Contents - Previous - Next

Energy requirements: general principles

Energy expenditure as the basis for estimating energy requirements
Methodology
Importance of body composition
Points of uncertainty requiring further research
Summary of research needs
References
Discussion
References

JVGA Durnin
Department of Human Nutrition, Yorkhill Hospitals, Glasgow G3 8SJ, Scotland, UK

Descriptors: Human energy expenditure: methodology, basal metabolic rate, physical activity, body composition

Energy expenditure as the basis for estimating energy requirements

The energy requirement of an individual, in a state of desirable equilibrium, is equal to the energy expenditure. In some clinical situations, where an improvement in nutritional status may be advisable, the energy requirement may be set at a higher level than the energy expenditure in order to produce, temporarily, a positive energy balance. In certain physiological states, such as during growth in children, or in pregnancy and lactation, the energy requirement may also be higher than the energy expenditure. At the other extreme, when dealing with an obese individual or an obese population, again energy requirements would be derived from the energy expenditure, with a reduction to produce a negative energy balance; the amount and the duration of the energy imbalance would determine the rapidity and extent of the weight loss.

A more difficult situation to judge is where the energy requirements might be construed as being inadequate, because energy expenditure was less than desirable due to low levels of physical activity. In the absence of very clear-cut evidence specifically related to the health advantages of physical activity and the clinical dangers of inactivity which, for adults do not presently exist in an uncontroversial and entirely persuasive way - it is problematical to take this factor into account in calculating energy requirements. This is not to say that physical activity may not be important for physical, mental, and cognitive development and maintenance, particularly in children: it simply appears very difficult to introduce it in a quantitative way in the present context.

Methodology

Energy expenditure is therefore the key to the assessment of energy requirements. It may be measured by several different standard 'direct' techniques. 'Direct' in this context is not equivalent to the classical term 'direct calorimetry', which refers to the direct measurement of heat output in a calorimeter; in the present usage 'direct' refers to the measurement of energy expenditure from O₂ or CO₂ output. It therefore includes the various classical techniques of assessing O₂ consumption and CO₂ output, as well as the doubly-labelled water technique and the use of a whole body calorimeter.

'Indirect' methods of measuring energy expenditure comprise extrapolating from values of total energy intake in food and from heart-rate recording.

A description of the techniques, together with a brief analysis of some of the problems, is given in Durnin (1992).

1. Timed record of activities and associated energy costs

The method which has probably been used most frequently consists of a combination of a timed activity record, i.e. the average total duration of each of the 'important' activities throughout the whole 24h of the day, and an energy value (in kcal or kJ/min) for each of these activities; 'important' is defined as either occupying a significant period of time or else involving considerable physical effort. These energy values may be derived either from published data or by actual measurement of oxygen consumption.

When values from the published literature are used to apply to a specific activity, it is sometimes unclear as to whether or not 'ancillary activities' or rest pauses have been included. For example, while pushing a wheelbarrow, the worker may stop to load bricks or sand into the barrow; or during digging a ditch, the worker will have occasional rests of varying duration. It is thus uncertain whether the 'activity' which has been allocated an energy value represents that activity only, or also includes some short-term diversions.

If no precise information is provided about whether or not the activity is 'pure' or 'adulterated', it is a matter of guess-work to make the correct decision. On the whole, if the activity is spasmodic and of short-term duration perhaps of only a few minutes - it is probable that it is a 'pure' activity, e.g. planting rice, or digging a ditch. If it continues for perhaps half-an-hour or longer, it will almost certainly include some rest pauses, and will thus be 'adulterated'. In deciding what energy value to apply to an activity which has not been observed and timed (such as planting rice), if the duration of the activity is long enough to imply that it refers to a prolonged period of work, such as a whole day of 'planting rice', then the appropriate factor should be one that would include the influence of rest pauses. If the duration appears to be limited to the actual activity, excluding any pauses or diversions, the relevant factor will be higher.

Although there is a great deal of information on the energy cost of separate 'activities', it is somewhat undigested and its exact usefulness in relation to energy requirements is in a confused state. There is also a lack of clarity and logic about the relationship of actual physical activity and so-called desirable (or discretionary) activity; when people lead very inactive lives, is it sensible to consider only what they actually do, or should an allowance be made for a possibly healthier life style involving more physical activity ? This is one of the major problems to be decided in the context of energy requirements.

