Dietary Diversity and Child Malnutrition in Ghana: Raymond Boadi Frempong, Samuel Kobina Annim

Received:
28 August 2016
Revised:
Dietary diversity and child
3 February 2017
Accepted:
21 April 2017
malnutrition in Ghana
Heliyon 3 (2017) e00298
Raymond Boadi Frempong a, * , Samuel Kobina Annim b

a
Bayreuth International Graduate School of African Studies, University of Bayreuth, Germany
b
Department of Economics, University of Cape Coast, Ghana
* Corresponding author at: Universität Bayreuth, Raum 1.0 02 146, RW I, Universitätsstraße 30, 95440 Bayreuth,
(Germany).
E-mail address: [email protected] (R.B. Frempong).
Abstract
The health of children in Ghana has improved in recent years. However, the current
prevalence rates of malnutrition remain above internationally acceptable levels.
This study, therefore, revisits the determinants of child health by using Ghana’s
Multiple Indicator Cluster Survey to investigate the effect of infant feeding
practices on child health. We used the World Health Organization’s Infant and
Young Children Feeding guidelines to measure dietary quality. The econometric
analyses show that dietary diversity may cause improvement in children’s health in
Ghana. This suggests that educational campaigns on proper infant feeding and
complementary dieting could be an effective means of improving the health of
children in Ghana.
Keywords: Medicine, Pediatrics, Public health
1. Introduction
The importance of nutrition has been emphasized for the intellectual and physical
development of children. It is, therefore, important to ensure that children have
adequate diet as it will ensure a healthy and productive population in future. The
benefits of adequate nutrition to the economy come directly in the form of reduced
public health expenditure on health care and indirectly through improved health,
which may lead to economic growth (Hoddinott et al., 2008). Because of the
importance of nutrition in human and economic development, pragmatic steps,
http://dx.doi.org/10.1016/j.heliyon.2017.e00298
2405-8440/© 2017 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
Article No~e00298
both at sub-national, national and international levels, have been taken to reduce
the incidence of malnutrition. The importance development practitioners attach to
nutrition is reflected in targets 1C and 4A of the first Millennium Development
Goal’s (MDG), which identify the need to reduce hunger and mortality among
children under five years of age. In spite of the various interventions, about 870
million people were malnourished globally between 2010 and 2012 (FAO, 2012),
with a sizeable proportion of this is found in sub-Saharan Africa and Asia
(UNICEF, WHO & World Bank, 2015). Also, women and children tend to be the
most affected (Bain et al., 2013). According to UNICEF, WHO & World Bank’s
(2016) joint child malnutrition estimates, about 32 percent of children in Africa are
stunted whilst another 8 percent are wasting. Malnutrition at the early stages of life
does not only affect health outcomes of the child, but it also has a serious adverse
impact on the determinants of their livelihoods, such as physical and intellectual
growth, school performance and eventual future earnings and productivity
(Hoddinott et al., 2008; Strauss and Thomas, 1998). As individuals get locked
up in a vicious cycle of poor health, lower learning capacity, reduced physical
activity, and lower productivity as a result of poverty, malnutrition and poor child
health nexus, the gains from previous economic growth may be threatened
(Bagriansky, 2010).
Some countries in Africa, including Ghana, have made marginal, but steady, gains
in reducing malnutrition among children less than five years (ICF Macro, 2010).
However, according to the last Multiple Indicator Cluster Survey (MICS4), 13
percent of children in Ghana are moderately or severely underweight, 23 percent
are stunted (too short for their age.), and 6 percent are wasting (too thin for their
height.) (Ghana Statistical Service (GSS), 2011). Fig. 1 presents the nutritional
status of children according to the Ghana Demographic Health Survey (GDHS,
[(Fig._1)TD$IG]
Fig. 1. Nutritional Status of Children under Five Years in Ghana from the GDHS I–V and Ghana
MICS-4. (Source: ICF Macro (2010) and GSS (2011)).
2 http://dx.doi.org/10.1016/j.heliyon.2017.e00298
Article No~e00298
rounds I −V) and MICS4. The graph shows varying trends for overweight,
stunting, underweight and wasting among children aged 5 to 59 months in Ghana.
The graph shows that child stunting and underweight have reduced since 2003,
even though the proportion of children who are stunted in Ghana is still higher than
the global average of 25 percent (UNICEF, 2013). On the other hand, child wasting
and overweight rose between 2003 and 2008. Both indicators increased by 1
percent between 2003 and 2008. Similarly, the GDHS estimate of child wasting is
higher than the global estimate of five (5) percent (UNICEF, 2013). Whilst these
figures are encouraging when they are benchmarked against other African
countries (UNICEF, WHO & World Bank, 2016), they are still above the World
Health Organization’s (WHO) classification of low prevalence. Thus, notwith-
standing this progress, there is the need to investigate the drivers behind
malnutrition in Ghana. This has become necessary given the fact that there has not
been a steady downward trend for some indicators.
Following UNICEF’s (1990) conceptual framework on the causes of malnutrition,

