Contents - Previous - Next

Guidelines for the study of mechanisms involved in the prevention or reversal of linear growth retardation in developing countries

The design of research on stunting
Approaches to the study of stunting
Sample size
Potential experimental models for clinical trials
Examples of useful measurements

L.H. Allen 1 and R. Uauy 2

1 Department of Nutrition, University of California, Davis, CA 95616-8669 USA and 2 Instituto de Nutricion y Tecnologia de los Alimentos (INTA), University of Chile, Casilla 138-11, Santiago, Chile


These guidelines are intended for investigators who plan to undertake research on the prevention or reversal of growth stunting in developing countries. They stem from the symposium sponsored by the IDECG on 'Causes and Mechanisms of Linear Growth Retardation', and consider the major conclusions reached by the group concerning the types of research questions that need to be answered and the methods that should be used to address them. The intent is to prioritize when (at what ages) and how (what outcome and confounding measures) the research should be done, rather than which independent variables should be studied. The point of proposing elements of a common general research protocol is to strengthen the external validity of the findings of individual projects, and to improve the opportunity to interpret results from different environments.

The main purpose of the suggested approaches to research is to improve our scientific understanding of how specific nutritional and other factors affect the body so that growth stunting results or is reversed. Further details concerning the choice of independent and outcome variables are to be found in this supplement.

Investigators studying causes and mechanisms of linear growth retardation and interested in following these guidelines are encouraged to inform the IDECG Secretariat (c/o Nestle Foundation, PO. Box 581, 1001 Lausanne, Switzerland), because IDECG is considering the possibility of organizing another workshop at which the results of such studies will be discussed.


The prevalence of linear growth stunting in children, defined here as a length-for-age more than 2 standard deviations below NCHS reference data, ranges from 10% to 50% in most developing countries. This is consistent with the 10-15 cm difference in height of the adult populations relative to adults in the industrialized world or in more privileged segments of the population in developing countries.

Linear growth stunting may be present at birth, but is usually a phenomenon that begins mainly between 6 and 18 months of age. Stunting is associated with delays in many functions, for instance mental and motor development, and with increased risk of morbidity and death.

The occurrence of stunting is conditioned by factors which may vary depending on the ecological setting. Although the main cause(s) remain unclear, the following main conditioning factors have been identified in field studies.

1. Genetic factors only account for a small proportion of stunting, as evidenced by the relative lack of growth retardation in privileged groups with the same genetic background, and the reduction in stunting prevalence that occurs when populations move from a poor to a better environment. The Asiatic and Amerindian populations may have a genetically determined lower stature in the order of -0.5 SD, although this remains a point of debate.

2. Altered fetal growth may result from genetic, congenital or environmental (including high altitude and radiation) factors, maternal disease, or nutritional deficits. The effect of maternal malnutrition on infant linear growth in utero may take more than one generation to resolve.

3. Insufficient postnatal nutrient supply will also cause growth stunting. The case for energy deficit alone is weak; most infants who fail to gain in length are adequate in weight-for-length, many of them have access to sufficient food and, if they did not, they would be lacking in other nutrients as well. Recent studies suggest that nutrients present in animal products such as milk may be important for normal linear growth in early life. The potentially relevant nutrients in these products include animal protein, zinc, iron, retinol, riboflavin, vitamin B12 and essential fatty acids among others. The actual growth-limiting deficiencies remain to be identified.

4. Growth stunting can result from an increased prevalence of infection or infestation, either symptomatic or subclinical in nature. The metabolic response to infection, which includes malabsorption and cytokine release, may be in part responsible for growth failure in young children. Presently the evidence indicates that controlling infections and parasitic disease may improve linear growth.

The mechanisms responsible for growth stunting are likely to include the interaction between hormonal responses (growth hormone, insulin, cortisol and others), growth factors (IGFs), their respective binding proteins and cellular receptors at the target site, specific nutrient deficiencies, and cytokine release during infectious disease episodes.

A target organ of particular interest is the growth plate in the epiphysis of long bones, and more research should be directed to understand how it is affected by undernutrition and infection.

The rate of linear growth declines with age, except at puberty, and the effects of stunting under 3 years of age are most significant. However, there is some evidence that stunting can be reversed by nutritional, social or environmental interventions at any time before the growth plate has fused, i.e. before linear growth stops.

