Abstract
It is well understood that undernutrition underpins much of child morbidity and mortality in less developed countries, but the causes of undernutrition are complex and interrelated, requiring a multipronged approach for intervention. This paper uses a subsample of 3853 children under age 5 from the most recent family health survey in El Salvador to examine the relationship between birth spacing and childhood undernutrition (stunting and underweight). While recent research and guidance suggest that birth spacing of three to five years contributes to lower levels of infant and childhood mortality, little attention has been given to the possibility that short birth intervals have longer‐term effects on childhood nutrition status. The analysis controls for clustering effects arising from siblings being included in the subsample, as well as variables that are associated with household resources, household structure, reproductive history and outcomes, and household social environment. The results of the multiple regression analyses find that in comparison to intervals of 36–59 months, birth intervals of less than 24 months and intervals of 24–35 months significantly increase the odds of stunting (<24 months Odds Ratio (OR) = 1.52; 95% confidence interval (CI): 1.21–1.92; 25–36 months OR = 1.30; 95% CI: 1.05–1.64). Other factors related to stunting and underweight include standard of living index quintile, child's age, mother's education, low birthweight, use of prenatal care, and region of the country where the child lives. Policy and program implications include more effective use of health services and outreach programs to counsel mothers on family planning, breastfeeding, and well child care.
Keywords: anthropometry, birth interval, child growth, international child health nutrition, stunting, wasting
Introduction
It is well understood that undernutrition underpins much of child morbidity and mortality in less developed countries (Lopez et al. 2006). In spite efforts to improve the nutritional status of high‐risk children, it is not unusual for a large percentage of them to suffer some type and degree of undernutrition, based on anthropometrics indicators (Chevassus‐Agnes 1999; Swinburn 2004). The damage of undernutrition, which can begin as early as pregnancy and last through 2 years of age, is largely irreversible and carries negative consequences through adulthood. Thus, the window for addressing the long‐term consequences of childhood undernutrition is critical and short – from pre‐pregnancy to 2 years of age (Shrimpton et al. 2001; Victora et al. 2008). While nutrition programs aimed to address food security and improved and increased food intake are a logical strategy for addressing undernutrition, there are other more complex strategies, such as access to and use of health services, improved hygiene, and poverty reduction, that are being addressed through new types of partnerships that confront the root causes of undernutrition (World Bank 2006).
One factor that may predispose children to childhood undernutrition that has received little attention is birth spacing, or the time elapsed between consecutive births of children. The relationship between birth spacing and infant and child mortality is well studied, but fewer studies have examined the link between birth spacing and childhood undernutrition. Evidence suggests that if pregnancies and births are spaced too closely, women may not have the opportunity to replenish their nutritional stores stressed during pregnancy and/or breastfeeding, which results in predisposing infants and children to adverse outcomes, including preterm birth, low birthweight, as well as stunting and underweight (Conde Agudelo 2002; Rutstein 2003). If in fact, birth spacing is significantly related to childhood undernutrition, efforts to increase support for family planning programs would also contribute not only to reductions in maternal and child mortality, but to improved child health and development as well.
This paper uses the 2002–2003 El Salvador National Family Health Survey [Encuesta Nacional de Salud Familiar (FESAL)] to determine the extent to which short birth intervals might be associated with stunting (low height‐for‐age) and underweight (low weight‐for‐age) among children under 5 years of age. Specifically, we hypothesize that short birth intervals may predispose children to a higher risk of childhood undernutrition. Our objectives are to explore how short birth intervals may increase the risk of stunting and underweight, and to identify policy and programmatic efforts that can contribute to reducing the risk of undernutrition, including healthy timing and spacing of pregnancies.
Background
As a region, Latin America and the Caribbean (LAC) has experienced decreasing levels of undernutrition over the last three decades. Despite advances, the prevalence of undernutrition is still of public health concern, as undernutrition plays an important role in childhood mortality, morbidity and ability to reach developmental potential (Schroeder & Brown 1994; Pelletier et al. 1995). According to the United Nations Children's Fund (2004), in the LAC region, 16% of children under the age of 5 years are stunted, 8% are underweight and 2% are wasted (low weight‐for‐height). While stunting has declined since 1980 in the sub‐regions of South America and the Caribbean, Central America has seen little progress in this regard (de Onis et al. 2000). In El Salvador, between 1988 and 2002, stunting dropped slightly from 23% to 19%; over the same time, prevalence of underweight remained at approximately half that of stunting and dropped from 11% to 10% [1994, 2000, 2004].
Although relatively few researchers have examined the impact of preceding birth intervals, defined as the time elapsed between the birth of a previous child and the birth of a subsequent ‘index child’, on a child's nutritional status, a review of the available evidence confirms that birth intervals can affect nutritional outcomes, though the degree to which the relationship holds varies substantially (Dewey & Cohen 2007). Research drawing on data from 17 countries shows children born after preceding birth intervals between 3 and 5 years experienced substantially reduced nutritional risks compared to shorter birth intervals of less than 2 years (Rutstein 2005), but the relationship may be more complex and require controlling for potential confounders.
Stunting (low height‐for‐age)
Short birth intervals may increase the risk for stunting, which captures the long‐term and cumulative effects of poor dietary intake and/or lengthy illness on nutrition status of the child (de Onis et al. 2000). Rutstein (2005) observed a strong relationship between stunting and birth interval, noting a nearly linear decrease in stunting with increasing birth interval length. In particular, children born following an interval of less than 18 months were 43% more likely to suffer from stunting than children born following an interval of 60 months or longer. Research by Boerma & Bicego (1992) on 17 Demographic and Health Surveys (DHS) surveys and Aerts et al.'s (2004) data from Brazil found that children with short preceding birth intervals were more likely to experience stunting than those with longer preceding birth intervals.
Longer intervals of more than 3 years can also provide protection against stunting. In particular, compared to intervals between 24 and 29 months, intervals between 36 and 41 months were associated with up to a 29% decrease in the risk of stunting (Rutstein 2003). Olinto et al. (1993) found similar results in Brazil, with children born after intervals of 3 years or longer experiencing a lower risk for stunting than their counterparts born after intervals shorter than 3 years.
Underweight (low weight‐for‐age)
Low weight‐for‐age represents a composite index of both weight‐for‐height and height‐for‐age and, thus, is a measure of both acute and chronic undernutrition. Short birth intervals can increase the risk for underweight, and as birth intervals grow longer, risk of underweight decreases. In Boerma & Bicego's (1992) analysis, the impact of short intervals on underweight was slightly larger than for stunting; however, birth intervals of more than 3 years offered a protective effect from underweight. Rutstein's (2003) findings were similar; in comparing intervals of 24–29 months and 36–41 months, the longer intervals were associated with a 29% decrease in underweight. Mozumder et al. (2000) also determined that birth intervals (either preceding or succeeding) of 2 years or less were associated with underweight in children.
