Page 564


Nomogram for determining BMI. To use this nomogram, place
a ruler or other straight edge between the body weight in
kilograms or pounds (without clothes) located on the left-hand column
and the height in centimeters or in inches (without shoes)
located on the right-hand column. The BMI is read from the middle of
the scale and is in metric units. Copyright 1978, George A. Bray. Used by permission.

most widely used tables of desirable body weight. Both the 1959 and the 1983 Metropolitan Life Insurance tables were based on data obtained from the pooled experience of the life insurance industry in the United States (Society of Actuaries, 1959, 1980a,b). Although these surveys of weight and stature among insured individuals provide data on nearly 5 million people, they suffer from a self-selection bias, i.e., they provide data only on people who choose to take out life insurance. The insured tend to have a longer life expectancy, to be healthier, and, on average, to weigh less than the general population.

A second data base has been generated by the National Center for Health Statistics (NCHS), which in the past 20 years performed five surveys, including measurements of weight and stature of a representative sample of Americans from  census tracts in the United States (Abraham et al., 1983). These surveys include approximately 20,000 people.

Appropriate weight standards can be determined in two ways. First, the normal distribution of

TABLE 21-1 Desirable Body Mass Index in Relation to Age

Age Group
















weight in relation to height can be arbitrarily divided into overweight and severely overweight groups. This approach has been used by the NCHS, which defines overweight as those in the 85th percentile of weight for height using as reference the weights of 20- to 29-year-olds. With this technique, a BMI higher than 27.8 kg/m2 for men and above 27.3 kg/m2 for women is considered overweight, and the top 5th percentile is severely overweight. This approach was used in the Surgeon General's Report on Nutrition and Health (DHHS, 1988) but not by the National Institute on Aging. There are several drawbacks with this approach. First, the standards change as the weight distribution of the population changes. Second, the 85th percentile values of the BMI, 27.8 kg/m2 and 27.3 kg/m2 for men and women, respectively, will be very difficult for health professionals and the public to remember or understand. Third, and more important, is the underlying assumption that average weight is a healthy or preferred weight. Lastly, it is assumed in this approach that optimal weights remain constant at different ages—an assumption that may not be justified (Andres, 1985).

Weight standards can also be based on the BMI associated with the lowest overall risk to health. The minimal death rate in several prospective studies is associated with a BMI of 22 to 25 kg/m2. Andres (1985) reanalyzed the Build and Blood Pressure Study of 1979 (Society of Actuaries, 1980a,b) and showed that the BMI associated with the lowest mortality increased with age. A similar increase in the BMI distribution curve with age is evident from a study conducted in Norway (Waaler, 1984). On the basis of these collated data, the ranges for BMI in relation to age proposed in Table 21-1 seem reasonable. Although the BMI is adjusted for age, the range overlaps. For example, the highest BMI for 19- to 24-year-olds is 24 kg/m2, which is the lowest for those over 65. A BMI above 25 kg/m2 was associated with increased

The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement