Abstract
Stunting, a form of undernutrition, is the best measure of child health inequalities as it captures multiple dimensions of children's health, development and the environment where they live. The aim of this study was to quantify the predictors of childhood stunting in Nigeria. This study used data obtained from the 2008 Nigeria Demographic and Health Survey (NDHS). A total of 28 647 children aged 0–59 months included in NDHS in 2008 were analysed in this study. We applied multilevel multivariate logistic regression analysis in which individual‐level factors were at the first level and community‐level factors at the second level. The percentage change in variance of the full model accounted for about 46% in odds of stunting across the communities. The present study found that the following predictors increased the odds of childhood stunting: male gender, age above 11 months, multiple birth, low birthweight, low maternal education, low maternal body mass index, poor maternal health‐seeking behaviour, poor household wealth and short birth interval. The community‐level predictors found to have significant association with childhood stunting were: child residing in community with high illiteracy rate and North West and North East regions of the country. In conclusion, this study revealed that both individual‐ and community‐level factors are significant determinants of childhood stunting in Nigeria.
Keywords: under nutrition, childhood stunting, multilevel analysis, under‐five children
Background
Linear growth retardation, commonly regarded as stunting, is the best measure of child health inequalities as it captures multiple dimensions of children's health, development and the environment where they live (World Health Organisation 1995; Pradhan et al. 2003). Stunting is a form of under‐nutrition in which the linear growth retardation is due to poor nutrition, infections and environmental problems, both prenatal and postnatal (Grantham‐McGregor et al. 2007). Early occurrence of stunting is associated with poor cognitive, motor and socio‐psychological development, and increased likelihood of mortality (Pelletier et al. 1993; Pelletier and Frongillo 2003). Children suffering from stunting are unable to reach their full growth potentials and often become stunted adolescents and adults later on (Martorell et al. 1994). The consequences of stunting in adulthood are often functional with its attendant reduced work capacity. (Spurr et al. 1977) Stunted women have increased likelihood of morbidity and mortality during childbirth for themselves and their babies (Habicht et al. 1973; Martorell et al. 1981; Royston & Armstrong 1989).
In 2004, it was reported worldwide that about one‐third of pre‐school children were stunted, although the global prevalence is on the decreasing trend from about 47% in 1980 to between 30–40% in 2004 with variations depending on geographic regions (de Onis et al. 2004). Most of the progress in stunting reduction that has been made in low‐ and middle income countries has been in Southeast Asia with little changes in Sub‐Saharan African, which bear the brunt of the problem (de Onis et al. 2000). Research carried out in 1990 showed that in Nigeria, 43% of the under‐fives were stunted (de Onis et al. 1993) while in 2008, the NHDS indicated that 41% of Nigerian children were stunted (NPC and ICF Macro 2009). Comparing these two studies, one would see that not much has changed in the trend and it should be obvious that stunting in the under‐fives in Nigeria is still a major problem needing solution. From studies carried out internationally, it is evident that childhood stunting is linked to a number of environmental and socioeconomic factors such as poverty (Madzingira 1995; Phimmasone et al. 1996; Hong et al. 2006; Van de Poel et al. 2007). Having access to health care services was also found to be significant in determining nutritional status of children as the children who did not have access to health care facilities suffered mostly from childhood stunting (Muhe et al. 1996; Van de Poel et al. 2007; Monteiro et al. 2009). Some studies also associated childhood stunting to poor sanitary condition of the households (El Taguri et al. 2009; Medhin et al. 2010).
Previously published literatures on childhood stunting in Nigeria are limited. Most of them examined few determinants [either socioeconomic factors (Odunayo & Oyewole 2006), cultural factors (Esimai et al. 2001) or environmental factors on their own, or individual and community factors]; hence, effects of confounding factors were not sufficiently looked into. The studies on children's nutritional status in Nigeria were mostly carried out in urban areas with little attention being paid to rural areas (Akaninwor et al. 1996; Ajayi and Akinyinka 1999; Ojo et al. 2000; Abidoye & Ihebuzor 2001; Esimai et al. 2001; Ijarotimi & Ijadunola 2007) where about 70% of the population lives. The 2008 Nigeria Demographic and Health Survey (NDHS) has national representation covering both rural and urban settings; it is new and unexplored by researchers, specifically with respect to stunting in under‐five children.
Therefore, the aim of this study was to examine individual‐ (child's factors, maternal/household factors) and community‐level factors associated with childhood stunting in a single analytical framework to provide reliable and accurate information for policy‐making and programme design that aims at addressing nutritional deficiencies in under‐five children.
Key messages
-
•
This study revealed the importance of examining individual and community‐level factors' influence on stunting as a measure of childhood nutritional status in Nigeria.
-
•
Almost 8% of the variance in the odds of childhood stunting could be attributed to the community‐level factors.
-
•
Individual‐ (such as child's age, sex, breastfeeding duration and household's socioeconomic status) and community‐level factors (such as literacy level and geopolitical region) were independently associated with childhood stunting, suggesting that interventions to reduce childhood stunting should focus more on high risk places as well as high risk groups of children.
Methods
Study design
This is population‐based cross‐sectional study which used data obtained from 2008 NDHS.
Data source/sampling technique
This study was be based on 28 647 under‐five children included in the 2008 NDHS from which data from 18 286 households were taken. The main survey included 24 880 households from rural areas and 11 418 households from urban areas. The country was divided by stratification into 36 states plus the Federal Capital Territory, which were further divided into 774 local government areas all within the six geopolitical zones to obtain a nationally representative sample (NPC & ICF Macro 2009). Domain was set up and each one consists of enumeration areas that was established by the general population and housing census conducted in 2006 (NPC & ICF Macro 2009). The sampling frame was made up of a list of all enumeration areas (clusters) (NPC & ICF Macro 2009). From each domain, a two‐stage probabilistic sampling method was used for the clusters selection. The first stage involved choosing of 888 primary sampling units (PSUs), 602 in the rural and 286 in the urban areas with a probability proportional to the size (NPC & ICF Macro 2009). A second stage of sampling followed the first stage, which involved the systematic sampling of households from the selected enumeration areas (NPC & ICF Macro 2009).
Ethical consideration
Approval was granted for secondary analysis of existing data after the removal of all identifying information of the respondents by the Ethics Committee of the ICF Macro at Calverton in the United States in conjunction with the National Ethics Committee of the Federal Ministry of Health in Nigeria.
Study variable
Dependent variable
The dependent variable of this study was stunting (yes = 1 or no = 0). Stunting was defined as height for age z‐score less than −2 standard deviations (HAZ < −2 SD) from the median of the reference population of World Health Organization (WHO Multicentre Growth Reference Study Group 2006; de Onis et al. 2009). It indicates skeletal growth reduction because of chronic under‐nutrition.
