The objective of this study is to identify the ways in which poverty could affect the nutritional health of the child and to analyze the strength of these links. On the whole, it appears that the relationship between poverty (measured by the wealth index) and health of the child (measured by an anthropometric index) is positive and highly significant.
The relationship between health and poverty is often described in the literature as very complex, still poorly identified and reciprocal. In the particular case that interests us, this relationship has a dual aspect [
To characterize the health status of children, previous research used anthropometric data (size, children’s weight); because these data are simple, accurate, and have been the subject of a consensus for estimating the malnutrition of children. In this anthropometric approach, we have focused on the delayed growth factor as measured by the long term indicator of height for age. With regard to poverty, a non-monetary approach is used: a wealth index is built using durable goods owned by households. The indicator thus obtained is considered as a proxy for long-term household income.
In a study of the determinants of the health of the child, the empirical work shows that there is a positive link between household resources and the nutritional status of the child. However, the intensity of this connection may vary from one country to another: for example, it is strong in the case of Benin [
This text is set forth as follows: The following chapter presents the origin of the data; it also describes in detail the variables of the study. In the third chapter, we specify the model chosen to study the relationship between poverty and nutritional health of the child. The fourth chapter provides the results of the econometric estimates.
This work is based on the Demographic and Health Survey of Congo carried out in 2005 by the National Centre for Statistics and Economic Studies with technical assistance from ORC Macro, an American cooperating institution in support of this type of investigation [
This being said, during the study, all children under five years present in the household selected, had to be weighed and measured. Thus, the results on the nutritional status are based on a sample of 4472 children.
We begin by first defining the nutritional status of the child, then by identifying the main determinants of this state grouped into three categories.
There is no single indicator of the nutritional status per se. We generally use approximate measures which provide information on the nutritional status through their involvement in various processes or physiological functions [
In our study, we have focused on the height for age index1. This index is considered by specialists [
Among the variables specific to the child we have retained a number (depending on the availability of data) that
Socio-demographic Characteristics | Height by Age | Number of children | Socio-demographic Characteristics | Height by age | Number of children | ||
---|---|---|---|---|---|---|---|
Percentage of <−3SD | Percentage of <−2SD | Percentage of <−3SD | Percentage of <−2SD | ||||
Children Age <6 6 - 9 10 - 11 12 - 23 24 - 35 36 - 47 48 - 59 | 1.5 3.0 3.2 16.1 11.0 13.9 11.5 | 4.7 10.2 17.4 34.3 28.1 30.1 30.9 | 470 359 136 858 959 910 780 | Weight at birth Very small Small Normal o big Not defined | 24.1 14.4 9.2 17.9 | 40.8 35.7 23.6 34.4 | 85 250 3477 183 |
Sexe Masculine Feminine | 12.3 9.3 | 27.6 24.3 | 2279 2194 | Place of residence Urban Rural | 9.2 12.1 | 22.1 29.2 | 2045 2427 |
Order of birth 1 2 - 3 4 - 5 6+ | 11.9 8.7 9.6 13.5 | 27.1 23.2 25.2 28.2 | 950 1627 851 581 | Region Brazzaville Pointe Noire South North | 11.7 5.0 11.5 11.9 | 23.7 19.2 27.6 30.3 | 1218 592 1723 940 |
Mother’s Age 15 - 19 20 - 24 25 - 29 30 - 34 35 - 49 | 10.2 13.2 9.8 8.3 11.5 | 26.0 28.5 24.4 22.4 27.4 | 1051 1176 914 689 643 | Mothers’ level of education None Primary Secondary 1st cycle Secondary 2d cycle or above | 19.4 12.2 7.7 5.6 | 33.9 29.3 21.2 22.0 | 385 1502 2193 66 |
Period between births (in months) First born <24 24 - 47 48+ | 12.0 13.0 11.5 6.3 | 27.2 30.1 27.0 19.6 | 960 388 1546 1115 | Quintiles of economic wellbeing Poorest Second Q Medium Fourth Q richest | 13.5 9.9 11.3 10.8 7.3 | 31.9 27.2 24.6 23.7 19.7 | 1041 1034 927 780 690 |
Total | 10.8 | 26.0 | 4472 |
we propose to define2.