researchers have attempted to explain the role of quality diet in reducing child
malnutrition. The empirical literature is replete with the assessment of
complementary feeding and its effect on children’s health and nutrition (Dewey
and Adu-Afarwuah, 2008; Saha et al., 2008 and Vaahtera et al., 2001). The results
from these studies suggest that proper complementary feeding could be effective in
improving the health of children below the age of five years. In line with these
findings, this paper studies the effect of dietary diversity as means of improving
child health in Ghana.
Whilst UNICEF’s (1990) conceptual framework links food security to child

nutrition, the relationship is not as simple as demonstrated by the framework.
Given the fact that food security as a concept is multifaceted, the transmission
mechanism of food security to better nutrition can either run from food availability
in terms of quantity and quality or sustainability. Thus, policy intervention aimed
at improving nutrition through food security may have different impacts, based on
which aspect is targeted. In addition, the relationship between food security and
child nutrition could be more complex, based on the level of the analysis and the
interactions of social, environmental and cultural practices. Government policy can
improve child nutrition by ensuring food security at the household or at the
individual (child and mother) level. However, depending on which level is
targeted, the transmission mechanism will differ and the desired impact may be
achieved at different time intervals. In view of these complexities, the current study
seeks to test the hypothesis that the consumption of a diversified diet leads to better
child nutrition. To the best of our knowledge, this study is different from other
studies conducted on the subject to the extent that we take care of the endogeneity
between dietary diversity and child health outcomes.
Article No~e00298
The next section of the paper discusses the methodology, then we present results
and discussion and finally draw conclusions from the study.
2. Methodology
2.1. Data
Data for the study was obtained from the fourth round of the Ghana Multiple
Indicator Cluster Survey (MICS4) collected in 2011. The MICS4 was conducted
by the GSS as part of a broader international household survey designed by the
UNICEF. The aim of the survey is to provide current information on the socio-
economic circumstances of children and women by measuring key indicators
relating to the MDGS (GSS, 2011). MICS4 is, therefore, a cross-sectional survey
that provides current information on the health, social and economic circumstances
of women, children and other household members. To obtain the sample for the
survey, rural and urban areas in the ten administrative regions of Ghana were used
as the main sampling strata. The final sample was selected in two stages. In the first
stage, enumeration areas were selected with the probability of selection being
proportional to the size. Then 15 households in each enumeration area were
systematically selected. Eventually, 12,150 households were successfully sampled
and interviewed. This consists of 11,925 households, 10,627 women aged 15 to 49
year and 7550 under the age of five years (GSS, 2011). After cleaning a merging
the various data files, we 6598 children remained for the regression analysis.
2.2. Measurement of dietary diversity

To measure dietary quality, we adopt the WHO’s Infant and Young Children
Feeding guidelines (IYCF), because they are designed to measure dietary diversity
for both breastfed and non-breastfed children. Also, the dataset used in the analysis
contains information on the food items that can be used to calculate this indicator.
The section of the survey on children collected information on food and liquids a
child consumed in the previous day. We categorized these food items into seven
major food groups based on the WHO’s IYCF guidelines (Swindale and Bilinsky,
2006). These food groups are: (i) grains, roots, and tubers; (ii) legumes and nuts;
(iii) flesh foods (meat, fish, poultry and liver/organ meats); (iv) eggs; (v) vitamin A
rich fruits and vegetables; (vi) dairy products (milk, yogurt, cheese); (vii) other
fruits and vegetables. If a child consumed at least one food item from a food group,
the group was assigned a value of one for that child. The group scores are then
summed to obtain the dietary diversity score, which ranges from zero to seven,
where zero represents non-consumption of any of the food items and seven
represents the highest level of diet diversification.
Article No~e00298
2.3. Definition of nutrition status (anthropometric indicators)

