Previous - Next
This is the old United Nations University website. Visit the new site at http://unu.edu
Rome, 12-14 October 1977
A Joint FAO/WHO Expert Consultation on Energy Intake and Protein Requirements was held in Rome from 12 to 14 October 1977. The meeting was opened by Dr. P. Lunven of the Food and Agriculture Organization, Rome.
The Consultation was convened as a first step to initiate a joint study on energy intake and protein requirements envisaged under the project INT 297 (DEN) and funded by DANIDA, the Danish International Development Agency. These studies are needed to answer, or at least examine, several questions that have been raised since the report of the Joint FAO/WHO Expert Committee on Energy and Protein Requirements was published in 1973 (1). Much of our concern arises from the fact that most of the work on human protein requirements has been done in well-fed subjects in developed countries. The aim of the agencies is to ensure that the most vulnerable groups in developing countries have adequate protein intakes, and therefore the definition of protein requirements must be related to the conditions that obtain in those countries.
1.1. Problem Areas
1.1.1. The Effect of Different Levels of Energy Intake on Protein Requirements
In 1975 a joint FAO/WHO informal gathering of experts 12) drew attention to studies, both old and new, showing that energy intakes above the requirement level increase nitrogen retention. The protein requirement is conventionally defined as the amount needed when the requirements for energy and other nutrients are fully met. Therefore, to be on the safe side, some investigators, in nitrogen balance studies on adults, have provided liberal energy intakes that permitted some weight gain. The effect of this treatment would be to minimize nitrogen losses. If results obtained under these conditions are then applied to people whose habitual energy intakes are rather low, the protein requirements for maintenance may be underestimated unless there is a compensatory adjustment of the protein-energy relationship, Similar considerations may apply to the interpretation of metabolic shifts in children.
A further question is posed by recent work suggesting that, when nitrogen balance is measured at high levels of energy intake, the quality of the protein fed may be overestimated (3, 4, 20). It may then be misleading to correct dietary allowances for protein quality on the basis of these measurements of Net Protein Utilization (NPU) if the energy intake from the diet in question is low.
The conclusion is that we need to re-examine protein requirements in terms of the habitual diets of developing countries when these diets are fed at natural or usual levels of energy intake.
1. 1.2. The Criteria of Protein and Energy Requirements
The concept that protein requirements must be related to the conditions that exist in groups at risk leads to the question of whether the criteria conventionally used are adequate. In young children, weight gain provides a useful index, and it is probably for this reason that the published estimates of protein requirements in this age group are widely accepted and are considered generally applicable. In adults, protein requirements have been determined from direct or indirect estimates of obligatory nitrogen losses or from measurements of nitrogen balance. However, the existence of nitrogen balance provides no real guarantee of an optimal state of protein nutrition. For example, it has often been suggested that overall nitrogen balance may conceal an imbalance of protein synthesis in one tissue compared with another. Moreover, with few exceptions (5, 6), balance studies in both adults and children are generally made over relatively short periods, and the conditions are artificial.
In the case of energy the situation is even worse. We have no adequate method for determining the energy requirement of individuals. The conventional criterion of maintenance of consistent body weight in adults or a normal rate of weight gain in children is too crude to be of much value except in rapidly growing infants. Current methods of measuring changes in body composition are insufficiently sensitive, direct measurements of energy expenditure are too complex for general use, and estimates based on the heart rate are subject to criticism. In fact, some workers consider that nitrogen balance is probably the most sensitive index of whether or not energy requirements are met. This re-emphasizes the need, discussed in the previous section, for further work on the interrelationships between protein requirements and energy intakes.
We recognize, however, that the final criteria of whether or not protein and energy requirements are met must be functional. For an intake to be considered adequate it must be capable of supporting normal mental and physical growth in children; desirable levels of physical activity; satisfactory performance in reproduction and lactation; adequate defence against infections, and recovery after episodes of disease or transient periods of food shortage. Long and complex field and intervention studies will be needed to assess requirements by these criteria.
