THE DEFINITION OF REQUIREMENTS AND RECOMMENDATIONS FOR PROTEIN WITH NOTES ON THE MAGNITUDE AND DEGREE OF UNCERTAINTY OF THE FACTORIAL COMPONENTS

P.R. Payne
Department of Human Nutrition
London School of Hygiene and Tropical Medicine

It is perhaps useful to begin by attempting to draw a clear distinction between a requirement and a recommended intake.

In recent years, nutritionists have become increasingly concerned with the problem of specifying recommended intakes of nutrients which might be considered optimal for different aspects of health and well being with, for example, the object of prolonging active life, or of achieving some desirable standard of growth and maturation rate in children.

Despite the importance of these problems, the first objective of the committee should be to establish minimum requirement levels which will be most efficient as a means of assessing the adequacy of diets, i.e. they should be regarded primarily as a diagnostic tool, leading to a better assessment of the number identity and location (in a socioeconomic as well as geographical sense) of the poorly nourished people in a community. The levels should be based upon what is known of physiological minimum needs. Raising these by the addition of too many arbitrary safety factors, allowances for 'stress', 'individual variation' etc. may satisfy humanitarian desires, or gratify a sense of scientific caution; but will only result in a reduction of diagnostic effectivness.

The formulation of recommended levels of intake is much more difficult. These should evidently be seen in the first place as an attempt to link diagnosis with treatment by providing a standard which is more than adequate for basic needs, but which represents a realistic and practical objective for a developing community. Again there may be a temptation to specify high recommendation, in the hope that this may serve as an extra stimulus to action, or even in some mysterious way to overcome problems of maldistribution. In fact no man has ever had his hunger satisfied by the recommendations of a committee; only if these form the basis for acceptable and practically feasible advice to individuals and governments will they lead to any improvement in nutrition. There is no very clear scientific evidence upon which such recommendations can be made. They should perhaps include some allowance for the effects of stresses and infections of everyday life, but there is little quantitative evidence on which these could be evaluated.

What is certain is that they cannot be based simply on a study of the self-selected diets of various communities. There is increasing evidence in the form of the incidence of degenerative diseases of middle and later life, that such diets are not optimal; in fact there is no reason to expect them to be since survival much beyond reproductive age has not been an object of evolutionary selection over significant periods of time. Neolithic man with an average life expectancy of about 20 years would have evolved simply those dietary preferences which gave the best chance of survival to the species as a whole, and not necessarily those which most enhance the longevity of the individual.

Whatever the basis chosen we should resist the temptation to cover ignorance by specifying high recommended intakes, or by adhering to dietary patterns which are socially or culturally accepted in certain communities. Whatever the basis, it must be remembered that if they are subsequently adopted as part of a development plan, recommendations must in order to be effective, exert an influence upon the pattern of food production and will therefore involve an economic cost to the community in terms of diversion of resources and changing labour patterns. Thus the scientific basis for the recommended levels must be adequate to justify the benefits in terms of improved health and productivity in relation to the costs involved.

The elements of the factorial method

In what follows, an attempt has been made to re-assess the separate elements of the factorial method in the light of the most recent measurements, indicating wherever possible the degree of uncertainty to be associated with the mean values. Addition of the separate elements gives an average physiological requirement, together with an overall estimate of uncertainty. Some of this variability is due to interindividual variation and some to intraindividual e.g. due to daily variations in urinary N output. It is not possible to separate these sources of variance, and we can only therefore regard the mean value plus twice the standard deviation as a safe level of intake below which any individual must be considered at risk.

The requirement levels calculated in this way should be regarded only as a means of assessing the adequacy of diets in order to determine the extent of malnutrition in a community. They may however serve as a basis for recommendations as discussed in the accompanying paper by Waterlow.

The component factors

1. Basal Urinary N per day

For adult males, the values given by Hawley et al (1948), by Mueller and Cox (1947) and by Young and Scrimshaw (1968) have been combined to give 19 subjects.

For adult females, the values given by Hawley et al and by Bricker and Smith (1951) total 31 subjects

All measurements were made after at least 6 days on N-free diet.

Urinary N is correlated with Wt^3/4, surface area and with body weight, r = 0.700, 0.620, 0.692 for males and 0.664, 0.669, 0.660 for females.

Correlations with basal calories are poorer, 0.371 for males and 0.640 for females, probably because of the greater error involved in determination of BMR, than in measurement of weight and height. There is in any case no advantage to be gained from expression as mg/basal calorie, since all that is needed is a means of relating urinary N to age or body size for the purpose of interpolation

For 3 - 4 year old children Fomon et al (1965) for the 4th to 6th day on N-free diets give

Accordingly 100 mgN/Kg^3/4 has been used to obtain the interpolated values in the table.

2. Metabolic fecal N

For adults, Young and Scrimshaw give 9.0 ± 1.1 mg/Kg for 8 male subjects. Bricker et al for 25 females give 8.7 ± 1.4 mg/Kg. Fomon et al give 23 ± 9.3 for 3 - 4 year old children, and Waterlow and Wills (1960) give 33 as representative of 1 year old infants.