In this study, the child’s nutritional status is measured by the WHO’s (2006)
anthropometric indicators: height-for-age (HAZ) is used to measure chronic
malnutrition due to prolonged food deprivation; weight-for-height (WHZ) captures
undernutrition due to recent food deprivation and malnutrition. Weight-for-age
which is measures the child’s body mass relative to her chronologic age is used as a
proxy for underweight.
2.4. Model specification and estimation technique

Based on the reviewed literature and the objective of this study we estimate the
following model:
yij ¼ β0 þ β1 sexi þ β2 agei þ β3 f everi þ β4 numchildren þ β5 urbani

þ β6 magei þ β7 medui þ β8 wealthi þ β9 wateri þ β10 toileti þ β11 f oodi
þ β12 Bmilk þ β13 ethnici þ β14 regioni þ εi (1)
where: sex = sex of child; age = age of child; numchildren = number of children in
household; urban = whether household is in an urban area; mage = mother’s age;
medu = mother’s education; wealth = wealth quintile; water = source of water;
toilet = type of toilet; food = food diversity score; Bmilk = child ever breastfed;
ethnic = ethnicity of household head; region = which administrative region
household is located in;yj ¼ ðWAZ; HAZ; WHZ Þ are the raw z-score of the
respective indicator; and i indexes individual children.
Estimating Eq. (1) with ordinary least squares (OLS) can yield a consistent
estimate of the effect of dietary diversity on child health if dietary diversity is
exogenous; that is uncorrelated with the error term in the model. However, this
may not be the case because of the bi-causal relationship between the two
variables. This bi-causal relationship is likely to arise because parents and
caregivers may adopt different feeding practices depending on the current health
status of the child. In this instance, the direction of causality could also run from
child health to dietary diversity. If this is the case, the OLS estimates will no longer
be consistent. To resolve this problem, we employed the Two Stage Least Square
(TSLS) estimator to identify the causal effect of dietary diversity on child health.
We used the number of chicken owned as well as household ownership of pigs as
instruments for dietary diversity. We assume our instruments have no direct effect
on child health outcomes but they indirectly influence it by improving the quality
of diet consumed. Based on these considerations we estimate the following
equations:
f ood i ¼ α0 þ α1 sexi þ α2 agei þ α3 f everi þ α4 numchildreni þ α5 urbani

þ α6 magei þ α7 medui þ α8 wealthi þ α9 wateri þ α10 toileti
þ α11 Bmilki þ α12 ethnici þ α13 regioni þ α14 pig þ α15 chickeni þ ψ i (2)
Article No~e00298
yij ¼ ϕ0 þ ϕ1 sexi þ ϕ2 agei þ ϕ3 f everi þ ϕ4 numchildreni þ ϕ5 urbani

þ ϕ6 magei þ ϕ7 medui þ ϕ8 wealthi þ ϕ9 wateri þ ϕ10 toileti þ ϕ11 f d
oodi
þ ϕ12 Bmilk þ ϕ13 ethnici þ ϕ14 regioni þ εi (3)
where f d
oodi is the linearly predicted food diversity score from Eq. (2); pig and
chicken are the number of pigs and chicken owner by the household.
3. Results and discussion