Metabolic studies are, in general, confined to more limited objectives. They have often been criticized because their conditions are highly artificial, but against this one has to balance the greater precision that can be obtained. Furthermore, metabolic studies can yield valuable information on the basic processes and mechanisms of the interaction between energy intake and protein requirement and on the factors associated with adaptation to low or high intakes. With careful control the conditions of a metabolic study can be made less artificial and can, to a certain extent, reproduce natural situations.
1.1.3. The Relevance of Existing Estimates to Conditions in Developing Countries
As mentioned above, much of our information on protein requirements and on energy-protein relationships is derived from studies on healthy young men in the United States. The results may be different if similar measurements are made in developing countries on subjects with habitually low energy and protein intakes. There is already some evidence that such people utilize dietary protein more efficiently than subjects accustomed to higher intakes (7). If this is confirmed, the further question arises, whether the increased efficiency should be regarded as a result of depletion or of adaptation, without an adverse effect on health and function.
The conclusion that these subjects are depleted rests on the fact that, when given generous amounts of protein, they retained nitrogen and gained weight. However, it has been repeatedly observed that even well-fed subjects appear to retain nitrogen over long periods when their protein intakes are greater than their requirements (8). This phenomenon, which remains unexplained, emphasizes the inadequacy of nitrogen balance as the sole criterion for the evaluation of protein requirements. The term "depletion" implies a pathological state, and it is not justifiable to apply it unless it can be shown that the adaptation has occurred at a cost in terms of functional capacity.
People in developing countries not only have habitually low intakes of protein and energy but. as all previous committees have noted, they are exposed to repeated infections that will increase their requirements. From recent community studies in Central America (91 and Africa (10) relating growth of children to the frequency of infection, it is possible to make some estimates of the increased amounts of protein and energy needed to restore normal weight. The general conclusion is that, although there is an increase in the protein requirement, the major limiting factor, at least in the case of children, is likely to be the total food intake. It is probable that the same considerations will apply to adults, although perhaps in lesser degree.
These matters have obvious implications for the design of metabolic studies. If the results are to be relevant to conditions in developing countries, is it logical, as is conventionally done, to select experimental subjects who are well nourished and healthy? The general strategy agreed upon by the group was that, in order to examine the interrelationships of energy and protein requirements, in the first phase of the programme, subjects should as far as possible be free from disease and not manifestly malnourished (i.e. normal weight for height). It was hoped that in a second phase it might be possible to examine the influence of these variables, to study the factors associated with adaptation to low intakes, and to develop functional indicators of nutritional status. The final phase will be to test the conclusions reached under conditions prevailing in the community. This will require an epidemiological approach, and without it the interpretation and relevance of metabolic studies will always be open to question.
One original aim of the meeting was to produce a standardized experimental protocol that could be followed by all the participating institutions, and perhaps by other groups as well. However, it became apparent that standardization is neither possible nor desirable. Some latitude must be allowed in the choice of subjects, according to the particular interest of the group, and in the choice of experimental design, e.g. long-term or short-term studies. We have therefore attempted no more than to define the priority questions, to examine the special problems of various age and sex groups, to discuss the pros and cons of different experimental approaches, and to lay down some guidelines on methods.
The priority questions for examination in metabolic studies were defined as follows:
1. Is the present "safe level" of protein intake adequate (a) at habitual levels of energy intake; (b) at levels of energy intake that correspond with present estimates of energy requirements?
2. Will habitual protein intakes be adequate if energy intakes correspond to present estimates of energy requirements?
The available evidence suggests that, in developing countries, the habitual protein intakes of adults are in general greater than the FAD/WHO safe intakes, and that the limiting factor is the energy intake. The results of surveys summarized by Waterlow and Rutishauser (11) suggest that the same is probably true for pre-school children, at least in most cases, although the margin of safety may be less. Therefore, it was agreed that the most appropriate experimental design is to fix the protein intake at either the safe or the habitual level, according to the particular facet of the problem being tackled, and to vary the energy intake. In all the studies the diets should be made up of the foods habitually consumed in the region. When studies are being made at the safe level of protein intake, the level should be that given in the 1973 FAO/WHO report (1), and the correction for protein quality should be calculated in the way recommended in that report as modified by the joint FAD/WHO Informal Gathering of Experts (2). Habitual intakes of protein and energy should be determined wherever possible from previous measurements of the intakes of those subjects who are going to take part in the studies.