Metabolic fecal N has been shown in a number of species to be related to the total dry matter intake in the food. Accordingly it seems reasonable for purposes of interpolation, to relate metabolic fecal N to total calorie intake. The above values give ratios of 0.22, 0.20, 0.23 and 0.24 mg N/calorie consumed. Metabolic fecal N losses for the adolescent in the table have been calculated on the basis of 0.25 mg N per dietary calorie (FAO calorie allowance).

3. Other possible routes of N losses are:

Integumental - skin, nails, hair etc. Possibly some N loss as flatus (though it is not known if this may properly be regarded as of exogenous or endogenous origin).

Sweat losses which may range from 2 mg/Kg (Sirbu et al 1967) to 6 mg/Kg for heavy work in hot conditions Ashworth et al 1967).

The sum total of all these possible losses is best estimated from the results of long continued N balance trials in adults. Mitchell (1949) gives a figure of 1.38 g N/day for 23 subjects over a 220 day period. It was suggested that in these active subjects 0.38 g could be accounted for a sweat loss above minimal levels. (Mitchell and Edman (1962)) so that excluding these greater than minimal losses, other routes accounted for 1.0 gN/day with an SD of 20%. In female subjects, Bricker et al. (1949) together with values reported by Johnson and McMillan (1952) give a total of 15 subjects averaging 0.67 g/day ± 22 % (a figure which includes menstrual losses). It would be expected that skin losses should be related in some degree to surface area, and in fact in both male and female groups there is a better correlation between N loss and surface area than there is with body weight. For the purposes of extrapolation for the adolescent and infants, a value of 44 mg N/Kg^3/4 has been used (corresponding to the adult male value) although it is certainly possible that this represents an underestimate, since minimal sweat losses for example might well be related to total calorie intake.

Sweat loss above the minimum is best regarded separately, since firstly there may be some compensatory balance between urinary and sweat losses (Sirbu et al.) and secondly it evidently depends upon work output and environmental conditions. The maximum value corresponds to an extra 0.04 g protein/day for the adult male i.e. just over 10% addition which could be made for high levels of calorie intake.

4. Growth requirements

The following table gives rates of N gain at different ages. Figures for growth velocity are from Tanner, Whitehouse, and Takaishi (1965) N contents from Fomon (1967).

Individual variation

The standard deviation will represent the combined variance due to experimental error, day to day variation within subjects, as well as between subjects variation. There is insufficient information to separate these, but it is notable that Mueller and Cox (1947) give daily values for urinary N over a 12 day period on N free diets, for four subjects. The variation between subjects was not greater than the daily variation which was about 12% (CV) over the last 6 days.

It would seem likely that individual variation is small compared with day to day changes and in this respect is probably similar to that found for BMR/sq. metre within a given age/sex group, namely 3.5% for male and 4.7% for female subjects. Berkson & Boothby (1936).

Individual variation of the growth component is indicated by the SDs shown above, and is calculated from the difference between 3rd - 27th percentile growth velocity standards.

It seems likely, therefore, that much of the total variability in the adult (N10%), to a large extent represents day to day fluctuations. In view of the fact that the 'skin' loss figures adopted almost certainly are overestimates because of the cumulative positive errors in balance experiments, the addition of 20% should provide an adequate safety factor. There are however no grounds for describing such an addition as an allowance for individual variation since the degree of such interindividual as opposed to interindividual variation can only be determined from a series of repeated observations on large groups of subjects such as those made for BMR by Berkson and Boothby. Such measurements have not been made for any of the components of N losses.

REFERENCE PROTEIN REQUIREMENTS

Figures in parentheses are calculated from Urinary N = 100 mg/Kg^3/4

Skin N = 44 mg/Kg^3/4

Fecal N = 0.25 mg/calorie

REFERENCES

(1) Berkson, J. and Boothby, W.M. J. Physiol. 121:669 (1938).

(2) Bricker, M.L., Shively, R.F., Smith, J.M., Mitchell, H.H., and Hamilton, T.S.J. Nutrition. 37:165 (1949).

(3) Bricker, M. and Smith, J. J. Nutrition. 44:553 (1951).

(4) Fomon, S.J. Infant Nutrition. W.B. Saunders. Philadelphia and London (1967).

(5) Fomon, S.J., De Maeyer, E.M., and Owen, G.M. J. Nutrition. 85:235 (1965).

(6) Hawley, E.E., Murlin, J.R., Nasset, E.S., and Ezymonski, T.A. J. Nutrition. 36:153 (1948).

(7) Johnston, F.A., McMillan, T.J. J. Nutrition. 47:425 (1952).

(8) Mitchell, A.H. Arch. Biochem. 21:335 (1949).

(9) Mitchell, H.H. and Edman, M. Amer. J. Clin. Nutr. 10:163 (1962).

(10) Mueller, A.G. and Cox, W.M. J. Nutrition. 34:285 (1947).

(11) Sirbu, E.R., Margen, S. and Calloway, D.H. Amer. J. Clin. Nutr. 20:1158 (1967).

(12) Tanner, J.M., Whitehouse, R.H., and Takaishi, M. Archs. Dis. Childh. 41:613 (1965).

(13) Waterlow, J.C. and Wills, V.G. Brit. J. Nutrition 14:183 (1960).

(14) Young, V.R. and Scrimshaw, N.S. Brit.J. Nutrition. 22:9 (1968).