Table 1 presents the bivariate relationship between children’s nutritional outcomes,
on the one hand, and individual and household characteristics, on the other hand.
We find that approximately 25 percent of male children are stunted as compared to
20 percent of females. Similarly, 7.6 percent and 15.6 percent of male children are
wasted and considered underweight respectively, whilst 5.2 and 11 percent of
female children are wasted and underweight. In all cases, the incidence of
malnutrition tends to be higher among male children than female children. This
situation could be attributed to differences in biological composition and
caregiving; and possibly due to daughter preference (Fuse, 2010).
One can also observe that the difference between the incidence of malnutrition
over the different age groups is statistically significant. It is important to note that
in all the three cases the proportion of malnourished children increases with age up
to age 35 months and then falls, thus, suggesting a non-linear relationship between
child age and nutritional status. Apart from stunting, we observe significant
differences in the prevalence of wasting and underweight among children who had
suffered from fever in the past two weeks and those who had not.
On the association between dietary diversity and child health, Table 1 shows that
malnourishment varies significantly for the different levels of consumption. About
a quarter of children who consumed at least four of the seven food items are
stunted: 5 percent of this group were wasted and another 12 percent were
underweight. The prevalence of wasting and child underweight begins to decline
after the intake of two or more food items. Thus, child health improves as the
intensity of dietary diversity increases.
Sanitation factors such as the type of toilet facility and the source of drinking water
also tend to have a significant association with child health in Ghana. About 30
percent of children found in households where the bush or bucket are used as toilet
facility is stunted, whilst only 10 percent of their counterparts in households with
flush toilets are stunted. Similarly, only 4 percent of children found in households
with flush toilets are wasted as compared to 7 percent of children found in
households which neither use flush toilet nor pit latrine. The incidence of
malnutrition decreases as the source of drinking water improves. For instance, the
prevalence of child underweight falls from 17 percent among those who use
Article No~e00298
Table 1. Bivariate analysis of socioeconomic variables and anthropometric

indicators.
Stunted Wasted Underweight

% No. of % No. of % No. of
Children Children Children
Sex
Female 20 699 5.2 182 11.1 389
Male 25.1 852 7.6 261 15.6 536
Chi square 24.83 (Pr = 0.000) 27.78 (Pr = 0.000) 15.78 (Pr = 0.000)
Age
0–5 7.8 62 11.3 89 6.6 52
6–11 12.5 88 11.3 80 15.9 113
12–23 26.5 365 7.9 108 16.6 231
24–35 29.2 412 5 70 14 199
36–47 26.8 373 2.7 38 13.1 185
48–59 20.7 251 4.8 58 11.9 145
Child ill with fever in last 2 weeks

No 22.1 1,235 5.9 330 12.3 693
Yes 24.1 315 8.7 113 17.6 231
Food diversity Score

None 10.4 92 10.3 92 8.8 79
Any one 18.9 91 11.8 57 16.1 78
Any two 27.1 145 10 53 21.6 116
Any three 24.2 427 5 88 14.2 252
At least four 24.7 796 4.8 154 12.3 400
Main source of drinking water

Pipe 15.8 478 5.9 178 10.5 318
Protected 25.8 621 7.1 172 14.8 359
Unprotected 30.8 452 6.3 93 16.6 247
(Continued)
Article No~e00298
Table 1. (Continued)
Type of toilet facility

Flush 9.7 83 4.4 38 6.6 56
Pit latrine 21.8 921 6.6 278 12.5 532
Bush, bucket and other 30.2 547 7 128 18.3 337
Number of children
1–3 20.6 830 6 242 11.6 475
4–6 23.6 574 6.8 166 14.4 352
7–9 33.7 115 8.5 29 23.2 80
More than 9 39.8 32 6.8 6 23.8 19
Mother's education
None 28.4 620 8.1 177 16.9 372
Primary 25 383 5.3 82 13.7 212
Middle/JSS 18.5 437 5.5 130 11.1 263
Secondary + 13.7 111 6.9 55 9.6 78
3.
Region
Western 23 157 7.6 53 14.5 100
Central 22.2 148 5.6 38 13.3 90
Greater Accra 13.5 137 3.4 35 8 81
Volta 22.1 121 9.1 50 10.9 60
Eastern 20.7 159 7 54 10.8 83
Asante 22.2 289 6.6 86 11.7 154
Brong Ahafo 18.4 115 3.4 21 12.3 78
Northern 37.1 286 8.4 65 23.9 185
Upper East 30.9 93 7.5 23 20.3 62
Upper West 22.5 46 9.7 20 15.1 32
Wealth index quintile
(Continued)
Article No~e00298
Poorest 33.1 514 7.9 124 19.9 314