The design of research on stunting

Because stunting occurs at a very early age and is associated with a wide range of functional deficits, it seems appropriate to give priority to research on the causes and mechanisms of stunting in infancy and early childhood, and to its prevention. On the other hand, there is also a need to test the extent to which stunting can be reversed in later childhood, and especially around the time of peak growth velocity in adolescence.

Many studies on stunting have already been done, yet our understanding of the causes and the underlying biological explanations is still relatively poor. Properly designed and conducted research on stunting can be expensive, long in duration, and fraught with confounding factors. It is imperative to ensure and document that a supplement is taken as designed, which involves supervision and a large expense. In the case of food supplementation it is usually necessary to also measure the effect on habitual dietary intake, again an expensive and difficult task. The conduct of intervention trials is so complex that the numbers of subjects should be as small and the period of measurement as short as is compatible with adequate statistical power. The possibility of seasonal differences in growth should be considered in the design.

The group came to the general conclusion that priority should be given to intervention studies rather than observational studies. This will enable us to differentiate between several possible causal factors. In order to identify specific nutrient deficiencies that cause stunting, or nutrients that will reverse it, interventions with specific nutrients or combinations of nutrients are needed. Because multiple micronutrient deficiencies are likely to occur in many settings, failure of growth response to intervention with a single nutrient may only reflect the fact that this nutrient deficiency is not the most growth limiting. Also, once this nutrient is supplemented another nutrient may become limiting and reduce the growth response to the first. These issues need to be considered carefully in the design and interpretation of intervention studies.

Interventions with specific foods may also be appropriate in some cases. For example, the testing of improved weaning foods is important if the goal is to alter local weaning practices to improve child growth. In the longer term, interventions with foods are more sustainable than interventions with single nutrients, but here our main intent is to improve scientific understanding of the specific nutrient deficiencies involved in growth stunting and the underlying biological mechanisms. Observational studies can also be useful if a special group can be investigated, e.g. children fed macrobiotic diets, or if specific behaviours need to be measured, e.g. maternal feeding practices or appetite.

An appropriate control group should be included and receive a placebo treatment; the assignment of treatments to individuals should be random and double-blind.

Approaches to the study of stunting

Three possible approaches are suggested here. The outcomes are perceived as the minimum information that should be collected. It should be noted that a considerable amount of additional time may be needed to recruit subjects of the right age and type (especially pregnant women), to stagger enrollment so that personnel can handle the measures, and to capture and analyse the data.

1. Research focused on the prevention of stunting

Subjects: Fetuses from conception up to children 2 years of age. This postnatal period usually involves three stages of feeding: breast-feeding, often relatively exclusive for 4-6 months; weaning, until the time when partial breast-feeding ends; and the post-weaning period. The nutritional causes of stunting may well differ among these periods. There are also two phases of growth in early postnatal life; an infancy phase (starting in mid-gestation and declining in importance up to 3-4 years after birth), and a childhood phase that starts around 8 months of life in well-nourished children and several months later in those who are malnourished (see paper by Karlberg et al., pp. 25-44).

Interventions: Select the kind of intervention that seems most likely to be effective in the particular setting, in the light of what is known about diet, disease incidence, etc. Examples might be supplementation during pregnancy, supplements to the infant of protein, trace elements, etc; there should be a separate group of children for each intervention. A control group is desirable of children from the same background, with the same risk of becoming stunted (this can be assessed approximately from the heights of mothers and sibs). Postnatal interventions should start as soon after birth as possible, preferably not later than two months. Outcome: Outcome measures should include length and lower leg length measurements made monthly for the first year and thereafter every three months, and growth velocity over 3- to 6-month intervals. Only with measurements of this frequency will it be possible to detect a delay in the entrance into the child phase. If the intervention is a supplement which has to be given every day under supervision, monthly measurements should not involve much extra work load. If the intervention is the elimination of parasites, longer intervals between measurements may be permissible, dictated by the intervals chosen for controlling for parasitic and infectious disease.

2. Research focused on the reversibility of stunting before puberty

Subjects: Preschool or school age children who are more than 2 SD below reference length for age.

