The relationship between SES and obesity is also mediated by behavioural factors such as diet (or nutrition) and physical inactivity, which play a role in the SES-body mass index (BMI) gradient [15-18]. However, in a population-based study in Sweden, only a part (18%–29%) of the association between educational level and obesity could be explained by the measured lifestyle factors [13].

In underprivileged zones and among the most disadvantaged groups, individuals (especially young people) are more inclined to smoke, which increases risk for overweight [19-21].

In Luxembourg, the prevalence of overweight and obesity was 36% and 17%, respectively, in 2005 [22]. A recent study [23] involving patients who underwent coronary angiography in 2008 and 2009 showed high rates of overweight (47.1%) and obesity (34.2%). Because the prevalence of these conditions is increasing in most developed countries, we studied their evolution and determinants between 1995 and 2007 in the general population in Luxembourg.

Resident nationality reveals changes in lifestyle and dietary habits associated with population migration. For example, obesity prevalence in Portugal was lower (14.1% of men and 16% of women in 2005 and 2006) [26] than among Portuguese residents in Luxembourg. In contrast, for French residents in Luxembourg, obesity prevalence was lower than the prevalence rates observed in France (12.5% of men and 13.6% of women) [25].

The extent of the relationship between SES factors and obesity prevalence examined with multiple logistic regression analysis confirmed the results of other studies [4,8-14,23,28,31-33].

The influence of education on obesity is significant in women because it clearly shows the social gradient; other studies have presented results showing an inverse relationship between education level and SES status [8,10,31]. The link between obesity and educational level is reflected in the multivariate logistic regression model. Educational level influences the ability to process information regarding healthy lifestyle and, more specifically, overweight and obesity [8,32]. According to Kenkel [32], education helps people choose healthy lifestyles by improving their knowledge of the relationship between health behaviours and health outcomes. In rural Appalachia, a study found that education has a significant influence on risk for obesity [4]. Cawley and colleagues [10] emphasized the role of information and education in the association between education and obesity. A low level of education was an obesity risk factor in our study, particularly among women. Women who had only reached the level of primary or secondary education were twice as likely to be obese compared to those who had achieved a higher level of education. In Greece, educational level was not associated with risk for obesity among men, but this relationship existed among women [28]. In Spain, a study conducted between 1987 and 1997 showed a predisposition to obesity that was much higher in people with a basic level of education, regardless of gender [11]. Overall, these results confirm the important role of education in the prevalence of overweight and obesity, as observed in other studies [4,9,10,28].

Some studies found that married people were more likely to be obese than unmarried people [28,33]. Our results showed that marital status was associated with obesity among women in 1995 and among men in 2007. In Greece, married men and married women were twice as likely to be obese than those who were not married [28].

Our results demonstrated a larger difference in obesity prevalence between agricultural workers, craftsmen, general workers, and those with no profession, and men working as managers and in intellectual professions. The results of our multivariate analysis showed that men working in lower professions were more than twice as likely to be obese than men working in top professions (e.g., managers and intellectual professions) in 2007. This result was similar to that found in England, where higher professional status was associated with lower risk for obesity [34]. Among women, this difference decreased through the study period but remained high. This trend also was observed in France between 1992 and 2003 [35].

Obesity distribution is associated with respondents’ residence location; this association was higher in 2007. These differences may be attributable to SES context, diet and physical activity, and environment (urban or rural). In Portugal, a cross-sectional study showed that people who lived in rural areas were at lower risk for obesity than those who lived in urban areas [36]. In contrast, our results suggest that men and women who live outside of the major town of Luxembourg (Luxembourg City) are more likely to be obese. The North and West areas are less populated and include rural areas, and obesity prevalence was higher among men and women in these 2 areas. It is noteworthy that even in a small country such as Luxembourg, spatial variations of obesity can be observed.

Many studies have found that physical activity is associated with obesity [37,38]. Our findings showed that physical activity was associated with obesity risk in both men and women, particularly in 2007.

Other studies found a statistically significant association between obesity and diet, but sometimes the results were mixed [39-41]. Our results suggest the relationship between obesity and diet was statistically significant in both men and women in 1995 and 2007. These results may be of interest for those involved in diet-related health promotion.