Bulletin of the World Health Organization

Progress towards Millennium Development Goal 1 in Latin America and the Caribbean: the importance of the choice of indicator for undernutrition

Chessa K Lutter a, Camila M Chaparro b & Sergio Muñoz a

a. Pan American Health Organization, 525 23rd Street NW, Washington, DC, 20037, United States of America (USA).
b. Food and Nutrition Technical Assistance II, Academy for Educational Development, Washington, USA.

Correspondence to Chessa K Lutter (e-mail: luttterch@paho.org).

(Submitted: 09 April 2010 – Revised version received: 26 July 2010 – Accepted: 10 August 2010 – Published online: 01 November 2010.)

Bulletin of the World Health Organization 2011;89:22-30. doi: 10.2471/BLT.10.078618

Introduction

Child growth is a gauge of individual and population-level well-being.13 Child height, in particular, reflects the cumulative effects of intergenerational poverty, poor maternal and early childhood nutrition and repeated childhood episodes of illness.46 It also reflects insufficient household purchasing power and poor access to education, housing, water and sanitation, and health services. Height not only tells the story of a nation with regard to maternal and child health and nutrition, but also of how equitably these have been distributed. This is particularly true in Latin America, which has some of the most pronounced inequity in the world.7

Maternal and child undernutrition contributed to more than one-third of all child deaths and to more than 10% of the total global disease burden in 2005.8 Of the nutritional factors leading to child death, stunting, severe wasting and intrauterine growth retardation together were responsible for 2.2 million deaths and 21% of disability-adjusted life-years. Therefore, improving nutrition in infants and young children is essential to achieving the Millennium Development Goals (MDGs) related to child survival (MDG 4) and the eradication of extreme poverty and hunger (MDG 1). Because of the intergenerational and far-reaching effects of early childhood nutrition on health and cognitive development,811 improving nutrition in infants and young children would indirectly contribute to progress towards achieving MDGs pertaining to universal primary education, gender equity, the empowerment of women and improved maternal health.

Growth in height and weight in accordance with the new child growth standards of the World Health Organization (WHO),12 which are internationally accepted, is used to evaluate the nutritional status of individual children and paediatric populations. Both weight gain and linear growth most directly reflect dietary intake and the effects of illness, as well as the interaction between these factors.5,6,13 However, linear growth is probably more indicative of nutrition in the intrauterine environment and of subsequent dietary quality. For population-level assessment, stunting is a better indicator than underweight because it reflects a cumulative growth deficit,14,15 which, unlike weight, cannot be reversed and is usually permanent when children remain in an environment marked by poverty.16

Because child height captures the effects of a broad range of economic and social influences on health, it is useful for monitoring progress towards several health and development objectives, including MDG 1. Unfortunately, underweight is the indicator of child undernutrition officially used to monitor progress towards achieving one of the targets linked to MDG 1, namely to reduce the prevalence of undernutrition by half between 1990 and 2015 throughout the world. This indicator has the disadvantage that it cannot distinguish between a drop in the prevalence of underweight resulting from improved linear growth or from increased weight-for-length/height. Increased weight-for-length/height across the population is not desirable because overweight in childhood predisposes to chronic non-communicable diseases in adulthood. The choice of indicator is decisive in determining whether countries are on track to meet the nutritional component of MDG 1 or not. In this paper we use both stunting and underweight as indicators to make this determination for countries in Latin America and the Caribbean and compare the results obtained with both indicators.