Also, most of the data (and indeed the outlook adopted in this paper) relate to healthy populations. The influence of malnutrition, disease, and disablement has been virtually ignored, and the importance of these conditions to energy requirements needs to be addressed in a comprehensive way, albeit at this stage in a preliminary fashion.

Useful methodological information. Even if a labour-intensive technique is used of apparently high validity, such as where activities are carefully categorized and their duration monitored, there is still the possibility of error of varying magnitude. To be able to assess the likely validity and precision of the method, certain descriptive information is needed. If a 'diary-record' is used to obtain detailed information on how the 24h of the average day is spent, it is necessary to know to what degree of exactitude this was done (e.g. in single minutes, in blocks of 5 to 10 min? etc.), and whether the diary was compiled by the subjects themselves or by observers. Who were the observers? How often was the accuracy of the diary checked, and by whom?

If indirect calorimetric measurements were made, were these done by a respirometer or using a Douglas bag, etc.? What was the duration of the measurement? Was the activity standardized or was it thought to be representative of the normal free-living situation? Some indication should also be given on the range of activities measured and the number of measurements on each activity, preferably in tabular form. How was the volume of the expired air measured? Were the instruments calibrated, and, if so, how? How was the sample of expired air collected for analysis? How long did it remain in the collecting container before analysis ? What instruments were used for analysis and how were they calibrated? If published or unpublished values of energy costs have been used, their source should be quoted and their adjustment for varying body size, etc. should be defined clearly.

This list of desirable bits of methodological information may appear somewhat excessive, but ought to have been collected in the first place, and the scientific worth of the experimental data would surely be increased by a brief description of these items.

Ancillary advantages of using this method. There are several subsidiary advantages to using this method with its attendant information on the duration and probable energy cost of the different daily activities. First, it provides a large amount of interesting physiological and social information on life styles.

Although it may be preferable actually to measure the energy expenditure of the important activities, there is no absolute necessity for this if values from the literature are used. There is therefore no need for sophisticated apparatus nor highly skilled technical assistance.

Secondly, it should allow a more accurate assessment of which overall value to attach to an activity factor that would be relevant to an extended period, perhaps at work or in leisure, or indeed over the whole 24 h day.

Thirdly, it provides information about the types and strenuousness of different forms of physical activity. Monitoring the duration and strenuousness of physical activity can also play an important role in detecting minor degrees of poor or under-nutrition, since reduced physical activity may be the first and possibly the major indication of an attempt at adaptation.

Available data on the energy cost of activities. The energy costs of many different activities were measured by Orr & Leitch (1938) and Durnin & Passmore (1955; updated 1967), and a short list, largely based on these earlier data, is given by James and Schofield (1990), but the information is still somewhat inadequate. Often, the number of individuals who supposedly supply the source of the data is ridiculously small e.g. the values for the energy cost of such simple household tasks as washing dishes, cleaning windows, and such leisure activities as playing bowls or playing golf, depend in each case in the James & Schofield (1990) tables of measurements made on only one individual. There is probably a fairly large reservoir of unpublished data of this nature, as well as some published results which have not been included in these tables. At the present time, an attempt is being made to collate them in a discriminating way, together with more information about how and where the data were obtained.

2. BMR multiplied by an appropriate activity factor

Average daily energy expenditure may also be estimated from the BMR, multiplied by an appropriate activity factor which will be dependent on the degree and duration of physical activity [BMR × A]. BMR may either be measured directly or else calculated from an equation. The activity factor may vary from about 1.2 to 1.4 for relatively inactive people, up to 2.0 or more in the case of people who are physically very active. The factor by which BMR must be multiplied may be gauged from information of different kinds; perhaps from a timed activity record, or from a questionnaire designed to provide information on habitual physical activity. (Heart-rate monitoring may also be used to calculate energy expenditure, and this will be discussed later.)

The most important constraint to the general technique of BMR × A is its basic dependence on (1) a valid value for the BMR and (2) the possibility to obtain a factor which may be applied to the BMR (perhaps 1.5 or 1.6) which will result in a calculated daily energy expenditure that is reasonably pertinent and accurate.

