Previous - Next
Abstract
1. Essential terminology and concepts1.1. Energy requirements
1.2. Recommended dietary allowance (RDA)2. Factorial approach to energy requirement
3. The breast-fed baby as a model for energy requirements
4. The doubly-labelled water method
5. Application of the doubly-labelled water method to estimate energy requirement
6. Validation studies employing doubly-labelled water6.1. Energy expenditure and milk intake in fast-growing preterm infants
6.2. Validation of dose-to-the-baby method for measuring milk intake
6.3. Determination of milk energy content using the doubly-labelled water method8. How logical is the current approach to estimating energy requirements?
9. Future directions
References
A. LUCAS *
* MRC Dunn Nutrition Unit, Downham's
Lane, Milton Road, Cambridge CB4 1XJ, U.K.
Recommended dietary allowances for energy intake in populations of infants and children are based on estimates of physiological requirements in individuals. Technical, physiological and conceptual problems in deriving such requirements are reviewed. Previous estimates were hampered by unsatisfactory methods. The 'factorial' approach of estimating separate components of energy requirement including the energy cost of growing to an internationally agreed standard, results in accumulated and uncertain errors. Alternatively, energy intake in healthy breast-fed babies has been assumed to reflect physiological requirements. Yet, intra-feed, diurnal and longitudinal changes in breast-milk energy make representative milk sampling difficult; and special sampling devices (nipple-shield system) interfere with normal lactation.
The doubly-labelled water method constitutes a major recent advance, permitting total energy expenditure, metabolisable energy intake (MEI), milk volume intake and energy density of breast milk (MEI/milk volume) to be estimated non-invasively in free-living subjects without recourse to milk sampling. For example, in 20 healthy 4- to 6-week-old breast-fed infants, energy intake was 88 kcal/kg compared with 116 recommended by FAO/WHO/UNU (1985) using previous information. These data reflect new, low estimates of breast-milk energy close to 60 kcal/100 mL.
Ideally, energy 'requirement' reflects physiological needs for optimal health. Current estimated requirements may be debated. Growth rate is subject to secular trends and the optimal rate is unknown. Moreover, energy expenditure may itself be determined by energy intake. Thus, it could be shown with doubly-labelled water that 30 formula-fed infants consumed 16% more energy than 20 infants fed by breast, yet expended 80% of this excess. Most important, energy requirements have never been based on outcome data. The possibility that early energy intake could have long-term consequences is considered. There is a need for formal long-term clinical outcome studies to estimate optimal energy intakes in early life.
For 40 years, multidisciplinary, national and international committees have made recommendations on the needs for energy in healthy individuals. In infants and children, these recommendations have three principal applications: firstly, to provide guidance to infant-food manufacturers on the composition of breast-milk substitutes suitable for use as a sole source of nutrition. Secondly, international recommendations have been used for medical and political purposes in assessing needs of infants and children at risk of malnutrition, for instance in developing countries. Finally, a knowledge of energy requirement at each age is essential information for deriving the special nutrient needs of sick babies and children. The latter area has received increasing interest with recognition of the potential importance of attending carefully to nutrition in a wide variety of clinical states including prematurity, malignancy, renal disease, cystic fibrosis, heart failure, cerebral palsy and so forth.
Whilst there is clearly a need to
provide well-founded recommendations for dietary energy, there have been major technical,
physiological and conceptual problems in doing so. The purpose of this article is to
pinpoint some of these problems, highlight recent advances and consider future research
objectives.
1.1. Energy requirements
1.2. Recommended dietary allowance (RDA)
Essential terminology has been
loosely and inconsistently applied. WATERLOW (1989) emphasised the important distinction
between a requirement and a recommended dietary allowance (RDA). Yet, attempts to define
these concepts precisely reveal the reasons for inconsistency.