Description of the independent variables
We included both individual‐ and community‐level predictors of stunting. The coding for each explanatory variable is given in Table 1. Full definitions of some of the variables were also stated in the same table. We used the term community to describe clustering within the same geographical living environment. Communities were based on sharing a common PSU within the DHS data. The unit of analysis for the multilevel model was chosen for two reasons. First, in the DHS sample, PSU was used to define the clusters. Second, it has been shown that for most of the DHS conducted, the sample size per cluster met the optimum size with a tolerable precision loss. The bias introduced by using cluster averages based on about 25 women as a proxy for the PSU population averages is very small. (Kravdal 2006)
Table 1.
Independent variables | Description |
---|---|
Individual‐level factors | |
Child factors | |
Age of child (months) | Categorized into (1) 0–11; (2) 12–23; (3) 24–35; (4) 36–47; and (5) 48–59. |
Sex of child | Categorized into (1) female and (2) male. |
Birthweight (g) | Categorized into (1) low < 2500 and (2) normal ≥ 2500. |
Type of birth | Categorized into (1) single and (2) multiple birth. |
Maternal/household factors | |
Maternal age in years | Categorized into (1) 15–24; (2) 25–34; or (3) 35–49. |
Educational level of mother | Categorized into (1) no formal education; (2) primary; (3) secondary; or (4) higher. |
Breastfeeding (months) | Categorized into (1) <6; (2) 6–12; (3) 13–24; or (4) >24. |
Immunization | Categorized into (1) incomplete or (2) complete. |
Mother's body mass index (kg/m2) | Categorized into (1) <18.5; (2) 18.5–24.9; or (3) ≥25.0. |
Occupation | Categorized into (1) not working; (2) manual; or (3) white collar. |
Birth interval (months) | Categorized into (1) ≥24 and (2) <24. |
Number of under‐fives | Categorized into (1) 1; (2) 2; (3) 3; or (4) ≥4. |
Ethnicity | Categorized into (1) major or (2) minor. |
Maternal health‐seeking behaviour | Categorized into (1) (first quantile) (Least); (2) (second quantile); (3)(third quantile); (4) (fourth quantile); or (5) (fifth quantile) (Highest). |
Type of family | Categorized into (1) monogamous or (2) polygamous. |
Head of household | Categorized into (1) male or (2) female. |
Wealth index | Categorized into (1) (first quintile) (Poorest); (2) (second quintile); (3) (third quintile); (4) (fourth quintile); or (5) (fifth quintile) (Richest) |
Community‐level factors | |
Residence | Categorized into (1) rural or (2) urban. |
Geographic region | Categorized into (1) North Central; (2) North East; (3) North West; (4) South East; (5) South South; or (6) South West. |
Poverty rate | Proportion of households living below poverty level (wealth index below 20%, poorest quintile). Categorized into (1) Low or (2) High. Median value serves as the reference for the low and high groups. |
Illiteracy rate | Proportion of people in the community with no formal education. Categorized into (1) Low or (2) High. Median value serves as the reference for the low and high groups. |
Unemployment rate | Proportion of people who are unemployed in the communities. Categorized into (1) Low or (2) High. Median value serves as the reference for the low and high groups. |
Proper sanitation | Categorized into (1) Yes or (2) No. |
Safe water | Categorized into (1) Yes or (2) No. |
Statistical analysis
Descriptive analyses involved the use of numbers and percentage for categorical variables to show the distribution of the outcome variables by the predictor variables. Multivariate multilevel logistic regression was used to analyse factors associated with childhood stunting because of the hierarchical nature of the data set. Four models were constructed for the analysis. The first model, an empty model, was without any explanatory variable i.e. simple component of variance analysis. The second model controlled for the individual‐level variables, the third model controlled for community‐level variables while the fourth controlled for both the individual‐ and community‐level variables simultaneously. P‐value of <0.05 was used to define statistical significance.
Fixed effects (measures of association)
The fixed effects i.e. measures of association have their results presented as adjusted odds ratio (OR) with their corresponding 95% confidence intervals (CIs) and P‐values.
Random effects (measures of variation)
Measures of random effects included intracluster correlation (ICC), median odds ratio (MOR) and proportional change in variance (PCV) (2005a, 2005b). The ICC was calculated by the linear threshold according to the formula used by (Snijders & Bosker 1999) while MOR is a measure of unexplained cluster heterogeneity. The method used for calculating MOR and PCV had been described elsewhere (Larsen & Merlo 2005); (2005a, 2005b).
The statistical analysis on the data was carried out with the use of Stata (xtmelogit routine) statistical software for Windows version 11.
Results
Descriptive analysis
Figure 1 shows the prevalence of childhood stunting as estimated from the data set, which indicated that 7322 (25.6%) of under‐five children have stunted growth across the 37 states. The percentage of childhood stunting ranged from as low as 8.2% in Enugu state to as much as 36.7% in Sokoto state.
Table 2 is a descriptive analysis that shows that the prevalence of stunting was 35.0% in the age group 24–35 months while it was 14.2% in the age group 0–11 months, as reported. There is significantly higher prevalence of childhood stunting among under‐five children who breastfed longer than usual, whose sex is male, did not complete immunization and those who were born with low birthweight. In the same vein, childhood stunting is highest in children born to mothers with low body mass index (BMI) status and whose mothers are not working. The prevalence is highest in the under‐fives whose parents were the poorest with respect to wealth index status and those that had no formal education. Children born to young mothers (15–24 years) had the highest level of stunting. Furthermore, stunting was more prevalent with under‐five children who reside in rural areas and communities with high poverty rate, unemployment rate and no access to safe drinking water.
Table 2.