Variability of BMR data and validity of predictive equations. To obtain some indication of the reliability of the value obtained for BMR, we need to analyse the relevance and importance of the intra- and inter-individual variability. Measurements of BMR on an individual by indirect calorimetry, each lasting 10 to 15 min and done on two or three consecutive occasions immediately following on one another, should show a variation of no more than 2-3%. If BMR is measured on an individual on many different days over a period of weeks or months, the 'intra-individual variability' is remarkably constant with a coefficient of variation (C.V.) of about 3% (Benedict & Cathcart, 1913; Loewy & Zuntz, 1916; Lusk & Du Bois, 1924; Benedict, 1935; Berkson & Boothby, 1938; Soares & Shetty, 1987; Henry et al, 1989). The influence of measurement error, when the technique is carefully controlled, is likely to be minimal (less than 1 %) and of no practical relevance.

The variable which has probably more importance in the present context of assessing energy requirements in populations, is the inter-individual variability in BMR. This appears to have a coefficient of variation of the order of at least 8% (Harris & Benedict, 1921; Henry et al, 1989).

The extent of this C.V. is large enough to introduce all sorts of confusing sequelae. As an illustration of some practical implications, with a C.V. of 3% for intra-individual variability the 95% probability of the value for the BMR of a 70 kg man would be expected to have a variance of twice the C.V.; i.e., the values, expressed as kcal/d, or MJ, would have a range from 1590 to 1790 kcal (6.7-7.5 MJ) per day. That range is large enough to introduce uncertainty about making tenable conclusions of acceptable precision. Moreover, it refers to actually measured BMR, with the inherent probability that if BMR is derived from equations the range will be greater still. Clearly, we should be careful about making simplistic deductions about relatively small differences in BMR. When we are dealing with energy requirements, where we have to fit a hypothetically derived 'activity' factor to the BMR, the scope for error is increased, since the variability will be considerably augmented. If we are considering the inter-individual variability in a group of men whose mean body mass was 70 kg, a C.V. of 8% implies that the 95% confidence limits are from 1420 to 1960 kcal/d (6.0-8.2 MJ). A C.V. of 10% results in the range becoming 1350-2030 kcal/d (5.7-8.5 MJ).

Part of this C.V. is the result of having to make some allowance for differing body mass; a population might include adult individuals with a range of body mass from 40 to 80 kg, and various equations have been formulated to take this into account (FAO/WHO/UNU 1985; James & Schofield, 1990; DH, 1991). However, these equations assume that there is a normalisation effect produced by introducing a simple dependence of body mass into the equation, and also that the composition of the body with respect to the fat-free mass (FFM) and the fat mass (FM) has a minimal influence on energy metabolism. Neither of these assumptions is likely to be correct in all circumstances, so that an error of unknown dimension is introduced by this procedure, which might well increase the C.V. to at least 10%. (This possible error is discussed later.)

It is probably sensible to apportion most of the inter individual variability to varying body mass. Even if body composition is more or less identical in the group, a large difference in body mass will inevitably lead to a comparable variability in BMR. There are also other factors which will affect variability such as hormonal influences, perhaps especially in thyroid function; it is theoretically possible that differences in body temperature may modify energy metabolism; physical fitness is unlikely to have much relevance in this context. It is difficult to see how any allowance for these influences could be introduced, especially in field situations.

Available data on BMR. Since BMR has considerable importance in the calculation of energy requirements, and since in the great majority of cases BMR will be calculated using published data and not actually measured, in order to minimize error it is rather critical that there should be an adequate volume of BMR measurements, the use of which would allow predictive equations to be calculated. There is some doubt about whether or not this adequacy of BMR data exists. Durnin (FAO/UNU 1981) made the first comprehensive attempt to determine whether or not formulae might be derived using sex, age and body mass alone, to calculate BMR. Subsequently, Schofield et al (1985) expanded the base-line data and produced new equations. These have later been modified in a minor way (James & Schofield, 1990).