Ideally, requirements may be defined as an individual's physiological needs (in this case, for energy) to maintain optimal health and function. Unfortunately, available data do not permit definition of energy intake in childhood in terms of optimal health. We therefore have had to adopt more pragmatic approaches. For example, energy requirement may be defined as the individual's physiological needs for each component of energy expenditure (including basal metabolism, activity and thermogenesis) and for growth at an internationally agreed 'standard' rate (NCHS Standard; HAMILL et al., 1977). This is not an entirely satisfactory definition, since neither optimal activity nor growth rate has been adequately related to health status. In this instance, then, the term 'requirement' would denote an internally consistent concept; if a level of activity and growth pattern are agreed upon, then the requirement is an energy intake that will allow these to be fuelled. This rather limited approach to the definition of a requirement is the one most commonly adopted for older children. However, in early infancy at least, another approach is possible. It could be argued that after 200 million years of mammalian evolution, intake of energy during exclusive breast-feeding had been naturally selected to be optimal and would therefore represent the infants' requirement. This argument is not entirely compelling. Patterns of breast-feeding practice have been heavily influenced by cultural background and medical opinion. Furthermore, we may presume that breast-feeding evolved as a compromise to meet the needs of mother and infant at a time when food supplies for the mother might be limiting. Our circumstances and biological targets have changed since that time, and without outcome data it is only an assumption that breast milk provides optimal energy intakes. Nevertheless, it is at least an assumption which is devoid of any value judgements on such factors as desirable rates and quality of growth. Arguably, however, beyond 3 to 4 months of age, the breast-fed baby can no longer be taken as a realistic standard, since a high proportion of mothers will have introduced weaning foods; and weaning practices, certainly in the industrialized countries, far more reflect current fashion than any biologically programmed event. Despite this, food intake data have been used as a measure of requirement throughout infancy, and indeed FAO/WHO/UNU (1985) used this approach up to age 10.
In summary, three ways of defining energy requirement operationally have been used:
1. calculation and summation of needs for components of energy expenditure (factorial approach)
2. to match energy intake
3. to support optimal health and function.
The RDA represents an attempt to
generalise the available physiological data on requirements and apply them to populations.
Healthy people, however, vary considerably in their requirements. Furthermore, in any
individual, growth rate is not constant, as illustrated in a case study of a healthy child
described by WHITEHEAD (United Nations University, 1979). In this study, measurements of
body weight were made at approximately 6-week intervals. Weight-gain velocity during any
one interval varied between -4 to +4 times the Boston 50th centile increments, yet average
increment over the whole period was on the 50th centile. Thus, energy needs are subject to
wide intra-individual as well as inter-individual variation. It is the average requirement
that has traditionally been adopted as the energy RDA. Defined in this way, a healthy
individual at any one time may have an energy intake that is far from the RDA. Thus, the
average requirement for energy (RDA) is a potentially useful population measure which can
only be used as an approximate guide for individuals. The value chosen for the RDA,
however, will depend on which approach, of those above, is chosen to define and determine
energy requirements in individuals. These approaches are now discussed.
In the factorial approach, individual components of energy expenditure, including the energy costs of tissue synthesis and energy storage, are summated to derive a requirement. The relevant equations are:
Equation 1: Metabolisable energy (ME) = gross energy intake gross energy loss (in stools + intestinal gases + urine)
where gross energy is the heat released after complete combustion. The fate of metabolisable energy is as follows:
Equation 2: Metabolisable energy (ME) = E basal + E thermogenesis + E activity + E tissue synthesis + E stored
where E basal (basal metabolic rate or basal energy expenditure) is the energy expenditure whilst awake but completely at rest, shortly after waking, in the thermoneutral state and in the postabsorptive state 12-14 hours after a meal; E thermogenesis includes the energy costs of maintaining body temperature and of diet-induced thermogenesis; E activity is the energy cost of physical activity; E synthesis is the energy cost of synthesizing new tissue during growth; and E stored is the energy content of the new tissue itself.
Clearly, basal metabolic rate (BMR), a highly standardised measurement, cannot be obtained in babies and young children. An alternative is resting energy expenditure (REE), sometimes called resting metabolic rate (RMR). Resting energy expenditure, even if measured a few hours after the last meal, will include some (unknown) component of the thermic effect of food. Sleeping energy expenditure (SEE) is probably closer to basal metabolic rate since the thermic effect of the previous meal may be offset by the lower energy expenditure found during sleep. Both resting and sleeping energy expenditure may include the energy cost of maintaining body temperature, especially in view of the difficulty in providing a thermoneutral environment during indirect calorimetry using a ventilated hood. In very early infancy, when the infant has always been recently fed, resting or sleeping metabolic rate will include a significant component of the energy cost of growth. Indeed, in rapidly growing neonates the cost is considerable; 30% of energy intake is stored in new tissue (LUCAS et al., unpublished).