Variables and their values | Stunted n = 7322 (25.6%) | Not stunted n = 21 325 (74.4%) | Missing values N (%) |
---|---|---|---|
Individual‐level factors | |||
Child factors | |||
Age of child (months) | 4289 (14.9) | ||
0–11 | 820 (14.2) | 4946 (85.8) | |
12–23 | 1819 (37.0) | 3102 (63.0) | |
24–35 | 1584 (35.0) | 2944 (65.0) | |
36–47 | 1631 (33.9) | 3180 (66.1) | |
48–59 | 1468 (33.9) | 2864 (66.1) | |
Sex of the child | – | ||
Male | 3866 (26.5) | 10 738 (73.5) | |
Female | 3456 (24.6) | 10 587 (75.4) | |
Type of birth | – | ||
Single birth | 7058 (25.5) | 20 627 (74.5) | |
Multiple birth | 264 (27.4) | 698 (72.6) | |
Birthweight (grams) | 664 (2.3) | ||
Low < 2500 | 1210 (28.5) | 3029 (71.5) | |
≥2500 | 6000 (25.3) | 17 744 (74.7) | |
Maternal/household factors | |||
Mother's age (years) | – | ||
15–24 | 1867 (25.8) | 5382 (74.2) | |
25–34 | 3617 (25.6) | 10 494 (73.4) | |
35–49 | 1838 (25.2) | 5449 (74.8) | |
Educational levels | – | ||
No formal education | 4287 (29.7) | 10 131 (70.3) | |
Primary | 1681 (25.7) | 4871 (74.3) | |
Secondary | 1181 (18.6) | 5157 (81.4) | |
Higher | 173 (12.9) | 1166 (87.1) | |
Breastfeeding duration (months) | 1679 (5.9) | ||
<6 | 357 (7.6) | 4365 (92.4) | |
6–12 | 1217 (19.4) | 5049 (80.6) | |
13–24 | 5119 (33.3) | 10 244 (66.7) | |
>24 | 293 (47.5) | 324 (52.5) | |
Immunization | – | ||
Incomplete | 3805 (30.5) | 8658 (69.5) | |
Complete | 3517 (21.7) | 12 667 (78.3) | |
Maternal health‐seeking behaviour index in quantiles | 1996 (7.0) | ||
Fifth quantile(highest) | 984 (18.8) | 4241 (81.2) | |
Fourth quantile | 1188 (24.1) | 3735 (75.9) | |
Third quantile | 1536 (27.0) | 4148 (73.0) | |
Second quantile | 1446 (28.9) | 3565 (71.1) | |
First quantile (least) | 1771 (30.5) | 4037 (69.5) | |
Mother's body mass index (kg/m2) | 680 (2.4) | ||
<18.5 | 1129 (34.0) | 2195 (66.0) | |
18.5–24.9 | 5006 (26.6) | 13 801 (73.4) | |
≥25.0 | 1090 (18.7) | 4746 (81.3) | |
Mother's occupation | 157 (0.6) | ||
Not working | 2387 (26.4) | 6648 (73.6) | |
Manual | 2522 (26.6) | 7615 (76.3) | |
White collar | 2370 (23.7) | 6948 (73.4) | |
Birth interval (months) | – | ||
≥24 | 5880 (25.3) | 17 392 (74.7) | |
<24 | 1442 (26.8) | 3933 (73.2) | |
Number of children under‐five | |||
One | 24 (2.2) | 1090 (97.9) | – |
Two | 1660 (22.8) | 5633 (77.2) | |
Three | 3049 (27.5) | 8039 (72.5) | |
≥Four | 2589 (28.3) | 6563 (71.7) | |
Ethnicity | 166 (0.6) | ||
Major | 4319 (25.7) | 12 519 (74.4) | |
Minor | 2954 (25.4) | 8689 (74.6) | |
Types of family | 1456 (5.1) | ||
Monogamous | 4427 (24.5) | 13 636 (75.5) | |
Polygamous | 2593 (28.4) | 6535 (71.6) | |
Head of household | – | ||
Female | 594 (22.3) | 2074 (77.7) | |
Male | 6728 (25.9) | 19 251 (74.1) | |
Wealth index | – | ||
Fifth quantile (richest) | 592 (15.6) | 3216 (84.5) | |
Fourth quantile | 1001 (21.0) | 3754 (79.0) | |
Third quantile | 1454 (25.9) | 4155 (74.1) | |
Second quantile | 1993 (29.0) | 4878 (71.0) | |
First quantile (poorest) | 2282 (30.0) | 5322 (70.0) | |
Community‐level factors | |||
Residence | – | ||
Rural | 5738 (27.3) | 15 296 (72.7) | |
Urban | 1584 (20.8) | 6029 (79.2) | |
Regions | – | ||
North central | 1336 (26.5) | 3710 (73.5) | |
North east | 1938 (29.6) | 4621 (70.4) | |
North west | 2402 (30.2) | 5545 (69.8) | |
South east | 327 (13.4) | 2123 (86.6) | |
South south | 611 (18.4) | 2716 (81.6) | |
South west | 708 (21.3) | 2610 (78.7) | |
Poverty rate | – | ||
Low | 3156 (21.9) | 11 267 (78.1) | |
High | 4166 (29.3) | 10 058 (70.7) | |
Illiteracy rate | – | ||
Low | 3031 (21.1) | 11 347 (78.9) | |
High | 4291 (30.1) | 9978 (69.9) | |
Unemployment rate | – | ||
Low | 3436 (23.9) | 10 934 (76.1) | |
High | 3886 (27.2) | 10 391 (72.8) | |
Proper sanitation | – | ||
Yes | 3553 (24.7) | 10 829 (75.3) | |
No | 3769 (26.4) | 10 496 (73.6) | |
Safe water | – | ||
Yes | 3286 (22.9) | 11 056 (77.1) | |
No | 4036 (28.2) | 10 269 (71.8) |
N, number of children; %, proportion (percentage).
Multilevel analysis
Table 3 shows the results of multilevel models for both individual‐ and community‐level factors. With all factors controlled for in the multilevel analysis, child's age was statistically significantly associated with the odds of childhood stunting. Male children were more likely to be stunted than female children (adjusted OR 1.18; 95% CI 1.10–1.26, P < 0.001). Children who are products of multiple births were 89% more likely to be stunted. Children born with low birthweight were 21% more likely to be stunted. Under‐five children of mothers with the highest health‐seeking behaviour index were 22% less likely to be stunted when compared with those from mothers with the least health‐seeking behaviour. Children of mothers with low BMI (<18.5 kg/m2) were 26% more likely to be stunted compared with the children of mothers with normal BMI. Generally, the odds of a child being stunted increases with increasing breastfeeding duration after 6 months. Children with shorter than usual birth interval were 19% more likely to be stunted. The odds of being stunted decreases with increasing wealth index level. When all the factors were controlled, the association between childhood stunting, mother's occupation and immunization status, number of under‐five children, and household head were statistically not significant. When other factors were controlled for in the model, community‐level illiteracy rate remained statistically significantly associated with childhood stunting. For communities with high illiteracy rate, the odds of a child being stunted (OR 1.49; 95% CI 1.19–1.88, P < 0.001) increased by 49%. Children whose parents reside in the North East region were 35% more likely to be stunted when compared with those from the North central region.
Table 3.