Methodology
Data
The following analysis draws on data collected as part of the 2002–2003 El Salvador National Family Health Survey (FESAL) conducted by the ADS, with technical assistance from the US Centers for Disease Control and Prevention. The survey sample was determined through random probability sampling methods. A total of 10 689 women of reproductive age (ages 15–49) completed the interview. Because the outcomes are related to infant and child mortality, the dataset for these analyses was organized so that the unit of observation is the child, rather than the mother. Each observation includes information about both the mother and the child. The FESAL includes information on 5868 children born in the 5 years prior to the survey.
Because the analysis examines the relationship between preceding birth spacing and dependent variables stunting and underweight, first births are excluded from the sample because they lack a preceding birth interval. Similarly, we restricted the subsample to singleton births, thus eliminating the confounding effect of multiple births, which experience higher rates of mortality and other adverse outcomes (Ibraham et al. 1994). Anthropometric data were available for a total of 3853 index children in this subsample of second‐ and higher‐order births.
The independent variables of stunting and underweight assessments are drawn from anthropometric measures in the 2002–2003 FESAL, which used normalized z‐scores for height‐for‐age and weight‐for‐age. Using the National Center for Health Statistics (NCHS)/World Health Organization standards (World Health Organization Working Group 1986), children are classified as stunted or underweight if their z‐score for the aforementioned categories is two or more standard deviations below the mean. Using these standards, 18.9% of the entire under‐five population in the survey was stunted and 10.3% was underweight. In the subsample used in this analysis, we observe 843 cases (21.8%) of stunting and 386 cases (10.0%) of underweight. In total, 939 children (24.4% of the sample) present some form of undernutrition (including wasting).
Birth interval length in El Salvador has increased during the last two decades. Median intervals reported in the 1993 and 1998 FESAL surveys (29.9 and 39.6 months, respectively) are much shorter than the median found in the 2002–2003 survey (53.8 months) (ADS 1994, 2000, 2004). Approximately 23% of the index children in the sample were born following birth intervals of less than 24 months (referred to as ‘short birth intervals’), 30% after intervals of 24–35 months, 26% after intervals of 36–59 months, and 21% after intervals of 60 months or longer (see Table 1).
Table 1.
Frequency distributions of predictor variables for stunting and underweight
Variable | n (n = 3853) | Percentage |
---|---|---|
Birth intervals | ||
Birth interval <23 months | 902 | 23.4 |
Birth interval 24–35 months | 1152 | 29.9 |
Birth interval 36–59 months ‡ | 998 | 25.9 |
Birth interval ≥60 months | 801 | 20.8 |
Household resources | ||
Lowest SLI quintile | 1610 | 41.8 |
Lower middle SLI quintile | 840 | 21.8 |
Middle SLI quintile ‡ | 578 | 15.0 |
Upper middle SLI quintile | 428 | 11.1 |
Highest SLI quintile | 397 | 10.3 |
Lives in rural area | 2431 | 63.1 |
Lives in urban area ‡ | 1422 | 36.9 |
No education | 855 | 22.2 |
Education 1–3 years | 902 | 23.4 |
Education 4–6 years ‡ | 929 | 24.1 |
Education 7–9 years | 624 | 16.2 |
Education 10 or more years | 543 | 14.1 |
Household structure | ||
Birth order 6 or higher | 616 | 16.0 |
Birth order 2 to 5 ‡ | 3237 | 84.0 |
0 to 11 months of age ‡ | 678 | 17.6 |
12 to 23 months | 713 | 18.5 |
24 to 35 months | 821 | 21.3 |
36 to 47 months | 790 | 20.5 |
48 to 59 months | 852 | 22.1 |
Reproductive history and outcomes | ||
Previous child died | 551 | 14.3 |
No previous child died ‡ | 3302 | 85.7 |
Low birthweight | 239 | 6.2 |
Normal birthweight ‡ | 2038 | 52.9 |
Missing birthweight data | 1576 | 40.9 |
Wanted pregnancy ‡ | 2844 | 73.8 |
Did not want pregnancy | 1009 | 26.2 |
Standard prenatal care ‡ | 1553 | 40.3 |
Below‐standard prenatal care | 1734 | 45.0 |
No prenatal care | 566 | 14.7 |
Delivery at home | 1557 | 40.4 |
Delivery at health facility ‡ | 2296 | 59.6 |
Social environment of household | ||
Occidental | 771 | 20 |
Central | 451 | 11.7 |
Metropolitan ‡ | 643 | 16.7 |
Paracentral | 1067 | 27.7 |
Oriental | 921 | 23.9 |
SLI, standard of living index. ‡indicates reference group in logistic regressions.
Analysis
The analysis of the association between stunting and underweight and the independent variables began by examining frequency distributions, using SPSS Version 12.0. Two‐way (bivariate) analyses of the independent categorical variables were conducted using chi‐squared tests to determine if statistical associations with stunting and underweight existed. In the third stage of the analysis, we used multiple logistic regression as the statistical model to examine the relationship between birth intervals and dependent variables stunting and underweight. We selected this approach because stunting and underweight are dichotomous variables. The subsample included 3853 children, and information on those children was obtained through interviews with 3032 mothers, which means that some of the mothers had more than one child in the subsample. Because 821 children in the subsample also had siblings in the subsample and because siblings in the subsample share some of the same background characteristics (i.e. household and maternal characteristics), they are not independent observations. We controlled for this clustering by using the ‘robust cluster’ subcommand of the ‘logistic’ command in STATA Version 9 to fit the logistic regression and obtain unbiased standard errors.
Predictor variables
In general, risk factors for adverse nutritional outcomes during infancy and childhood are largely determined by a child's prenatal and post‐natal environment and health behaviours practiced within the environment, including socio‐economic position, breastfeeding practices and size of household. This analysis considers similar predictor variables that are organized into similar conceptual categories similar to those used by Heaton et al. (2005): household resources, household structure, reproductive history and outcomes, and the social environment of the household (see Table 2 for predictor variables and evidence of their impact on nutritional outcomes). Additional variables such as breastfeeding, mother's exposure to violence, child's sex, mother's age and civil status were initially examined but are not statistically significant, and are therefore not included in subsequent analyses. Table 3 provides a description and treatment of the variables used in the analysis.
Table 2.