Theoretically, a period of indirect calorimetry in the non-active state would measure simultaneously: E basal + E thermogenesis + E synthesis + E growth. These can be termed, collectively, 'maintenance energy expenditure' (a concept more meaningfully applied to non-growing, sedentary adults). Thus:
Equation 3: Metabolisable energy = E maintenance + E activity
Usually, indirect calorimetry can only be used to measure metabolic rate in children for short periods. Given diurnal variation in diet-induced thermogenesis and new tissue synthesis (ASHWORTH, 1969), it is unlikely that a representative measure of maintenance energy expenditure could be obtained. However, when energy expenditure is measured in sleep and as long as possible after the last feed, which is the most practical time for studies in infants and small children, energy utilisation for growth is likely to be minimal since new tissue synthesis has been shown to occur in bursts after meals (ASHWORTH, 1969). It is apparent that, aside from technical problems of making measurements of energy metabolism in children which will be discussed later, there are problems in deciding what to measure.
In premature babies it has been possible to estimate components of the equations above with reasonable confidence since these infants spend their lives in the fixed, thermally-controlled environment of the incubator, which facilitates balance studies, intensive physiological monitoring and relatively long-term indirect calorimetry. In full-term infants and children living in the community, the problems of providing accurate estimates of individual components of energy metabolism are legion.
Consider Equation 1. The RDA for energy is usually presented as metabolisable energy, i.e., a correction has been applied to adjust gross energy for energy losses in stools and urine assessed by combustion. Conversion factors have been derived in balance studies on adults (SOUTHGATE and DURNIN, 1970) and in babies (SOUTHGATE and BARRETT, 1966). However, given the difficulty of performing balance studies in healthy infants and young children in the community, there are insufficient data on the metabolisable energy of foods consumed in early life and on how their digestibility is influenced by postnatal age and the nature of the diet. Studies in premature babies, in whom balance data are more easily obtained, indicate that some factors have a profound influence on metabolisable energy. In one study (WILLIAMSON et al., 1978), pasteurising breast milk, which destroys milk lipase, resulted in a fall to 45% when the milk was boiled. The appropriateness of currently employed Atwater factors continues to be debated, though stable isotope probes now open new possibilities for reinvestigation, as will be discussed later.
In Table 1 components of energy requirement estimated for babies in the first 6 months are listed (after WATERLOW, 1989). Yet estimates of each of the five components in this Table could be challenged as follows:
1. Basal metabolic rate, as discussed, cannot be measured in babies and young children. Sleeping or resting metabolic rate may include other components of energy expenditure.
2. Physical activity cannot be quantified accurately in energetic terms in early life. Heart rate and activity monitoring, for instance, are tests with uncertain error. Experts have often resorted to guesswork. Thus, WATERLOW (1989) states: "As a guess, since there is no hard information, and by analogy with adults, we might take the cost of activity at 3 months as +0.2 BMR for 12 hours out of 24, and at 6 months +0.4 BMR for 12 hours in the day".
3. The thermic effect of food is difficult to evaluate in infants since part of this thermogenesis is the energy cost of tissue deposition, which occurs maximally in the postprandial phase (ASHWORTH, 1969). Thus, BROOKE and ASHWORTH (1972) noted that in infants and young children recovering from malnutrition the rate of weight gain and postprandial thermogenesis were linearly related. To avoid this problem, it is often assumed, without confirmation, that data from non-growing adults apply to babies and children.
4. The energy cost of growth (synthesis + deposition) cannot be calculated from any 'desired' rate of weight gain; it is essential to decide on an acceptable standard for body composition since the energy cost of weight gain will depend on the proportions of fat, protein and water. Whilst it is relatively easy to measure growth, measurements of body composition are more difficult to make. Our studies show that simple anthropometric techniques that have been useful in adults and older children, such as body mass index (weight/height2) and skinfold thicknesses, do not predict body fat in infancy (DAVIES and LUCAS, 1989 and unpublished). A number of inventive procedures have now been employed including: deuterium and 18O dilution, whole-body 40K counting, body density by gas displacement, bioelectrical impedance, insonation techniques, ultrasound investigation of fat depots and, more recently, magnetic resonance imaging. A problem with many of these methods is that accuracy is insufficient for the purpose. Thus, a relatively small measurement error becomes critically important when small changes in body composition over time need to be quantified. Nevertheless, population data based on more than one method are likely to give satisfactory answers. But we are still left with the problem of deciding on what body composition is desirable.