1) The age and gender of the child
For a child of a given gender, age is an important determinant of the individual growth. As age increases, the nutritional status of children in the developing countries deteriorates because of the cumulative effects of the lack of nutritional intake [
Recent research [
2) The birth rank
Some studies take into account the order of births as an explanatory factor of child malnutrition [
3) The interval of births
A short interval between births can cause a physiological impairment of the mother, such that the child may have a delay in weight and size at birth. The more closely spaced the births, the lower the breast milk quality, particularly under the effect of physical exhaustion of the mother. It is obvious that mothers who must raise two children at the same time give them less care. Accordingly, it is therefore expected that the interval of birth can have a significant impact on the levels of malnutrition.
4) The presence of a twin
It has been found that the absence of a twin significantly improves size. We explain this biological fact in the following manner: each twin often suffers at birth from a handicap that must be compensated for by food and appropriate care [
5) The number of children in the household
We want to know if the integration of the child with its siblings has an impact on its growth. To this effect, two variables are tested: the number of children under five years of age and the number of children in the household3. A priori, the expected effect of an increase in these variables would be to deteriorate the child’s health. The presence of a high number of children under five years of age increases the mother’s load in terms of care and therefore should have a negative impact. However, one could imagine, in the case of the variable “number of children in the household” that the older children can take care of the younger children when they are not yet working, providing resources for parents if they work [
1) The mother’s age at the first birth
The expected effect of the mother’s age at the first birth upon the size of the child and on its probability of having a normal growth is ambiguous. From a biological point of view, one would assume that the physical conditions of a young mother are better than those of an older mother. In these conditions, a positive relationship is suspected. If age is considered as a variable approached from the accumulation of experience in terms of care, one might think that a mother who is too young is probably less mature and less experienced. In this case, one can expect a negative relationship4.
2) The mother’s health
Among the variables characterising the state of the mother’s health, certain previous work continues [
3) The mother’s family situation
Three situations are distinguished: the monogamous family, the polygamous family and the single parent family (the mother lives alone). Concerning the polygamous family, one might suspect a negative impact of this variable on the nutritional status of the child. In effect, one might think that polygamous fathers would have more of a burden than others, meaning many more children and adults to feed. One could also imagine a woman living alone has fewer resources. In this case, one might expect a negative impact on the growth of the child. However, in the study already cited, Morrison and Liskens [
4) The parents’ education
There is extensive literature on the positive role of parents’ education of on the health of children. With respect to the work of Schult [
In this study, we chose the mother’s education, because many studies have shown that the number of years of education of the husband/spouse had little effect on the health of the child. Moreover in Africa, it is the mother who has the primary responsibility for childcare. In fact, the most decisive aspect for the mother is that she knows how to read and write. If this is the case, there would be no correlation between the child’s health and the mother’s educational level. This hypothesis will therefore be tested. Its non-acceptance would mean that an educated woman would develop greater capability for childcare in particular if she has training in nutrition. In this context, we have introduced a variable of the mother’s access to information (access to at least one media). This variable enables checking the likely knowledge of the mother in the area of nutrition and childcare.
