Chapter 1: Stunting and Young Child Development

Has there been progress in reducing stunting?
Trends in stunting and underweight move in parallel
Concepts and indicators
Consequences of stunting
Born malnourished
The issue of age
Understanding regional differences in stunting
Further considerations
Conclusions
References

INTRODUCTION

Beginning in 1987, the ACC/SCN has periodically examined the trends in the prevalence of child underweight. In this chapter, for the first time, we report on trends in stunting, i.e., poor growth in the length of infants and the height of children. This has been made possible by the increased availability of national data on height for countries in all regions, especially over the past five years.

As in previous reports in this series, this chapter provides information on trends occurring in six regions: Sub-Saharan Africa, Near East/North Africa, South Asia, South East Asia, Middle America/Caribbean and South America The primary purpose is to estimate the trends in the prevalence of stunting by region, determine if any of the regional trends are speeding up or slowing down and determine if the regions are progressing differently. Information is also presented on rates of low-birth-weight, the relationship of stunting to other national factors and the ages at which children are susceptible to faltering in height.

In the Second Report on the World Nutrition Situation (ACC/SCN, 1992, p.10),underweight was found to affect about one-third of all children in developing countries in 1990. The percentage of children who were underweight fell in the 1980s from around 37.8% in 1980 to 34.3% in 1990. In contrast to other regions, the underweight prevalence in Sub-Saharan Africa generally deteriorated or remained static in the 1980s. South Asia was found to be improving slowly, but the underweight prevalence in this region was the highest in the world, and more than half of the underweight children in the world lived there. Two subsequent reports (Update on the Nutrition Situation, 1994 and Update on the Nutrition Situation, 1996) reinforced these findings.

The Update 1996 suggested a slowdown in the rate of progress in reducing underweight. These three reports also examined country trends and relationships of rates of underweight to other country-level social and economic factors.

This chapter focuses on stunting for several reasons. First, as stated earlier, trends in the prevalence of stunting have not previously been reported. Second, child health goals for the early part of the next century have specifically targeted improvements in the rates of stunting. One of the five health outcome targets (out of eleven total targets for health) given by the World Health Organization in its recently revised Health for All in the 21st Century is 'the percentage of children under five years who are stunted should be less than 20% in all countries and in all specific subgroups within countries by the year 2020' (WHO, 1998, p.22). Third, stunting may be a better cumulative indicator of well-being for populations of children in countries than is underweight, because underweight is affected by weight recovery for some children between two and five years of age and by some children being overweight.

Has there been progress in reducing stunting?

To assess progress in child stunting, data compiled by the WHO Nutrition Programme in its Global Database on Child Growth and Malnutrition were used. Data were available from 61 countries for the estimation of trends in stunting, and from 95 countries for the estimation of prevalence of stunting (Table 1). Countries contributed to the estimation of trends in stunting if they had two or more surveys. Countries contributed to the estimation of prevalence in stunting if they had at least one survey.

Figure 1: Trends in Stunting (height-for-age <-2 SD below the reference median) over Time in Six Regions

Sub-Saharan Africa

South East Asia

Near East/North Africa

Middle America/Caribbean

South Asia

South America

Note: Heavy lines indicate regional trends, weighted by population. In the case of Sub-Saharan Africa, two trend lines are included: one for the countries where stunting is improving, and one for the countries where stunting is deteriorating.

Region	Trends in Stunting			Prevalence of Stunting
	Countries Covered		Population Covered	Countries Covered		Population Covered
	%	Number	%	%	Number	%
Sub-Saharan Africa	50	25	67	74	36	85
Near East/North Africa	35	7	55	85	17	94
South Asia	75	6	98	88	7	98
South East Asia	23	5	33	55	11	78
Middle America/Caribbean	83	10	95	85	11	95
South America	47	8	84	75	12	99
Across all regions	44	61	77	74	95	91

Figure 1 presents graphs of the available data for each country in each of the six regions. In the graphs, data from each country are tied together by lines so that specific countries can be distinguished. It can readily be seen that the variability in the trends across countries is much greater in the Sub-Saharan African region than in other regions. Further information can be found in Annex 1, where prevalence rates are presented for each country.

Trends in stunting over time for children under five years old were estimated for the period 1980 to 1995 for each of the six regions and for the regions combined. A few surveys that were available before 1980 were also included in the analysis. The analytical method used was a generalization of linear regression that combined information on the trends for each country, assuming a straight-line relationship between stunting overtime. All the results presented in this chapter are from models that included the population of each country as weights, so that larger countries contributed more to the estimates for each region than did smaller countries. The methodology and data sources are described in Annex 2. Trends in the prevalence of stunting are reported as the absolute prevalence change in percentage points per year (i.e., not as the percentage change from initial value). Overall, stunting during the 1980 to 1995 period declined globally at the rate of 0.54 percentage points per year.

Table 2: Estimated Trends in Stunting for the Period 1980 to 1995 for Children Under Five Years Old by Region

Region	Trend % Points per Year	S.E. for Trend	P-Value for Trend
Sub-Saharan Africa	0.130	0.145	0.374
Near East/North Africa	-0.635	0.166	0.002
South Asia	-0.837	0.068	0.000
South East Asia	-0.903	0.118	0.000
Middle America/Caribbean	-0.259	0.190	0.188
South America	-0.807	0.089	0.000
Across all regions	-0.539	0.065	0.000

As shown in Table 2 and Figure 2, the trends in the prevalence of stunting differed by region. Sub-Saharan Africa had an increase in the average prevalence of stunting of 0.130 percentage points per year, whereas each of the other regions showed a statistically significant decrease in stunting, with trends ranging from -0.259 to -0.903 percentage points per year. South Asia, South East Asia and South America showed the greatest progress. Near East/North Africa showed substantial progress and Middle America/Caribbean showed modest progress. The trends for each of the regions were estimated with sufficient precision to be able to contrast these different trends with statistical confidence.