Predictor variables and evidence for inclusion in model
Variable | Impact on child nutrition status | Reference |
---|---|---|
Household resources | ||
Low standard of living index* | Poverty is associated with poor nutritional outcomes. | Nestel et al. 1999; Ricci & Becker 1996; Filmer & Pritchett 2001 |
Rural residence | Associated with undernutrition given seasonality of food availability | Madzingira 1995; Adair & Guilkey 1997; Larrea & Kawachi 2005 |
Low maternal education | Children are more likely to suffer from malnutrition when their mothers have lower levels of education; risk of child malnutrition decreases with mother's education. | Mozumder et al. 2000; Ojofeitimi et al. 2003; Nestel et al. 1999 |
Household structure | ||
Large family size; Higher‐order births | Children of larger families are more likely to be malnourished. With more children, nutritional status of all children, particularly higher‐order children worsens, especially in families with ≥5 children. | Jeyaseelan & Lakshman 1997; Mozumder et al. 2000; Paknawin‐Mock et al. 2000; Reichenheim & Harpham 1990; de Carvalho Lima et al. 2004 |
Child age under 3 years | Period between the second half of infancy (6 months)–3 years, children are vulnerable to malnutrition, which may be related to weaning practices. Children <3 years suffer more from diarrhoea and respiratory infections than older children, which can cause, result from or exacerbate poor nutrition | Nestel et al. 1999 |
Reproductive history and outcomes | ||
Death of older sibling | Children living in homes where there has been a previous child death experienced an increased risk of malnutrition. Maternal stress brought about by the death of a child had negative consequences on health and well‐being of surviving children | Reichenheim & Harpham 1990; Van den Broeck et al. 1996 |
Low birthweight (≤2500 gm) | Determinant of future nutritional status, especially during the first 6 months of life. Risk for low birthweight was highest among children born short and long intervals (i.e. <21 months and >69 months) | Arifeen et al. 2000; Kolsteren et al. 1997; Adair & Guilkey 1997; de Carvalho Lima et al. 2004; Conde‐Agudelo 2002; 1999, 2001 |
Unplanned pregnancy | An unplanned pregnancy may affect prenatal care seeking and other services, affecting child health outcomes. In the Dominican Republic, children from unplanned pregnancies were more likely to be stunted. | Joyce et al. 2000; Jensen & Ahlburg 1999; Safonova & Leparsky 1998; Montgomery et al. 1997 |
Receipt of prenatal care | Health and nutrition promotion during pregnancy via prenatal care can influence the health and nutrition of the newborn and throughout childhood. Late initiation of/lack of prenatal care can impact fetal development and impact birthweight and other outcomes. | Engle et al. 1999; Gülmezoglu et al. 1997 |
Home birth | Giving birth at home may indicate limited access to health services, which is associated with stunting and poverty. | Larrea & Kawachi 2005 |
Social environment of the household | ||
Geographic region of country | Geographic region may affect the social environment of the household and may provide a means of identifying priority areas for nutritional problems | Context specific |
Socio‐economic status is estimated with standard of living index (SLI) quintiles (Filmer & Pritchett 2001). The SLI is calculated for the entire sample of women in the Encuesta Nacional de Salud Familiar, with 20% of the sample in each quintile regardless of whether they have had children in the past five years. In this subsample, SLI quintiles are not evenly distributed because the children are assigned the SLI value of their mothers.
Table 3.
Predictor variable description and treatment in analysis
Variable | Description | Treatment |
---|---|---|
Birth spacing | Months between birth of index child and previous child | 4 categories; 3 dummy variables |
Household resources | ||
Low standard of living | Index based on household characteristics and assets | Mothers' responses serve as basis for creating 5 quintiles; 4 dummy variables |
Rural residence | Urban or rural location of mother's home | 2 categories, 1 dummy variable |
Low maternal education | Mother's highest level of educational attainment | 5 categories, 4 dummy variables |
Household structure | ||
Large family size; higher‐order births | Low (2 to 5) order and high (6 and higher) order | 2 categories, 1 dummy variable |
Child age | Age of index child (in months) at time of interview | 5 categories, 4 dummy variables |
Reproductive history and outcomes | ||
Death of older sibling | Whether or not the index child has had an older sibling die | 2 categories, 1 dummy variable |
Low birthweight (≤2500 gm) | Index child weighed 2500 g or less at birth or during the first week of life | 3 categories, 2 dummy variables* |
Unplanned pregnancy | Mother's report of whether or not she wanted to become pregnant with index child | 2 categories, 1 dummy variable |
Receipt of prenatal care | Extent to which mother received prenatal care when pregnant with index child (El Salvadoran Ministry of Health standard of care, some care, and no care) | 3 categories, 2 dummy variables |
Home birth | Whether or not the index child was born at home or at a health facility | 2 categories, 1 dummy variable |
Social environment of the household | ||
Geographic region of country | Region of country where mother resides | 5 categories, 4 dummy variables |
Because of missing data for birthweight, a dummy variable for ‘missing’ was created so as not to exclude those cases from the analysis.
Results
Stunting
Frequencies for the predictor variables are shown in Table 1. Results of bivariate analysis and multiple logistic regression on stunting are shown in Table 4.
Table 4.