Interventions: Test whether nutritional improvement can enhance linear growth without accelerating bone age maturation beyond chronological age. Suggested examples are supplementation with specific minerals, vitamins or key amino acids for anabolism.

Outcome: Linear growth velocity over a 2-year period, changes in bone age.

3. Research focused on the reversibility of stunting at the time of peak linear growth velocity at puberty

Subjects: Children 10-11 years of age who are more than 2 SD below reference length-forage.

Interventions: Test whether nutritional improvement or physical exercise can enhance linear growth without accelerating bone age maturation beyond chronological age or length age. Potential interventions include nutrients such as zinc or protein, or physical exercise, which might modify the hormonal responses that enhance linear growth during this critical period.

Outcomes: Final adult height, or height in the middle of the first 1-year interval after age at peak growth velocity (see paper of Karlberg et al., pp. 25-44).

Sample size

The research should take the form of a controlled trial involving a sufficient number of children to detect a 0.5 to 1.0 SD difference in length, with a significance of 0.05 and a power of 0.8. If two groups are to be compared, this implies the following sample size per group:

Difference in length SD

N per group













If more than two groups are compared this increases the sample size needed for each group; when more comparisons are made the chance of a Type 1 error is increased.

In order to reduce the influence of confounding factors it is often useful to make the groups as similar as possible in terms of, for example, initial height-for-age, sex, ethnicity, etc. Larger differences in length velocity may be seen more readily at younger ages, when growth is more rapid.

Studies should be conducted in ecological settings that have a high prevalence of stunting or where a useful factor could be isolated as a variable for testing (e.g. altitude, iron deficiency anaemia, etc.).

Potential experimental models for clinical trials

Independent variables:

1. Micronutrients, e.g. zinc, iron, vitamin A among others.
2. Infection and parasite control (especially gastrointestinal).
3. Total protein or selected amino acids.
4. Psychosocial stimulation and other environ mental enrichment.
5. Other controlled interventions, e.g. improved maternal nutrition or improved weaning foods.

Dependent variables:

1. Absolute length-for-age.

2. Growth velocity in length.

3. Other measures of linear gain, knemometer to measure lower leg length, long bone growth (segmental anthropometry), age at onset of childhood growth phase (see paper of Karlberg et al., pp. 25-44).

4. Relevant biochemical markers.

5. Relevant hormonal markers.

Confounding or intermediate variables:

1. Genetic/ethnic factors.
2. Parental factors.
3. Size at birth.
4. Other nutrient deficiencies.
5. Responses to infections, cytokines.
6. Damage to the gastric mucosa.
7. Clinical episodes of infection including diarrhoea.
8. Appetite.
9. Malabsorption or other diseases affecting growth.
10. Breast-feeding, weaning practices and weaning foods.
11. Physical activity.
12. Motor development and maturation.

Examples of hypotheses to test:

1. Modifying the independent variable during the first year of life in infants born at risk for stunting will prevent linear growth retardation as measured by length at 2 years.

2. The effect of the independent variable on growth in a specific age group is mediated by growth hormone and insulin-like growth factor-I activity.

3. The effect of the independent variable on linear growth is mediated by a decreased incidence of infections.

4. The effect of the independent variable on growth is mediated by changes in appetite.

Examples of useful measurements

The following measurements may detect or be related to a growth response to intervention, or explain the mechanism of stunting or its reversibility.

1. Size and growth velocity during the first 2 years of life measuring recumbent length, lower leg length, weight and head circumference. Length should be measured monthly for the first 5 months of life, every 2 months in later infancy, and every 3 months thereafter for detecting the age at the onset of the childhood growth phase (see paper by Karlberg et al. pp. 25-44).

2. Measurement of the tibial growth plate to explore its sensitivity to the independent variable and its usefulness as an index of linear growth velocity.

3. Changes in trace element status and haematological response to supplementation.

4. Pyridinoline (PYD) and deoxypyridinoline (DPD) in urine as indices of bone collagen turnover.

5. Growth hormone and IGF-I levels in plasma and growth hormone in urine as indices of the hormonal regulation of growth.

6. Incidence of infectious morbidity (gastrointestinal, respiratory, skin).

7. Cytokine response (TNF, IL-1 and IL-6).

8. Appetite and feeding behaviour.

Contents - Previous - Next