5) Household income
Income is the central variable (or variable of interest) of our study. This is one of the most significant variables of child health [
The effects of the environment (or the community) on the child’s health are well documented in the theoretical literature. The famous model of Mosley-Chen [
Two functional forms are often used. In the first, the probability for a child of having a risk to growth is described by a logistic model:
child
The linear model can be estimated by ordinary least squares. However, this regression suffers from a few statistical problems likely to create bias in the estimates. In fact, in our sample, only the children alive at the time of the survey could have been measured: there is therefore a selection bias to the extent that it can be assumed that there is no total independence between the fact of being alive and health status [
where
As can be seen, one is confronted with a system of simultaneous equations, in which one of them can only be estimated on a sub-sample depending on a system determined by the other. By involving a fully parametric characterisation of the system, assuming the joint normality of error terms of the two equations, the model can be estimated by the maximum-likelihood method [
It should first be noted that we used the Nakamura test, to check the endogeneity of some variables. The lack of necessary instruments has led us to consider some of them as exogenous. This is, for example, the case of the variables related to the household composition. The endogenous nature of other variables has been taken into account by applying the two stage regression procedure. Moreover, the problem of heteroscedasticity has been resolved using White’s correction. Finally, the Heckit procedure in Stata software has allowed us to reject the hypothesis of a selection bias in our sample. In the analysis of the results, it can be seen that the explanatory powers of the models measured by
If one takes into account the breakdown of the index by gender, and for significant values (at the threshold of 1%), particularly for the poorest class, one sees a substantial gap between the coefficient of the girls and that of the boys. In effect, an increase of one unit of the wealth index deteriorates the height for age of 0.56 units for boys and by 0.44 units among girls.
All | Boys | Girls | ||||
---|---|---|---|---|---|---|
Coeff. | t-stat | Coeff. | t-stat | Coeff. | t-stat | |
Wealth index | 0.6345185 | 6.06*** | 0.6270308 | 4.43*** | 0.6472413 | 4.15*** |
Constant | −0.9181392 | 33.53*** | −9,803,026 | −25.68*** | −0.8517949 | −21.67*** |
N = 3824 | N = 1 973 | N = 1851 | ||||
F(1, 3822) = 36.69 | F(1, 1971) = 19.62 | F(1, 1849) = 17.23 | ||||
R2 = 0.0089 | R2 = 0.0089 | R2 = 0.0091 |
***, **, * represent significant coefficients at the 1%, 5% and 10% respectively.
All | Boys | Girls | ||||
---|---|---|---|---|---|---|
Coeff. | t-stat | Coeff. | t-stat | Coeff. | t-stat | |
Poorest | −0.5036724 | −4.91*** | −0.5622544 | −3.98*** | −0.4381149 | −2.93*** |
Poor | −0.2899294 | −2.83*** | −0.4450123 | −3.16*** | −0.1259365 | −0.84 |
Average | −0.2983505 | −2.53** | −0.4338292 | −2.71*** | −0.1463372 | −0.84 |
Rich | −0.0588918 | −0.52 | −0.2567768 | −1.66* | 0.1481544 | 0.88 |
Constant | −0.6220886 | −6.90*** | −0.5676048 | −4.62*** | −0.6831544 | −5.15*** |
N = 3824 F(4, 3819) = 10.04 R2 = 0.0101 | N = 1973 F(4, 1968) = 4.72 R2 = 0.0088 | N = 1851 F(4, 1846) = 6.65 R2 = 0.0142 |
***, **, * represent significant coefficients at the 1%, 5% and 10% respectively.
In what will follow, we want to assess the net effect of poverty on the child’s nutritional health. To do this, we introduced other known control variables. These are in fact the characteristics of the child, the parents, or of the household, community and the environment that we have previously specified. Three regression models are proposed (
The results obtained for the age are consistent with the literature. We found the coefficients significant at the 1% threshold: negative for the age and positive for age squared. As regards the gender, one obtains significant negative coefficients at the threshold of 1%. This result thus confirms the results of Svedberg on the absence of bias against girls.
We found that the presence of a twin significantly deteriorates the size. This result is consistent with those of previous works.
An interval between the birth of the child studied and that of the previous child, has a significant and positive effect on its size.
Our study shows that the birth rank has no impact on the child’s size. Finally, we note that the integration of the child with its siblings has no impact on its growth.
With respect to the characteristics of the mother, the mother’s state of health represented by body mass index, has a significant negative impact on the child’s growth index. This could reflect the fact that household food security is not guaranteed and deteriorates accordingly the child’s nutritional status.