Covariates | Model 1 empty | Model 2 individual | Model 3 community | Model 4 individual & community | ||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|
OR | 95% CI | P‐value | aOR | 95% CI | P‐value | aOR | 95% CI | P‐value | aOR | 95% CI | P‐value | |
Individual‐level factors | ||||||||||||
Child factors | ||||||||||||
Child's age (months) | ||||||||||||
0–11 | 1.00 | 1.00 | ||||||||||
12–23 | 2.24 | (1.94–2.60) | <0.001 | 2.36 | (2.04–2.74) | <0.001 | ||||||
24–35 | 2.01 | (1.72–2.34) | <0.001 | 2.12 | (1.82–2.48) | <0.001 | ||||||
36–47 | 1.99 | (1.71–2.32) | <0.001 | 2.09 | (1.79–2.44) | <0.001 | ||||||
48–59 | 1.97 | (1.68–2.30) | <0.001 | 2.06 | (1.76–2.41) | <0.001 | ||||||
Sex | ||||||||||||
Female | 1.00 | 1.00 | ||||||||||
Male | 1.17 | (1.10–1.25) | <0.001 | 1.04 | (1.10–1.26) | <0.001 | ||||||
Type of birth | ||||||||||||
Single birth | 1.00 | 1.00 | ||||||||||
Multiple birth | 1.84 | (1.49–2.26) | <0.001 | 1.89 | (1.52–2.32) | <0.001 | ||||||
Birthweight (g) | ||||||||||||
≥2500 | 1.00 | 1.00 | ||||||||||
Low <2500 | 1.21 | (1.10–1.33) | <0.001 | 1.21 | (1.10–1.33) | <0.001 | ||||||
Maternal/household factors | ||||||||||||
Mother's age (years) | ||||||||||||
15–24 | 1.00 | 1.00 | ||||||||||
25–34 | 0.98 | (0.90–1.07) | 0.635 | 0.99 | (0.92–1.09) | 0.982 | ||||||
35–49 | 0.88 | (0.80–0.97) | 0.007 | 0.91 | (0.82–0.99) | 0.044 | ||||||
Birth interval (months) | ||||||||||||
≥24 | 1.00 | 1.00 | ||||||||||
<24 | 1.18 | (1.08–1.28) | <0.001 | 1.19 | (1.09–1.30) | <0.001 | ||||||
Maternal education | ||||||||||||
No education | 1.00 | 1.00 | ||||||||||
Primary | 0.98 | (0.89–1.08) | 0.707 | 0.98 | (0.78–1.07) | 0.267 | ||||||
Secondary | 0.83 | (0.74–0.94) | 0.002 | 0.84 | (0.64–0.97) | 0.024 | ||||||
Higher | 0.73 | (0.58–0.92) | 0.008 | 0.75 | (0.35–0.86) | 0.001 | ||||||
MHSBI | ||||||||||||
First quantile (least) | 1.00 | 1.00 | ||||||||||
Second quantile | 0.97 | (0.88–1.08) | 0.613 | 1.09 | (0.94–1.26) | 0.990 | ||||||
Third quantile | 0.92 | (0.82–1.03) | 0.131 | 0.96 | (0.85–1.24) | 0.321 | ||||||
Fourth quantile | 0.88 | (0.78–0.99) | 0.038 | 0.92 | (0.82–1.10) | 0.238 | ||||||
Fifth quantile (highest) | 0.70 | (0.60–0.80) | <0.001 | 0.83 | (0.66–0.92) | <0.001 | ||||||
Mother's BMI (kg/m2) | ||||||||||||
18.5–24.9 | 1.00 | 1.00 | ||||||||||
<18.5 | 1.27 | (1.15–1.39) | <0.001 | 1.26 | (1.14–1.39) | <0.001 | ||||||
≥25.0 | 0.79 | (0.71–0.85) | <0.001 | 0.79 | (0.72–0.87) | <0.001 | ||||||
Mother's occupation | ||||||||||||
Not working | 1.00 | 1.00 | ||||||||||
Manual | 1.02 | (0.93–1.11) | 0.720 | 1.03 | (0.94–1.12) | 0.557 | ||||||
White collar | 1.00 | (0.90–1.08) | 0.768 | 1.03 | (0.94–1.14) | 0.452 | ||||||
Breast feeding (months) | ||||||||||||
<6 | 1.00 | 1.00 | ||||||||||
6–12 | 2.13 | (1.83–2.48) | <0.001 | 2.14 | (1.84–2.49) | <0.001 | ||||||
13–24 | 2.66 | (2.24–3.17) | <0.001 | 2.52 | (2.12–3.00) | <0.001 | ||||||
>24 | 3.74 | (2.90–4.83) | <0.001 | 3.52 | (2.73–4.55) | <0.001 | ||||||
Immunization | ||||||||||||
Complete | 1.00 | 1.00 | ||||||||||
Incomplete | 1.05 | (0.98–1.13) | 0.158 | 1.06 | (0.95–1.17) | 0.284 | ||||||
Number of under‐fives | ||||||||||||
≥Four | 1.00 | 1.00 | ||||||||||
Three | 0.93 | (0.54–1.55) | 0.726 | 0.84 | (0.51–1.42) | 0.888 | ||||||
Two | 0.92 | (0.54–1.54) | 0.736 | 0.83 | (0.49–1.41) | 0.778 | ||||||
One | 0.85 | (0.50–1.43) | 0.529 | 0.78 | (0.46–1.32) | 0.745 | ||||||
Ethnicity | ||||||||||||
Major | 1.00 | 1.00 | ||||||||||
Minor | 0.97 | (0.89–1.06) | 0.506 | 0.99 | (0.89–1.10) | 0.873 | ||||||
Family type | ||||||||||||
Monogamy | 1.00 | 1.00 | ||||||||||
Polygamy | 1.12 | (0.93–1.21) | 0.056 | 1.10 | (0.96–1.19) | 0.123 | ||||||
Household head | ||||||||||||
Male | 1.00 | 1.00 | ||||||||||
Female | 0.97 | (0.85–1.10) | 0.602 | 1.01 | (0.89–1.16) | 0.832 | ||||||
Wealth index | ||||||||||||
Fifth quantile (richest) | 1.00 | 1.00 | ||||||||||
Fourth quantile | 1.20 | (1.04–1.39) | <0.001 | 1.21 | (1.04–1.41) | 0.014 | ||||||
Third quantile | 1.52 | (1.30–1.78) | <0.001 | 1.51 | (1.27–1.79) | <0.001 | ||||||
Second quantile | 1.64 | (1.39–1.92) | <0.001 | 1.58 | (1.31–1.91) | <0.001 | ||||||
First quantile (poorest) | 1.71 | (1.45–2.03) | <0.001 | 1.64 | (1.33–2.01) | <0.001 | ||||||
Community‐level factors | ||||||||||||
Residence | ||||||||||||
Urban | 1.00 | 1.00 | ||||||||||
Rural | 1.21 | (0.91–1.36) | 0.099 | 1.27 | (0.94–1.42) | 0.336 | ||||||
Region | ||||||||||||
North central | 1.00 | 1.00 | ||||||||||
North west | 1.13 | (1.02–1.23) | <0.001 | 1.26 | (1.14–1.29) | <0.001 | ||||||
North east | 1.28 | (1.22–1.46) | 0.024 | 1.35 | (1.27–1.43) | 0.036 | ||||||
South east | 0.45 | (0.38–0.54) | <0.001 | 0.49 | (0.39–0.56) | <0.001 | ||||||
South south | 0.66 | (0.57–0.77) | <0.001 | 0.76 | (0.78–0.89) | <0.001 | ||||||
South west | 0.86 | (0.74–0.99) | 0.040 | 0.97 | (0.72–1.00) | 0.044 | ||||||
Poverty rate | ||||||||||||
Low | 1.00 | 1.00 | ||||||||||
High | 1.01 | (0.93–1.08) | 0.921 | 1.05 | (0.89–1.10) | 0.524 | ||||||
Illiteracy rate | ||||||||||||
Low | 1.00 | 1.00 | ||||||||||
High | 1.37 | (1.06–1.76) | <0.001 | 1.49 | (1.19–1.88) | <0.001 | ||||||
Unemployment rate | ||||||||||||
Low | 1.00 | 1.00 | ||||||||||
High | 1.00 | (0.95–1.06) | 0.964 | 0.95 | (0.81–1.11) | 0.521 | ||||||
Proper sanitation | ||||||||||||
Yes | 1.00 | 1.00 | ||||||||||
No | 1.05 | (0.99–1.12) | 0.091 | 1.21 | (0.89–1.35) | 0.528 | ||||||
Safe water | ||||||||||||
Yes | 1.00 | 1.00 | ||||||||||
No | 1.06 | (1.00–1.13) | 0.046 | 1.08 | (0.74–1.15) | 0.087 |
aOR, adjusted odds ratio; CI, confidence interval; BMI, body mass index; MHSBI, maternal health‐seeking behaviour index.