In spite of these large-scale surveys, a completely satisfactory analysis of the data has not yet been carried out. Although the C.V. due to measurement error is very small when the methodology is carefully and strictly controlled, the possibility of experimental error in measuring BMR is always present. The strict experimental requirements (10-12h post-prandial, completely relaxed, in a thermally-neutral environment, etc.) are sometimes difficult to organize, and the measurement of oxygen consumption (or energy expenditure) has not always been done with the required strict care and attention to procedure (e.g. calibration of equipment, suitable mouthpiece and valve, face-mask, or ventilated hood). Many of the published studies have not explained adequately exactly how the measurements were done, and doubt exists about the validity of some of the data. There is also an inadequate volume of information on BMR in many populations in developing countries, especially in children, adolescents and elderly people.

Derivation of the factor with which to multiply BMR. In making use of the BMR to calculate energy requirements, an overall factor denoting the physical activity level (PAL) to apply to the BMR is needed: i.e. energy expenditure or energy requirement equals

BMR × PAL.

This activity factor is usually applied only to groups or populations and not to individuals. An example might refer to a group of rural women undertaking daily work in the fields:

10 h of light agricultural work at BMR × 3.0.

Equally this approach could be used for the whole 24h of the day, for, as an example, a group of moderately active people:

BMR × 1.6.

The activity factor may also be applied to single 'activities' such as BMR × 1.2 for 'sitting quietly', BMR × 2.5 for 'household tasks', or BMR × 4.0 for 'walking', etc. This procedure makes it possible to calculate energy requirements for individuals if sufficient information is available on the life style.

The FAO/WHO/UNU (1985) Report gives values for three different levels of occupational activity: 'light', 'moderate' and 'heavy'. This sub-division is expanded a little in the DH (1991) Report, which, as well as having three categories of occupational activity, also had three levels of non-occupational activity (Table 1). In theory, this should allow a finer distinction to be made between different groups.

It should be possible to formulate some intelligent guesses about which value would be most appropriate for any particular group. The degree of concordance between the guess and reality would depend upon the accuracy of other information, such as might be gathered from activity questionnaires. In general, corroborative data should be obtained about the apparent activity levels for both occupational and non-occupational time. This should allow a reasonable estimate to be obtained, as long as the investigator making the guess has an adequate knowledge of the different relationships between the degree and duration of physical activity and energy expenditure.

3. Doubly-labelled water (DLW)

The so-called doubly-labelled water method of measuring energy expenditure makes use of the stable isotopes ¹⁸O and ²H. No attempt will be made here to discuss the likely validity of the method except to say that there may still be some reservations in accepting that the estimate of converting CO₂ production to heat production involves minimal error. Blaxter (1989) tabulated the varying heat equivalents of differing substrates (lipid, protein, carbohydrate) in relation to oxygen consumed and CO₂ produced. It is clear that while the range of oxygen consumed is relatively small (19.2-22.7 kJ/litre), that for CO₂ is much greater (17.5-27.8 kJ/litre). He concludes that, even in a state of energy equilibrium, the error involved in the conversion of CO₂ to heat could be as much as + 10%, and if the individual being measured were losing or gaining weight, the error could be much higher.

While the theoretical advantages of this method are considerable, such as the ability to provide data on free living individuals in almost any context related to age, environment, etc., there are counter-balancing drawbacks. The technicalities of the analyses of CO₂ output are such that, due to the possibility of using unsatisfactory equipment and the necessity for highly skilled and expensive technical assistance, many of the most experienced and distinguished laboratories in the field of energy balance' have been unable to obtain more than a very small amount of reliable data. The isotope ¹⁸O is also very expensive. Lastly, the data give information only on total CO₂ output during a period of several days, data which are then converted to total energy expenditure: there is no breakdown of any kind.

Table 1 Calculated physical activity level (PAL) of three adults at three levels each of occupational and non-occupational activity for men [M] and women [F]

It seems as if, unique among methods of measuring energy expenditure, the DLW technique provides no information other than on total energy expenditure over a period of several days. If we measure energy expenditure indirectly by energy intake we learn not only about energy expenditure but about diet, eating patterns, nutrient intakes, etc. If we measure energy expenditure using the timed activity and energy cost method, we obtain knowledge on life-style, activity patterns, relative strenuousness of work and leisure, etc. If we measure energy expenditure from BMR and an 'activity factor' we gather information which again includes material other than simply related to energy expenditure. Similarly with calorimetry, information on the whole pattern of the 24h of the day is also gathered. Only with the DLW method does one acquire nothing except total energy expenditure.