Rate of stunting by predictor variables and odds ratios of predictor variables on stunting
Variable | n (n = 3852) | Rate per 100 children | Odds ratios | ||
---|---|---|---|---|---|
Value | P‐value* | Value | P‐value † | ||
Birth intervals | |||||
Birth interval <24 months | 902 | 28.3 | <0.001 | 1.52 | <0.001 |
Birth interval 24–35 months | 1153 | 25.4 | 1.30 | 0.018 | |
Birth interval 36–59 months | 995 | 18.3 | 1.00 | ||
Birth interval ≥60 months | 802 | 14.1 | 0.95 | 0.696 | |
Household resources | |||||
Lowest SLI quintile | 1611 | 32.7 | <0.001 | 2.11 | <0.001 |
Lower middle SLI quintile | 840 | 20.5 | 1.31 | 0.096 | |
Middle SLI quintile | 580 | 13.4 | 1.00 | ||
Upper middle SLI quintile | 426 | 9.9 | 0.83 | 0.405 | |
Highest SLI quintile | 394 | 6.3 | 0.62 | 0.058 | |
Lives in rural area | 2431 | 26.9 | <0.001 | 0.83 | 0.131 |
Lives in urban area | 1422 | 13.3 | 1.00 | ||
No education | 856 | 32.8 | <0.001 | 1.51 | 0.001 |
Education 1–3 years | 900 | 27.4 | 1.32 | 0.024 | |
Education 4–6 years | 929 | 20.1 | 1.00 | ||
Education 7–9 years | 623 | 15.1 | 0.96 | 0.808 | |
Education 10 or more years | 545 | 6.2 | 0.47 | 0.001 | |
Household structure | |||||
Birth order 6 or higher | 617 | 30.3 | <0.001 | 0.97 | 0.803 |
Birth order 2 to 5 | 3236 | 20.3 | 1.00 | ||
0 to 11 months – child age | 676 | 8.4 | <0.001 | 1.00 | |
12 to 23 months – child age | 714 | 24.8 | 3.81 | <0.001 | |
24 to 35 months – child age | 820 | 23.8 | 3.52 | <0.001 | |
36 to 47 months – child age | 791 | 24.3 | 3.63 | <0.001 | |
48 to 59 months – child age | 852 | 26.1 | 4.07 | <0.001 | |
Reproductive history and outcomes | |||||
Previous child died | 551 | 27.2 | 0.001 | 0.97 | 0.818 |
No previous child died | 3302 | 21.0 | 1.00 | ||
Low birthweight | 237 | 30.8 | <0.001 | 1.97 | <0.001 |
Normal birthweight | 2039 | 18.4 | 1.00 | ||
Missing data on birthweight | 1577 | 25.0 | 0.93 | 0.484 | |
Unwanted pregnancy | 1011 | 24.4 | 0.022 | 1.08 | 0.402 |
Wanted pregnancy | 2842 | 21.0 | 1.00 | ||
Standard prenatal care | 2301 | 18.3 | <0.001 | 1.00 | |
Below standard prenatal care | 984 | 25.3 | 1.27 | 0.018 | |
No prenatal care | 568 | 30.5 | 1.01 | 0.920 | |
Delivery at home | 1567 | 29.3 | <0.001 | 1.16 | 0.126 |
Delivery at health facility | 2286 | 16.8 | 1.00 | ||
Social environment of household | |||||
Western region | 770 | 28.6 | <0.001 | 1.05 | 0.782 |
Central region | 451 | 24.2 | 0.82 | 0.306 | |
Metropolitan region | 645 | 16.6 | 1.00 | ||
Paracentral region | 1066 | 22.7 | 0.73 | 0.047 | |
Eastern region | 921 | 17.9 | 0.58 | 0.001 |
SLI, standard of living index. *Clustering was not controlled for in the bivariate analyses (shown as rate per 100 children), which means that the estimates of standard errors and P‐values associated with the chi‐squared tests are biased. † P‐values for bivariate analyses based on chi‐squared tests and P‐values for odds ratio based on multiple logistic regression model that includes all predictor variables and takes account of clustering.
Birth intervals
Of central interest in this analysis is the relationship between preceding birth intervals and nutritional outcomes. As length of the birth interval increases, the risk of stunting decreases. Among children born after an interval of less than 24 months, 28% are stunted. Prevalence of stunting decreases as birth interval increases, and among children born after 60 months or more, only 14% are stunted. In the multiple logistic model, being born after an interval of less than 24 months increases the odds of stunting by more than 1.52 times, in comparison to intervals of 36 to 59 months. Similarly, intervals of 24 to 35 months are associated with an increase of 1.30 times the odds of stunting, in comparison with 36 to 59 months. There is little evidence that intervals of 60 months or more have significant association with stunting.
Household resources
The relationship between poor standard of living index (SLI) status (poorest 40% of population) and stunting is statistically significant and strong. Almost 33% of children in the lowest SLI quintile and 20% in the lower–middle quintile were stunted; as the SLI quintile increases, the risk of stunting decreases. In comparison to the middle SLI quintile, being in the lowest quintile more than doubled the odds of stunting. On the other hand, being in the highest quintile had a marginally significant effect of reducing the odds of stunting. Children living in rural areas have a higher prevalence of stunting, but the association was not significant in the multiple logistic regression.
Maternal education is also associated with stunting. In the bivariate analysis, as mother's education increases, the risk of stunting decreases; almost 30% of children whose mothers had 3 or fewer years of education experienced stunting. In the regression, children whose mothers had never attended school experienced a 1.51 times increase in the odds of stunting compared to children whose mothers had 4 to 6 years of schooling. Similarly, children whose mothers had 1 to 3 years of schooling had a 1.32 times increase in the odds of stunting. At the other end of the spectrum, children whose mothers had 10 or more years of school had a significantly lower odds of stunting that was 0.47 times that of children whose mothers had 4 to 6 years of schooling.
Household structure
Of the factors related to household structure, child age had a significant association with stunting. In comparison to infants, the odds of stunting were between 3.52 and 4.07 times higher for older children. Also related to stunting is high‐birth order; in the bivariate analysis, higher‐birth‐order children experienced a higher odds of stunting, but in the multiple logistic regression, the result was not significant.
Reproductive history and outcomes
In the bivariate analyses, most of the factors related to reproductive history and outcomes increased the prevalence of stunting: the death of a previous child, low birthweight, an unplanned pregnancy, use of prenatal care and giving birth at home. In the multiple logistic model, children who had a low birthweight experienced almost a doubling in the odds of stunting compared to normal‐birthweight children; bivariate analyses indicated that prevalence of stunting was also almost twice as high among low‐birthweight children. Prevalence of stunting among children who received some prenatal care, but below Ministry of Health (MOH) norms, was higher than children whose mothers received MOH standard prenatal care; prevalence of stunting was highest among those children who did not receive any prenatal care. In the multiple logistic regression, below‐standard prenatal care increased the odds of stunting by 1.27 times in comparison to children who received care according to MOH norms. Children who had an older sibling die and children born at home had higher prevalences of stunting, but neither was significant in the multiple logistic model.
Social environment of household
The social environment of the household was also associated with stunting. Children in the Western region have the highest prevalence of stunting (29%), while children in Metropolitan San Salvador had the lowest prevalence of stunting (17%). In the multiple logistic model, children living in the Paracentral and Eastern regions experienced a significantly lower odds of stunting than children living in Metropolitan San Salvador.
Underweight
Results of bivariate analysis and multiple logistic regression on underweight are shown in Table 5.
Table 5.