Measures of variations
As shown in Table 4, with respect to the empty model (the null model), there was a significant variation in the odds of childhood stunting (τ = 0.282, P < 0.001) across the communities. The ICC indicated by the estimated intercept component variance showed that 7.9% of the variance in the odds of childhood stunting could be attributed to the community‐level factors. The variations in childhood stunting across communities remained statistically significant, even after adjusting for individual and community‐level factors (full model). As determined by percentage change in variance, the full model accounts for about 46.0% in the odds of childhood stunting across the communities.
Table 4.
Measures of variation | Model 1* | P‐value | Model 2 † | P‐value | Model 3 ‡ | P‐value | Model 4 § | P‐value |
---|---|---|---|---|---|---|---|---|
Community‐level | ||||||||
Variance (SE) | 0.282 (0.025) | <0.001 | 0.176 (0.024) | <0.001 | 0.163 (0.023) | <0.001 | 0.155 (0.021) | <0.001 |
Explained variation (%) | Reference | 37.6 | 42.4 | 46.0 | ||||
ICC (%) | 7.9 | 5.1 | 4.8 | 4.6 | ||||
MOR | 1.66 | 1.50 | 1.49 | 1.46 | ||||
Model fit statistics | ||||||||
DIC (−2 log likelihood) | 32 055 | 22 538 | 22 487 | 22 466 |
SE, standard error; ICC, intracluster correlation; MOR, median odds ratio; DIC, deviation information criterion. *Model 1 is the empty model, a baseline model without any determinant variable. †Model 2 is adjusted for individual‐level factors. ‡Model 3 is adjusted for community‐level factors. §Model 4 is final model adjusted for both individual‐ and community‐level factors.
Results of the MOR also confirmed evidence of community‐dependent phenomenon modifying childhood stunting. The MOR for stunting was 1.66 in the empty model; this relatively low MOR suggests that the clustering effect was moderate. The unexplained community heterogeneity in stunting decreased to an MOR of 1.46 when individual‐ and community‐level factors were added to the empty model. Thus, there are very little variations between communities in the predisposition for being stunted.
Discussion
This study examined individual‐ and community‐level factors as significant determinants of childhood stunting using nationally representative survey data. It confirms the importance of community variations with respect to childhood stunting. Both individual‐ and community‐level factors in the final model accounted for about 46% of the variations observed for stunting. Children of age group 12–24 months have the highest odds, which subsequently declined after 24 months. This finding supports the findings of two previous studies conducted by Kabubo‐Mariara et al. (2009) and Shrimpton et al. (2001), which revealed a rapid fall in children's height from birth to 24 months; although stunting processes after 24 months still continue, but at a much slower rate. This could be as a result of weaning and lower breast milk intakes, which make them prone to childhood stunting.
The result of this study indicated that the sex of a child is a strong determinant of stunting in under‐five children. Previous literature has reported an inconsistent association between the sex of the child and childhood stunting. While some studies have reported that male children were more likely to be stunted, others found that female children were more likely to be stunted. However, one recent meta‐analysis of DHS from 16 sub‐Saharan countries found that male children were more likely to be stunted in most of the countries studied (10 out of 16) (Wamani et al. 2007). In this study, we found male children had higher odds of being stunted compared with female children. Previous studies in neonatology and in cohorts of preterm under‐fives showed both morbidity and mortality to be consistently higher in males than females in early life even after adjusting for gestational age and body size, and this being more pronounced in the preterm under‐fives (Chen et al. 1993; Kilbride & Daily 1998; Elsmen et al. 2004). An alternative explanation could be as a result of preferential treatment of females as a result of a high value placed on women for agricultural labour in some cultures and this supports the findings of other studies.(Cronk 1989; Sevedberg 1996). Findings of this current study is consistent with the findings of other previous studies (Madzingira 1995; Reed et al. 1996; Hong et al. 2006; Kabubo‐Mariara et al. 2009), which found that children of multiple births are more likely to be stunted than those of single births. This could be attributed to inadequate breastfeeding, low birthweight and competition for nutritional intake, which happen more in children of multiple births than those of single births.
The findings of this study support those of other similar studies, which indicated that maternal education has positive effect on childhood stunting (Hong et al. 2006; Odunayo & Oyewole 2006; Pongou et al. 2006; Wamani et al. 2006; Ijarotimi & Ijadunola 2007; Van de Poel et al. 2007; Semba et al. 2008; El Taguri et al. 2009; Kabubo‐Mariara et al. 2009; Monteiro et al. 2009). Mothers with formal education may be knowledgeable of what to do to prevent childhood stunting from occurring or lessen the degree if it occurs. Maternal health‐seeking behaviour index in this study was found to have positive effect on childhood stunting, which is consistent with findings of other similar studies that examined the predictors of childhood stunting (Pongou et al. 2006; Uthman 2009). Mother's BMI in this study revealed a significant association with childhood stunting; this suggests that under‐fives whose mothers have low BMI are more likely to suffer from childhood stunting. Similar finding has been documented in some studies (Bhargava 1999; Uthman 2009). However, one may argue that the mother's height will be a better parameter for assessing this association; to the best of our knowledge, we did not find any literature to support this. Prolonged breastfeeding, particularly beyond infancy (>12 months), in this study was associated with childhood stunting. This finding could be attributed to the fact that most households are poor in developing countries and are unable to feed the children with adequate and quality complementary foods. Hence, they continue to practice solely exclusive breastfeeding beyond 12 months without supplementation with complementary foods (Larrea and Kawachi 2005; Hong 2007; Van de Poel et al. 2007). There is also the possibility that some of the children that are breastfed for longer than normal duration refused other foods apart from breast milk as evidenced by a study conducted in Ghana (Brakohiapa et al. 1988).