Table 3: Estimated Prevalence of Stunting (%) and Numbers of Children Affected for 1980, 1985, 1990 and 1995 by Region

Region	Prevalence Stunting				Numbers Stunted (in millions)				% Increase (+) or Decrease (-) In Numbers from 1980 to 1995
	1980	1985	1990	1995	1980	1985	1990	1995
Sub-Saharan Africa	37.4	38.1	38.7	39.4	26.255	30.832	36.248	42.590	+62
Near East/North Africa	30.8	25.9	23.0	22.2	11.397	10.991	10.865	10.913	-4
South Asia	66.1	61.9	57.7	53.5	88.873	93.237	91.520	89.877	+1
South East Asia	51.9	47.3	42.8	38.3	35.581	32.862	30.119	30.206	-15
Middle America/Caribbean	31.6	30.4	29.1	27.8	5.398	5.467	5.631	5.626	+4
South America	25.0	21.0	16.9	12.9	8.285	7.309	5.965	4.644	-44
China (1992)			31.4				36.068
Across all regions (excluding China)	48.8	45.6	42.5	39.9	175.789	180.698	180.348	183.856	+5

Note: These estimates were derived assuming a linear relationship between stunting and year. The only region for which there was evidence of a non-linear relationship was Near East/North Africa. For this region, a quadratic model was used to approximate the non-linear relationship. The estimated prevalence values for this region were from this model.

From the statistical models, the estimated prevalence of stunting was determined for each region and for the regions combined for each of four years: 1980, 1985, 1990 and 1995 (Table 3). In 1995, the overall prevalence of stunting across the six regions (excluding China) was 39.9%. South Asia had the highest prevalence (53.5%), followed by Sub-Saharan Africa (39.4%), South East Asia (38.3%), Middle America/Caribbean (27.8%), Near East/North Africa (22.2%) and South America (12.9%). No relationship was found between prevalence rate and rate of progress across regions.

The prevalence of stunting in China was 31.4% in 1992, the only national survey estimate for that country.

The numbers of children under five years old who were stunted during 1980 to 1995 were also estimated for the six regions and for the regions combined (Table 3). These estimates were derived using the prevalence values of stunting and the total population under five of each region from the United Nations' estimates of 1996. Thus, the estimate of the numbers of children under five who were stunted for each region covered all countries in each region, including those countries that did not have a survey to contribute to the estimation of prevalence values. The combined number of children who were stunted in the six regions increased from 175.8 million in 1980 to 183.9 million in 1995. China's prevalence rate of 31.4% means that some 36 million children in China are stunted. The number of children affected in Sub-Saharan Africa increased from 26.3 to 42.6 million, an increase of 62%.

In this region, both the prevalence of stunting and the population of children under five years old increased. In the other regions, population size increased while the prevalence of stunting decreased. This resulted in either a decrease or no change in the number of stunted children.

Trends in stunting and underweight move in parallel

Weight-for-age is commonly used as an indicator for malnutrition because weight is easier to measure than height. Weight-for-age reflects linear growth and weight accumulation achieved pre- and post-natally over a long term as well as weight accumulation in the short term. Low weight-for-age may reflect either normal variation in growth or a deficit in growth. Underweight is usually defined as weight <-2 SD below that expected on the basis of the international growth reference. The prevalence of underweight in developing countries is 31%, ranging from 6.5% in South America to around 50% in South Asia (WHO, 1997, p.38). Thus, both stunting and underweight are most prevalent among South Asian children.

Trends in the prevalence of underweight, as reported in the Update 1996, are similar to those for stunting. Sub-Saharan Africa had an increase in the prevalence of underweight from 1985 to 1995, just as it did for stunting (Table 3). Other regions had decreasing trends in the prevalence of both underweight and stunting. Stunting is more prevalent than underweight in all regions. The prevalence of stunting is, on average, 11.5 percentage points¹ higher than the prevalence of underweight. This can be explained by the fact that children who are stunted in early life may attain normal weight later on but remain short.

¹ The regression line for the relationship between stunting and underweight is: stunting prevalence = 11.5 + (0.889) × (underweight prev)

Concepts and indicators

Malnutrition 'results from the interaction between poor diet and disease and leads to most of the anthropometric deficits observed among children in the world's less developed countries' (WHO, 1995a). The conceptual frame work developed by UNICEF and reproduced here on page 79 summarizes current thinking that the immediate causes of malnutrition are poor diet and disease. Poor diet and disease result from the underlying causes of food insecurity, inadequate maternal and child care, and poor health services and environment. The basic causes are social structures and institutions, political systems and ideology, economic distribution, and potential resources.

Height-for-age is one of three anthropometric indices commonly used as an indicator for malnutrition. A deficit in height-for-age does not establish the specific processes that lead a particular child or a group of children to be malnourished. Height-for-age reflects linear growth achieved pre- and post-natally, and its deficits indicate long-term, cumulative effects of inadequacies of health, diet, or care. For children up to about two years of age, height is measured by recumbent length. For older children, height is measured by stature while standing.

Shortness in height refers to a child having low height-forage. Shortness may reflect either normal variation in growth or a deficit in growth. Stunting refers to shortness that is a deficit, i.e., linear growth that failed to reach genetic potential as a result of suboptimal health or nutrition conditions. Beaton et al, (1990) have argued that stunting is a population proxy for multifaceted deprivations. Recent research has found that linear bone growth occurs in an episodic or saltatory process such that a stasis period of one or more days of no growth is punctuated by a daily saltation of growth (Lampl et al., 1992). This research suggests that stunting must result from a decreased frequency of growth events, a decreased amplitude of growth when an event occurs, or both.