However, it must be accepted that it will be extremely difficult to obtain a realistic estimate of an individual's habitual intake, even with accurate dietary measurements, because the intake may vary at different seasons. Similarly, because of the wide range of individual variation it is impossible to predict the energy requirement of an individual subject. Therefore, the best approach that can be used is to measure nitrogen balance at a range of energy intakes, within which it is reasonable to expect that an individual subject's habitual intake and requirement will lie. The choice of the range will depend upon the circumstances. If the previous intakes of individuals cannot be measured, the best guideline will be the typical or average intake of the age-sex group to be examined. If this typical intake is, for example. 90 per cent of the estimated average requirement for such subjects (FAO/WHO. 1973 (1)), then tests might be made at energy intakes 80, 90, and 100 per cent of the estimated average requirement. In other situations it might be useful to have either higher or lower levels of energy intake.
This approach will contribute to answering all the questions posed above. In this way we hope to find out, for a range of individual subjects, what is the energy intake at which the safe and the habitual levels of protein intake are adequate, the criterion of adequacy being N balance, or, in the case of children, satisfactory N retention and growth.
Other priority questions are:
3. What are the factors associated with adjustment to low energy intakes?
4. Are habitual protein and energy intakes adequate to meet requirements under conditions of high prevalence of infections?
Both questions are extremely important, and should be tackled in a second phase, in the light of the answers obtained to priority questions 1 and 2. For question 4, a design may be needed that involves an open community or a convalescent home rather than a conventional metabolic ward.
Each age and sex group presents special problems. The groups that have been most intensively studied up to now are adult men and young children, but this does not mean that no further work on them is needed.
3.1. Infants and Young Children
The studies of Fomon (12) on infants up to nine months old fed breast-milk ad lib from a bottle have provided reliable information on protein and energy intakes which are adequate for normal growth over this age range. However, when infants are fed at the breast with no supplements, the volume of milk is likely to become limiting by about four months, particularly when lactating mothers are themselves undernourished (13). The problem of measuring milk output is a very difficult one; several teams are working on it and therefore it was not considered at this meeting.
Although there have been fairly numerous balance studies on young children from six months to two years old, not all the questions have been answered. According to the estimates of FAD/WHO (1), which take account of data from balance studies, the safe protein intake at one year, in terms of milk, is 1.27 9 per kg. Some workers (e.g. Graham, unpublished data) regard this as unacceptably low. More studies are needed on whether the safe level of intake. with appropriate adjustment for NPU, is adequate when the protein is derived from local sources and not solely from milk. The pioneer work of Hansen (14), De Maeyer and Vanderborght (15), and Bressani and colleagues (16) covered only some of the wide range of weaning diets found in different countries. We also need more information about the range of energy requirements of infants, and of older pre-school children who are becoming active physically. The interrelationships of nitrogen balance, energy intake, and physical activity have been very little studied in children. Recent work at INCAP suggests that physical activity enhances growth of the lean body mass (Town and Viteri, unpublished data). In many countries it is likely that the energy intake of the preschool child is. Iimiting because of the low caloric density of the food. Better information about energy requirements will focus attention on the need for new approaches to this problem.
3.2. Older Children, Adolescents, and Adult Women
Less work has been done on the protein requirements of older children, adolescents. and adult women. For this reason the 1973 FAD/WHO Committee (1) found it necessary to compute their requirements indirectly, on the assumption that the obligatory nitrogen loss per basal kcal is the same in all age and sex groups. This procedure must be regarded as an interim measure until verified by direct measurements.
It is of great importance for the child that the mother has adequate intakes of protein and energy during pregnancy and lactation. Women in developing countries do not gain as much weight during pregnancy as their counterparts in Europe and North America. Studies in the United States (17) suggest that the FAO/WHO (1) estimate of the additional requirement for protein during the last trimester of pregnancy may not be adequate. This subject needs further study in developing countries. The protein supplement for lactation recommended by FAD/WHO (1) is 16 9 per day, as milk or egg. It is likely that many lactating women in developing countries do not receive this extra amount, and they must therefore be producing milk at the expense of their own stores unless their habitual protein intake exceeds their normal physiological needs by this amount. More information is needed about the relationship of protein and energy intakes to the volume and duration of milk production.
