The researchers from the David Geffen School of Medicine at UCLA said that the ratio of waist size to hip size may be a better indicator when it comes to people over 70 years of age.
They found that the waist-to-hip circumference ratio was a better yardstick for assessing obesity in high-functioning adults between the ages of 70 and 80, presumably because the physical changes that are part of the aging process alter the body proportions on which BMI is based.
"Basically, it isn't BMI that matters in older adults - it's waist size. Other studies have suggested that both waist size and BMI matter in young and middle-aged adults and that BMI may not be useful in older adults; this is one of the first studies to show that relative waist size does matter in older adults, even if BMI does not matter," said Dr. Preethi Srikanthan, UCLA assistant professor of endocrinology and the study's lead investigator.
The researchers used data from the MacArthur Successful Aging Study - a longitudinal study of high-functioning men and women between the ages of 70 and 79.
They examined all-cause mortality risk over 12 years by BMI, waist circumference and waist-hip ratio.
They adjusted for gender, race, baseline age and smoking status. The average age of participants was 74.
The study authors said that obesity is very often linked with premature mortality because it leads to an increased risk of diabetes, heart attack, stroke and other major health problems.
However, the researchers found no association between all-cause mortality and BMI or waist circumference-the link was only with waist-hip ratio.
In women, each 0.1 increase in the waist-hip ratio was associated with a 28 percent relative increase in mortality rate (the number of deaths per 100 older adults per year) in the group sampled.
Thus, if the waist-hip ratio rose from 0.8 to 0.9 or from 0.9 to 1.0, it would mean a 28 percent relative increase in the death rate. Put another way, if hip size is 40 inches, an increase in waist size from 32 to 36 inches signalled a 28 percent relative death-rate increase.
But the relationship was not graded in men. Instead there was a threshold effect-the rate of dying was 75 percent higher in men with a waist-hip ratio greater than 1.0 (men whose waists were larger than their hips) relative to those with a ratio of 1.0 or lower.
There was no such relationship with either waist size or BMI.
However, the authors said that the study may have some limitations-BMI may be underestimated because height and weight were self-reported and older adults tend to report those numbers from their younger, peak years.
Besides, waist-hip ratios, waist circumference and BMI numbers were based on single measurements, limiting the researchers' ability to gauge how changing body size in old age can affect mortality risk.
The study has been published online in the peer-reviewed journal Annals of Epidemiology.