In addition, findings of this study revealed that under‐five children from less wealthy households have greater odds of being stunted than under‐fives from the wealthy households. Similar results were documented in previous studies carried out in different developing countries, which gives further proof that poverty is a significant predictor of childhood stunting (Hong et al. 2006; Odunayo & Oyewole 2006; Pongou et al. 2006; Hong 2007; Van de Poel et al. 2007; Hien & Kam 2008; Semba et al. 2008; Kabubo‐Mariara et al. 2009; Monteiro et al. 2009; Ramli et al. 2009). Levels of socioeconomic status of a household are vital as it often determines the availability of good and nutritious foods for the growth and development of children. Moreover, under‐five children from poorer households in most developing countries like Nigeria where a publicly funded health care system is not practised would lack access to good and basic health care services when they fall ill.
Using multilevel method of analysis for this study allowed variations in childhood stunting among different communities to be fully accounted for by contextual effects (area or neighbourhood of residence effects) rather than compositional effects (individual characteristics effects), although the variations among different communities with respect to the odds of having childhood stunting were found to originate mainly from variations in individual‐level factors. These findings are consistent with most studies that have tried to differentiate contextual effects from compositional effects (Frohlich et al. 2002; Subramanian et al. 2003) and supports a major role for community‐level phenomenon as a strong influence on childhood stunting. This study therefore supports the growing body of evidence advocating that community‐level factors, on their own, exert a significant influence on individual‐level factors (Macintyre et al. 1993; Pickett and Pearl 2001).
Study strengths and limitations
One of the strengths of this study is the representativeness of the data from the multistage sampling technique used, which makes the findings of the study to be relevant to the study population and also generalizable if applied to similar populations. An important strength of the study is it being population‐based study with a large sample size and a participants' response rate of over 98%. Another important strength of this study is the small amount of missing data in virtually all the variables used apart from the child's age group category. In addition, another important strength of this study is using multilevel analysis in this study has made it possible to reveal other factors beyond individual‐level factors that were responsible for the variations in childhood stunting; which the typical one‐level model would not be able to reveal. Lastly, the NDHS data set, being the best data available in Nigeria and is used by both national and international agencies, gives credibility to the worthiness of this study.
An important limitation of this study is the cross‐sectional nature of the study, which cannot be used as a good measure of causal relationship. Hence, strong conclusions cannot be drawn with respect to the aetiology of childhood stunting. The selection of multilevel units in this study could be a source of selection bias. We included only two levels (children – level 1 and community – level 2). The multilevel analysis can be performed at more finite levels, namely child (level 1), maternal (level 2) and community (level 3). However, the DHS was not designed for such stratification. Another limitation of this study is the difficulty in obtaining income and expenditure data for measuring wealth status in developing countries like Nigeria; therefore, an asset‐based index is the only feasible alternative for measuring household wealth status. Lastly, there is an absence of dietary consumption data and other health care indicators such as morbidity data to substantiate the findings.
Policy implications
The findings from this study have some relevant policy implications. There is a clear need for intervention to reduce economic inequalities and ultimately poverty among the populace. Adult literacy programmes with special focus on child's health and nutrition should be organized particularly for women in communities with a high illiteracy rate as a short‐term solution aimed at increasing low literacy level in Nigeria. In addition, governments of developing countries should ensure that female children are given appropriate formal education as a long‐term measure. All of these will ultimately help equip uneducated mothers with knowledge on how to prevent stunting from occurring. Any intervention by governmental and non‐governmental organizations that aim at improving under‐five children's nutritional status should consider regions with a high rate of childhood stunting so as to avert under‐coverage of the regions that deserve it. Nutritional intervention programmes like the therapeutic feeding and nursery school feeding programmes should be established and be directed specifically towards higher risk groups such as male under‐fives, children born to non‐educated or less‐educated mothers, children who are products of multiple birth, and children who reside where there is a high illiteracy rate. This will at least lessen the degree of linear growth retardation in those who are already experiencing childhood stunting. Furthermore, interventions directed at subsidizing consumptions particularly for under‐five children should be instituted by both governmental and non‐governmental agencies in Nigeria and other developing countries. Establishment of public service fees especially health service fees for the less wealthy and poor in the society who are unable to pay for services would also be of tremendous help at lessening the degree of growth retardation in the sufferers. Moreover, provision of free health care for pregnant mothers, new mothers and young children, as being carried out in Sierra Leone, can serve as important policy alternative. Lastly, future studies should be conducted, which can be used for establishing causal relationships and will also include morbidity factors.
Source of funding
The authors have no support or funding to report.
Conflicts of interest
The authors declare that they have no conflicts of interest.
Contributions
VTA was involved in the conception of the study. VTA carried out data extraction. VTA conducted statistical analysis with contributions from OAU and GAK. VTA drafted the paper with contributions from OAU and GAK. All authors read and approved the final manuscript.
Acknowledgement
The data used in this study were made available through MEASURE DHS Archive. The data were originally collected by the ICF Macro, Calverton, USA. Neither the original collectors of the data nor the Data Archive bear any responsibility for the analyses or interpretations presented in this project. The authors thank the two anonymous reviewers for the critical review of an earlier version of the manuscript.
