the obesity socioeconomic gradient. Since effects of SES are different for men and women, socioeconomic health inequalities should be addressed also from a gender perspective.

Introduction

Overweight and obesity have more than doubled worldwide since 1980, becoming the main epidemic of the most developed countries. In 2014, 39% of world adults aged 18 and over were overweight, and 13% were obese (WHO, 2015). Obesity has become a health problem associated with a significant sum of comorbidities, a poorer quality of life and a higher use of healthcare resources (WHO, 2000). This is why obesity is considered to be this century’s epidemic. Spain is also experiencing an acute increase in adult obesity (Cámara and Spijker, 2010, García-Goñi and Hernández-Quevedo, 2012). The percentage of obese adults in Spain doubled from 1987 to 2006; besides, Spain shows the second highest prevalence of child obesity and overweight in Europe (García-Goñi and Hernández-Quevedo, 2012). This has raised great concern among Spanish health authorities. In response, the Spanish Ministry of Health, through the Spanish Food Safety Agency, launched in 2005 NAOS (Nutrition, Physical Activity and Obesity) strategy, basically focused on children and adolescents, which is being deployed since then.

Several studies have shown the existence of a socioeconomic gradient in obesity in different countries, including Spain (Cohen et al., 2013, Costa-Font et al., 2014, Costa-Font and Gil, 2008, Devaux and Sassi, 2011, Gutiérrez-Fisac et al., 1996, Gutiérrez-Fisac et al., 2003, Roskam et al., 2010). According to their findings, prevalence tends to be higher in socioeconomically disadvantaged groups, i.e. those with lower levels of education, low income and belonging to a lower social class. Despite the increase in obesity in all sociodemographic groups, some authors highlight a trend of increasing social inequality since the late eighties (Gutiérrez-Fisac et al., 2003), while the increase in prevalence has tended to focus on individuals of lower socioeconomic status (SES) as a result of two main factors: the change in eating habits (resulting from the high availability of cheap high-energy food), and the evolution undergone in physical activity patterns (the increase of inactivity, particularly among the most socially disadvantaged individuals). The onset of the economic crisis in 2008 has also been noted as a probable cause of the most recent increase in social inequality linked to obesity (Sassi, 2010, OECD, 2014). Low-income families have been forced to cut their food budgets, which has increased the purchase of cheaper and less healthy food products (to the detriment of others, such as fruits and vegetables) (OECD, 2014). Therefore, it is assumed that energy-dense diets are more affordable than diets based primarily on healthy foods (Drewnowski and Specter, 2004).

It is well known that socioeconomic environment has a significant impact on the prevalence of a high number of diseases, including obesity. But more evidence is needed about the mechanisms through which SES affects health outcomes in order to effectively combat health problems as well as socioeconomic inequalities in health. In this paper we estimate the socioeconomic gradient of obesity in Spain by using corrected concentration indices, and provide new evidence about how SES affects obesity. Path analysis may be very helpful in this task, since it contributes to disentangle direct and indirect effects of a variable of interest on another. As path analysis helps to identify which are the mechanisms that may mediate the effect of the SES on obesity, it may supply some guidance about how to address socioeconomic inequalities in the prevalence of this condition. Until now, this technique has been applied to the study of childhood and adolescent obesity (Dollman et al., 2007, Janssen et al., 2006), but not to the adult population. The analysis will be applied to the Spanish adult population and conducted separately for men and women, since obesity shows different patterns by gender (Devaux and Sassi, 2011). Hence, the results here obtained could also be useful to better understand and address gender inequalities in health.