Chapter 3 BMI as an indicator of CED
The optimum range of BMI which is compatible with good health has been based on data available in apparently healthy populations such as army recruits of developed countries. These ranges were described in the FAO/WHO/UNU report (1985) as 18.7 to 23.8 for women and 20.1 to 25.0 for men. These values were taken from the calculations made from the U.S. Metropolitan Life Insurance Tables by the Royal College of Physicians (1983). A simplified use of BMI in the classification of obesity involves normal values of 20-25, a range used for both men and women (Garrow, 1981) which was subsequently adopted in a WHO Report (1990)
The issue is whether a range designed for the assessment of obesity in affluent societies can be applied to developing countries. The difference in the lower range for women between < 18 and < 20 is unimportant in relation to obesity, yet it is very important in the assessment of CED in developing countries. Therefore, it was necessary to decide whether a lower value of BMI (i.e. <20.0) could be used. First, the lower limits of BMIs of a large sample of healthy, young British soldiers were examined. The BMI and body composition data obtained from 6,000 members of the U.K. armed forces (about 5,000 men and 1,000 women) are summarized in Table 3.1. More than two-thirds of the men in this group and all the women were aged between 17 to 34 years. The advantage of this highly selective population of young adults was that every individual was healthy and fit and was subjected to exacting medical tests twice a year. If the mean minus 2 standard deviations of the range of BMIs is taken as the cut-off point for the lower limit of an acceptable range, then the weighted mean lower limit proved to be 18.5 for men and 17.6 for women.
Other evidence must be considered before accepting these as generalizable cut-off points. In this healthy adult U.K. population, 12 percent of the individuals had a BMI <20.0. This is probably explained by some reduction in lean tissue as well as in fat stores compared with those with BMIs between 20.0 and 25.0. In this particular population group, the BMI at the lower end of the optimum range must represent an individual's natural "habitue" since there is ho shortage of food. It should be noted that some of the men and women were in their teens and may not be fully grown. It is, however, crucial to realize that apparently healthy individuals may exist in developed and developing countries with BMIs below the suggested lower limit of 18.5 and that the cut-off suggested is arbitrary but based on available data from developed countries. The overlap and hence inclusion of some healthy individuals in the distribution of BMIs below 18.5 does not detract from the usefulness of this cut-off or its acceptability.
TABLE 3.1 Mean BMI, LBM and percent fat in a population of healthy young UK soldiers (*)
Categories |
Age |
BMI |
LBM (kg) |
% fat |
Men |
20-24 |
23.4 |
60.4 |
16.6 |
25-29 |
24.2 |
62.0 |
17.4 | |
30-34 |
24.8 |
60.4 |
21.1 | |
35-39 |
24.9 |
60.7 |
21.1 | |
Women |
20-24 |
22.8 |
43.9 |
28.1 |
25-29 |
22.5 |
43.8 |
27.2 | |
30-34 |
22.9 |
41.1 |
29.8 |
Source: Durnin et al., 1984
(*) Civilians tend at each age and sex to be 1-2 kg lighter than the soldiers, mainly at the expense of fat-free mass. (OPCS, 1984)
In most of the adult groups from the developing countries considered by Eveleth and Tanner (1976), the average BMI ranged from 19.0 to 21.0 excluding nomadic tribes such as the Samburu of Kenya and the Dinkas of Sudan. The values have been calculated from mean weights and mean heights. However, considering the large sample sizes this should not introduce any significant error. For urban Indians, there was an average BMI of less than 20 among the younger members of the higher social group, while among the slum dwellers, average BMI was less than 19 (Gopalan, 1987). The rather low fat-free mass of Indian males has been noted elsewhere in this text and this will give lower values of BMI. This may explain why Indians, who were presumably well-fed, still had mean BMIs less than 20.0.
Logically, there can be no defensible single cut-off point without the review of very extensive data from many areas of the world. On the basis of the available evidence, the upper limit for the diagnosis of CED using BMI has been taken as less than 18.5, since a BMI above this has been established as compatible with health both in healthy male soldiers and normal women in the U.K. and in men and women of high socio-economic status from developing countries. A lower limit of 20 would thus be inappropriate
Ideally, determining the lower limit of cutoff of BMI to indicate the presence of CED should be based on information at the population level. It should be based on the relationships between low BMI and the morbidity or health status of adults. It would also be necessary to show causality, i.e. that a low BMI predisposes to morbidity, rather than the BMI simply reflecting the fact that an individual was ill and had lost weight. Information in this area is largely lacking.
TABLE 3.2 The simple classification of adult chronic energy deficiency
BMI |
< 16.0 |
16.0-16.9 |
17.0-18.4 |
> 18.5 |
CED grade |
III |
II |
I |
Normal |
Source: Ferro-Luzzi et al., 1992.
The lower limit of acceptable BMI and hence of both fat mass and lean body mass, might partially depend on the level of physical activity that has to be maintained. Thin and tall adults with a rather low BMI may be fit and healthy. However, the issue is one of judging their physical effectiveness and their ability to withstand periods of semi-starvation. An extreme example of the problem arises with young women with anorexia nervosa who, in affluent societies, may appear well despite having a low BMI and being physically hyperactive. Yet, it is difficult to imagine that women with this BMI would be capable of doing the kind of agricultural work that is often required of rural women in developing countries. Women with anorexia nervosa also have other functional problems, such as amenorrhoea, and undue proneness to infection. Therefore, these low weights seem incompatible with health despite the intense activity of these women. Male Indian labourers were also found to be physically fit by standard tests when they had mean BMIs < 17.0 (Shetty, 1984); whether they were capable of sustained and optimum work output is not known. This concern that modestly thin but active healthy adults may be categorized as being CED led the IDECG Working Party to demand additional proof of a low energy turnover along with a BMI below 18.5 before individuals could be classified as having particular grades of CED.
For the present time, this activity criterion is not applied and only BMI is now considered in developing the index of CED.
In Table 3.2 the limits proposed after taking into account both the BMIs and the compositional studies noted so far are set out. By having three grades of CED, the system conforms to the use of three grades of BMI in classifying obesity (Garrow, 1981) and it relates to the original three grades of infantile malnutrition set out by Gomez et al. (1956).
The choice of the same BMI values for both adult men and women is a notable feature of the standard which needs to be justified It was recognized that women naturally have more body fat than men. In physiological terms, this implies that a CED classification based exclusively on body energy or body fat would mean that women can have a lower BMI than men to obtain the same proportion of energy reserves per unit height. The argument based on weight as a measure of physical power was also important because women are usually shorter than men and therefore, disadvantaged if weight dependent activities are demanded. In spite of the knowledge that women tolerate lower BMIs than men, it was considered reasonable to assign the same BMI cut-off for diagnosing CED as that chosen for men because of the additional energy needed by women to sustain pregnancy and lactation and the recognized major burden they bear in child raising, household care, and agricultural activities in developing countries. An additional advantage is the simplicity of having one criterion for all adults, male and female, for nutritional monitoring and surveillance purposes.