This quite serious drawback of the method, in conjunction with the considerable expense required, makes it desirable to assess carefully its relative advantages and disadvantages. It is not simply that we need to consider only the ability to measure energy expenditure. Because of the considerable intra- and inter-individual variability of energy expenditure, we may often need much ancillary data to help us make sensible decisions on what are frequently complex problems.

On the other hand, the DLW method could, with benefit, be used to corroborate the validity of some of the low activity factors (such as 1.4, 1.3, or even 1.2 × BMR) which are sometimes found in individuals without apparent reasonable explanation.

4. Calorimetry

Another method to be considered is the use of a respiration chamber or of a direct calorimeter. To obtain reliable data with either of these techniques involves an experimental set-up which is both expensive and technically complex. Relatively few of these chambers exist and their usefulness in the present context is restricted to specific basic problems which do not require a natural free-living environment; for example studies on relationships between energy metabolism and heart rate, on diet-induced thermogenesis, on validation of the use of DLW to measure energy expenditure in varying situations, on the influence of varying proportions of energy-supplying nutrients on energy metabolism, etc.

5. Heart rate

Extrapolating from heart rate to energy expenditure is a method which has been widely believed to be valuable and reasonably valid. The technique is fairly practicable, there are several instruments on the market which are not very expensive, and it is probably the method of choice in some population groups, such as young children, old people, and ill people. This is not the place to give a detailed critique of the methodology, other than to say that it must be used with circumspection and an awareness of the variable relationships of heart rate and energy expenditure.

Heart rate recording is also a potentially useful tool in some of the other methods of measuring energy expenditure as a means of obtaining an acceptably accurate value for the factor (A) with which to multiply BMR.

6. Energy intakes

Although the basis of calculating energy requirements depends upon obtaining values for energy expenditure, much of the assessment up to very recent times has been indirectly derived from data on energy intakes. This has often fallen out of favour in the last few years because of criticism of the likely accuracy of the energy intakes, or of the measured intakes being representative of the true intakes of the population. This is another area of energy metabolism where superficial criticisms may reflect little more than the authors' limitations.

To allow a correct interpretation of the results, it is critical to have a very precise description of how food intake was measured, what exactly was measured, and by whom, and how experienced were the observers. For example, if a 24-h recall was used, it is well known that sometimes errors can be enormous and will not be distributed in a random fashion around a more-or-less correct mean. It is helpful to know who carried out the 24h recall, where it was done, how long it took, if a standardized procedure was used, what was the experience of the observer, whether repetitions were made on other days, if the method was validated using other techniques or cross-over questioning by different observers, etc.

When food intake was measured and recorded, what was the exact way in which the measurement was done? What was the precision of the frequently quoted 'household measures'? What kinds of balances were provided? To what degree of exactitude were they read? Were they calibrated and how? Were other utensils (plates, containers, etc.) provided? What sorts of log-books were used, and how did the subjects record each item of food (e.g. in relation to composite dishes; and if sauces, etc., were taken, were they measured separately?).

Were instructions in the methodology given to the subjects in their own homes, in clinics, by writing, etc.? Was any supervision used during the study, and for how long and how often? How accurately assessed were food and drink consumed outside the home? What was the exact duration of the investigation on each individual? When this varied, it is not satisfactory to say things such as 'the investigation was carried out for periods of between 3 and 42 days on the subjects'. Was the period of the measured food intake apparently representative of the normal or not?

As far as the calculation of the intake of energy and other nutrients, which tables were used and how? For example, some well-known tables give energy values for carbohydrate, fat, and protein which do not include losses in digestion and absorption. Were allowances made for this? Sometimes duplicate samples of the diet have been measured for energy content by bomb calorimetry, but of course this can be a spuriously accurate procedure if it is not realized that the values obtained do not at all represent the available energy in the food as far as the body is concerned.

This information if indeed it were even obtained, is seldom quoted. However, where the procedure of measuring energy intake has been done in an acceptable fashion by experienced observers, it has been shown on many occasions to be non-significantly different from energy expenditure, simultaneously measured.

