Results

Discussion

These nationally representative AusDiab data show alarming rates of overweight and obesity. The obesity rates reported here are considerably higher than those of previous Australian urban studies in 1980, 1989 and 1995.13-15 In 1980, 7.1% of the population aged 25–64 years living in major cities in Australia were obese.13 Using the same age and geographical restrictions, the prevalence (defined by BMI) in our study was 18.4%, showing a 2.5-fold rise over 20 years. Across the four surveys in the last 20 years, the prevalence appears to have stabilised in men since 1995, but a continuing rise is apparent for women.

The consequences of high rates of overweight and obesity are likely to be profound, with the AusDiab data also showing that Australia now has one of the highest rates of type 2 diabetes in the developed world.4

The high prevalence of obesity found is still considerably lower than the reported rate for 1999–2000 from the United States2 (27.5% in men and 33.4% in women), but comparable with those reported from the UK16 (17% in men and 21% in women) and (West) Germany17 (19.4% in men and 20.9% in women), ranking Australia as one of the most severely affected Europid populations.

Waist circumference provides an alternative measure of adiposity that correlates reasonably well with BMI,18 but appears to be a better indicator of visceral fat, type 2 diabetes and cardiovascular disease.19 There are few published studies of the age and sex distribution of obesity according to waist circumference. In the second MONICA survey20 only two centres (Germany, Czech Republic) had higher mean waist circumferences for men than in our study, and four centres (Germany, Czech Republic, Spain and Yugoslavia) had higher mean waist circumferences for women.

The age and sex pattern of BMI is consistent with a number of other studies.2,21 Both the prevalence of obesity and mean BMI are lower in young women than in young men, but a more rapid rise in BMI in women results in women overtaking men by the age of 35–44 years for obesity prevalence and by age 55–64 years for mean BMI. The impact of age on waist circumference is almost identical for men and for women, suggesting that the factors leading to increasing waist circumference with age are much more similar between men and women than are the factors affecting BMI. Increasing peripheral fat in women related to childbirth and the menopause may well be the critical difference.1

We found that physical activity time and television viewing time were the strongest correlates with obesity. Television viewing time showed a significant positive association with both measures of obesity in men and women, after adjustment for physical activity time and other risk factors. A corresponding negative association was seen for increased physical activity time, although it was not a significant predictor of BMI in men. Of particular note is the strength of the relationship of obesity with television viewing time — even those in the top tertile of physical activity showed a high risk of obesity if they were also in the top tertile of television viewing time. This dominant effect of television viewing time has been reported previously in another population-based Australian study.22 The relative imprecision of recall of physical activity time in comparison with television viewing time may explain part of this finding. Another explanation may be the reduction in incidental (non-structured) physical activity associated with television viewing, which, in an inactive society, has the potential to significantly reduce total energy expenditure.23 The association between television viewing time and obesity is important for health education and public health programs. While most such programs focus on increasing the time spent engaged in physical activity, it may be more achievable to recommend reducing the time spent in completely sedentary activities such as watching television. Indeed, a recent interventional trial has shown that measures to limit television viewing in children can be effective in controlling obesity.24

Together with television viewing time and physical activity time, energy intake has an important impact on obesity. From 1983 to 1995, Australian data show that there has been a significant increase in energy intake among adults and children.25 Television viewing itself has been linked to higher energy intake (take-away meals) among women,26 and this may relate to habits of eating during television viewing.

Of the socioeconomic factors examined, lower educational attainment showed the most consistent relationship with obesity. This finding is supported by other studies,21 although the cause is not clear. The association of obesity with income was sex specific. In men, minor trends for middle-income groups to be more obese and the least affluent to be thin were observed, although these were not significant. Women, by contrast, showed a strong positive graded association between income and obesity.

When interpreting these results, some caution should be exercised. As the AusDiab study was cross-sectional, causality cannot be determined from the associations observed. For example, obese people may be less active as a consequence of their obesity. This is unlikely to be the entire explanation for the associations reported, as decreased physical activity has been linked to obesity in prospective studies.27 The level of response to the study should also be considered, as well as the small differences between responders and non-responders.7 Additionally, even though AusDiab was designed to provide estimates representative of the adult Australian population, the exclusion criteria may have resulted in under-representation of some population groups (eg, Indigenous and rural Australians).

In conclusion, Australia has been shown to have alarming rates of both central and general obesity. This urgently demands action on many levels to prevent further rises in the prevalence of diseases such as type 2 diabetes. The strong relationship we found between obesity and surrogate indices of energy expenditure needs to be confirmed in prospective studies, but suggests that reducing time spent in sedentary activities could be an important target for preventing and treating obesity.