Stunting is defined as height <-2 SD below that expected on the basis of the international growth reference. National prevalences of stunting in developing countries range up to 64.2% (WHO, 1997, p.18). There are often large disparities within countries. For example, in Mozambique, 34.0% of children in Maputo province and 74.0% in Zambezia province were stunted according to a national survey done in 1995. Where the prevalence of stunting is high, it can be assumed that most short children are stunted because of environmental reasons. It also follows that where stunting rates are high, the majority of children (and not only those below the traditional cut-off point) are not reaching their growth potential. The process that leads to stunting is thought to occur pre-natally and post-natally, primarily during the first two to three years of life. The cause of stunting probably varies in different settings depending on which nutrient (or nutrients) may be limited and the frequency of infection. Protein as well as energy, zinc and iron have been implicated, as has prolonged infection (Allen, 1994).

The prevalence of stunting has several uses in populations of children under five years of age:

à to identify the most vulnerable areas to target nutrition, health, social and economic interventions;

à to determine priorities for allocation of resources;

à to assess response to interventions in order to make decisions about effectiveness, improvement, discontinuation and modification;

à to promote the survival of children at nutrition risk through appropriate interventions (WHO, 1995).

Consequences of stunting

Malnutrition is an important problem on its own firstly because good nutrition is an essential determinant for well-being, secondly, because good nutrition is a fundamental right (Jonsson, 1996), and thirdly because of the consequences associated with malnutrition. The consequences of malnutrition as indicated by anthropometry include childhood morbidity and mortality, poor physical and mental development and school performance, and reduced adult size and capacity for physical work (WHO, 1995a).

Malnutrition potentiates the effects of infection (Pelletier et al., 1993). Malnourished children have more severe diarrhoeal episodes as measured by duration, risk of dehydration or hospital admission, and associated growth faltering. They also have a higher risk of pneumonia. Exposure to pathogens in the environment can affect the growth of children through several mechanisms. One is the reduction of food intake and poor use of ingested nutrients. In addition, the body has an inflammatory response to many infections. Inflammation may reduce the length of bones because of systemic and local disturbances of normal growth (Sherry, 1994). An unsanitary environment may have broader effects on children than just those associated with particular bouts of overt illness. Children living in poor conditions are constantly exposed to pathogens that cause persistent tow-level challenge to their immune systems. Children who are not apparently infected, that is, who show no clinical illness, may nonetheless have an immunological response that diverts specific nutrients from normal growth and thus restricts length gain (UNICEF, 1997, p. 26).

Malnourished children have a high mortality rate in early life. Those with severe malnutrition are most likely to die. However, because only a small percentage are severely malnourished, most deaths from malnutrition occur among children with mild-to-moderate malnutrition (Pelletier et al., 1993). Malnutrition acts synergistically with disease by increasing case-fatality rates. Stunting is associated with impaired mental development and poor school performance. This association is not a simple causal one because complex environmental, social and economic factors affect both physical growth and mental development (see Box 1).

Stunting in childhood leads to reduced adult size and reduced work capacity. This in turn has an impact on economic productivity at the national level (WHO, 1995a, p. 180). Women of short stature are at greater risk of obstetric complications because of smaller pelvic size. Small women have a greater risk of delivering an infant with low birth weight. This leads to an intergenerational effect since tow-birth-weight infants tend to attain smaller size as adults. As presented later in this chapter, those countries with high rates of stunting also have high rates of low birth weight.

Born malnourished

Stunting is a cumulative process that can begin in utero and continue to about three years after birth. Low-birth-weight (LBW, defined as <2500 g at birth) is an important indicator of foetal/intrauterine nutrition and a strong predictor of subsequent growth and well-being. A recent analysis using data from the WHO Database on Low Birth Weight has quantified the magnitude and described the geographical distribution of this problem for the first time (de Onis et al., 1997). Whereas rates of LBW in industrialized countries are in the range of 6% to 8%, in developing countries LBW is much more common: in South Asia 33%, in Sub-Saharan Africa 16%. In least developed countries, 23% of babies are born with LBW. These average national figures mask enormous variation within countries. In India, for example, rates range from about 10% amongst privileged high-income families to 56% in poor urban slums (UNICEF, 1997a, p. 25). LBW is common in countries with prevalent stunting (Figure 3).

Box 1: Stunting and Mental Development (extracted from a short paper prepared by SM Grantham-McGregor and LC Fernald for the ACC/SCN's Commission on the Nutrition Challenges of the 21st Century. December, 1997) Stunting in poor populations is usually associated with poor mental development. However, the many socio-cultural and economic disadvantages that coexist with stunting (Martorell et al., 1988) may also detrimentally affect mental development. This makes it difficult to determine whether the poor development of stunted children is due to nutrition deficiency or whether stunting is just an indicator of poverty. Thus, it is important to control for social background as much as possible in study design and statistical analysis. Most cross-sectional studies have found significant associations between height-forage and children's cognitive development in preschool and school-age children. Even after controlling for socio-economic conditions, investigators have found significant associations between height-for-age and IQ, cognitive function and school achievement levels in school-age children in many countries. Significant associations have also been found between stunting and poor psychomotor development, fine motor skills and neuro-sensory integration. In populations with high levels of stunting, height in early childhood also predicts IQ at school age. Stunted children's cognitive function is more likely to be detrimentally affected by short-term hunger than non-stunted children (Simeon and Grantham-McGregor, 1989). The only supplementation study aimed specifically at stunted children was conducted with stunted and non-stunted Jamaican children aged nine to 24 months (Grantham-McGregor et al., 1991). The stunted children received nutritional supplementation for two years with or without psychosocial stimulation. Supplementation and stimulation produced independent benefits to the children's mental and motor development. The benefits from a combination of supplementation and stimulation were additive, and only the children receiving both treatments caught up to the non-stunted control group in development levels. The implications of these findings are that at least part of the deficit in the development of stunted children is due to poor nutrition. However, both stimulation and supplementation are necessary to improve the development of stunted children to culturally appropriate levels. The precise mechanism linking stunting to poor mental development is unknown. It is possible that the mechanism varies according to which nutrients are deficient, or that several mechanisms could act together. One possibility is that undernutrition causes poor motor development and apathy which in turn reduce a child's ability for environmental exploration and skill acquisition (Levitsky, 1979). Reduced activity has been described in iron, zinc and energy deficiencies. Another possible mechanism is that the children's small size could lead adults to treat them like younger children and not provide age-appropriate stimulation. Undernutrition could have a direct effect on children's central nervous system. Stunted children have smaller heads than non-stunted children, and in one study, head size in early childhood was a stronger predictor of IQ at seven years of age than other previous or current anthropometric measures (Grantham-McGregor et al., 1997). A more speculative explanation is that raised anxiety levels, as evidenced by heightened cortisol, could contribute to poor cognition and behaviour.