In the light of the occasional difficulty and expense of measuring energy expenditure, there is still a place for using energy intake, properly assessed, as the basis.

Importance of body composition

Whether or not body composition plays, not a significant but an important role in estimating energy requirements, is a disputed question. The question may perhaps be simplified to enquiring whether using body mass, or fat-free mass, as the reference makes an important difference to the estimate.

We can consider the problem at two levels of fatness.

1. Moderate levels of fatness

This includes adults aged from 20 y up to 50-60 y, with a fat mass of, for men, up to 30% of the total body mass, with the equivalent for women of up to about 35% (or a BMI of 30 or so).

If the proportionate fat mass is only moderate there is little reason to expect that it will specifically influence energy metabolism, either at rest or while physically active, in ways which are unrelated to the actual body mass. The biochemical sources of energy to the body will not differ over this sort of range of fatness nor will those levels of fatness, within the range of average and normal activity, markedly influence the mechanical efficiency and thus the energy cost of movement.

The theoretical justification for the above statements is that adipose tissue has a metabolic rate which is not grossly dissimilar from the energy metabolism of the total fat-free mass. Therefore, within the above limits of fatness, the metabolic rate of a moderately overweight person of 60 kg body mass, would not be expected to be very different from a moderately lean, or from a 'normal' individual of the same body mass. The variations in relative fatness will not influence, in an important way, the energy metabolism per kg body mass. An adequate analysis of the likely influence of varying levels of fatness on energy metabolism has still not been carried out exhaustively and this is an important area for future research.

In the case of very lean populations, there may be confusing influences. If the leanness is of semi-permanent or long-term duration, but is compatible with a level of nutritional status which does not inhibit a 'normal' lifestyle, particularly with regard to physical activity, then BMR and energy requirements are unlikely to be influenced in any important way. On the other hand, if leanness has resulted from a comparatively recent negative energy balance, BMR may be significantly lowered, and it is at least possible that such a reduction would be one of the results of seasonal deficiencies in food availability (Durnin et al, 1990; Ferro-Luzzi, 1990; Schultink et al, 1990).

2. Obesity

With greater levels of fatness, the situation is likely to become more complicated. There will certainly be an effect on the mechanical efficiency of movement but, although the biochemical energy transformations at rest may involve some basic differences from the normal, the possibility is that sources of variation both within and between individuals may mask any clear-cut effect. The considerable extent of the 95% confidence limits on inter-individual variability of BMR should always be kept in mind in relation to studies purporting to show the possible influences of body composition.

It is obviously necessary to take account of body mass, and as long as we are aware of the naively of expressing our data on BMR as energy per kg of body mass, the simplicity of the conversion and the fact that it will be unlikely to differ from energy per kg of fat-free mass (Garby et al, 1988; Henry et al, 1989) makes it a convenient and reasonable reference. It may appear that there are biologically sound reasons for making an allowance for differing body compositions and, at the simplest level, expressing data as energy per kg fat-free mass, but the probability is that this manoeuvre has, in fact, little biological justification and makes no important difference to any conclusions which may be drawn from the data. (No attempt has been made here to take into account more complex differences in body composition, i.e. related to organs or tissues, largely because of inadequate experimental information). A recent paper by Carpenter et al (1995), dealing with a meta-analysis of 13 studies using doubly-labelled water to measure total energy expenditure (TEE), seemed to show little relationship between TEE and adiposity, and a lower resting metabolic rate in women than in men, although the latter statement is made on the basis of only a very small sample of the published data. They also demonstrated the statistical dangers of using a simple ratio of TEE and body mass to compare data, but it is not clear from their paper how great an error is produced if the simple ratio is used.

In summary, in conditions of physical rest it is unlikely that body composition plays an important role in energy requirements. However, its role with regard to physical activity is more complex. Even if its influence on tissue energy metabolism is not of great practical importance, its influence on the amount of physical activity and on the energy cost of that activity might be appreciable.

3. Body mass of populations

Since obviously energy requirement, or at least BMR, is directly influenced by body mass, and populations of greater body mass will therefore have higher energy requirements, a careful appraisal of the actual or desirable body mass of all the relevant populations (infants, children, adults) becomes very important. This problem has not been considered as part of this analysis of 'general principles'.

Contents - Previous - Next