Methods

We applied the new WHO Child Growth Standards to anthropometric data provided by the Centers for Disease Control and Prevention in Atlanta, United States of America (USA), or downloaded, with permission, from the Demographic and Health Survey (DHS) web site (http://www.measuredhs.com/accesssurveys/start.cfm). All nationally-representative, publicly available data sets with anthropometric data for countries in Latin America and the Caribbean were included. Multiple data sets were available for 10 countries (the Plurinational State of Bolivia, Brazil, Colombia, the Dominican Republic, El Salvador, Guatemala, Haiti, Honduras, Nicaragua and Peru). Data from published reports, which used the new WHO Child Growth Standards, were extracted for the Plurinational State of Bolivia (2008),18 Brazil (2006),19 Costa Rica (1982, 1996, 2008),20 El Salvador (1966, 1988, 2008)20,21 the Dominican Republic (2007),17 Guatemala (1966)20, Honduras (1966, 1987),20 Mexico (1988, 1999, 2006),22 Nicaragua (1966, 1993, 2008)20,23 and Panama (1966, 1997, 2003).20 The preliminary report of the most recent national survey in Guatemala (2008) used the growth reference population of the National Center for Health Statistics (NCHS) rather than WHO to report prevalence figures and thus could not be included in the present analysis. Only countries with nationally-representative surveys were included in the analysis. This limited the analysis to the subset of 13 countries with trend data represented in this study. The surveys are identified by country name, the year in which they were performed and sample size (Table 1). The most recent Peruvian survey, Peru 2004–2008, is ongoing and is being completed in several cycles. The anthropometric data used for all analyses for Peru, except for the subregional analyses, were collected in 2005. The data for the subregional analyses were collected in 2005, 2007 and the first round of 2008 (expanded survey). For the DHS data sets, starting in approximately 1999 (the fourth phase of surveys) all children less than 5 years of age in the household, and not just the children of the respondent woman, were measured for height and weight.

Analyses were performed using Statistical Analysis Systems for Windows Version 9.1 (SAS Institute, Cary, USA) or STATA for Windows Version 11.0 (StataCorp LP, College Station, USA). Individual z-scores for three anthropometric indices (weight-for-age, length/height-for-age, weight-for-length/height) in accordance with the WHO Standards12 were calculated using the SAS macro downloaded from the WHO web site (www.who.int/childgrowth/software). From the country survey data available we derived summary statistics to calculate the overall prevalence of underweight, stunting, wasting and overweight, along with the corresponding 95% confidence intervals (CIs), as well as mean z-scores nationally and within particular subgroups (e.g. subregions, age categories). We defined underweight as a weight-for-age more than 2 standard deviations (SD) below the median; stunting as a length/height-for-age more than 2 SD below the median; wasting as a weight-for-length/height of more than 2 SD below the median, and overweight as a weight-for-length/height more than 2 SD above the median.

We used independent linear regression models and the prevalence figures for each country in Table 1 to study prevalence trends over time for stunting and underweight. Using regression coefficients, we estimated the prevalences for 1990 and 2015 and computed the yearly percent reduction in prevalence between 1990 and 2015. We then compared the estimated prevalences of stunting and underweight for 2015 with the target established for MDG 1(half of the estimated 1990 baseline prevalence). To evaluate progress we applied the classifications of “on track” and “insufficient,” used in the Countdown to 201524; countries were deemed to be “on track” if they were within 2 percentage points of the target prevalence.

Results

Stunting is the most common growth deficiency in Latin America and Haiti among children aged less than 5 years. Its prevalence in this group ranges from 5.6% in Costa Rica to 54.5% in Guatemala (Fig. 1). In contrast, the prevalence of underweight is less than 9% in all countries except Guatemala (18.0%) and Haiti (19.2%). In approximately half of the countries, the prevalence of wasting is lower than expected in a population with a normal distribution of weight-for-length/height. It is highest in Haiti (10.3%) and lowest in Honduras (1.4%), respectively. Overweight ranges from about 4% in Colombia and Haiti to 9% or more in Argentina, the Plurinational State of Bolivia, the Dominican Republic, Panama and Peru.

Fig. 1. Estimated prevalence of underweight,a stunting,b wastingc and overweightd according to the WHO Child Growth Standards, most recent survey data for 15 countries of Latin America and the Caribbean, 2002–2008
Fig. 1. Estimated prevalence of underweight,<sup>a</sup> stunting,<sup>b</sup> wasting<sup>c</sup> and overweight<sup>d</sup> according to the WHO Child Growth Standards, most recent survey data for 15 countries of Latin America and the Caribbean, 2002–2008
a Weight-for-age 2 or more standard deviations below the median.b Length/height-for-age 2 or more standard deviations below the median.c Weight-for-length/height 2 or more standard deviations below the median.d Weight-for-length/height 2 or more standard deviations above the median.e Data for overweight are not available for Brazil, Costa Rica and Nicaragua.