References
- Abidoye R.O. & Ihebuzor N.N. (2001) Assessment of nutritional status using anthropometric methods on 1–4 year old children in an urban ghetto in Lagos, Nigeria. Nutrition and Health 15, 29–39. Available from: PM:11403370. [DOI] [PubMed] [Google Scholar]
- Ajayi I.O. & Akinyinka O.O. (1999) Evaluation of the nutritional status of first year school children in Ibadan, Southwest Nigeria. African Journal of Medicine and Medical Sciences 28, 59–63. Available from: PM:12953989. [PubMed] [Google Scholar]
- Akaninwor J.O., Abbey B.W. & Ayalogu E.D. (1996) Profile of protein energy malnutrition amongst children under four years in urban areas of Rivers State. West African Journal of Medicine 15, 50–55. Available from: PM:8652441. [PubMed] [Google Scholar]
- Bhargava A. (1999) Modeling the effects of nutritional and socioeconomic factors on the growth and morbidity of Kenyan school children. American Journal of Human Biology 11, 317–326. Available from: PM:11533953. [DOI] [PubMed] [Google Scholar]
- Brakohiapa L.A., Yartey J., Bille A., Harrison E., Quansah E., Armar M.A., Kishi K. & Yamamoto S. (1988) Does prolonged breastfeeding adversely affect a child's nutritional status? Lancet 2, 416–418. Available from: PM:2900352. [DOI] [PubMed] [Google Scholar]
- Chen S.J., Vohr B.R. & Oh W. (1993) Effects of birth order, gender, and intrauterine growth retardation on the outcome of very low birth weight in twins. The Journal of Pediatrics 123, 132–136. Available from: PM:8320607. [DOI] [PubMed] [Google Scholar]
- Cronk L. (1989) Low socioeconomic status and female‐biased parental investment: the mukogodo example. American Anthropologist 91, 414–429. Available at: 10.1525/aa.1989.91.2.02a00090. [DOI] [Google Scholar]
- El Taguri A., Betilmal I., Mahmud S.M., Monem A.A., Goulet O., Galan P. & Hercberg S. (2009) Risk factors for stunting among under‐fives in Libya. Public Health Nutrition 12, 1141–1149. Available from: PM:18789172. [DOI] [PubMed] [Google Scholar]
- Elsmen E., Hansen P.I. & Hellstrom‐Westas L. (2004) Preterm male infants need more initial respiratory and circulatory support than female infants. Acta Paediatrica 93, 529–533. Available from: PM:15188982. [DOI] [PubMed] [Google Scholar]
- Esimai O.A., Ojofeitimi E.O. & Oyebowale O.M. (2001) Sociocultural practices influencing under five nutritional status in an urban community in Osun State, Nigeria. Nutrition and Health 15, 41–46. Available from: PM:11403372. [DOI] [PubMed] [Google Scholar]
- Frohlich K.L., Potvin L., Gauvin L. & Chabot P. (2002) Youth smoking initiation: disentangling context from composition. Health & Place 8, 155–166. Available from: PM:12135639. [DOI] [PubMed] [Google Scholar]
- Grantham‐McGregor S., Cheung Y.B., Cueto S., Glewwe P., Richter L. & Strupp B. (2007) Developmental potential in the first 5 years for children in developing countries. Lancet 369, 60–70. Available from: PM:17208643. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Habicht J.P., Yarbrough C., Lechtig A. & Klein R.E. (1973) Relationship of birthweight, maternal nutrition and infant mortality. Nutrition Reports International 7, 533–546. Available from: PM:12306500. [PubMed] [Google Scholar]
- Hien N.N. & Kam S. (2008) Nutritional status and the characteristics related to malnutrition in children under five years of age in Nghean, Vietnam. Journal of Preventive Medicine and Public Health 41, 232–240. Available from: PM:18664729. [DOI] [PubMed] [Google Scholar]
- Hong R. (2007) Effect of economic inequality on chronic childhood undernutrition in Ghana. Public Health Nutrition 10, 371–378. Available from: PM:17362533. [DOI] [PubMed] [Google Scholar]
- Hong R., Banta J.E. & Betancourt J.A. (2006) Relationship between household wealth inequality and chronic childhood under‐nutrition in Bangladesh. International Journal for Equity in Health 5, 15. Available from: PM:17147798. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Ijarotimi O.S. & Ijadunola K.T. (2007) Nutritional status and intelligence quotient of primary schoolchildren in Akure community of Ondo State, Nigeria. Tanzania Health Research Bulletin 9, 69–76. Available from: PM:17722408. [DOI] [PubMed] [Google Scholar]
- Kabubo‐Mariara J., Ndenge G.K. & Mwabu D.K. (2009) Determinants of children's nutritional status in Kenya: evidence from demographic and health surveys. Journal of African Economies 18, 363–387. Available at: http://jae.oxfordjournals.org/cgi/content/abstract/18/3/363. [Google Scholar]
- Kilbride H.W. & Daily D.K. (1998) Survival and subsequent outcome to five years of age for infants with birth weights less than 801 grams born from 1983 to 1989. Journal of Perinatology 18, 102–106. Available from: PM:9605298. [PubMed] [Google Scholar]
- Kravdal Ø. (2006) A simulation‐based assessment of the bias produced when using averages from small DHS clusters as contextual variables in multilevel models. Demographic Research 15, 1–20. Available at: http://www.demographic-research.org/volumes/vol15/1. [Google Scholar]
- Larrea C. & Kawachi I. (2005) Does economic inequality affect child malnutrition? The case of Ecuador. Social Science & Medicine 60, 165–178. Available from: PM:15482876. [DOI] [PubMed] [Google Scholar]
- Larsen K. & Merlo J. (2005) Appropriate assessment of neighborhood effects on individual health: integrating random and fixed effects in multilevel logistic regression. American Journal of Epidemiology 161, 81–88. Available from: PM:15615918. [DOI] [PubMed] [Google Scholar]
- Macintyre S., Maciver S. & Sooman A. (1993) Area, class and health: should we be focusing on places or people? Journal of Social Policy 22, 213–234. Available at: http://journals.cambridge.org/action/displayAbstract?fromPage=online&aid=3308796&fulltexttype=RA&fileId=S0047279400019310. [Google Scholar]
- Madzingira N. (1995) Malnutrition in children under five in Zimbabwe: effect of socioeconomic factors and disease. Social Biology 42, 239–246. Available from: PM:8738549. [DOI] [PubMed] [Google Scholar]
- Martorell R., Delgado H.L., Valverde V. & Klein R.E. (1981) Maternal stature, fertility and infant mortality. Human Biology 53, 303–312. Available from: PM:7309018. [PubMed] [Google Scholar]
- Martorell R., Khan L.K. & Schroeder D.G. (1994) Reversibility of stunting: epidemiological findings in children from developing countries. European Journal of Clinical Nutrition 48 (Suppl. 1), S45–S57. Available from: PM:8005090. [PubMed] [Google Scholar]
- Medhin G., Hanlon C., Dewey M., Alem A., Tesfaye F., Worku B., Tomlinson M. & Prince M. (2010) Prevalence and predictors of undernutrition among infants aged six and twelve months in Butajira, Ethiopia: the P‐MaMiE Birth Cohort. BMC Public Health 10, 27. Available from: PM:20089144. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Merlo J., Chaix B., Yang M., Lynch J. & Rastam L. (2005a) A brief conceptual tutorial of multilevel analysis in social epidemiology: linking the statistical concept of clustering to the idea of contextual phenomenon. Journal of Epidemiology and Community Health 59, 443–449. Available from: PM:15911637. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Merlo J., Yang M., Chaix B., Lynch J. & Rastam L. (2005b) A brief conceptual tutorial on multilevel analysis in social epidemiology: investigating contextual phenomena in different groups of people. Journal of Epidemiology and Community Health 59, 729–736. Available from: PM:16100308. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Monteiro C.A., Benicio M.H., Konno S.C., Silva A.C., Lima A.L. & Conde W.L. (2009) Causes for the decline in child under‐nutrition in Brazil, 1996–2007. Revista de Saude Publica 43, 35–43. Available from: PM:19169574. [DOI] [PubMed] [Google Scholar]