Rate of underweight by predictor variables and odds ratios of predictor variables on underweight
Variable | n (n = 3853) | Rate per 100 children | Odds ratio | ||
---|---|---|---|---|---|
Value | P‐value* | Value | P‐value † | ||
Birth intervals | |||||
Birth interval <24 months | 902 | 11.5 | <0.001 | 1.21 | 0.245 |
Birth interval 24–35 months | 1153 | 12.1 | 1.26 | 0.115 | |
Birth interval 36–59 months | 995 | 8.8 | 1.00 | ||
Birth interval ≥60 months | 802 | 6.7 | 0.94 | 0.725 | |
Household resources | |||||
Lowest SLI quintile | 1611 | 15.3 | <0.001 | 1.86 | 0.002 |
Lower middle SLI quintile | 840 | 8.5 | 1.10 | 0.644 | |
Middle SLI quintile | 580 | 6.6 | 1.00 | ||
Upper middle SLI quintile | 426 | 4.7 | 0.80 | 0.422 | |
Highest SLI quintile | 396 | 2.8 | 0.55 | 0.084 | |
Lives in rural area | 2431 | 12.3 | <0.001 | 0.90 | 0.522 |
Lives in urban area | 1422 | 6.2 | 1.00 | ||
No education | 856 | 14.6 | <0.001 | 1.31 | 0.106 |
Education 1–3 years | 900 | 13.1 | 1.31 | 0.092 | |
Education 4–6 years | 929 | 9.4 | 1.00 | 0.000 | |
Education 7–9 years | 623 | 6.6 | 0.88 | 0.517 | |
Education 10 or more years | 545 | 2.8 | 0.46 | 0.011 | |
Household structure | |||||
Birth order 6 or higher | 617 | 13.0 | 0.008 | 0.93 | 0.650 |
Birth order 2 to 5 | 3236 | 9.5 | 1.00 | ||
0 to 11 months – child age | 676 | 3.6 | <0.001 | 1.00 | |
12 to 23 months – child age | 714 | 12.9 | 4.04 | <0.001 | |
24 to 35 months – child age | 820 | 12.7 | 3.95 | <0.001 | |
36 to 47 months – child age | 791 | 10.0 | 2.95 | <0.001 | |
48 to 59 months – child age | 852 | 10.2 | 3.09 | <0.001 | |
Reproductive history and outcomes | |||||
Previous child died | 551 | 11.4 | 0.232 | 0.92 | 0.565 |
No previous child died | 3302 | 9.8 | 1.00 | ||
Low birthweight | 237 | 16.5 | <0.001 | 2.27 | <0.001 |
Normal birthweight | 2039 | 7.9 | 1.00 | ||
Missing data on birthweight | 1577 | 11.8 | 1.12 | 0.373 | |
Unwanted pregnancy | 1011 | 11.0 | 0.236 | 1.08 | 0.544 |
Wanted pregnancy | 2842 | 9.7 | 1.00 | ||
Standard prenatal care | 2301 | 9.2 | 0.096 | 1.00 | |
Below standard prenatal care | 984 | 11.5 | 1.06 | 0.636 | |
No prenatal care | 568 | 10.9 | 0.76 | 0.124 | |
Delivery at home | 1567 | 13.4 | <0.001 | 1.18 | 0.209 |
Delivery at health facility | 2284 | 7.7 | 1.00 | ||
Social environment of household | |||||
Western region | 770 | 14.3 | <0.001 | 1.12 | 0.577 |
Central region | 451 | 11.6 | 0.85 | 0.480 | |
Metropolitan region | 645 | 8.3 | 1.00 | ||
Paracentral region | 1066 | 9.5 | 0.61 | 0.013 | |
Eastern region | 921 | 8.5 | 0.68 | 0.066 |
SLI, standard of living index. *Clustering was not controlled for in the bivariate analyses (shown as rate per 100 children), which means that the estimates of standard errors and P‐values associated with the chi‐squared tests are biased. † P‐values for bivariate analyses based on chi‐squared tests and P‐values for odds ratio based on multiple logistic regression model that includes all predictor variables and takes account of clustering.
Birth intervals
In the bivariate analysis, short birth intervals were significantly associated with underweight. As birth interval lengthened to 3 years and longer, the prevalence of underweight decreased. However, in the multiple logistic model, birth interval length was not significant.
Household resources
SLI quintile were associated with underweight, as well as the place of residence and the mother's education. In the bivariate analysis, as SLI quintile increased, the prevalence of underweight children decreased. However, only children in the lowest quintile experienced a significant increase in the odds of underweight; the odds of underweight among children in the lowest quintile were 1.86 times that of children in the middle quintile. The prevalence of underweight children living in rural areas was almost twice as high as that of children living in urban areas, but the relationship was not significant in the multiple logistic regression. Mother's education was also associated with underweight: children whose mothers had 3 or fewer years of schooling were most likely to have underweight children; as educational attainment increased, the proportion of underweight children decreased. In the multiple logistic model, the only significant effect was among women with 10 or more years of schooling: in comparison to children whose mothers had 4 to 6 years of education, the odds of underweight was 0.46, suggesting that higher levels of education reduces the odds of underweight.
Household structure
In the bivariate analyses, both child age and birth order were associated with underweight. In comparison to infants, older children experienced prevalences of underweight that were three to four times as high. In the multiple logistic model, the size and significance of the effects of being between ages 12 and 59 months indicated that age is a very important risk factor for underweight. In comparison to infants, children ages 12 to 23 months and 24 to 35 months experience very large increases in the odds of underweight (4.04 and 3.95, respectively). Although higher‐order births experienced a higher prevalence of underweight in the bivariate analysis, birth order was not significant in the multiple logistic regression.
Reproductive history and outcomes
The only reproductive history or outcome variables associated with the prevalence of underweight were low birthweight and delivery at home. In the bivariate analysis, children with low birthweight had a prevalence of underweight that was twice as high as the rate among normal‐weight children (16.5% vs. 7.9%). In the multiple logistic model, low birthweight remained large and significant: those children experienced an odds of underweight that was 2.27 times that of normal‐weight children. Children born at home had a higher prevalence of underweight, but the variable was not significant in the multiple logistic model.
Social environment of the household
Again, region of country had a significant predictor. In the bivariate analysis, children in the Western region had the highest proportion of underweight children (14.3%), followed by the Central region (11.6%). In the multiple logistic model, being from the Paracentral region and, to a lesser degree, the Eastern region was associated with a significant reduction in the odds of underweight in comparison to children in the Metropolitan San Salvador region.
Discussion
Stunting and underweight are the two most common forms of undernutrition in El Salvador, with stunting being approximately twice as common as underweight. Stunting is associated with long‐term factors and reflects the cumulative effects of chronic undernutrition and frequent illness since birth and even in the womb. As an indicator for policy and program development and planning, stunting is very sensitive to socio‐economic inequalities, which provides insight into some of the findings in this analysis. Underweight, which is a composite indicator, fluctuates more over time than stunting, and reflects current and acute as well as chronic undernutrition (INDEPTH Network 2006). As indicators that draw on children's height, weight, and age, stunting and underweight are intrinsically linked. Although the analyses yield more statistically significant results for stunting, the findings and their implications may also be meaningful in addressing underweight, given the qualitative similarity in trends and patterns for stunting and underweight.