National prevalence estimates mask enormous within-country differences that are most pronounced in the case of stunting (Fig. 2). These within-country differences are sometimes larger than the differences observed between different countries in the region. Peru showed some of the largest within-country differences; the prevalence of stunting for the country as a whole was 29.8% but subregional estimates ranged from a low of 6.7% to a high of 60.1%. In general, within-country differences in the prevalence of underweight followed a similar pattern; regions with the highest prevalence of stunting also had the highest prevalence of underweight.

Fig. 2. Within-country differences in the prevalencea of stunting, survey data for 12 countries of Latin America and the Caribbean, 1996–2006
Fig. 2. Within-country differences in the prevalence<sup>a</sup> of stunting, survey data for 12 countries of Latin America and the Caribbean, 1996–2006
a Each point represents the prevalence in each region; the country mean prevalence is represented by the horizontal line.

Although stunting has decreased gradually in all countries over the period covered by the surveys, different patterns emerge (Table 1). Brazil, Colombia, El Salvador, Honduras, Mexico and Nicaragua have experienced large declines in stunting between surveys. Brazil, with the largest decline, has seen a drop in stunting prevalence from 19.9% in 1989 to 6.8% in 2006. Costa Rica stands out for its very low prevalence figures throughout the measurement period. In the Plurinational State of Bolivia, stunting declined significantly between 1989 and 1994, remained stagnant between 1994 and 2003 and then declined again between 2003 and 2008. In Guatemala, stunting declined significantly between 1987 and 1995 and again between 1995 and 1999, but not between 1999 and 2002. Peru also had a large decline in stunting between 1992 and 1996, but not in the two subsequent surveys. El Salvador showed important declines until the last two surveys but none since. Haiti had a decline of almost 10 percentage points between 1995 and 2000 but stagnated between 2000 and 2005. Panama had a reduction of 13 percentage points between the first two surveys but had an increase of 7 percentage points between the last two. Overall, the average annual decline between the earliest and the most recent surveys (1966–2008) for all countries combined ranged from 0.89 percentage points in El Salvador to 0.25 percentage points or less in Costa Rica, Guatemala and Panama (Table 1).

Underweight declined following a pattern generally similar to that of stunting, with several exceptions (Table 1). In the Plurinational State of Bolivia, underweight declined significantly between 1994 and 1998, while stunting remained unchanged. Both Colombia and Guatemala experienced important declines in stunting between surveys, but not in underweight. Only Haiti displayed a significant increase in the observed prevalence of undernutrition; underweight showed a large increase during the last survey interval (2000–2005), while stunting did not change.

If underweight (the official MDG indicator) is used as the indicator to monitor undernutrition, all 13 countries analysed are on track to meet the target prevalence established for achievement of MDG 1 (Table 2). In contrast, if stunting is used as the indicator, only 6 of the 13 countries (Brazil, Colombia, Dominican Republic, El Salvador, Mexico and Nicaragua) are on track. Another two countries, Costa Rica and Haiti, are within 2 percentage points of the goal (Table 3) and are therefore considered on track as well. At the current predicted trends in stunting, the remaining 5 countries (the Plurinational State of Bolivia, Guatemala, Honduras, Panama and Peru) are not on track to reach their goal, though the Plurinational State of Bolivia is within 3 percentage points of being on track.

Discussion

The choice of indicator for undernutrition clearly affects the conclusions about which countries are on track to reach the targets for reducing hunger related to MDG 1. Guatemala and Peru, the only two countries in the Americas among the 36 countries where 90% of the world’s stunted children live,8 illustrate the influence the choice of indicator can exert. Both countries are on track to achieve MDG 1 if underweight is used as the indicator, but not if stunting is used. If underweight is used as the indicator for undernutrition, countries will appear to be on track towards achieving the targeted prevalence established under MDG 1, but the large remaining burden of stunting will be ignored. In addition, as stunting is directly related to child mortality, actively monitoring its prevalence is also necessary to assess progress towards achieving MDG 4.