3.3. Adult Men and the Elderly
Since the 1973 FAO/WHO report (1), a considerable number of papers have been published on the protein requirements of adult men, measured by nitrogen balance. The greater part of this work has been done in the United States (4, 18,19, 20), though important studies have also been made in Japan (3), Taiwan (21), India (22), and Nigeria (7, 23). As mentioned in the Introduction, some of the experiments carried out in the United States (4, 19, 20) show that the FAO/WHO safe intakes of egg protein for adults were not adequate for nitrogen balance either in the short or the long term, even at energy intakes that were considered to be unduly high. These studies also cast doubt on conventional estimates of protein quality. The key question is the extent to which this conclusion can be applied to people who are accustomed to less generous intakes. Therefore, further studies on adult men in developing countries are undoubtedly needed.
A final group is the elderly, who are becoming more numerous as life expectancy increases. Studies in the United States (24, 25) have shown that the protein requirement of old people per kg body weight is the same as that of younger adults, at least when determined according to the procedure adopted by FAO/WHO (1). Again, these results need to be substantiated in developing countries.
This short review shows that all age and sex groups present special problems. It was agreed that, in the first phase of the programme, priority should be given to studies on adult men and on young children, for the very reason that information is already available from these age groups that can be used for comparison.
The knowledge so gained on protein-energy interactions can then be applied to the design of studies on those groups who have not been adequately investigated, particularly adult women, non-pregnant, pregnant, and lactating.
4.1. Cross-sectional versus Longitudinal Studies
Metabolic studies can be cross-sectional or longitudinal. Some questions can be answered by either approach and others only by one of them. Cross-sectional experiments usually allow the study of a larger number of subjects. However, there is so much individual variability in factors that can affect protein and energy requirements and interactions, such as gastro-intestinal function, physical activity. metabolic rate, and so on, that if different treatments are being compared in different groups of subjects, the groups must be very large or very homogeneous, or the changes that are expected to result from the experimental treatment must be greater than the known variability.
Longitudinal metabolic studies can control inter-individual variability better by analysing the changes caused by the treatments within each individual. This approach is to be preferred whenever it is feasible.
4.2. Short- Versus Longer-Term Studies
The longer the duration of the study. the more variables can be tested, or the more replicate measurements can be obtained in the same subject to allow for intraindividual variability. Also, the longer the testing periods during which conditions are maintained the same, the more practicable it becomes to use indicators such as growth of children, physical fitness, physical activity pattern, or body composition. However, long-term metabolic studies may be limited in some instances by the availability and co-operation of experimental subjects, prevailing family and social structures, and the technical and physical facilities of the metabolic unit or the investigative group.
Other important differences between short- and long-term metabolic experiments relate to the number of subjects studied and the replication of measurements to attain higher precision. When time and resources are limited, the longer the testing period, the smaller the number of subjects that can be studied, although the results will be more precise.
The decision on whether to follow a short or a long protocol must rest on practical considerations, such as whether the particular question being examined can be better answered by a small number of very precise observations with functional indicators, or by a large number of observations that allow less precision, probably with nitrogen retention as the main indicator. For example, if the aim is to assess the effect of varying energy intake on nitrogen retention, the protein intake being fixed, it may be appropriate to choose short studies that, when repeated with different individuals over a given period of time, will cover a large number of subjects. If, however, the aim is to assess the effect of varying energy intakes on body composition or on rates of growth, a longer experimental design will be needed to evaluate changes that occur slowly, or that must be measured repeatedly to account for short-term variability.
4.3. Suggested Guidelines
As mentioned in the Introduction, the Group did not consider it practicable to propose a standardized protocol. However, there are certain guidelines that should be followed in so far as possible, whatever the experimental design.
4.3.1. Sample Size
Statistical or simulation techniques (26, 27) should be used whenever possible to determine the size of sample needed to produce results that can be interpreted. If this is not possible, the sample size should be selected on the basis of the available evidence on intra- and interindividual variability.