Although the FESAL offers the opportunity to conduct a preliminary examination of the relationship between birth spacing and undernutrition, it has a number of limitations due in part to being a cross‐sectional general health survey. It does not include maternal characteristics, such as height or weight at the time of each pregnancy; that type of information is only available at the time of the survey. Recorded birthweight is based on the mother's recall and not verified by any external source, which explains the missing data (41%) on the variable. Information on breastfeeding was only collected on the last birth; therefore, information on breastfeeding is missing on 20% of the children because they did not represent the mother's last birth in the 5 years preceding the interview. In addition, the survey does not include specific information on dietary intake that would normally be part of a nutrition survey. Population‐based surveys, such as the FESAL, have been improving collection and measurement of infant and young child feeding, including breastfeeding, so additional data will be available in future surveys. Without these data restrictions, the results of the analysis would be stronger. Nevertheless, the existing data are very useful for exploring the research hypothesis. In addition to these limitations with the data, we did not control for clustering during bivariate analysis and, as a result, have biased estimates of standard errors and P‐values associated with the chi‐squared tests; this issue, however, was addressed when fitting the multivariate logistic regression.
Short birth intervals are associated with an elevated risk of stunting, which is the most common form of childhood undernutrition in El Salvador. Intervals of 24 to 35 months are also associated with an increased risk of stunting. As a cross‐sectional survey, the FESAL does not provide evidence to support a causal relationship. Nevertheless, the findings do suggest that birth intervals of less than 3 years are either directly related to stunting or act as a proxy for a risk factor that is either unobserved or not included in the analysis. These findings support previous studies that find an association between short intervals and undernutrition. Unlike some previous studies, we did not find that longer birth intervals are also associated with an increase in the odds of undernutrition.
Child age is also an extremely significant predictor of the adverse nutrition outcomes, and supports findings in the scientific literature related to children's risk of undernutrition increasing with age. In general, after infancy, our findings indicate that the odds of stunting are quite high and relatively consistent. As the literature suggests, part of the association between age and undernutrition may be related to breastfeeding during infancy, which provides children with high‐quality nutrition. The results indicate that, in comparison to infancy, the odds of stunting and underweight are significantly higher, especially between ages 12 and 23 months. This relationship may be due to weaning and inappropriate or inadequate complementary feeding practices that contribute to lower resistance to pathogens related to diarrhoeal disease and acute respiratory infections – two common illnesses that contribute to and compound childhood weight loss. While breastfeeding status per se was not significant in the bivariate analyses, the FESAL data indicate that 94% were breastfed for some period of time.
A third predictor of stunting and underweight is low birthweight, which is associated with a broad range of adverse consequences on infants and children. We found statistically significant and independent relationships of both low birthweight and short birth intervals on stunting. Our findings corroborate previous studies in which low birthweight is associated with poor nutritional status during infancy and early childhood and support the need to intervene during pregnancy (or prior to) to prevent low‐birthweight infants. The relationships between low birthweight, short birth intervals and child undernutrition are complex and further research is needed to better understand them.
Other significant associations with stunting, such as low maternal education, being in the lowest SLI quintile, and insufficient use of prenatal care may all be interrelated, as suggested earlier, reflecting ways that stunting is associated with socio‐economic inequalities via limited access to health services, household food insecurity, and the likelihood of poor environmental conditions (access to clean water, hygienic conditions for food preparation) increasing the likelihood of illness. These results corroborate the scientific literature, with one exception that merits closer examination: while insufficient use of prenatal care was associated with stunting, the complete absence of prenatal care was not. By relying on the norms established by the Ministry of Health of El Salvador to guide our variable specification, we may have missed a better way of examining the association between undernutrition and use of prenatal care. Our findings also suggest that children living in the Paracentral and Eastern regions of El Salvador have a lower odds of being stunted and underweight. However, based on the results published in the FESAL final report, the Paracentral and Eastern regions do not necessarily have the highest levels of development nor the most advanced infrastructure (ADS 2004).
In the bivariate analyses for stunting, four additional variables (living in a rural area, sixth or higher birth order, death of a previous child and unwanted pregnancy) were statistically significant. However, when the multiple logistic regression was fit, these predictors were no longer statistically significant, likely due to confounding by including other predictor variables in the model and covariance between predictors. For example, the significant odds ratio for mothers not having attended school is likely to be associated with rural residence and birth order of six or higher. Similarly, death of a previous child and unwanted pregnancy are likely to be associated with being in the lowest SLI quintile. It is very difficult to tease out the relationships between several of the predictor variables: a poor child living in a rural area, born to a mother who never attended school and who has already had many children is likely to characterize many children at risk for undernutrition in El Salvador. In spite of the possible interrelatedness of the predictors in the multivariate model, the association between short intervals and stunting remains consistent and significant.
If there is a direct association between short birth intervals and undernutrition, then many of our findings have implications for the quality of care provided by the health care system and opportunities for outreach services and community health workers. The window for intervention in undernutrition is short (before pregnancy to 2 years of age), and the findings suggest opportune points of time to reach women of reproductive age to provide counselling and promote health behaviours to prevent child undernutrition risks – preconception, prenatal, post‐partum, infancy and early childhood (2 years). Improving utilization of health facility and outreach services and promoting specific health behaviours such as breastfeeding, birth spacing, improved hygienic practices (clean water and food preparation), recognition and treatment of childhood diarrhoeal and respiratory infections could mediate many of the factors affecting stunting and underweight. For example, only 60% of children in the subsample received the level of prenatal care proscribed by MOH norms and only 48% of women seek postpartum care with 6 weeks of delivery. If use of these basic but important types of care was increased, providers could take advantage of more opportunities to promote health behaviours among mothers of young children. At the same time, although 67% of married women use family planning, approximately half of all users have chosen sterilization, which does not offer any benefits related to the spacing on future pregnancies (ADS 2004). Higher acceptance of temporary methods, such as oral contraceptives, injectables, condoms, intrauterine device, could, however, contribute to longer and healthier birth intervals.
Because many of the risk factors for stunting and underweight are also tied to poverty, an effective approach that could be used to address undernutrition is a strategy that targets the poor and provides education about and access to health services and incentives to use them. Since the completion of the 2002–2003 FESAL, the Government of El Salvador has begun implementing a conditional cash transfer program called the ‘Solidarity Network’. Directed towards extremely poor families in priority areas, one of the program's objectives is the improvement of the health and nutritional status of children under the age of 5 years through improving food intake for children and pregnant and lactating women, vaccinations against childhood diseases, and growth monitoring. Although El Salvador's program was not initiated until 2006, conditional cash transfer programs in Latin America have generally been successful in stimulating use of health services, among other priority targeted behavioural changes (Rawlings & Rubio 2003). During the first year of implementation, use of health services for infants increased by 83% and use of post‐natal care increased by 62% in the 15 poorest municipalities (Red Solidaria 2006). Conditional cash transfer programs have had positive effects on child nutritional status in Nicaragua (Maluccio & Flores 2004) and in Colombia (Attanasio et al. 2005). These advances may, in time, be reflected in national health surveys such as the FESAL. Nevertheless, by focusing on improved use of health services and nutritional status, the program addresses some of the critical risk factors identified in this analysis, including opportunities to promote healthy timing and spacing of pregnancies.