5. Finally, estimated energy losses in the stool, as discussed, have been based on insufficient data.
Table 1. Factorial estimates of energy requirements in normal infants (values in kcal/kg/d). After WATERLOW (1989)
Age (months) |
|||
1 |
3 |
6 |
|
1. Basal metabolic rate |
48 |
52 |
54 |
2. Physical activity |
2.4 |
5.2 |
11 |
3. Thermic effect of food |
5.0 |
4.5 |
4.5 |
4. Growth |
39 |
21 |
11 |
5. Losses |
5.0 |
4.5 |
4.5 |
Total estimated requirements |
99 |
87 |
84 |
Clearly, using the factorial
approach, the potential accumulated errors for total estimated energy requirement could be
very large and, indeed, unquantifiable.
Whilst we may choose the breast-fed infant as a model for energy requirement in early life, it has been notoriously difficult to define the breast-fed infant's intake of some nutrients, especially those which vary in their concentration in milk during the course of a feed. The measurement of milk fat, and therefore energy intake, has been particularly fraught with methodological difficulties.
Traditionally, energy consumed by breast-fed infants has been assessed by matching data on human milk composition, obtained by manual or mechanical expression of the breast, with daily milk intake measured by test-weighing or the use of stable isotope tracers (e.g., deuterium oxide) given to the baby (COWARD et al., 1979; LUCAS et al., 1987a) or mother (COWARD et al., 1982).
The technique chosen for the estimation of milk volume intake requires careful consideration. Test-weighing is clearly unsuitable for mothers who feed frequently. In some cultures, breast-feeding may take place on up to 20 occasions per day and may be virtually continuous over some periods. Test-weighing under these circumstances is socially and biologically invasive, and unidirectional weighing errors at each feed can result in serious cumulative error. In the industrialized countries, with infrequent feeds and with mothers who are motivated to do their own test-weighing, the technique can be satisfactory. Some concerns have been expressed over the 'dose-to-the-baby' isotope-kinetic method, though, as discussed below, satisfactory validation has been achieved in formula-fed babies (LUCAS et al., 1987a).
All available methods for measuring milk volume intake are subject to errors, but these are likely to be small if a method appropriate to the experimental circumstances is selected (COWARD, 1984). The principal difficulty, however, has been one of obtaining representative breast-milk samples. Two problems exist. First, milk fat, a major determinant of energy content, varies considerably among individuals and changes diurnally and throughout lactation. Moreover, both milk fat concentration and milk flow rate may change continuously during the course of a feed from each breast (LUCAS, LUCAS and BAUM, 1980). Therefore it is difficult to calculate energy intake from isolated samples of 'foremilk' or 'hindmilk'. Second, milk that has been obtained by manual or mechanical expression (EBM) of the whole breast may not have the same composition as milk obtained during normal suckling (so-called suckled breast milk (SBM); see below). For example, since fat concentration rises as milk is ejected, energy content may be overestimated if more milk is expressed than the infant would have consumed spontaneously. A possible mismatch between EBM and SBM may be especially marked for the second breast, if the infant normally takes a relatively small volume, leaving a significant residue of high-fat hindmilk that could be removed by expression. Thus, the energy content of EBM obtained by complete expression of the first breast, as reported by the DHSS (1977), may not reflect the average energy content of SBM from both breasts. Entire 24-hour collections of EBM may be analysed, but this approach is subject to the objection that during such a prolonged 'experimental' period, the normal biology of lactation (e.g., the hormonal milieu) could be altered, perhaps affecting milk flow or even composition; in addition, this procedure is generally unacceptably invasive. Other techniques have included foremilk and hindmilk sampling (PRENTICE et al., 1986; SAINT, MAGGIORE and HARTMANN, 1986) or midmilk sampling (NEVILLE et al., 1984) and analytic data have been linked to an equation to derive average milk fat. However, in our experience the pattern of changes in milk fat and flow rate are very variable between individuals, making such techniques difficult to validate.