The relationship between stunting and underweight25 and between stunting and wasting shows regional differences26 that may affect the extent to which the analysis presented in our paper is applicable to other world regions. However, Latin America is not the only region in the world where stunting has a higher prevalence than underweight. An analysis of all recent surveys (2003–2004 to 2008) from 32 African and Asian countries using the Demographic and Health Survey STATComplier (accessed 4 March 2010) showed that all countries but four (Bangladesh, 2007; India 2005–06; Nepal, 2006 and Senegal, 2005), or 88%, had a prevalence of stunting that exceeded that of underweight. Therefore, the results of our study, which demonstrate the importance of the indicator used to determine whether countries are on track to meet the nutritional targets related to MDG 1, may be relevant to other world regions, although this requires empirical testing.

Our analyses have several limitations. Because no country had a survey in 1990, the baseline year for the MDGs, we used regression analysis to estimate the 1990 prevalence figures. Our predicted trends are also based on past trends and we cannot foresee whether the same rate of progress will continue until 2015, particularly since many of the governments of the countries included in the analysis have put reducing child stunting high on the political agenda with concomitant investment in resources. In addition, our analysis assumes that the progress between the first and last survey was linear, which was not always the case. Many countries experienced sharper declines during earlier surveys than in more recent years.

Regardless of which indicator is used for undernutrition, making it an objective to reduce the mean prevalence of undernutrition by half at the national level ignores the enormous within-country differences in the prevalence of undernutrition illustrated in Fig. 2 and will fail to identify regions whose lack of progress may be masked by the attainments of other regions. Thus, an additional objective under MDG 1 should be to set specific prevalence targets by subregion. The goal, for example, could be to reduce by half the prevalence of stunting in each subregion identified in the survey. This goal is particularly relevant in countries having the widest internal disparities in stunting.

Effective interventions to reduce stunting are available.27 Such interventions can also bring about improvements in human capital28 measured in terms of educational attainment, productivity and income and a reduced risk of developing chronic, non-communicable diseases.29 However, platforms for the delivery of these interventions, such as primary health- care networks, are weak,30,31 particularly in settings where stunting is a common problem. In the words of an expert, “we have the silver bullets to reduce child undernutrition but lack the rifles” (Victora CG, personal communication, 2010). Investment in improving delivery platforms and operations research to maximize the impact of this investment are urgently needed to maximize efficiency and deliver nutrition interventions known to be effective.

Efforts to address the underlying determinants of stunting can also bring about remarkable reductions in prevalence when accompanied by political leadership and investment, as shown by the recent decline of nearly 50% in Brazil (from 13.5% to 6.8% between 1996 and 2006).19,32 Two-thirds of this reduction is attributable to four factors: 25.7% to improved maternal schooling; 21.7% to an increase in the purchasing power of families; 11.6% to the expansion of health care; and 4.3% to improved sanitation. Importantly, the reductions were greatest in the poorest areas of Brazil, with a resultant decrease in disparities in stunting prevalence and in its detrimental consequences for individual and national development.33 Mexico has also achieved remarkable reductions through a concerted government effort targeting the poorest rural communities, where the demand for health and education services is created through conditional cash transfers.22,34 Although addressing the underlying determinants of stunting has been described as a long route to improving nutrition, the remarkable achievements by the governments of Brazil and Mexico over a relatively short period should trigger a re-evaluation of the timeframe required for strategies that address underlying determinants to achieve results. It should also make us reflect on how best to reduce stunting through an integrated strategy that implements the effective interventions identified in the recent Lancet series on maternal and child nutritional health27 while addressing the underlying determinants of undernutrition.

Countries throughout the world and the United Nations have made an important commitment to achieve the MDGs by 2015. Their actions should be guided by those policy frameworks, strategies and interventions that are most effective and the results should be evaluated vis-à-vis the indicator that best reflects the intended goal. As our results show, stunting rather than underweight should be included as a target for assessing progress towards achieving MDG 1, since, unlike underweight, it reflects the cumulative effects of undernutrition and predicts health and well-being in adulthood.


Competing interests:

None declared.

References

Share