The main finding of this study was the high prevalence of obesity in all groups including the reference group, subjects born in Sweden, and regardless of outcome. We found higher age-adjusted odds of obesity in Middle Eastern women than in Swedish women regardless of outcome. Men of other European origin had higher odds of BMI obesity than Swedish men. Most associations between country of birth and obesity remained in the full model.
Although the finding of a high prevalence of BMI obesity in immigrant groups is in line with previous research, e.g. Sundquist et al [8
] and Gadd et al [11
], it is important to note that the prevalence established in our study was much higher, being up to three times as high. The prevalence of BMI obesity among men and women from the Middle East was, however, higher than in another Swedish study based on a sample from the entire population [11
]. A similar pattern was evident among native Swedes with a much higher prevalence of BMI obesity than in a previous population-based study [8
]. A Turkish study with a representative national sample [38
] classified 21% of men and 40% of women as BMI obese.
A nationally representative study from Greece based on objectively measured WC [39
] reported that 26.6% of men and 35.8% of women were WC obese, which is in accord with the prevalence of WC obesity established in our study. An interesting result is that 50% of the women of Other European origin had WC abdominal obesity, while only 14.3% were categorized as BF% obese. Thus, it seems that the use of several objective indices of body fatness can be useful when studying obesity in adults from different ethnic groups as WC and BF% provided different pictures of the prevalence of obesity. However, these results should be interpreted with caution since the Other European group was small, e.g. there were only 14 women.
We found higher odds of obesity in Middle Eastern women regardless of objectively measured outcomes than in Swedish women and higher odds of BMI obesity in men of other European origin than Swedish men. It is important to note that the Swedish reference group also had a very high prevalence of obesity, which influences the odds ratios.
In the analysis of sociodemographic correlates, individuals reporting economic difficulties several times during the past year had higher odds of obesity than those without such difficulties, which is in line with previous research on socioeconomic variables [1
]. Surprisingly, educational status was not associated with obesity, which is in contrast to a previous Swedish study [8
] and most other recent research in countries in Western Europe [1
]. This lack of association may be due to differences in data collection, i.e., self-reported data in previous studies versus objectively measured indices of obesity in this study. The result may also be due to the fact that many immigrants living in deprived neighbourhoods could be highly educated but with few realistic opportunities to reach higher socioeconomic strata.
In terms of objectively measured indices of obesity, few studies have examined specific communities or deprived neighbourhoods. A Spanish study [40
] using randomly selected adults estimated the prevalence of BMI obesity to be 23.6% for men and 36.5% for women in the Canarian community of Guía. A recent American study conducted in deprived neighbourhoods classified 45% of the sample as BMI obese [41
Several studies have found associations between neighbourhood deprivation and obesity [12
]. The mechanisms behind neighbourhood environments and obesity are not well understood. Recent studies have suggested that the associations between neighbourhood disorder and obesity are mediated by psychological stress [44
]. Feeling unsafe in neighbourhoods is associated with obesity in young mothers [45
]. Lower socioeconomic status has been associated with being less health conscious and having several unhealthy lifestyle habits [46
]. Previous studies have demonstrated relationships between the built-up environment and weight-related outcomes [47
]. However, studies in ethnically diverse neighbourhood environments based on objective measurements are few in number.
The key strength and novel contribution of the present study is the use of three objectively measured indices of obesity. To the best of our knowledge, no previous study has objectively measured obesity in random samples of residents from deprived neighbourhoods, taking sociodemographic correlates into account. This study also has some limitations. For example, the high non-response rate might introduce a selection bias. However, we have, to some extent, taken a possible non-response bias into consideration by including a weighting system based on gender and age in the statistical analysis. In addition, we achieved a representative sample in terms of ethnicity (31% were of Swedish origin compared to 33% in the population register in the study neighbourhoods). Another limitation concerns the possible socioeconomic selection bias caused by the exclusion of residents without a listed correct phone number. This possible selection bias is, however, most probably a minor problem in Sweden, where social inequalities are less pronounced than in many other countries. This means that socioeconomic differences are unlikely to affect access to a telephone. A third limitation is the relatively small sample (n = 33) in the immigrant group Other European origin resulting in wide confidence intervals in Table , which suggests that only large effects might be captured.