Second 26 367 7.3 103 14.3 203
Middle 22.3 318 6.4 90 13.4 190
Fourth 15.9 202 5 64 10.7 138
Richest 11.3 135 5 59 5.8 70
164.96 (Pr = 0.000) 16.91 (Pr = 0.002) 106.60 (Pr = 0.000)
Ethnicity of household head

Akan 20 564 5.6 159 11.7 332
Ga/Dangme 19.4 106 5.5 30 12.8 70
Ewe 18.7 171 6.6 61 9.8 90
Guan 27.4 67 5.4 13 18.8 46
Gruma 37.3 158 6 26 18.3 78
Mole-Dagbani 25.6 344 8.2 110 16.9 230
Grusi 31.4 87 7.3 20 17.6 50
Mande 15 18 7 8 7.5 9
Total 22.5 1,550 6.4 444 13.3 925
Source: Authors’ Computation from MICS4, 2011.
unprotected water sources to 11 percent among those who use pipe-borne water.
Good sanitation and water supply improve health by reducing infections and
malnutrition (Cuesta, 2007).
We also find that wealthier households have children with better nutritional status
than poorer households. About a third of children born to poor parents are stunted.
Equally, one-fifth of children born into poor households are underweight compared
to only 6 percent of those in the richest households. A similar pattern was observed
by Urke et al. (2011), who found that wealth status and maternal education are
positively associated with child health outcomes. In addition, we observe a positive
relationship between maternal education and child health outcomes. The effect of
dependence and competition for care is also felt through the number of children in
the household. Table 1 shows that child nutritional status worsens as the number of
children increases. This could be because of competition for care and resources in
the household.
Article No~e00298
[(Fig._2)TD$IG]
Fig. 2. Feeding Pattern in Ghana by child age groups. (Source: Authors’ Computation from MICS,
2011).
3.1. Feeding pattern in Ghana by child age groups

Fig. 2 provides a description of dietary diversity across different child age groups
and it shows that in Ghana a majority of children (47%) consume a minimum of
four out of the seven food groups. About a quarter of them had three food groups
whilst 13 percent had none of the food groups. In addition, more than 80 percent of
infants, aged zero to five months, can be assumed to be undergoing exclusive
breastfeeding since they consumed none of the food items. Observable from Fig. 2
is how food diversity increases with child age. For instance, whilst only 0.5 percent
of children who are under 6 months of age had at least four of the food items, the
proportions increased to 49 percent among those aged 12 to 23 months and 61
percent among those aged 48 to 59 months. The pattern portrayed by this graph
indicates that parents and caregivers see the need to vary their children’s diet
within the first five years.
3.2. Maternal education and food diversity

In Table 2 we test whether dietary diversity for children is statistically different
across levels of maternal education. The table shows that children born to parents
with higher levels of education had a more diversified diet. We found that children
whose mothers have primary, middle/Junior High School (JHS) or at least
secondary school education are significantly likely to consume more diversified
diet than those whose mothers have no education. However, we fail to find any
statistical difference in child feeding among those whose parents had received
some form of formal education. Thus, as far as child feeding is concerned the
feeding pattern of children who are born to parents with only primary education is
Article No~e00298
Table 2. Dietary diversity across levels of maternal education.
None Primary Middle/JHS Secondary Mean
None 0.00 3.00

***
Primary -0.22 0.00 3.18
Middle/JHS -0.27*** -0.04 0.00 3.22
Secondary sch. -0.32*** -0.10 -0.06 0.00 3.28
+
p < 0.1, *p < 0.05, p < 0.01,
**
p < 0.001. Source: Authors’ Computation from MICS, 2011.
***
not statistically different from those who are born to mothers with either middle/
JHS or secondary school education. This suggests that intensifying nutrition
education at basic/primary school level could also be an effective approach to
fighting malnutrition.
3.3. Regional distribution of food diversity among children