Total low-birth-weight figures combine term and pre-term births. In most developing countries, most low-birth-weight infants are full-term, but have intrauterine growth retardation (IUGR). IUGR is a major clinical and public health problem in developing counties. It is defined as a deficit in weight of the foetus relative to that expected for gestational age. IUGR is an important cause of stunting, because small foetal size leads to LBW, which in turn is highly related to small postnatal size.

'IUGR contributes to closing the intergenerational cycle of poverty, disease and malnutrition'

(de Onis et al., 1997)

Figure 3 shows that LBW is very common in countries with prevalent stunting. The relationship between these two indicators has been analyzed empirically by Osmani (1997). He concluded that the very high prevalence of LBW lies behind the high rate of undernutrition in South Asia.

Figure 3: Prevalence of Low Birth Weight (LBW) and Stunting

In many countries, these high rates of impaired foetal growth exceed the recommended levels for triggering public health action. IUGR in excess of 20% has been recommended by WHO as the cut-off; in the absence of information on gestational age, a prevalence of >15% LBW may be used. Population-wide interventions are needed in these countries. A recent review of randomized controlled studies evaluated the effectiveness of interventions to prevent or treat impaired foetal growth (Gülmezoglu et al., 1997). Interventions likely to be beneficial include balanced protein/energy supplementation where diets are deficient, smoking cessation and anti-malarial chemoprophylaxis in primigravidae. Zinc, folate and magnesium supplementation during gestation were flagged as meriting further research. The same authors emphasize that appropriate combinations of interventions should also be evaluated (de Onis et al., 1997a). Research in the Gambia showed that low birth weight can be reduced by about 40% and infant mortality by 50% through improved food intake during pregnancy (Ceesay et al., 1997).

The issue of age

Susceptibility to poor linear growth changes as the young child ages. The velocity of linear growth is highest during the first months of life, hence this is a period of particularly increased susceptibility. Exclusively breastfed infants are protected against early post-natal stunting (WHO, 1995). Consequently, for most infants in the developing world, the post-natal period that is most susceptible to poor linear growth is after 3-6 months and up to 24-36 months. After this time, it is thought that poor conditions have less of an effect on linear growth because growth velocity is much lower.

This age differential in the susceptibility to poor linear growth after birth has important implications for assessment and monitoring. Children aged from birth to 59 months do not form a homogenous group. Prevalence estimates are affected by the distribution of ages included in the survey. Prevalence estimates may be higher if older children are included than if they are not, and may be tower if infants from birth to three months are included than if they are not. This implies that separate estimates of the prevalence of stunting should be made for younger (i.e., birth to 24 months) and older children for the purposes of describing trends over time. This issue is discussed further in Annex 3. For this Report there were insufficient data to estimate trends in stunting or regional prevalences by age group.

There is now convincing evidence (as reviewed by Scrimshaw, 1997) that foetal malnutrition lays the foundation for adult chronic disease. This has enormous consequences for policy-setting in countries where LBW is a public health problem.

The pre-pregnancy determinants of LBW, as well as adequate nutrition throughout pregnancy, require priority attention.

Understanding regional differences in stunting

Table 4 summarizes the numbers of countries by region showing increasing or decreasing trends in stunting over the period 1980-1995. There were 61 countries with more than one survey; stunting decreased in 43 of these over the period 1980-1995. Stunting increased in 18 countries. This section discusses factors associated with the prevalence and trends in stunting for the six regions. Further information about the regions can be found in Update on the Nutrition Situation, 1996 (ACC/SCN 1996), the Human Development Report 1997 (UNDP, 1997) and Malnutrition in South Asia: A Regional Profile (UNICEF, 1997a).

Table 4: Numbers of Countries with more than one Survey in each Region Showing Increasing or Decreasing Trends

Region	Number of Countries Decreasing	Number of Countries Increasing
Sub-Saharan Africa	13	12
Near East/North Africa	5	2
South Asia	6	0
South East Asia	4	1
Middle America/Caribbean	8	2
South America	7	1
Total	43	18

Sub-Saharan Africa

The trend in Sub-Saharan Africa for stunting is very disturbing. For the region as a whole, no progress has been made in reducing the prevalence of child malnutrition over the past 15 years, and there is some indication that the prevalence has increased. Because of population growth, the number of children who are stunted has been increasing substantially. Two countries, Nigeria and Ethiopia, accounted for about half (52%) of the stunted children in this region in 1995.