In an attempt to resolve some of these difficulties, SBM has been studied directly (LUCAS, LUCAS and BAUM, 1980). A thin nipple shield, worn during a feed, can be used to collect sequential SBM samples. For population studies, changes in milk composition during a feed can be matched with cross-sectional data on the pattern of milk flow from mother to infant, estimated by a test-weighing protocol that avoids repeated interruption of the feed. Using this approach, LUCAS and colleagues (1980) showed that, early in lactation, SBM had a lower fat and energy content than had been reported previously in EBM. These workers also studied SBM in individual mother-infant pairs: a microminiaturized ultrasonic flowmeter was incorporated into the tip of the nipple shield (How et al., 1979) allowing changes in milk flow and composition to be measured simultaneously throughout the feed. Although these methods have considerable potential for investigating the physiology of lactation, they impede milk flow and may interfere with normal feeding behaviour due to the presence of a device and an investigator.
The advent of the doubly-labelled
water method has made it possible to study energy expenditure, energy intake and the
energy content of human milk without recourse to milk sampling. The impact of this
technology is discussed in the following sections.
The doubly-labelled water method was originally developed by LIFSON and coworkers (1955) for use in small mammals and was later applied to man (SCHOELLER and VAN SANTEN, 1982; COWARD et al., 1984; KLEIN et al., 1984). The principle of the method is that two stable isotopes of water (H218O and 2H2O) are administered simultaneously, and their initial enrichment and subsequent disappearance rates in a body fluid (e.g., urine or saliva) are monitored by isotope ratio mass spectrometry. The initial increase in enrichment of either isotope reflects body-water pool size (from the principle of dilution) and permits an estimation of body composition.
Subsequently, the disappearance rate of 2H2O reflects water output (and hence water intake); that of 2H218O reflects water output plus CO2 production, because 18O is free to interchange between water and CO2 through the action of carbonic anhydrase (see Figure 1). The difference between the two disappearance rates is therefore a measure of CO2 production rate. With additional knowledge of the subject's respiratory quotient (measured or derived from previous studies on a similar population, or calculated from food intake), oxygen consumption and hence energy expenditure can be calculated using a standard equation (WEIR, 1949).
k = experimentally-determined rate constant;
r = production rate
In addition, the technique also
allows water intake and hence milk intake to be assessed, and total body-water pool size,
derived from the method, gives an estimate of body composition from which energy storage
can be calculated.
Difficulties described in relation both to the factorial method and the breast-milk intake model can be circumvented to a considerable extent by application of the doubly-labelled water method. It had not been possible previously to measure total energy expenditure in free-living subjects over prolonged representative periods (7-14 days). Calorimetry, which involved isolating the child from its parents was certainly not suitable for this purpose.
Equation 2, required for the factorial method, can now be simplified to:
Equation 4: Metabolisable energy = total energy expenditure + energy stored in new tissue
Since both energy expenditure and energy storage can be estimated using the doubly-labelled water method, an assessment can be made of the energy intake that is physiologically available. Theoretically, the method could be used to validate currently used Atwater factors, without the need for a balance study, as follows:
Equation 5: Energy losses = gross energy intake - metabolisable energy where gross energy intake is recorded accurately in an experimental setting and metabolisable energy is measured by the doubly-labelled water method.
Moreover, the doubly-labelled water method may be used to validate traditional methods of estimating food intake. Weighed food intake and dietary recall data have been shown to be unacceptably inaccurate in some studies on adults. If such methods are to be used for deriving requirements in children, as was the case in the FAO/WHO/UNU (1985) report, formal validation studies are urgently required.
With regard to the breast-fed infant model, it should be possible, in principle, to derive a value for the metabolisable energy content of breast milk from the following equations:
Equation 6: Metabolisable energy intake = energy expenditure a + energy stored b in new tissue
Equation 7: Breast milk energy content = metabolisable energy intake/milk volume intake c
Since a, b and c (energy
expenditure, energy stored, and milk volume intake) may be obtained from the
doubly-labelled water method (as indicated above), Equation 7 can be solved, thus
providing a value for breast-milk energy content without resort to milk sampling. (The
term breast-milk energy content used here and subsequently in this article, refers to
metabolisable, 'physiologic', energy content and not gross energy.)
6.1. Energy expenditure and milk intake in fast-growing preterm infants
6.2. Validation of dose-to-the-baby method for measuring milk intake
6.3. Determination of milk energy content using the doubly-labelled water method
The important question, in view of the foregoing arguments, is whether the measurements of energy expenditure, milk volume intake, and energy content of breast milk, can be made with sufficient precision in fast growing infants.