Contents - Previous - Next

Introduction: the challenge of adult chronic energy deficiency

Introduction
Indices of chronic energy deficiency in adults
The FAO initiative
A simplification in the approach to adult CED
The use of BMI alone in CED diagnosis
Challenge to come
The functional significance of modest reductions in BMI
Applications of BMI monitoring
Causes of CED
Conclusions
References
Discussion

W. P. T. James

The Rowett Research Institute, Greenburn Road, Bucksburn, Aberdeen AB2 95B, UK

In Guatemala, in 1987, the term 'chronic energy deficiency' (CED) was being used as a term to indicate an inadequate household food supply. Since then, attempts to define, measure and assess CED have evolved, using the body mass index (BMI or weight/height²) of individuals as the index of CED. An estimate of food intake or physical activity was incorporated into the early analyses, but has since been discarded as difficult and unnecessary because so many people of normal BMI were found to be inactive. Provisional cut-off points for low BMI were developed to define grades of CED in the same way as Garrow used higher levels of BMI to define different grades of obesity. The same BMI cut-off points were proposed for men and women in the absence of suitable data for proposing sex-specific cut-off points. Since then, we have been collecting and sifting worldwide data to test the validity of BMI as a measure of CED. FAO has helped in making data available with extensive information being obtained from China, India, Ethiopia and Zimbabwe. There is now clear evidence that those with a low BMI have more sickness, a lower work capacity, limited social activity, and a lower income. Mothers with a low BMI also have a greater proportion of low birth weight babies compared with those of normal BMI.

Three challenges remain before confirming the use of BMI as the key to specifying a state of CED: Is it valid? Does it apply to individuals as well as populations? What are the causes of CED? We currently recognize that BMI as a measure of CED is a crude tool, and that it may need to be refined in the light of future work and the papers that follow.

Introduction

This Workshop is an interesting illustration of the development of ideas. When many of us took part in one of the first IDECG Workshops, held in Guatemala in 1987, we little thought that the attempt to integrate the interests of physiologists, doctors, anthropologists, psychologists, epidemiologists and those involved in development programmes in the Third World would find themselves suddenly confronting our ignorance of how to specify the inadequacy of food on a household basis and whether we might find more sensitive indices of nutritional improvement when food supplements were given or other measures applied to a community. The conjunction of different disciplines and a willingness to interact constructively led to an acceptance that the term 'chronic energy deficiency' (CED) was being applied to adults in what often seemed to be clinically subjective or socio-economic terms. The value of these approaches was not questioned but the issue became one of recognizing that there was the potential for refining our understanding. Physiological principles might contribute to a clear and more effective understanding of how best to approach problems of survival when food was in short supply. Suitable indices could also prove valuable in rural development projects where current measures of effective outcome are lacking.

At the Guatemala Conference we were very unclear how to proceed. Waterlow was pessimistic because many unsuccessful attempts had previously been made. Taking account of studies on obesity I proposed that some index of weight, such as body mass index (BMI), might be useful or that physical activity expressed as the physical activity level (PAL) might be needed. The IDECG Committee then proposed that John Waterlow, Anna Ferro-Luzzi and I be charged with solving the problem, or at least with setting out the issues more clearly. Scrimshaw was anxious to see progress because effective promotion of nutrition in deprived areas needs a clear specifying of the handicaps and in good conditions, measures improvement. To maintain the adage that people need enough to eat to maintain their health, activity and contribution to the community, is insufficient; we would do our nutritional cause a disservice if clear measures could not be developed.

Indices of chronic energy deficiency in adults

There followed IDECG-sponsored meetings of the three of us in Rome and Lausanne which were wrongly seen by some of our amazed colleagues as gladiatorial battles. These began to draw crowds of onlookers intrigued to see our ideas range far and wide with a ferocity of debate which passes for tranquil, after-dinner musing in the colleges of English Oxbridge! We debated how to distinguish acute from chronic deficiency, energy depletion from other nutrient deficiencies, the storage state of the body from the capacity of organs to use energy, and the usefulness of tedious but accurate measurements conducted on individuals rather than the development of crude indices appropriate to populations. Without initially resolving these issues, we were soon in a debate about the validity of current data on energy requirements - which changed first when insufficient food was available? Did physical activity in behavioural, mechanical, physiological or biological efficiency terms alter fist, or did body weight start falling first only for these other changes to take over when body weight fell below critical thresholds? What now seems reasonably clear was at that stage very uncertain; confusion persisted as to whether the real need for a definition of CED was in the public health or economic development aspects of nutrition promotion programmes in the Third World.