4.3.2. Number of Treatments
The number of test periods will depend on the experimental design. If the data are to be evaluated by regression analysis (e.g. Ievel of energy intake (x) versus nitrogen retention (y) ), there should be at least three treatments. If the experiment is designed to compare habitual with currently recomended intakes, two treatments only may be enough.
4.3.3. Randomization of Treatments
Where applicable, it is preferable to randomize the order in which the experimental treatments are applied. A Latin square design is convenient where there are doubts about the effect that the preceding treatment may have had on the treatment being tested. It must be remembered, however, that in this design each treatment must also be preceded by itself.
Limitations of subjects, time, and cost may not allow a Latin square design. An alternative is for half the subjects to follow an "ascending" sequence, in which the amount of the independent variable is gradually increased, and the other half a "descending" sequence.
4.3.4. Duration of Treatments
Each treatment or test period should be long enough to evaluate the indicators (dependent variables) used. When changes occur slowly (e.g. Iongitudinal growth of children), or there is a large day-to-day variability (e.g. changes in body weight), the treatment period must be long enough to allow detection of changes beyond the lower limit of precision of the measuring techniques or greater than the daily variability.
4.3.5. Replication of Measurements
This is highly desirable, but most variables can be measured repeatedly only in relatively long-term experiments. When the duration of each treatment is long enough, replicates should be obtained of those indicators that are most variable or that require greater precision. If the number of experimental subjects is large enough, as may be the case in shortterm studies, replication of measurements is not so critical.
4.3.6. Independent Variables
The simplest experimental design is to test only one independent variable at a time. In principle it would be economical if efforts were made to use a more complex design that permits simultaneous study of several interacting variables. Methods for doing this should be examined. It must also be borne in mind that some apparently independent variables may have significant interactions.
A great many factors may influence the results of metabolic studies, apart from the particular variable being examined. Some of these factors cannot be controlled, and must therefore be identified and described. Others can be controlled within limits or eliminated.
5.1. General/ Characteristics that Must be Described
These are: location of the study; climatic and demographic characteristics of the area; racial origin of the subjects; seasonal variations in food intake and in physical activity. It may also be useful, as part of the general background, to have information on prevailing infectious disease and incidence of morbidity in the population.
If, as recommended below, subjects are chosen whose nutritional and health status is as nearly normal as possible, it is clear that they will probably not be representative of the population as a whole. However, as discussed in section 3, they will provide a baseline for further work on subjects who are more representative.
5.1.1. Characteristics of the Subjects
As much information as possible should be obtained on the subjects' previous nutritional and health history. The age and sex of the subjects chosen will be determined by the interest of the investigator. With young children it is legitimate to combine the sexes. In studies on adolescent girls and adult women, account must be taken of the timing of test periods in relation to the menstrual cycle, or stage of pregnancy or lactation.
5.1.2. Nutritional Status
It was agreed that in the first phase of the programme (see section 1\ the subjects should be in a "normal" nutritional state, i.e. of normal weight for height. Western standards, particularly for adults, are biased by the inclusion of people who are overweight or even obese. There, a more appropriate definition may be the weight for height accepted as normal by the investigator under the conditions prevailing in his country. For children, a lower limit of what is acceptable might be 90 per cent of standard (Harvard) weight for height and an upper limit of 110 per cent. A separate question is the effect, if any, of the degree of stunting (deficit in height for age), which reflects past nutritional history.
5.1.3. Past Nutritional History
In many previous balance studies. e.g. at INCAP (16) and in Jamaica (32), children who have recovered from malnutrition and who have achieved expected weight for height (though not expected weight for age) have been regarded as nutritionally normal and used as controls. Work at the National Institute of Nutrition in India (28) has shown that such children do not have normal immunological responses. Opinion in the group was divided on whether or not such children can be regarded as nutritionally normal and acceptable for balance studies. It was considered preferable, if possible, to study children who are representative of the community. When children who are recovering from malnutrition are investigated, they should ideally have been within +10 per cent of expected weight for height for one month before they enter the metabolic study.