under 6 years
In terms of regional differences in dietary pattern, Fig. 3 shows that children in the
Volta region had the most diversified diet. This is followed by Greater Accra,
which is the national capital. One would have expected regions like Eastern and
Western to have better food diversity than Upper West and Upper East regions
since the poverty rates in the former are lower than the latter region; however, the
graph portrays a different picture. This could be as a result of the fact that most of
the households in the Upper East and West regions are peasant farmers who grow a
wide range of crops and, rear different domestic animals which serve to improve
dietary diversity. The average dietary diversity of children in the Northern,
[(Fig._3)TD$IG]
Fig. 3. Distribution of Food Diversity by Administrative Regions in Ghana. (Source: Authors’Compu-

tation from MICS, 2011).
Article No~e00298
Western and Eastern regions are less than the national average of 3.12. The
implication of this of this pattern on children’s nutrition is seen in Table 1, where
the regions with low dietary diversity are also the ones with a high incidence of
child malnutrition.
3.4. Estimation results

This paper set out to study the effects of dietary diversity on child health outcomes.
To this end, anthropometric indicators of child health were regressed on dietary
diversity, maternal education, ethnicity and other covariates of child health. The
OLS and TSLS result of the effect of dietary diversity on child anthropometric
indicators are presented in Table 3. The first stage results for the TSLS are reported
in column 7. Table 3 indicates that dietary diversity has a positive effect on child
health outcomes. This relationship is observed in all the models, except for the IV
model for weight-for-age. Thus, Increasing the number of food groups a child
consumes by one causes HAZ and WAZ to increase by 0.65 and 0.52 units
respectively. Thus, we find a consistent positive association between dietary
diversity and better child health outcomes.
On the effect of ethnicity on child health, we find in Table 3 that children from the
Ewe, Grusi, Guan and Mole/Dagbani ethnic groups have better anthropometric
scores than those from the Akan ethnic group. This result shows that ethnicity may
have an important influence on child health in Ghana. This may be attributed to
differences in feeding and post-partum practices across the different ethnic groups.
Elsewhere, Adedini et al. (2015) and Annim and Imai (2014) have found that child
health outcomes differ among children of different ethnic descent in Lao People’s
Democratic Republic.
We also observe that an increase in the number of children in the household

worsens the child's health. The table shows that an additional child in the
household is associated with a 0.2 lower WHZ and a 0.3 lower scores in both WAZ
and HAZ. Annim, Awusabo-Asare, Amo-Adjei (2013) argued that as the number
of children in a household increase, children may have to compete for both
household resources and caregiving. This may cause each child to receive sub-
optimal care and resources to achieve the desired nutritional status. In line with our
observation in Table 1, male children tend to have lower weight-for-age and
height-for-age than their female counterparts. The observed relationship between
sex and nutritional status could be because of biological differences as well as
socio-cultural differences. Socio-culturally, gender preferences, and preferential
treatment may account for the observed coefficient. Indeed, Fuse (2010) has shown
that daughter preference is slightly higher than son preference in Ghana.
In addition, children who had diarrhea have lower anthropometric scores compared
to those who did not. Infections that lead to diarrhea and fever adversely affect
Article No~e00298
Table 3. Effects of food diversity, mother's education and ethnicity on child health outcomes in Ghana (all
children under 5 years).
(1) (2) (3) (4) (5) (6) (7)

Weight-for-Height Height-for-Age Weight-for-Age First stage results
(OLS) (IV) (OLS) (IV) (OLS) (IV) IYCF
IYCF score 0.029** 0.179 0.028** 0.613** 0.036*** 0.518**