John Waterlow produced our first draft which emphasized our agreement on the need to exclude states of progressive energy imbalance, which we termed acute energy deficiency. We surmised that in this state of changing energy stores we might have many more functional problems than that found when subjects had come once more into energy balance at a reduced rate of energy turnover. Waterlow drew on his experience of anthropometric data collected by John Durnin on British Army recruits, on Gopalan's and Shetty's data on Indian BMIs and on unpublished data by Pryer (1990) from the London School who had collected behavioural measures of illness in adult Bangladeshi men with different weights. There then followed the vexed question of whether adults of the same BMI had similar body composition in all parts of the world and how we could assess energy turnover in relation to the need for economically effective work, for desirable social activities or simply for a sense of wellbeing. This took us back to the debates in the 1981 Committee on Energy Requirements when Anna Ferro-Luzzi had set out these options. She then had to recalculate data from several previous surveys to find body compositional data and physical activity measurements appropriate to our use. By this stage the three of us were convinced that we needed to exclude remarkably thin but superbly fit athletes whom we guessed might abound in many developing countries where physical activity seemed an incessant part of daily life. We therefore chose BMI as the criteria and linked this with some measure of energy turnover which we instinctively saw as a PAL of 1.4 since this had featured in our FAO/WHO/UNU (1985) debate on energy requirements.

Further meetings by Anna Ferro-Luzzi and I then involved endless recalculations of data on men and women to see whether we were justified in having different thresholds for the two sexes. In the end, in the absence of sufficiently clear data, we plumped for the same cut-off points on the basis that biologically women were fatter than men at equivalent weights-for-height. We had already been responsible for proposing to FAD/WHO/UNU that in relation to energy requirements one could take standardizing figures for BMI of 18.7-23.8 for women and 20-25.0 for men, based on retrospective actuarial analyses of the post-war Metropolitan Life Insurance Tables. This difference had, however, been ignored by Garrow in simplifying the BMI classification for obesity by specifying a 20-25 range as normal for both women and men. We presumed that were we to adjust women's weights so that their body energy content was the same as men, then there might be serious physiological consequences and we surmised that the reproductive ability of women would certainly be impaired. Table 1 summarizes our initial proposals. At the time we were concerned about how best to link the BMI values with the measures of energy turnover. It seemed reasonable, however, to conform to the long-standing nutritional principles of having three grades of malnutrition, of increasing severity. This system was used by Gomez in his first assessment of childhood malnutrition (Gomez et al., 1956), and three grades were also proposed by Garrow (1981) for classifying obesity. It seemed reasonable to have a balance between a low BMI and a low energy turnover since we were unsure of their relative importance, so a BMI of 17-18.4 with a low energy intake might be as deleterious as a BMI of 17-17.9 but where energy was sufficient to provide more than the minimum of 1.4 times the basal metabolic rate (BMR).

Table 1. The original classification of chronic energy deficiency (CED) in adults

^aPAL (physical activity level) referred in the publication (James, Ferro-Luzzi & Waterlow, 1988) not only to the PAL expressed as a ratio of the basal metabolic rate but also as a general index of energy turnover. Thus it was envisaged (incorrectly - see text) that food intake could be used as an index of adequacy.

The FAO initiative

Having established the proposals, Ferro-Luzzi and I began to delve into data from different studies to see whether this scheme was practical and relevant to conditions in different countries. We were concerned that we had too little information on both BMI and on energy turnover, preferably assessed by physical activity. Proposals were therefore formulated for FAO with whom we were engaged in developing a more practical approach to energy requirements (James & Schofield, 1990). IDECG was already aware that FAO's 5th World Food Survey was somewhat controversial since it had specified malnutrition or food insufficiency on the basis that energy intake was below either 1.2 or 1.4 times our estimates of BMI (Schofield, Schofield & lames, 1985). The response to food insufficiency was also deemed to be a costless biological adaptation if adults' weights simply fell towards the lower limit of FAO's now normal range (see Table 2). We also recognized that FAO's statistical approach to assessing population rates of undernourished from the food balance sheets not only depended on a value for intra-household variability in the provision of food but also on what body weights and heights one assigned the population in order to develop estimates of population energy needs with which to compare food availability.