There are important differences amongst African countries. Trends defined two broad groups of countries: those that improved over time and those that worsened over time. Of the 25 countries with more than one survey in this region (noted in Annex 1), rates of stunting are on the decline in about half. These 13 countries had an estimated prevalence of stunting in 1995 of 35.2% (equivalent to over 15 million children), with the trend in stunting prevalence decreasing on average by 0.442 percentage points per year (p<0.0001). Countries achieving the highest rates of decline included Nigeria and Zimbabwe. For those Sub-Saharan African countries where stunting is increasing, the estimated prevalence in 1995 was 48.3% (equivalent to over 11 million children). The rate of increase for this group was on average 0.837 percentage points per year (p<0.0004). Countries in this group included Ethiopia and Madagascar.

Sub-Saharan Africa has been characterized by falling gross national product per capita. Contributors to this fall in economic resources have been drought, civil wars and political instability, world-wide recession, falling commodity prices and weaknesses in structural adjustment programmes. Although stunting is not restricted to drought-prone areas in Sub-Saharan Africa (Pelletier et al., 1995), seasonality is an issue in this region. LBW is less prevalent in Sub-Saharan Africa than in South Asia. However, LBW increases when food is short and the demands for agricultural labour are high. Debt service requirements reduce available external resources, limiting national expenditures for social services. Furthermore, the poverty rate is about 40% in this region, and poverty is increasing. Only about one-third of rural Africans have access to safe water and adequate sanitation. Immunization rates are still far below target levels in many countries; the proportion of fully immunized children is just over half (56%) in Sub-Saharan Africa as a whole. Illiteracy is still high amongst African women, especially in rural areas. The fertility rate is the highest of any region, but is declining and is now just below 6.0 births per woman.

Near East/North Africa

The trend in Near East/North Africa shows a decrease early in the period, but no progress at the end of the period. The number of children who are stunted had been decreasing, but then did not change between 1990 and 1995. Nevertheless, this region now has the second lowest prevalence of stunting (22.2%) among the six regions. Three countries, Egypt, Iran and Turkey, accounted for about half (51%) of the stunted children in this region in 1995.

As in Sub-Saharan Africa, country trends in the prevalence of stunting in this region were highly variable. Of the seven countries with more than one survey in this region, two had increased (Algeria and Turkey) and five had decreased prevalence of stunting over the period.

Near East/North Africa had a GNP per capita of US$ 1710 (1995) with a negative rate of growth during 1980-91. Economies vary greatly in this region, and oil prices have great influence on the economies. Other factors influencing economic growth are civil wars and political instability, worldwide recession, economic sanctions and, at least in the short term, structural adjustment programmes. The poverty rate in this region is low. Access to safe water is over 80% and immunization coverage is over 85%. Dietary energy supply is very high in this region. Although overall literacy rates continue to improve, there are still significant gender differences (in 1995, 70% for males and 47% for females). The fertility rate stands now at 4.5 births per woman.

Box 2 presents a perspective on stunting in Central Asia and the Caucasus.

South Asia

South Asia showed the second highest improvement in the prevalence of stunting among the six regions, with a decline of 0.84 percentage point per year. However, this region has by far the highest prevalence of stunting among the six regions at 53.5% in 1995. Two countries, India with 68.2 million and Pakistan with 11.4 million, together accounted for 88% of the stunted children in this region in 1995. All of the six countries with more than one survey in this region had decreased prevalence of stunting over the period.

South Asia has been characterized by a slowly rising gross national product per capita. However the poverty rate is high; some 47% of the population live on <$1 per day. South Asia has about two-fifths of the income-poor of the developing world. Access to safe water is, on average, quite high at around 80%. However, poor hygiene and sanitation, linked with overcrowding, are important determinants of stunting in South Asia. Immunization coverage is 77% for the region as a whole, but significantly lower for Nepal at 45%. Dietary energy supply (now over 2200 kcals/person/day) and access to health services have improved at the national level, but, there are very high rates of underweight among women, poor weight gain during pregnancy, and associated LBW. Bangladesh, India and Pakistan have the highest rates of LBW in the world (50%, 33% and 25% respectively). Poor child feeding practices or poor child care also require priority attention.

The primary school enrolment ratio for girls is still under two-thirds (63%). The fertility rate has decreased slowly from 4.3 to 3.5 in the last five years. The low status of women and the quality of care offered to women and children in this region are matters of serious concern to nutrition (ul Haq, 1997; Ramalingaswami et al., 1996 and Osmani, 1997).

Box 2: Stunting in Central Asia and the Caucasus Prepared for the ACC/SCN by H. Peimani In its 1998 State of the World's Children Report, UNICEF called attention to the rising rate of stunting amongst Russian children. In fact, stunting is widespread in many countries of the former Soviet Union, and gives cause for concern. Central Asia comprises five new states with a combined preschool child population of about 7 million. Kazakhstan and Uzbekistan are the two most populous and industrialized Central Asian states, with GNP per capita of US$ 1330 and US$ 970, respectively. A national survey in Kazakhstan in 1995 reported that 15.8% of preschool children were stunted. Disparities within the country were significant. Rates range from 7% in the northern districts along the Russia border - where the population is mainly of Russian origin - to 22.7% in the south where incomes are lower and the population is mainly ethnic Kazakh. Uzbekistan reported in 1996 a national prevalence rate of 31.3% in children up to three years of age. One severely affected district reported 20.3% <-3 SD and 39.8% <-2 SD. In the capital city, Tashkent, more than one in five children is stunted. Tajikistan is Persian-speaking, mainly agrarian and very poor with a per capita GNP of US$ 340 in 1995. A 1996 survey showed that stunting affects the majority of young children in some districts: 51.3% in Aini and 53.9% in Varzob. There is no national prevalence rate for the country. Tajikistan has experienced a civil war (now over), which has damaged the existing infrastructure and agricultural resources. Only 60% of the population have access to safe water, and only 49% in the rural area. Its neighbour to the south, Afghanistan has even higher stunting rates in some provinces: 66% in three provinces surveyed in 1995. Prevalence rates of stunting in preschool children in the Caucasus are: 11.9% in Armenia (1993) and 22.2% in Azerbaijan (1996). Internally displaced children in Azerbaijan are markedly worse off, with a prevalence rate of 30.5%. The civil war in Azerbaijan and the ten-year conflict between Armenia and Azerbaijan have damaged ailing infrastructures. Destruction of farms and other food-producing centres during military engagements have reduced the production and availability of food and sharply increased food prices. Unemployment is very high in these two countries. The break-up of the Soviet Union into 15 independent states did not remove the problems of the centrally-controlled economy. Transition from the old system to a type of free-enterprise economy has resulted in a severe short-term fall in living standards. These countries have seen significant reductions in industrial and agricultural production, rising unemployment and underemployment, low wages, shortages of food, poor distribution networks, high prices and partial or total removal of government subsidies. Access to health care has declined due to reduced funding, closures and cutbacks.

