Contents - Previous - Next

Relationship of gain in height to gain in weight


J.C. Waterlow

15 Hillgate Street, London W8 7SP, UK

There are some interesting, although not entirely consistent, observations in the literature on the temporal patterns of gains in height and in weight. Brown and his colleagues in Bangladesh made a longitudinal study of the growth of children over a period of 14 months (Brown, Black & Becker, 1982). Weight gains were minimal in August, towards the end of the monsoon, and then, as food became available, reached a peak in February. Height gains were minimal in January and reached peak in April-May. Thus height gains followed weight gains by 3-4 months. Very similar seasonal effects on growth were observed by Nabarro et al. (1988) in Nepal. Again, growth in height lagged behind growth in weight by about 3 months. On the other hand, in The Gambia both height and weight gains were lowest in the rainy season, between June and November, and both increased in December, when the dry season began, with no separation in time (McGregor et al., 1968). In Kenya, Wiersinga & van Rens (1973) made detailed study of six children; they also found that the peaks and troughs of weight velocity and height velocity coincided. The amplitude of the swings in height velocity varies with age, and seems to be greatest in the second year of life. In the studies mentioned, the maximum height velocity was about three times the minimum. If anything, it appears to be slightly less in The Gambia, which is surprising because there the high prevalence of diarrhoeal disease which occurs in the wet season is accompanied by a massive increase in malaria transmission (McGregor et al., 1956), so that one would expect a greater depression of growth, followed, perhaps, by a greater degree of catch-up. It is notoriously difficult to compare different countries, but the data from Asia suggest that the time-lag there is a real phenomenon.

Children recovering from malnutrition provide further evidence of such a relationship. Walker & Golden (1988) analysed the records of 369 children recovering from malnutrition at the Tropical Metabolism Research Unit, Jamaica. Their average stay in the ward was 31 days, which is not very long for recording increases in length, and indeed the length gains were very variable. However, two points stand out from their analysis: first, that the rate of gain in length was negatively correlated with the initial length, so that the more stunted children grew the fastest. Secondly, in the majority of children linear growth did not begin until they had achieved at least 85% of expected weight-for-length. The observations of Costello (1989) in Nepal are relevant here. Fig. 1 shows that in children studied over a period of 6 months weight gain and height gain were related in diametrically opposite ways to initial weight for length. There was very little height gain, but maximal weight gain in children who were initially underweight. Of course one might suppose that the overweight children who grew so well in height came from better-off families and therefore naturally grew better, but this does nothing to confound the positive correlation between height gain and weight.

Fig. 1. Effect of a child's initial nutritional status on subsequent growth velocity.

To me this implies a physiological relationship. The only explanation that I can suggest is that it is well established that circulating levels of IGF-1 are reduced in acute malnutrition (Unterman et al., 1985; Clemmons et al., 1985), and rapidly rise when energy and protein are provided (Isley, Underwood & Clemmons, 1983). These levels may have to be maintained for some time before the results become manifest in an increase in linear growth. I admit that this is pure speculation. My point is that the relationship to body weight may provide a clue to the regulation of linear growth. It is as if there were a bag which has to be filled up before it can increase in size, as suggested by Millward (personal communication).


Brown KH, Black RE & Becker S (1982): Seasonal changes in nutritional status and the prevalence of malnutrition in a longitudinal study of young children in rural Bangladesh. Am. J. Clin. Nutr. 36, 303-313.

Clemmons DR, Underwood LE, Dickerson RN, Brown RO, Hak LJ, MacPhee RD & Heizer WD (1985): Use of somatomedin-C/insulin-like growth factor 1 measurements to monitor the response to nutritional repletion in malnourished patients. Am. J. Clin. Nutr. 41, 191-198.

Costello AM de L (1989): Growth velocity and stunting in rural Nepal. Arch. Dis. Child 64, 1478-1482.

Isley WL, Underwood LE & Clemmons DR (1983): Dietary components that regulate serum somatomedin-C concentrations in humans. J. Clin. Invest. 71, 175-182.

McGregor IA, Gilles HM, Walters JH, Davies JH & Pearson FA (1956): Effects of heavy and repeated malarial infections on Gambian infants and children. Br. Med. J. ii, 686-692.

McGregor IA, Rahman AK, Thompson B, Billewicz WZ & Thomson AM (1968): The growth of young children in a Gambian village. Trans. R. Soc. Trop. Med. Hyg. 62, 341-352.

Nabarro D, Howard P, Cassels C, Pant M, Wijga A & Padfield N (1988): The importance of infections and environmental factors as possible determinants of growth retardation in less-developed countries. Vevey: Nestle Nutrition/New York: Raven Press.

Unterman TG, Vasquez RM, Slas AJ, Martyn PA & Phillips LS (1985): Nutrition and somatomedin XIII. Usefulness of somatomedin-C in nutritional assessment. Am. J. Med. 78, 228-234.

Walker SP & Golden MHN (1988): Growth in length of children recovering from severe malnutrition. Eur. J. Clin. Nutr. 42, 395-404.

Wiersinga A & van Rens MM (1973): The simultaneous effect of protein-calorie malnutrition on weight and height velocity. Env. Child Health (Special Issue).