5.1.4. Health Status
In the first phase the subjects should be free from diseases that could affect the results of a metabolic study, e.g. tuberculosis, schistosomiasis. giardiasis, anaemia. It must be left to the judgement of the investigator to decide on the importance of minor infections and infestations such as skin sepsis, ascariasis, etc.. but detailed information on such matters should be given. Work on the metabolic effects of common infections can be planned as a second stage of the programme.
5.2. Conditions of the Study
Apart from the diet, the main factors that have to be controlled are physical activity and infection.
5.2.1. Physical Activity
This should be standardized at the level habitual for the subjects being studied. This is clearly essential in any investigation in which the independent variable is the energy intake, but it is also very difficult. particularly since habitual expenditure may vary according to season. One approach might be to replicate the measurements at two different levels of physical activity. representing roughly the upper and lower limits of normal activity. This. of course, complicates the experimental design. In the case of children, "freeliving" activity should be encouraged as far as possible. It will seldom be feasible to measure the level of activity, but at least it should be observed and described.
5.2.2. Infections
These alter nitrogen balance, and therefore must be eliminated in short-term studies. If even a minor infection occurs, the results are likely to be valueless. No further tests should be done until after an interval that will depend on its severity - perhaps ten days after a minor infection. By contrast, in longterm studies, in which replicate measurements are made on the same individuals, the occurrence of an infection may provide useful information. It may. in fact. be desirable to allow infections to occur naturally in order to reduce the artificiality of the conditions of study. Experiments of this kind would form a bridge between what we have described as phases 1 and 2 of the programme (see section 1).
5.3. Diets
As emphasized in previous sections, the diets in these metabolic studies should be based on the foods habitually consumed. The method of assessing the diet must depend on the judgement of the investigator and the resources available to him. It must be clearly documented. since it may affect the conclusions drawn. A cross-sectional survey at one point of time may give a different picture from repeated studies at different times of the year. Because meal patterns may influence the utilization of protein and energy, the number and timing of meals should, as far as possible, be that to which the subjects under study are accustomed.
The diet should provide adequate amounts of all nutrients other than the variables being studied. It may, therefore, be necessary to add vitamins and minerals.
The definition of safe and habitual levels of protein intake. derived from foods naturally consumed, was discussed in section 2.
The priority questions (see section 2) require the protein intake to be fixed and the energy intake to be varied. This poses great difficulties of experimental design, since the habitual intakes of individual subjects are likely to vary over quite a wide range. The best that can be recommended is that, if the average habitual energy intake is. say 80 per cent of the estimated average requirement, tests might be done at 80, 90, and 100 per cent of the requirement level. Some modification of the habitual diet will be necessary to provide additional energy without extra protein, e.g. by adding sugar or oil, but the test diets should, as far as is possible, contain the same sources of energy (carbohydrates and fat) as in the habitual diet.
5.4. Protein Quality
It is important to have a common reference point with which to compare the protein quality of the diet used in a given study. This will allow the results of the investigations to be applied to populations that have different dietary protein sources.
The preferred way of assessing the protein quality of the diet is by direct comparison with a reference protein (milk or whole egg) tested under the same conditions, either simultaneously in a control group or at a different time in a similar group of subjects with the same experimental design.
If this is not possible, the amino acid score should be determined as recommended by the FAO/WHO Joint Expert Committee (1).
Since the protein to be given is derived from indigenous diets that may vary widely in digestibility, the adjustment suggested by the 1975 Joint FAD/WHO Informal Gathering of Experts is recommended (2). This gives an approximate figure for net protein utilization based on correcting the amino acid score of the diet by its digestibility, as follows:
approximate NPU = amino acid score x (% digestibility)/100
The digestibility is automatically determined in the course of the balance study.
The amino acid score of the diet should be calculated from actual determinations of its amino acid pattern. It is recommended that a reference laboratory check the amino acid analysis or provide analytical services to investigators who lack the technical facilities to perform them.
It is desirable that the protein digestibility and NPU (standard) of the diet be measured by rat assay, to determine how far estimates of NPU derived from human and rat data are comparable. This should also be done in a central reference laboratory if local facilities are not available.