South East Asia

South East Asia showed the sharpest improvement in the prevalence of stunting among the six regions, with a decline of 0.90 percentage points per year. Given the dramatic decline in poverty in this region, one might expect a lower overall prevalence of stunting (38.3% in 1995). Two countries, Indonesia with 9.2 million and Viet Nam with 4.7 million, together accounted for slightly less than half (46%) of the stunted children in this region in 1995. Four of the five countries with more than one survey in this region had decreased prevalence of stunting over the period. The region will be influenced significantly by future trends in Indonesia, which has the largest child population in this region and for which there is only one survey now. The prevalence rate in Indonesia was 42.2% in 1995.

South East Asia has had rapid economic expansion, as seen in quickly rising gross national product per capita, especially until 1992. The poverty rate is about 14%, substantially less than in South Asia and Sub-Saharan Africa. This region's current financial crisis resulting in widespread unemployment will have a negative impact on child nutrition unless social safety nets are in place. Access to safe water is around two-thirds, but access to sanitation is low (35%), especially in the rural areas (18%). Total fertility rates are nearly down to two births per woman. Immunization coverage exceeds 90%.

Middle America/Caribbean

Of the five regions that had improvements in the prevalence of stunting, Middle America/Caribbean had the least improvement, with a reduction of only 0.26 percentage points per year. There was no change in the number of stunted children over the period. Mexico with 3.9 million accounted for two-thirds of the stunted children in this region in 1995. Eight of the ten countries with more than one survey in this region had decreased prevalence of stunting over the period. In Guatemala, the national rate of stunting is very high at about 48%. This rate stands in stark contrast to that of Costa Rica (1.6%) and Panama (5.7%).

Middle America/Caribbean has had rapid economic expansion, as seen in quickly rising gross national product per capita, especially since 1988. However, poverty is rising in spite of overall economic growth. Immunization coverage improved substantially and reached 75% in Mexico, 78% in the Dominican Republic and Nicaragua.

Dietary energy supply was relatively high and remained constant. Female education increased slowly, while the fertility rate (about three births per woman) has been decreasing slowly.

South America

South America showed a reduction of 0.81 percentage points per year. By the end of the period this region had the smallest number of stunted children among the six regions. Brazil with 1.9 million accounted for 41% of the stunted children in this region in 1995. Seven of the eight countries with more than one survey in this region had decreased prevalence of stunting over the period, while Venezuela had an increase. Several countries in this region have significantly high rates of obesity among children.

South America has had rapid economic expansion, as seen in quickly rising gross national products per capita. The poverty rate in Latin America as a whole is 24%, but income maldistribution persists in spite of overall economic growth. Access to safe water and access to health care services remain high. Immunization coverage has improved and is relatively high. Dietary energy supply was relatively high and has remained constant. Female education has increased substantially, while the fertility rate (about three births per woman) has been decreasing slowly.

China

The prevalence of stunting in China in 1992 was 31.4%. This rate translates into some 36 million Chinese children who are stunted. The national trend in China could not be examined because there is only one national survey available. Beaton (1993, p. 39) documented a strong decreasing secular trend in stunting in China between 1975 and 1985. The trend in the prevalence of underweight was about -0.8 percentage points per year from 1985 to 1995 (ACC/SCN, 1996). China has achieved huge reductions in human poverty and rapid economic growth, although some of these faltered in the late 1980s. Dietary energy supply is relatively high and increasing. Access to safe water in rural areas is still low (56%) and access to adequate sanitation is very low (7%). On the other hand, immunization coverage is nearly universal (97%). About half of girls are enrolled in secondary education, and the proportion is increasing. The total fertility rate has fallen to below two children born per woman.

Further considerations

This chapter has examined both the prevalence of stunting from a cross-sectional perspective and the trends in the prevalence of stunting from a longitudinal perspective. A recent cross-sectional study found that about three-quarters of the variability in national prevalence of stunting could be explained by national and regional factors (Frongillo et al, 1997). The study shows that higher food availability, female literacy and gross product were the most important national factors associated with lower prevalence of stunting. Regional differences persisted, even after accounting for national factors. Social and human development factors were important even after accounting for economic factors.

There is substantial variability among countries in the prevalence of stunting; there is also substantial variation within countries. Most stunting is found in the poorest regions of a country, so those with reasonably good economic and social statistics on average can have high rates of stunting if there is substantial inequity across the country. Mexico is just one example of a country that has undertaken substantial economic changes, resulting in important benefits for some and costs for many others (UNDP, 1997). As a result, Mexico has a higher prevalence of stunting than would be expected on the basis of its average national statistics (Frongillo et al., 1997).

A recent synthesis of eleven detailed country studies in How Nutrition Improves (1996) suggested some explanations for how improvements in child malnutrition occur. Economic development is positively related to nutritional improvement, and equitable growth strategies seem to be better for reducing poverty and improving nutrition than compensatory poverty alleviation programmes. Economic growth is important, but nutritional improvements can move in advance of that growth. Investments in health and education (particularly of women) can help to improve nutrition if the investments reach the malnourished, are of good quality and are well-distributed. Policies encouraging community-based programmes accelerate improvement. Box 3 illustrates that significant reductions in child stunting rates will take at least two to four years. The impact of this improvement on educational achievement and economic growth may take decades before it becomes evident.