Virtually all the works confirmed the role of the mother’s education. This is also the case in our research where this variable has a very significant positive effect on the child’s health. An increase of one year in the mother’s years of education increases the growth score, all other things being equal, by 0.04. Moreover, it must be emphasized that the fact that the mother knows how to read has no effect on the child’s nutritional status: the coefficient of this variable is negative and non-significant (regression model 2). One can interpret this result as
Model 1 | Model 2 | Model 3 | ||||
---|---|---|---|---|---|---|
Coeff. | t-stat | Coeff. | t-stat | Coeff. | t-stat | |
Characteristics of children | ||||||
Child’s age in months | −0.0603463 | −10.02*** | −0.0602642 | −10.01*** | −0.0603277 | −100.01*** |
Square of the child’s age in months | 0.0006875 | 6.80*** | 0.0006859 | 6.79*** | 0.000689 | 6.81*** |
Child’s gender | −0.1347898 | −2.56** | −0.1355429 | −2.57** | −0.1314197 | −2.49** |
Child’s birth rank | −0.0169985 | −0.25 | −0.0165796 | −0.24 | −0.013153 | −0.19 |
Interval in months separating the child considered from the next oldest child | 0.004184 | 3.72*** | 0.0041986 | 3.73*** | 0.0041768 | 3.70*** |
The presence of a twin | −0.8650292 | −5.23*** | −0.8665762 | −5.23*** | −0.8491915 | −5.12*** |
The number of children in the household | 0.0004628 | 0.03 | 0.0004461 | 0.03 | −0.0003524 | −0.02 |
The number of children under age 5 in the household | −0.0328394 | −0.65 | −0.0324841 | −0.65 | −0.0315158 | −0.63 |
Characteristics of the household | ||||||
Mother’s education (years of studies) | 0.0361313 | 3.98*** | 0.0344436 | 3.63*** | 0.0362409 | 3.99*** |
The wife has access to at least one media | 0.033885 | 0.45 | 0.0174828 | 0.23 | ||
The wife knows how to read | −0.0258781 | −0.35 | ||||
The mother’s age at the first birth | 0.0131173 | 1.09 | 0.013266 | 1.11 | 0.0136326 | 1.13 |
The mother’s age (at the time of the birth) | 0.0030219 | 0.31 | 0.0027995 | 0.29 | 0.0030985 | 0.32 |
The mother’s health status (body mass index) | −0.0043082 | −2.53** | −0.0043145 | −2.54** | −0.0041726 | −2.46** |
Family situation (single parent family) | −0.0545196 | −0.75 | −0.0547773 | −0.75 | −0.0486887 | −0.67 |
Family situation (polygamous family) | −0.0924776 | −1.15 | −0.0918426 | −1.14 | −0.0917358 | −1.13 |
Wealth index (household income) | 0.4357158 | 2.86*** | 0.4432275 | 2.91*** | ||
Socio-economic classification | ||||||
Poorest | −0.2930593 | −2.19** | ||||
Poor | −0.1416679 | −1.12 | ||||
Average | −0.1607942 | −1.26 | ||||
Rich | 0.0144374 | 0.12 | ||||
Characteristics of environment/community | ||||||
The household has access to drinking water | −0.0776559 | −1.05 | −0.0768514 | −1.04 | −0.070348 | −0.95 |
The household has electricity | −0.0828471 | −1.06 | −0.0813175 | −1.05 | −0.0594063 | −0.76 |
Type of toilet in the household | 0.1608071 | 2.10** | 0.161577 | 2.10** | 0.1723039 | 2.26** |
Type of floor in the household | 0.0404869 | 0.54 | 0.0394338 | 0.53 | 0.0284875 | 0.38 |
- Place of residence | ||||||
Rural | −0.269182 | −2.88*** | −0.2713925 | −2.91*** | −0.2618085 | −2.81*** |
- Region of residence | ||||||
Brazzaville | −0.1898967 | −2.33** | −0.188974 | −2.32** | −0.2012345 | −2.47** |
South | 0.2215025 | 2.17** | 0.2223461 | 2.17** | 0.2082142 | 0.04** |
North | 0.1044212 | 0.97 | 0.1065899 | 0.99 | 0.0796651 | 0.74 |
Constant | −0.0858426 | −0.29 | −0.0380752 | −0.13 | 0.0489823 | 0.15 |
N = 3824 | N = 3824 | N = 3824 | ||||
F(24, 3799) = 17.48 | F(24, 3799) = 17.45 | F(27, 3796) = 15.61 | ||||
R2 = 0.0948 | R2 = 0.0949 | R2 = 0.0953 |
***, **, * represent significant coefficients at the 1%, 5% and 10% respectively.
follows: “knowing how to read and write in a language without attending a formal school is not enough to put into practise the lessons received in the area of nutrition and child care”.
