FATHOM: Sidebars

From Columbia University
Role of Fat Distribution in Obesity

To consider different measurements of obesity, we need to determine how they correlate with one another. Waist circumference and waist-to-hip ratio (WHR) are strongly correlated at 0.65. Waist circumference is even more strongly correlated with BMI at 0.82. It's very hard to have a high BMI and not have a high waist circumference. There is a much weaker relationship between WHR and BMI, with a correlation coefficient of 0.34. This means that WHR may be capturing different information than BMI.

What happens if we accept BMI as a risk factor and look for an additional risk factor on top of that? Will it change how we think about these two measures?

Let's look at WHR within each BMI grouping. Within each BMI, a higher WHR clearly increases the risk for coronary heart disease. It more than doubles the risk of coronary heart disease to have a high WHR within a low BMI. The same is true even on the upper end of the BMI spectrum. The WHR may add additional information about risk.

Women and fat distribution
The Nurses Health Study compared both WHR and waist circumference to subsequent coronary heart disease. A WHR of about 0.87 showed a relative risk about two and a half times higher than a WHR of less than 0.72. A similar effect occurred with waist circumference, although a bit more of a threshold effect exists in women with a waist circumference of more than about 30 inches. But again, even after adjusting for BMI, a significant trend and relative risk of three exists for a waist circumference greater than or equal to 38 inches.

What about the "normal" BMI group? Even in women with a BMI of less than 25, a WHR of 0.88 increased the relative risk to about four and a half, as compared with those who have a lower WHR and the same BMI. For women, these two measures--BMI and WHR--are not all that correlated. Women can have identical BMIs and very different WHRs.

Men and fat distribution
In examining fat distribution, it is not just fashionable but essential to look at men and women completely separately. The distributions of waist circumference and WHR barely overlap for men and women. The smallest WHR grouping for men was 0.90 versus 0.72 for women--barely overlapping distributions.

The Physicians Health Study looked at the relationship between WHR and coronary heart disease in men. The multivariate-adjusted estimate shows a significant increase of 1.53 for a greater WHR. However, after body mass index is taken into account, this finding becomes much weaker: 1.3; 1.27; and 1.3. Similar results have been seen in a number of studies, with WHR not adding much information after BMI is accounted for. Waist circumference shows a relatively similar pattern in men. An initial relative risk of 1.61 is reduced to a nonsignificant trend and a point estimate of 1.1 once BMI is taken into account.

The elderly and fat distribution
BMI is generally a stronger predictor in men under 65. In our research among men in the Physicians' Health Study, the relative risk estimate for a BMI greater than 27.6 is about 2.4 in younger men, versus barely 1.5 for older men. Given the natural decline in muscle mass as we grow older and increase in adipose (fat tissue), perhaps BMI does not measure total fat as well in the elderly. This led researchers to examine whether WHR is a better predictor in elderly adults.

The Health Professionals Study showed a fairly weak association between WHR and heart disease, with a relative risk of about 1.3 (already adjusted for BMI) for men under 65. Among men over 65, the relationship was much stronger. This finding posed an interesting argument as to whether we should be measuring WHR versus BMI in older adults. When we examined the question through the Physicians Health Study, we did not see a similarly strong relationship in older men. While the WHR performs better than the BMI in older men, it's still not as predictive as in younger men and women.

Although the Nurses Health Study used a slightly younger cut point of 60, the results remain similar in women: WHR is quite a strong predictor for people under 60 years old, but nonsignificant for those over 60.