We therefore proposed that rather than looking at a more refined statistical approach to world food needs with a reworking of a more valid measure of inter-household variations in food supply, FAO might take the initiative in developing a world-wide database on CED by establishing surveys of physical activity to be undertaken by Ferro-Luzzi and of BMI by myself. Already it was becoming apparent that our original CED proposals were controversial on four grounds. First, the demand for measurements of both BMI and energy turnover was considered too demanding. Secondly, investigators not involved in the early discussions sought to highlight the intra-individual differences in body fat content at the same BMI. There was also considerable uncertainty about the validity of this approach to classifying a state of energy deficiency. Finally there was almost an emotional rejection of the implications from our proposals that the world's nutritionists had been missing a potentially major nutritional problem by concentrating their whole attention on Third World children as the principal vulnerable group.

Table 2. Changing BMI values for optimum weight

FAO was anxious to improve its methods for assessing malnutrition but was understandably sensitive to any suggestion that it might prove inappropriate in future to rely on food balance sheets. It therefore supported a major effort at the Rowett Research Institute to assess the usefulness of BMI as an index of CED because a consensus on this would be needed before a UN Agency could change course or materially add to its own approaches to undernutrition. This set us off collating and sifting data, some of which proved very hard to obtain: one of the consequences of the last half-century of nutritional endeavour is that scientists in Third World countries have become so sensitive to releasing data even for general UN use. Some preferred to sequester unpublished data at home rather than joining in a research proposal. Alternatively, they demanded payment for work to be done by them on site. These responses, of course, ignored the benefit which could come from joint academic endeavour and indeed implicitly rejected the opportunity for a new look at the nutritional condition of their country. Given these surprises it was as well that we concentrated on a sifting through existing data with the help of Dr Ge from China and Mr Naidu from India. A data-bank was also available to François, Weisell and Sizaret in the Nutrition Division in FAO. On this basis Shetty and I drew up a report which is now published (Shetty & James, 1994).

A simplification in the approach to adult CED

While we were sifting pre-existing data at the Rowett, Ferro-Luzzi was organizing a series of meticulous studies with collaborators in India, Ethiopia and Zimbabwe as part of her approach to assessing food needs worldwide. During a visit to Ethiopia in December 1989 we decided to tackle the problem of how to simplify the indices of energy turnover because the classification set out in Table 1 would prove unhelpful if FAO or a nutritional group had to undertake exhaustive studies on food intake or expenditure as well as BMI before coming up with a national assessment of CED. I was also impressed by the observation that the leaders in this field, such as Durnin, Ferro-Luzzi and Shetty, rarely, if ever, did detailed community studies on energy expenditure. Therefore CED monitoring was beginning to look unrealistic without some simple measure of intake or expenditure. Any simple tool would, of course, need to be evaluated.

We soon realized that our original proposal to allow the monitoring of food intake plus BMI for grading CED was inappropriate because we were bound to use only combined measurements from individuals. We recognized that the BMR is normally highly variable between individuals and has a coefficient of variation of +10%. This was sufficient to allow, in those with a constitutionally low BMR, substantial extra energy for physical activity. Thus we would need to recognize a substantial error term in our assessment using food intake when BMR was assumed. If our prime aim was to ensure that sufficient energy was available for activity, then we would need to have measures of activity rather than of intake before effective indices of CED were to be established. Fortunately Ferro-Luzzi had just completed an exhaustive study on Ethiopian women who were monitored in detail almost every month for more than a year. She also had simpler measures of physical activity which could be used to assess our classification in both India and Ethiopia.

Two principal features emerged from this analysis. First there was an instability in PAL which meant that whether or not an individual was graded as CED seemed more of a question of chance rather than reflecting meaningful responses to a fall in food intake (which was also measured in this intensive study). Second, we were surprised to find that a substantial proportion of adults were very inactive. So widespread was the phenomenon that on routine measurement it was difficult to show lower PAL values in those with low BMIs. Only below a BMI of 16.0 was there a discernible extra proportion of adults in India with marked inactivity and in Ethiopia the thinnest women were just as active as the heaviest. If these issues of PAL variability and lack of grading in relation to BMI were ignored, then we were confronted with a further problem: when we applied our initial classification we had a curious lumping of the data in what seemed a biologically implausible distribution (James, Ferro-Luzzi & Waterlow, 1988). If, however, we ignored the requirement to monitor PAL at all, then we found the smooth, almost gaussian distribution of BMIs which might therefore plausibly be used on their own as an index of CED. The effect of including the level of PAL in the classification is illustrated in Fig. 1.