Through facilitating appropriate and healthy birth intervals, family planning may contribute to improving childhood nutrition, as well as reduce the number of unplanned pregnancies. Birth spacing can also contribute to a lower proportion of children born with low birthweight, which was found to be a significant predictor of both stunting and underweight. While the underlying causes of childhood undernutrition are numerous and complex, existing health programs can be better used to effectively promote healthy timing and spacing of pregnancies and contribute to a range of positive health outcomes for children.
Acknowledgements
Support for this research was provided by the United States Agency for International Development (USAID) under the terms of the POLICY II Project, Contract No. HRN‐C‐00‐00‐00006‐00, beginning July 7, 2000 and ending June 30, 2006. The views expressed in this publication do not necessarily reflect those of USAID or the US government.
The authors wish to thank Carlos Melendez, Wilfido Clara, Jorge Cruz Gonzales, Maricarmen de Estrada, Michelle Prosser, Charlotte Cabali, Ronette Briefel, two anonymous reviewers, and participants in the presentation of research results for their thoughtful comments on earlier drafts. We also wish to thank Steven McCracken, Paul Stupp and Mary Goodwin for their assistance in using the FESAL dataset, and Toshiko Kaneda and Mona Steffen for their valuable assistance with data analysis.
None of the authors have either a personal or financial conflict of interest with the research presented in this paper.
References
- Adair L.S. & Guilkey D.K. (1997) Age‐specific determinants of stunting in Filipino children. Journal of Nutrition 127, 314–320. [DOI] [PubMed] [Google Scholar]
- Aerts D., Drachler M.L. & Giugliani E.R.J. (2004) Determinants of growth retardation in southern Brazil. Cadernos de Saude Publica 20, 1182–1190. [DOI] [PubMed] [Google Scholar]
- Arifeen S.E., Black R.E., Caulfield L.E., Antelman G., Baqui A.H., Nahar Q. et al (2000) Infant growth patterns in the slums of Dhaka in relation to birth weight, intrauterine growth retardation, and prematurity. American Journal of Clinical Nutrition 72, 1010–1017. [DOI] [PubMed] [Google Scholar]
- Asociación Demográfica Salvadoreña (ADS) (1994) Encuesta Nacional de Salud Familiar 1993. San Salvador: El Salvador. [Google Scholar]
- Asociación Demográfica Salvadoreña (ADS) (2000) Encuesta Nacional de Salud Familiar 1998. San Salvador: El Salvador. [Google Scholar]
- Asociación Demográfica Salvadoreña (ADS) (2004) Encuesta Nacional de Salud Familiar FESAL 2002/03. San Salvador: El Salvador. [Google Scholar]
- Attanasio O., Battistin E., Fitzsimons E., Mesnard A. & Vera‐Hernández M. (2005) How Effective are Conditional Cash Transfers? Evidence from Colombia. Briefing Note no. 54. The Institute for Fiscal Studies: London.
- Boerma J.T. & Bicego G.T. (1992) Preceding birth intervals and child survival: searching for pathways of influence. Studies in Family Planning 23, 243–256. [PubMed] [Google Scholar]
- De Carvalho Lima M., Motta M.E., Santos E.C. & Pontes da Silva G.A. (2004) Determinants of impaired growth among hospitalized children – a case‐control study. Sao Paulo Medical Journal 122, 117–123. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Chevassus‐Agnes S. (1999) Anthropometric, Health and Demographic Indicators in Assessing Nutritional Status and Food Consumption. SD Dimensions. Sustainable Development Department, Food and Agriculture Organization of the United Nations: Rome. [Google Scholar]
- Conde‐Agudelo A. (2002) Association between Interpregnancy Intervals and Maternal/Paternal Outcomes in Latin America. OBSI Regional Workshop: Washington, DC. [Google Scholar]
- Dewey K.G. & Cohen R.J. (2007) Does birth spacing affect maternal or child nutritional status? A systematic literature review. Maternal and Child Nutrition 3, 151–173. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Engle P.L., Menon P. & Haddad L. (1999) Care and nutrition: concepts and measurement. World Development 27, 1309–1337. [Google Scholar]
- Filmer D. & Pritchett L.H. (2001) Estimating wealth effects without expenditure data – or tears: an application to educational enrollments in states of India. Demography 38, 115–132. [DOI] [PubMed] [Google Scholar]
- Gülmezoglu M., De Onis M. & Villar J. (1997) Effectiveness of interventions to prevent or treat impaired fetal growth. Obstetrical and Gynecological Survey 52, 139–149. [DOI] [PubMed] [Google Scholar]
- Heaton T.B., Forste R., Hoffman J.P. & Flake D. (2005) Cross‐national variation in family influences on child health. Social Science and Medicine 60, 97–108. [DOI] [PubMed] [Google Scholar]
- Ibrahim S.A., Babiker A.G., Amin I.K., Omer M.I. & Rushwan H. (1994) Factors associated with high risk of perinatal and neonatal mortality: an interim report on a prospective community‐based study in rural Sudan. Paediatric Perinatal Epidemiology 8, 193–204. [DOI] [PubMed] [Google Scholar]
- INDEPTH Network (2006) Anthropometric Status Indicators. INDEPTH Resource Kit for Demographic Surveillance Systems (Beta Version 0.9). Available at: http://www.indepth-network.org. Accessed February 11, 2008.
- Jensen E. & Ahlburg D. (1999) A Multicountry Analysis of the Impact of Unwantedness and Number of Children on Child Health and Preventive and Curative Care. The POLICY Project, Futures Group: Washington, DC. [Google Scholar]
- Jeyaseelan L. & Lakshman M. (1997) Risk factors for malnutrition in South Indian children. Journal of Biosocial Science 29, 93–100. [DOI] [PubMed] [Google Scholar]
- Joyce T.J., Kaestner R. & Korenman S. (2000) The effect of pregnancy intention on child development. Demography 37, 83–94. [PubMed] [Google Scholar]
- Kolsteren P.W., Kusin J.A. & Kardjati S. (1997) Pattern of linear growth velocities of infants from birth to 12 months in Madura, Indonesia. Tropical Medicine and International Health 2, 291–301. [DOI] [PubMed] [Google Scholar]
- Larrea C. & Kawachi I. (2005) Does economic inequality affect child malnutrition? The case of Ecuador. Social Science & Medicine 60, 165–178. [DOI] [PubMed] [Google Scholar]
- Lopez A.D., Mathers C.D., Ezzati M., Jamison D.T. & Murray C.J.L. (2006) Global and regional burden of disease and risk factors, 2001: systematic analysis of population health data. Lancet 367, 1747–1757 [DOI] [PubMed] [Google Scholar]
- Madzingira N. (1995) Malnutrition in children under five in Zimbabwe: effect of socioeconomic factors and disease. Social Biology 42, 239–246. [DOI] [PubMed] [Google Scholar]
- Maluccio J. & Flores R. (2004) Impact Evaluation of a Conditional Cash Transfer Program: The Nicaraguan Red de Protección Social. FCND Discussion Paper no. 184. Food and Nutrition Division, International Food Policy Research Institute, Washington, DC.