Conclusions

This examination of stunting shows that the six regions of the developing world have very different prevalences of stunting, and that the regions have progressed quite differently since 1980. South Asia has the highest prevalence of stunting but, with South East Asia and South America, has had the fastest rate of decrease in stunting. Sub-Saharan Africa as a whole has made no progress in reducing the prevalence of stunting and has, in fact, a higher prevalence of stunting in 1995 than in 1980.

During the 1980 to 1995 period, the trends for improvement or deterioration in stunting have been stable; that is, there is no evidence for a change in the rates of progress. The only exception is Near East/North Africa which shows a slowdown in progress of reducing stunting over the period.

The patterns of stunting across the regions reported here and the patterns of underweight reported in Update on the Nutrition Situation, 1996 are very much the same for both the prevalence and trends in the prevalence. The prevalence of stunting is uniformly higher than the prevalence of underweight. Stunting is a cumulative process of poor growth that primarily occurs before the age of three years and is not easily reversed. In contrast, underweight reflects a cumulative process of poor growth modified by the effects of periods when the older preschool child may have access to more food that results in weight gain. Thus, for purposes of differentiating regions and countries, stunting is a better indicator for quantifying the number of children who suffer the consequences of poor conditions for young children than is underweight. Stunting is a stricter standard because it is most sensitive for determining malnutrition in countries and because it separates early poor nutrition from later excess relative weight. However, both stunting and underweight equally capture regional differences in prevalence and trends in prevalence.

The pattern of child malnutrition across the regions is paralleled by patterns of underweight in adults. Data compiled by the WHO Programme of Nutrition show that the percentage of adults who are underweight in Africa, Asia and Latin America are 5.1, 9.4 and 3.8, respectively (Bailey, personal communication). Underweight was assessed by the percentage of adults with body mass index (BMI - weight in kg divided by height in meters squared) less than 17. Thus, Asia has by far the highest prevalence of adult underweight, just as it does for child stunting and underweight.

Box 3: Reducing child stunting in the best circumstances Prepared for the ACC/SCN by S Morris and M Ruel, IFPRI A simulation exercise was carried out to estimate the rate of response of different age groups to a hypothetical improvement in general conditions. A hypothetical data set with 100 children per age group (six-month intervals) was generated, with a pattern of height-for-age Z-scores that is typical for children growing up in developing countries. Average length-for-age at birth is close to zero (corresponding to the 50th percentile of the WHO/NCHS reference). Height-for-age Z-score at six months was set at approximately -0.5, and decreases to reach -1 at about 12 months and -2 at about 30 months of age. The mean height-for-age Z-score stabilizes at this low value thereafter. The simulation examined how long it would take for stunting to be appreciably reduced amongst these children if sustained improvements were to be achieved over a five-year period, such that all children would grow at the expected velocity for their age (median of growth velocity references). The figure below shows the patterns of growth of this cohort over five years, expressed in mean height-for-age Z-scores. The intention was not to simulate a short-term achievable goal, but rather to illustrate what can be expected in terms of rate of response, under the very best circumstances. It does not take into account factors such as poor foetal growth or maternal malnutrition which may limit children's post-natal growth potential and prevent them from growing at the reference median. Children younger than 24 months of age responded much more rapidly to the improvement than older children. This is because of the high rate of growth in young infants. In year one after the intervention, the mean height-for-age Z-score of 0-24 month old children was between -0.77 to -0.16 (Figure 4). The prevalence of stunted children dropped from 12.5% to 1.3% among this age group (not shown). This can be compared with a mean Z-score as low as -1.94 among 24-60 month old children in year 1 after the intervention. The prevalence of stunting among children 42 months and older is not improved even after one year of optimal growth velocity. This is because the expected growth rates at this age are much slower than at younger ages. It took four years of sustained improved growth for stunting to finally be eliminated among the older age groups, and five years for the whole growth curve to be at or above the reference throughout the age range. Figure 4: Mean height-for-age Z-score by age and year following improvement in age-specific growth rates to NCHS median levels, simulated dataset

The results reported here are consistent with those compiled by de Onis and Blössner and published recently by WHO (1997). The data used for both reports were essentially the same, but different analytical methods and regional classifications were used. The analytical method used for the examination of stunting in this chapter was chosen as the best for the estimation of the trends in the prevalence of stunting. The fact that different analytical methods yielded the same global picture of stunting increases confidence in the findings.

WHO has proposed that the goal for reducing stunting be:

'the prevalence of stunted children in any country (and within specific subgroups) should be less than 20% by the year 2020...'

Of 61 countries with trend data on stunting, only about 16 - barely a quarter - meet the target now, i.e. have a prevalence rate of around 20% or less. If past positive trends continue, only some nine additional countries might achieve the goal by 2010 and possibly one-half of these 61 countries would have prevalence rates below 20% by 2020. Yet countries as diverse as Brazil, Chile, Colombia, Costa Rica, Jamaica, Oman, Panama, Sri Lanka, Tunisia and Zimbabwe appear to have made good progress.

To reduce malnutrition rapidly requires focused and systematic action in the areas of health, food security, and child and maternal care. Access to education, health care and safe water, protection from illness and ensuring adequate micronutrient intakes are key elements, together with some system for community follow-up and support of severely and moderately malnourished children.

To have these elements in place is not easy, but when they are, experience shows that rates of malnutrition can be reduced rapidly. This was shown in Oman, Thailand, Uruguay, Viet Nam, and Zimbabwe. Rapid reduction of malnutrition is possible, and is urgently needed. This should be a specific component of all anti-poverty efforts.