The use of BMI alone in CED diagnosis

This shift in emphasis, set out in detail elsewhere (Ferro-Luzzi et al., 1992), now raises the issue of our original concern lest we include athletes in our diagnostic criteria for CED. Yet the Indian and Ethiopian data provided little evidence that we were dealing with people who were physically very active but thin. Indeed, if we looked at the practical effect of the switch in our classification we were reclassifying only 5% of the population by including everybody with a PAL of 1.4 or more in CED grades I and II even though our choice of the PAL cut-off of 1.4 constitutes not an index of athleticism but of substantial inactivity.

We have therefore moved a long way down the road of simplification since the confused days of our Guatemalan Conference, but a cautionary note must be struck. We really do not know whether the whole population being studied by Ferro-Luzzi and her colleagues in India and Ethiopia were all in a state of relative energy deficiency. Perhaps there is a societal acceptance of relatively low physical activity as the cost of survival, but this could also be at the expense of development which places a demand on physical activity. This is a big issue which must not be neglected and may require a further diagnostic level. Perhaps we will need to consider some people as energy-constrained rather than suffering from CED. How energy constraint is specified may eventually depend on some judgment of behaviour or require novel biochemical tests which signify the neurohumoral response to limited energy availability for desired physical activity.

Challenge to come

Table 3 sets out some of the challenges now to be considered under three headings: first the validity of using the BMI classification for CED; second, its application on either an individual or population basis; and third the issue of whether BMI simply reflects inadequate food supply or has more complex causes. Most of the validity issues have been dealt with in our report (Shetty & James, 1994) and are amplified in this issue. We are unlikely as yet to have a clear picture of whether acute energy deficiency superimposed on a chronic deficient state has different effects from those produced in well-nourished adults. The adaptive processes which one might expect in CED could, in theory, lead to very different responses. As yet we do not know whether protein turnover is lower in CED subjects than normal, but if true we might expect a sudden acute fall in food intake to produce less gluconeogenesis in a CED subject with a sharper rise in ketogenesis and a minimization of weight loss as lean tissue stores are maintained. In the well-nourished person, however, glycogen stores may be ample, so a sharp decline in weight is seen first as glycogen. Water and solute is lost with a subsequent inflow of amino acids from muscle and other peripheral tissues to maintain a high glucose turnover until adaptive changes progressively occur over ]-2 weeks. If this sequence of events is correct, then this may in part explain the intriguing features of Ferro-Luzzi's data on seasonal cycling whereby those with the lowest BMIs show the least falls in weight in response to seasonal deprivation.

Table 3. Challenges in the use of BMI for classifying chronic energy deficiency

The functional significance of modest reductions in BMI

We will probably come to recognize that defining whether a BMI of 17.0-18.5 is of functional significance or whether a cut-off point of 18.0 is more appropriate will prove difficult because the gradient in morbidity is likely to be modest. Thus very large studies will be required to evaluate this issue appropriately. It already seems clear that behavioural criteria are likely to be unrewarding because activity patterns, if depressed, seem to be low at supposedly normal BMIs (see above). Perhaps therefore instead of studying morbidity we should assess a more sensitive risk factor linked to illness rather than illness itself. To this end Ferro-Luzzi, Shetty and I are now embarked on an assessment of CD₄ lymphocyte levels since these are known to fall when nutrient supply is impaired as well as in patients with human immunodeficiency viral infection (Chandra, 1991). Even in HIV infection part of the immunological response may reflect inadequate food supplies.

The problem will then be to consider whether we can refine or add to the BMI measure to improve the prediction of problems at low BMI. This approach is emerging, for example, with the additional use of the waist/hip ratio in evaluating the link between obesity and increased morbidity. One might expect eventually to find an analogous simple measure at low BMIs. The most likely refinement might be a simple measure of body fat - not because it provides a better index of energy stores but simply because the greater the fat content of the body the less likely is the individual to lose lean tissue (Henry, Rivers & Payne, 1988). Thus, in our previous analyses, we showed that New Guinean adults have a greater proportion of lean tissue at equivalent BMIs than Italians or UK citizens: they may then be more likely to have an impaired function at low BMIs since they will have lost more lean tissue. This also implies that some refined measure of function, e.g. CD₄ lymphocytes, is more likely to be depressed in those people whose low BMI is associated with a low body fat content.