The first part of the discussion focused on the difficulties in establishing temporal relations between weight gain and height gain. First, there is the physical difficulty of measuring gains in height, which over short periods are inevitably small. Even the knemometer does not resolve this problem; for example, running for an hour can produce a change in lower leg length equivalent to one month's growth (Karlberg).

Then there is the question of variability, which has many sources. Even in well-to-do communities there are normal children who are tall and thin; it is perfectly possible for a child to grow normally in length even through he/she is only 85% of expected weight-for-height, the point at which, in children recovering from malnutrition, growth takes off (Walker & Golden, 1988). Whether such differences in body shape are genetically determined is not clear; there certainly seem to be ethnic differences in shape, as exemplified, for example, by the pastoralists of East Africa (Ulijaszek).

Another source of variability that affects shorter term measurements is that growth occurs in spurts, so that a child may actually be growing in length for only 5-10% of the time (Allen) (see also paper by Ulijaszek and discussion after the paper of Nilsson et al.). Weight gain also occurs irregularly. Are the spurts in height and weight gain related, or do they simply represent random biological variations?

Growth appears also to be influenced by a biological clock. In Europe it was shown many years ago that children gain more height in the spring and more weight in the autumn (Prentice; Lodeweyckx; see also paper by Ulijaszek). In blind children the seasonality of growth is not synchronized with the light-dark cycle, and in developed countries it is reversed (Karlberg). It seems, therefore, to be an unresolved question how far the seasonal variation is innate and how far it is determined by seasonal factors such as food, disease, or possibly sunlight and its effect on vitamin D metabolism. A few observations were contributed about the mechanism of the seasonal effect. In the cat it has been found that the light-dark cycle was accompanied by variations in the secretion of growth hormone (Karlberg). In a longitudinal study of bone growth in Italy, contrary to what one might expect, the peak excretion of cross-links, which may be taken as a marker for bone growth (see paper by Robins), occurred in the winter, coinciding with the peak of weight gain, while height gain was greatest two months later (Branca). This discrepancy in timing might be explained by the difficulty mentioned earlier, of measuring short-term changes in height.

A further complicating factor is the possibility that physical activity acts as a stimulus for bone growth through the pull of muscles on the ends of the long bones (see paper by Golding). Golden, for instance, said that he had been looking at the output in the urine of cross-links, which provide a measure of bone turnover (Branca et al., 1992; see paper by Robins). He related this to the creatinine-height index, which is a measure of muscle mass. Both in normal and in malnourished children bone turnover and height gain were related to muscle mass. One hypothesis to explain these findings would be based on physical activity: as children gain weight and put on muscle, they become more active. This stresses the bones and stimulates linear growth. Experiments need to be done to separate this mechanical effect from that of increased food intake (Jéquier).

In spite of these sources of variation, there are clinical findings which support Golden's conclusions, that weight gain precedes height gain. The example of coeliac disease is well known from the early work of Prader. In a study in Italy on children with this disease there was a lag of 11/2-21/2 months between height gain and weight gain (Karlberg). Some observations suggest that bone growth may occur at the expense of soft tissue growth. In a study in Ethiopia, the greatest seasonal gain in height was in those children who had the highest weight: height ratios, but these were also the children with the greatest seasonal loss of weight. Observations of the same kind have been made in Italy. Similarly, when obese children are put on a reducing diet, they show a rapid increase in height; maturation is accelerated, and they come to puberty at an earlier age (Lodeweyckx). Nothing is known about the way in which these relationships depend on hormones, their receptors and binding proteins.

One general hypothesis advanced the possibility of competition between bone and soft tissues. One object of such competition might be sulphate, which is essential for the growth of cartilage. Indeed, the uptake of labelled sulphate by cartilage fragments was the basis of the original bioassay of insulin-like growth factors. Not much seems to be known about the dietary intake of sulphate as such; it is probable that most of it is derived from the sulphur amino acids. In the child recovering from malnutrition one could suppose that if these amino acids are not in generous supply, there is an order of precedence of partitioning in their uptake, muscle having priority over cartilage, and visceral organs over muscle (Waterlow, 1990). One could visualize a series of switches that come into operation at different stages of catch-up growth, just as in ontogeny. The question then would be how far the switching sequence is affected by the amino acid supply. In this context mention was made of some unpublished studies in both children and experimental animals, in which the ratio of sulphur to nitrogen in the urine was reduced on a low protein diet, as if sulphur was being preferentially conserved (Waterlow).


Branca F, Robins SP, Ferro-Luzzi A & Golden MHN (1992): Bone turnover in malnourished children. Lancet 340, 1493-1496.

Walker SP & Golden MHN (1988): Growth in length of children recovering from severe protein-energy malnutrition. Eur. J. Clin. Nutr. 42, 395-404.

Waterlow JC (1990): Energy-sparing mechanisms: reductions in body mass, BMR and activity: their relative importance and priority in undernourished infants and children. In Activity, energy expenditure and energy requirements of infants and children, eds B Schürch & NS Scrimshaw, pp. 239-251. Lausanne: IDECG/Nestlé Foundation.

Contents - Previous - Next