- Muhe L., Byass P., Freij L., Sandstrom A. & Wall S. (1996) A one‐year community study of under‐fives in rural Ethiopia: health and behavioural determinants of morbidity. Public Health 110, 215–219. Available from: PM:8757702. [DOI] [PubMed] [Google Scholar]
- National Population Commission (NPC) & ICF Macro (2009) Nigeria: Demographic and Health Survey, 2008 – Final Report Abuja, Nigeria. 26‐11‐2009. Ref Type: Online Source. Available at: http://www.measuredhs.com/pubs/pdf/FR222/FR222.pdf
- Odunayo S.I. & Oyewole A.O. (2006) Risk factors for malnutrition among rural Nigerian children. Asia Pacific Journal of Clinical Nutrition 15, 491–495. Available from: PM:17077064. [PubMed] [Google Scholar]
- Ojo O., Deane R. & Amuna P. (2000) The use of anthropometric and clinical parameters for early identification and categorisation of nutritional risk in pre‐school children in Benin City, Nigeria. The Journal of the Royal Society for the Promotion of Health 120, 230–235. Available from: PM:11197450. [DOI] [PubMed] [Google Scholar]
- de Onis M., Monteiro C., Akre J. & Glugston G. (1993) The worldwide magnitude of protein‐energy malnutrition: an overview from the WHO Global Database on Child Growth. Bulletin of the World Health Organization 71, 703–712. Available from: PM:8313488. [PMC free article] [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, 1222–1233. Available from: PM:11100617. [PMC free article] [PubMed] [Google Scholar]
- de Onis M., Blossner M., Borghi E., Morris R. & Frongillo E.A. (2004) Methodology for estimating regional and global trends of child malnutrition. International Journal of Epidemiology 33, 1260–1270. Available from: PM:15542535. [DOI] [PubMed] [Google Scholar]
- de Onis O.M., Garza C., Onyango A.W. & Rolland‐Cachera M.F. (2009) [WHO growth standards for infants and young children]. Archives of Pediatrics 16, 47–53. Available from: PM:19036567. [DOI] [PubMed] [Google Scholar]
- Pelletier D.L. & Frongillo E.A. (2003) Changes in child survival are strongly associated with changes in malnutrition in developing countries. The Journal of Nutrition 133, 107–119. Available from: PM:12514277. [DOI] [PubMed] [Google Scholar]
- Pelletier D.L., Frongillo E.A. Jr & Habicht J.P. (1993) Epidemiologic evidence for a potentiating effect of malnutrition on child mortality. American Journal of Public Health 83, 1130–1133. Available from: PM:8342721. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Phimmasone K., Douangpoutha I., Fauveau V. & Pholsena P. (1996) Nutritional status of children in the Lao PDR. Journal of Tropical Pediatrics 42, 5–11. Available from: PM:8820613. [DOI] [PubMed] [Google Scholar]
- Pickett K.E. & Pearl M. (2001) Multilevel analyses of neighbourhood socioeconomic context and health outcomes: a critical review. Journal of Epidemiology and Community Health 55, 111–122. Available from: PM:11154250. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Pongou R., Ezzati M. & Salomon J.A. (2006) Household and community socioeconomic and environmental determinants of child nutritional status in Cameroon. BMC Public Health 6, 98. Available from: PM:16618370. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Pradhan M., Sahn D.E. & Younger S.D. (2003) Decomposing world health inequality. Journal of Health Economics 22, 271–293. Available from: PM:12606146. [DOI] [PubMed] [Google Scholar]
- Ramli K.A., Agho K.E., Inder K.J., Bowe S.J., Jacobs J. & Dibley M.J. (2009) Prevalence and risk factors for stunting and severe stunting among under‐fives in North Maluku province of Indonesia. BMC Pediatrics 9, 64. Available from: PM:19818167. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Reed B.A., Habicht J.P. & Niameogo C. (1996) The effects of maternal education on child nutritional status depend on socio‐environmental conditions. International Journal of Epidemiology 25, 585–592. Available from: PM:8671560. [DOI] [PubMed] [Google Scholar]
- Royston E. & Armstrong S. (1989) Preventing Maternal Deaths. World Health Organization: Geneva, Switzerland. [Google Scholar]
- Semba R.D., de P.S., Sun K., Sari M., Akhter N. & Bloem M.W. (2008) Effect of parental formal education on risk of child stunting in Indonesia and Bangladesh: a cross‐sectional study. Lancet 371, 322–328. Available from: PM:18294999. [DOI] [PubMed] [Google Scholar]
- Sevedberg P. (1996) Undernutrition‐gender bias in Sub‐Saharan Africa. The Journal of Development Studies 32, 933–943. [Google Scholar]
- Shrimpton R., Victora C.G., de O.M., Lima R.C., Blossner M. & Clugston G. (2001) Worldwide timing of growth faltering: implications for nutritional interventions. Pediatrics 107, E75. Available from: PM:11331725. [DOI] [PubMed] [Google Scholar]
- Snijders T. & Bosker R. (1999) Multilevel Analysis – An Introduction to Basic and Advanced Multilevel Modelling. SAGE publications: Thousand Oaks, CA. [Google Scholar]
- Spurr G.B., Barac‐Nieto M. & Maksud M.G. (1977) Productivity and maximal oxygen consumption in sugar cane cutters. The American Journal of Clinical Nutrition 30, 316–321. Available from: PM:842483. [DOI] [PubMed] [Google Scholar]
- Subramanian S.V., Lochner K.A. & Kawachi I. (2003) Neighborhood differences in social capital: a compositional artifact or a contextual construct? Health & Place 9, 33–44. Available from: PM:12609471. [DOI] [PubMed] [Google Scholar]
- Uthman O.A. (2009) A multilevel analysis of individual and community effect on chronic childhood malnutrition in rural Nigeria. Journal of Tropical Pediatrics 55, 109–115. Available from: PM:18845589. [DOI] [PubMed] [Google Scholar]
- Van de Poel E., Hosseinpoor A.R., Jehu‐Appiah C., Vega J. & Speybroeck N. (2007) Malnutrition and the disproportional burden on the poor: the case of Ghana. International Journal for Equity in Health 6, 21. Available from: PM:18045499. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Wamani H., Astrom A.N., Peterson S., Tumwine J.K. & Tylleskar T. (2006) Predictors of poor anthropometric status among children under 2 years of age in rural Uganda. Public Health Nutrition 9, 320–326. Available from: PM:16684383. [DOI] [PubMed] [Google Scholar]
- Wamani H., Astrom A.N., Peterson S., Tumwine J.K. & Tylleskar T. (2007) Boys are more stunted than girls in sub‐Saharan Africa: a meta‐analysis of 16 demographic and health surveys. BMC Pediatrics 7, 17. Available from: PM:17425787. [DOI] [PMC free article] [PubMed] [Google Scholar]
- World Health Organisation (1995) Physical status: the use and interpretation of anthropometry Report of a WHO Expert Committee. WHO Technical report series No.854, Geneva. [PubMed]
- World Health Organisation (WHO) Multicentre Growth Reference Study Group (2006) WHO Child Growth Standards based on length/height, weight and age. Acta Paediatrica Supplement 450, 76–85. Available from: PM:16817681. [DOI] [PubMed] [Google Scholar]