By including the variable “access to at least one media”, we wanted to confirm the argument of Thomas, Strauss and Henriques [
We found that the family status of women does not intervene in the child’s development. The fact that the woman lives alone, or in a polygamous union, has no impact on the child’s nutritional status.
The wealth index always appears as a major determinant of the child’s nutritional status. Its coefficient remains high and very significant. As of now, an increase in the household wealth of 10%, would reduce chronic malnutrition by 4.3%. This result is close to that obtained by [
Some studies have found a reasonably solid connection between access to water and electricity and the nutritional status of the child. In contrast to these studies, in the sample from the Demographic and Health Survey of Congo, these two variables are not significant factors in the child’s growth. Most of the coefficients on these two variables are negative, that is to say they do not have the expected sign. It should be noted that access to safe drinking water and electricity depends in general on the state through their national companies for distribution of water and electricity. One can therefore be connected to these distribution networks and not have water or electricity during a good period of the year.
With regard to the two variables on the housing, characterising the household living conditions, only the provision of a modern toilet has a positive and significant impact on the child’s growth. The coefficient of the variable type of floor (cement), although having the expected signs is not significant at the thresholds selected.
We found that the children in rural areas are disadvantaged in size by 0.3 SD compared to those in an urban environment.
Finally, we note the existence of a regional dimension of malnutrition. If you take the city of Pointe-Noire (economic capital of the country) as a reference, we found that the children living in the southern part of the country are favoured. An opposite effect is observed in Brazzaville (political capital of the country) whose children are disadvantaged by 0.2 SD compared to those in Pointe-Noire.
The objective of this study was to explore the relationship between poverty and the nutritional health of the child based on the data from the first Demographic and health survey of Congo carried out in 2005. Beyond this relationship, we wanted to analyze the determinants of the child’s nutritional status and to base it on the specific characteristics of the child, the household and the parents, and on the characteristics of the environment or the community. To this effect, regression models have been proposed and have shown that several variables had a significant impact on the child’s nutritional status. The public authorities could therefore take advantage of these results to, in particular, combat malnutrition of children and, in general, fight against poverty. From this study, and in view of the econometric tests performed, we could draw the main conclusions as follows:
・ firstly, the study has shown that an increase of the wealth index of households very significantly improved the nutrition of children; that this strong improvement will benefit girls a little more than boys. In the light of the breakdown of this index, it could be suggested that a public policy of transfer, that is to say, which would alter the distribution of income in favour of the poorest (the poorest quintile) could be most effective;
・ secondly, the results of the study suggest reflections focused on the health and education of women. First, the study has highlighted the negative role of the mother’s health (through the body mass index) on the child’s growth, testifying to the importance of household food insecurity. We know that the repetition of closely spaced births weakens the mother. Increasing the gaps between the births would therefore be a fundamental element for improving the health of the child and the dissemination of contraception means would be a key objective in this case. Then, the study has confirmed the role of maternal education in reducing child malnutrition because an educated mother better understands the teachings on child nutrition. The public authorities could therefore improve the living conditions of children by adopting a well-targeted policy concerning certain expenditures for health and education; finally, the study revealed a regional dimension of malnutrition. A policy of intervention aimed at improving the conditions of community life in rural areas and Brazzaville would be desirable.
SamuelAmbapour,Jean ChristopheOkandza,Hylod ArmelMoussana, (2015) Poverty and Nutritional Health of the Child: Some Evidence from 2005 Demographic and Health Survey of Congo. Health,07,1466-1476. doi: 10.4236/health.2015.711161