References

ACC/SCN (1992) Second Report on the World Nutrition Situation. ACC/SCN, Geneva.

ACC/SCN (1994) Update on the Nutrition Situation, 1994. ACC/SCN, Geneva.

ACC/SCN (1996) Update on the Nutrition Situation, 1996. ACC/SCN, Geneva.

ACC/SCN (1996a) How Nutrition Improves. ACC/SCN, Geneva.

Allen LH (1994) Nutritional influences in linear growth: a general review. European Journal of Clinical Nutrition, 48: S1, S75-S89.

Beaton GH, Kelly A, Kevany J, Martorell R and Mason J (1990) Appropriate Uses of Anthropometric Indices in Children: A Report Based on an ACC/SCN Workshop. ACC/SCN State-of-the-Art Series Nutrition Policy Discussion Paper No. 7.

Beaton G (1993) Which age groups should be targeted for supplementary feeding? In: Nutritional Issues in Food Aid. ACC/SCN Symposium Report. Nutrition Policy Discussion Paper No. 12. Geneva.

Ceesay S M, Prentice A M, Cote T J, Ford F, Weaver L T, Poskitt E M E and Whitehead R G (1997) Effects on birth weight and perinatal mortality of maternal dietary supplements in rural Gambia: 5 year randomised controlled trial. British Medical Journal, 315, 786-790.

de Onis M, Blössner M, and Villar J (1997) Levels and patterns of intrauterine growth retardation in developing countries. European Journal of Clinical Nutrition, 52, S1, 5-15.

de Onis M, Villar J, Gülmezoglu M (1997a) Nutritional interventions to prevent intrauterine growth retardation: evidence from randomized controlled trials. European Journal of Clinical Nutrition, 52:S1, 83-89.

Frongillo E A Jr, de Onis M and Hanson K M P (1997) Socio-economic and demographic factors are associated with worldwide patterns of stunting and wasting. Journal of Nutrition, 127. In press.

Grantham-McGregor S, Powell CA, Walker SP and Himes JH (1991) Nutritional supplementation, psychosocial stimulation, and mental development of stunted children: the Jamaica study. Lancet 338, 1-5.

Grantham-McGregor S, Walker SP Chang SM and Powell CA (1997) Effects of early childhood supplementation with and without stimulation on later development in stunted Jamaican children. American Journal of Clinical Nutrition, 66, 247-253.

Grantham-McGregor S and Fernald LC (1997) Stunting and mental development. Paper prepared for the ACC/SCN Commission on Nutrition in the 21st Century.

Gülmezoglu M, de Onis M and Villar J (1997) Effectiveness of interventions to prevent or treat impaired fetal growth. Obstetrical and Gynecological Survey, 52(2), 139-149.

IFPRI (1995) A 2020 vision for food, agriculture, and the environment: the vision, challenge, and recommended action. International Food Policy Research Institute, Washington, DC.

Jonsson U (1996) Nutrition and the convention on the rights of the child. Food Policy. 21:41-55.

Lampl M, Veldhuis J D and Johnson M L (1992) Saltation and stasis: a model of human growth. Science, 258, 801-803.

Levitsky DA (1979) Malnutrition and hunger to learn. In: Malnutrition, Environment and Behavior, Eds Levinsky, Ithaca N.Y.

Martorell R, Mendoza F and Castillo R (1988) Poverty and stature in children. In: Linear growth retardation in less developed countries, Nestles nutrition workshop series, Vol. 14, Ed: Waterlow JC. New York.

Osmani S R (1997) Poverty and nutrition in South Asia. Text of the Abraham Horwitz Lecture delivered at the ACC/SCN Symposium on Nutrition and Poverty held at the ACC/SCN Session in Kathmandu (UNICEF Regional Office for South Asia), 17-18 March 1997.

Pelletier D L, Frongillo E A Jr and Habicht J P (1993) Epidemiological evidence for a potentiating effect of malnutrition on child mortality. American Journal of Public Health, 83, 1130-1133.

Pelletier DL, Deneke K, Kidane Y, Haile B and Negussie F (1995) The food-first bias in nutrition: lessons from Ethiopia. Food Policy. 20, 279-298.

Ramalingaswami V, Jonsson U and Rohde J (1996) Commentary: the Asian enigma. In: The Progress of Nations 1996. United Nations Children's Fund, New York.

Scrimshaw N (1997) The relation between fetal malnutrition and chronic disease later in life. British Medical Journal, 315, 825-826.

Sherry T M (1994) The effects of the inflammatory response on bone growth. European Journal of Clinical Nutrition, 48, S190-S198.

Simeon DT and Grantham-McGregor S (1989) Effects of missing breakfast on the cognitive functions of school children of different nutritional status. American Journal of Clinical Nutrition, 49, 646-653.

ul Haq M (1997) Human Development in South Asia 1997. Oxford University Press and the Human Development Centre, New York.

UNDP (1997) Human Development Report 1997. Oxford University Press and United Nations Development Programme, New York.

UNICEF (1997) The State of the World's Children 1998. Oxford University Press and United Nations Children's Fund, Oxford.

UNICEF (1997a) Malnutrition in South Asia: a regional profile. UNICEF, Kathmandu.

WHO (1995) Working Group on Infant Growth: An evaluation of infant growth: the use and interpretation of anthropometry in infants. Bulletin of the World Health Organization, 73, 165-174.

WHO (1995a) Physical status: the use and interpretation of anthropometry. Report of a WHO Expert Committee. WHO Technical Report Series 854. WHO, Geneva.

WHO (1997) Global Database on Child Growth and Malnutrition. WHO, Geneva.

WHO (1998) Health for all in the 21st Century. EB101/8, Geneva.