Applications of BMI monitoring

Many examples of those listed in Table 3 are likely to emerge, but one neglected area is the use of adult BMIs to re-evaluate the basis for childhood malnutrition. Table 4 provides a framework which we are currently exploring to see whether assessing both child and parental weights-for-height will provide new insights into the differential role of parental care and food deprivation in explaining the poor growth of children. This could prove fruitful in policymaking because it might help on a community level to emphasize the role of health promotion in terms of water supply and maternal education rather than concentrating for example on the issue of supposed generalized food deprivation.

One feature which is already apparent from using the BMI as an index of nutritional state is its value in placing in perspective a large number of claims. Thus it is useful to know that the BMIs of societies are very different -from the studies on overweight women involved in the USAID project in Cairo - to the women studied by Naidu in India (Naidu, Neela & Rao, 1991). The finding that Gambian adults also have a BMI of about 22 immediately highlights the likelihood of their seasonal stresses being an intermittent issue rather than CED being a problem.

Perhaps more valuable is the usefulness of BMI in predicting biological outcomes of fundamental importance, e.g. the link between maternal BMI and birth weight. The preliminary Indian analyses (Shetty & James, 1994) are alarming in showing that 50% of mothers with a BMI <16.0 have babies weighing <2.5 kg. Later in this symposium we will discover whether this reflects the inclusion of young adolescent girls with the older mothers in the analyses. Adolescent pregnancy is likely to prove a fascinating area for study because it could be argued that the phase of pubertal growth is one where endogenous factors, such as insulin growth factor-I (IgF-I), are likely to direct nutrients to the developing peripheral tissues of the mother in preference to the normal process of preferential maternal channelling of nutrients to the fetus. The distinction between the effects of a low BMI in adult pregnant women from a low BMI in an adolescent could therefore prove to be important.

Causes of CED

To many it may be obvious that CED depends on a low intake of food. This does not necessarily mean, however, that the primary basis for a low intake is a lack of available food. We still have a very primitive understanding of appetite control but it is recognized that intestinal infestation can produce anorexia and as yet we do not know, until longitudinal studies are undertaken, whether the higher mortality rates found by Satyanarayana (personal communication) in Indians of low BMI reflect the impact of pre-existing infections or whether the expected immunodeficiency with a low BMI makes these adults prone to succumb from infections. In experimental studies on zinc deficiency it is well recognized that animals show amazing changes in appetite with cyclical 4-5 day swings being observed in rodents fed marginally low intakes (Williams & Mills, 1970). In children on low zinc intakes unusual pica is also evident (Cavdar et al., 1980). Gopalan (1992) has suggested that the Indian subcontinent is becoming denuded of minerals as a result of inappropriate intensive agriculture and there are indeed analyses of sustained nutrient imbalances in Indian agriculture (Shekar, Habicht & Latham, 1991). It is not inconceivable therefore that the low BMIs of the rural Indian reflects the outcome of prolonged subtle zinc deficiency which would impair lean tissue growth. This perhaps might also explain Shetty's findings of an unusual body composition in Indians who seem to have a surprising amount of residual fat at these low BMI values. McNeill, studying adults in Tamil Nadu, found a higher percentage body fat (22.0) in Indian women at a BMI of 19.1 than in UK women (21%) with a BMI of 24.5 (Gillespie & McNeill, 1992). This compositional change may prove to be a partial compensation for a poor growth in lean body mass during adolescence in association with a population-wide reduction in adult BMIs in India.

Table 4. A new potential approach to evaluating the policy implications of childhood malnutrition also measuring the BMI of parents

Conclusions

Our initial tentative steps towards enunciating a new approach to adult malnutrition now seem to have been unduly timid. We are bound to find limitations to the use of BMIs in adults but, if nothing else, the developments of the last 5 years have opened up a new phase where we can now proceed to analyse objectively the many vague statements which are made about food insufficiency on a regional or global basis. I would also expect us to be able to unravel progressively the metabolic and nutrient basis of CED in adults: if micronutrient deficiencies prove to be important this will give a totally new emphasis to agriculture and food policies. An evaluation of maternal and child anthropometry may also provide a new way of refining development programmes so that they become more effective. If only a few of these studies bear fruit I believe we can now claim to be on the point of moving the Third World priorities to nutrition. An emphasis on the welfare of adults may well prove to be warranted after decades of exclusive support for children's programmes.

Contents - Previous - Next