Discussion of papers by Allen, Neumann & Harrison and Dagnelie et al.

The CRSP studies in Mexico and Kenya and that on vegans in Holland suggested a relation between stunting and a diet of poor quality. As a preliminary, technical questions were raised about the nature of the correlations: do they apply across individuals as well as groups? are the dietary intakes normally distributed? would the correlations be improved by log transforms? In the vegan study the comparison was made with matched controls, but it is impossible to match for more than a small number of variables: for example, the macrobiotic children had lower birth weight. Therefore the conclusion can only be a negative one: that the particular variables that were matched did not explain the differences between the groups.

Granted that the correlations are real, the term 'poor quality diet' covers many potential deficiencies. The discussion centered first round the question of animal protein. In Allen's report it was stated that stunting occurred in spite of adequate intakes of protein and amino acids. It was pointed out, however, that adequacy of protein for children has always been calculated on the basis of the amount needed for weight gain. No account has been taken of height gain; in fact, we simply do not know what the requirements are for growth in height. Large numbers of people in the world live on mainly vegetable diets, and it has been argued that it is not necessary to have any animal protein in the diets of children (Town). Again the question arises: necessary for what? Apart from effects on linear growth, it was pointed out that in the CRSP studies measurements of cognitive function repeatedly showed a relation with animal products in the diet (Neumann).

If animal protein is necessary, how much? The Mexican children received only about 80 kcal/d from milk, eggs and meat (Allen). In the vegan children, breast-feeding seemed to delay the onset of stunting. The nutrient content of the mother's milk was the same as that of the controls at 3 months, but it was lower at 1 year (van Dusseldorp). It was when supplementary foods were given that growth began to fall off. Nevertheless, it cannot be assumed that breast-fed children never become stunted; it depends on whether lactation is adequate in undernourished mothers-a subject that is still somewhat controversial.

The discussion moved on to the question of how the requirements for linear growth could be established. As far as concerns protein, the classical nitrogen balance method is clearly not applicable. The only solution would seem to be experimental intervention. Early feeding trials were recalled, going back to those of Corry Mann, Boyd Orr and Aykroyd in the 1930s. All these showed some effect on linear growth of foods containing good quality protein, particularly milk. Indeed, it was these observations that suggested the sulphur amino acid hypothesis referred to in discussion of the paper by Golden. However, except for the study in India by Aykroyd & Krishnan (1939), in which calcium lactate was given, all the trials showing an effect on growth involved foods such as milk or meat, and not single nutrients. Therefore the stimulation of growth could not be attributed unequivocally to good quality protein; it could equally result from the micronutrients associated with protein, such as iron, trace elements and vitamin B₁₂. Moreover, amino acids may enhance the bioavailability of some minerals, and vegetable diets are likely to contain more phytate and fibre which have the opposite effect. For example, the macrobiotic diet provided more iron than the control diet, but a higher proportion of the children were anaemic because they received no haem iron. In Mexico the diet was very high in calcium but probably very little of it was absorbed.

The question of bioavailability was considered to be particularly important in relation to zinc, especially since in the CRSP studies zinc intakes were frequently low. The question was raised, but not resolved, of the possibility of absorption of zinc from the colon. Another point is that zinc plays a special role in the response to infections, and it has often been suggested that infections and infestations are a major factor in the production of stunting. The requirements of nutrients for growth may be different from the requirements for counteracting infection, so it could be dangerous to extrapolate to the Third World estimates of requirements obtained in affluent countries. It was also pointed out that in many experimental studies male animals were found to be more susceptible to zinc deficiency than females (Golden). In the CRSP studies, girls seemed to do better than boys and get fewer infections in the preschool years (Neumann).

Lastly, it was suggested that fat intake, which is low in the vegan and Third World diets, may be important because there is evidence of obligatory fat oxidation in the postabsorptive state (Jéquier).

The conclusion from these three studies, and from earlier intervention trials, is that stunting occurs in children whose diets are low in animal protein and in factors associated with animal protein in foods. This conclusion can best be tested by intervention studies. The question was not discussed, whether there would be any value in setting up studies designed to elucidate which particular nutrients are involved. From a scientific point of view such studies would be very important. From the practical point of view, since people eat foods not nutrients, the lesson would be that there is a need for improving the quality as well as the quantity of the diets of preschool children.