- Montgomery M.R., Lloyd C.B., Hewett P.C. & Heuveline P. (1997) The Consequences of Imperfect Fertility Control for Children's Survival, Health and Schooling. Demographic and Health Surveys Analytical Reports no. 7. Macro International: Calverton, MD. [Google Scholar]
- Mozumder A.B.M.K.A., Khuda B.E., Kane T.T., Levin A. & Ahmed S. (2000) The effect of birth interval on malnutrition in Bangladeshi infants and young children. Journal of Biosocial Science 32, 289–300. [DOI] [PubMed] [Google Scholar]
- Nestel P., Melare A., Rosado J. & Mora J.O. (1999) Undernutrition among Honduran children 12–71 months old. Revista Panamericana de Salud Pública 6, 256–265. [DOI] [PubMed] [Google Scholar]
- Ojofeitimi E.O., Owolabi O.O., Aderonmu A., Esimai A.O. & Olasanmi S.O. (2003) A study on under five nutritional status and its determinants in a semi‐rural community of Ile‐Ife, Osun State, Nigeria. Nutrition and Health 17, 21–27. [DOI] [PubMed] [Google Scholar]
- Olinto M.T.A., Victora C.G., Barros F.C. & Tomasi E. (1993) Determinants of malnutrition in a low‐income population: a hierarchical analytical model. Cadernos de Saude Publica 9, 14–27. [PubMed] [Google Scholar]
- De Onis M., Frongillo E.A. & Blossner M. (2000) Is malnutrition declining? An analysis of changes in levels of child malnutrition since 1980. Bulletin of the World Health Organization 78, 1222–1231. [PMC free article] [PubMed] [Google Scholar]
- Paknawin‐Mock J., Jarvis L., Jahari A.B., Husaini M.A. & Pollitt E. (2000) Community‐level determinants of child growth in an Indonesian tea plantation. European Journal of Clinical Nutrition 54, S28–S42. [DOI] [PubMed] [Google Scholar]
- Pelletier D.L., Frongillo E.A., Shroeder D.G. & Habicht J.P. (1995) The effects of malnutrition on child mortality in developing countries. Bulletin of the World Health Organization 73, 443–448. [PMC free article] [PubMed] [Google Scholar]
- Rawlings L.B. & Rubio G.M. (2003) Evaluating the impact of conditional cash transfer programs: lessons from Latin America. World Bank Policy Research Working Paper 3119. World Bank: Washington, DC. [Google Scholar]
- Red Solidaria (2006) Red Solidaria: El Salvador conditional cash transfers, Red Solidaria and government of El Salvador. Presentation at the Human Development Forum, Washington, DC, October 2006.
- Reichenheim M.E. & Harpham T. (1990) An intra‐community profile of nutritional deficiency: a study of children under 5 in a low income community in Rio de Janeiro (Brazil). Revista de Saúde Pública 24, 69–79. [DOI] [PubMed] [Google Scholar]
- Ricci J.A. & Becker S. (1996) Risk factors for wasting and stunting among children in Metro Cebu, Philippines. American Journal of Clinical Nutrition 63, 966–975. [DOI] [PubMed] [Google Scholar]
- Rutstein S. (2003) Effect of Birth Intervals on Mortality and Health. Measure/DHS+/Macro International: Calverton, MD. [Google Scholar]
- Rutstein S. (2005) Effects of preceding birth intervals on neonatal, infant, and under‐five years mortality and nutritional status in developing countries: evidence from the demographic and health surveys. International Journal of Gynecology and Obstetrics 89, S7–S24. [DOI] [PubMed] [Google Scholar]
- Safonova T. & Leparsky E.A. (1998) The unwanted child. Child Abuse and Neglect 22, 155–157. [DOI] [PubMed] [Google Scholar]
- Schroeder D.G. & Brown K.H. (1994) Nutritional status as a predictor of child survival: summarizing the association and quantifying its global impact. Bulletin of the World Health Association 72, 569–579. [PMC free article] [PubMed] [Google Scholar]
- Shrimpton R., Victora C.G., De Onis M., Costa Lima R., Blössner M. & Clugston G. (2001) The worldwide timing of growth faltering: implications for nutritional interventions. Pediatrics 107, e75. [DOI] [PubMed] [Google Scholar]
- Swinburn B. (2004) Undernutrition and childhood obesity together deliver potent punch worldwide. Global Health and Environment Monitor 12, 1–2. [Google Scholar]
- United Nations Children's Fund (2004) State of the World's Children. UNICEF: New York. [Google Scholar]
- Van den Broeck J., Eeckels R. & Massa G. (1996) Maternal determinants of child survival in a rural African community. International Journal of Epidemiology 25, 998–1004. [DOI] [PubMed] [Google Scholar]
- Victora C.G., Adair L., Fall C., Hallal P.C., Martorell R., Richter L. et al (2008) Maternal and child undernutrition: consequences for adult health and human capital. The Lancet 371, 340–357. [DOI] [PMC free article] [PubMed] [Google Scholar]
- World Bank (2006) Repositioning Nutrition as Central to Development: A Strategy for Large‐Scale Action. The International Bank for Reconstruction and Development/The World Bank: Washington, DC. [Google Scholar]
- World Health Organization Working Group (1986) Use and interpretation of anthropometric indicators of nutritional status. Bulletin of the World Health Organization 64, 929–941. [PMC free article] [PubMed] [Google Scholar]
- Zhu B.P., Rolfs R.T., Nangle B.E. & Horan J.M. (1999) Effect of birth interval between pregnancies on perinatal outcomes. The New England Journal of Medicine 340, 589–594. [DOI] [PubMed] [Google Scholar]
- Zhu B.P., Haines K.M., Le T., McGrath‐Miller K. & Boulton M. (2001) Effect of the interval between pregnancies on perinatal outcomes among black and white women. American Journal of Obstetrics and Gynecology 185, 1403–1410. [DOI] [PubMed] [Google Scholar]