(0.012) (0.190) (0.013) (0.301) (0.011) (0.236)
* *** ***
Child is male -0.051 0.005 -0.109 -0.052 -0.076 -0.001 -0.078**
(0.027) (0.036) (0.030) (0.048) (0.025) (0.039) (0.035)
** *** ***
Child had diarrhea -0.082 -0.042 -0.130 -0.089 -0.132 -0.079 -0.043
(0.040) (0.047) (0.044) (0.061) (0.038) (0.051) (0.053)
Child had fever -0.186*** -0.219*** 0.095** 0.111** -0.079** -0.092** -0.014
(0.032) (0.038) (0.037) (0.050) (0.031) (0.041) (0.042)
* * *** ** *** ***
No. of child. in HH -0.011 -0.015 -0.027 -0.025 -0.025 -0.027 0.000
(0.007) (0.008) (0.008) (0.010) (0.007) (0.008) (0.008)
Urban residence 0.020 0.052 -0.041 -0.133* -0.012 -0.039 0.025
(0.043) (0.056) (0.049) (0.071) (0.041) (0.060) (0.059)
Mother’s age -0.002 -0.002 0.009*** 0.011*** 0.004** 0.006** -0.002
(0.002) (0.002) (0.002) (0.003) (0.002) (0.003) (0.003)
*** ** **
HH own agric. land -0.114 -0.103 -0.016 0.091 -0.080 -0.002 -0.147***
(0.034) (0.049) (0.038) (0.069) (0.032) (0.054) (0.045)
Child ever breastfed -0.305** -0.255 0.048 0.255 -0.189 -0.040 -0.011
(0.153) (0.206) (0.179) (0.272) (0.145) (0.203) (0.230)
Child Age (0–5 = 0)
6–11 -0.465*** -0.777** -0.444*** -1.427*** -0.546*** -1.369*** 1.707***

(0.073) (0.335) (0.076) (0.519) (0.065) (0.409) (0.080)
12–23 -0.294*** -0.828 -1.224*** -3.020*** -0.713*** -2.215*** 3.050***
(0.071) (0.584) (0.079) (0.921) (0.064) (0.722) (0.054)
*** *** *** ***
24–35 -0.050 -0.584 -1.433 -3.487 -0.710 -2.383 3.544***
(0.073) (0.678) (0.079) (1.066) (0.066) (0.838) (0.045)
36–47 0.075 -0.444 -1.354*** -3.478*** -0.635*** -2.340*** 3.555***
(0.071) (0.678) (0.078) (1.072) (0.065) (0.841) (0.045)
*** *** *** ***
48–59 -0.004 -0.499 -1.326 -3.455 -0.705 -2.400 3.563***
(0.072) (0.682) (0.079) (1.072) (0.065) (0.842) (0.046)
Mother’s Educ. (None = 0)
Primary 0.016 -0.004 0.021 0.024 0.024 -0.000 0.113**

(0.040) (0.052) (0.044) (0.071) (0.038) (0.058) (0.053)
(Continued)
Article No~e00298
(1) (2) (3) (4) (5) (6) (7)
Middle/JSS -0.034 -0.009 0.029 -0.064 -0.013 -0.058 0.120*

(0.046) (0.062) (0.053) (0.082) (0.043) (0.066) (0.062)
Secondary and above -0.123* -0.176 0.133* 0.057 -0.021 -0.125 0.247**
(0.070) (0.117) (0.079) (0.161) (0.065) (0.127) (0.117)
Wealth quintile (Poorest = 0)
Second 0.023 -0.003 0.123** 0.075 0.086** 0.039 0.139***

(0.043) (0.054) (0.048) (0.072) (0.040) (0.059) (0.051)
Middle -0.056 -0.145** 0.127** 0.036 0.045 -0.064 0.061
(0.057) (0.070) (0.063) (0.089) (0.053) (0.074) (0.076)
* *** *** ***
Fourth -0.076 -0.165 0.412 0.310 0.196 0.073 0.101
(0.067) (0.091) (0.072) (0.114) (0.063) (0.097) (0.098)
Fifth 0.065 0.029 0.493*** 0.375** 0.356*** 0.264* 0.145
(0.091) (0.147) (0.096) (0.187) (0.081) (0.156) (0.170)
Water source (Pipe = 0)
Protected -0.004 -0.026 -0.036 -0.052 -0.033 -0.055 -0.045

(0.046) (0.059) (0.053) (0.076) (0.044) (0.064) (0.065)
Unprotected 0.071 0.028 -0.089 -0.112 -0.004 -0.046 -0.046
(0.050) (0.062) (0.056) (0.080) (0.047) (0.067) (0.069)
Toilet type (Flush = 0)
Pit latrine -0.034 0.082 -0.168** -0.157 -0.113* -0.022 -0.091

(0.071) (0.131) (0.076) (0.181) (0.065) (0.146) (0.153)
Bush/bucket etc. -0.047 0.104 -0.144* -0.125 -0.108 0.018 -0.151
(0.079) (0.138) (0.086) (0.191) (0.074) (0.155) (0.159)
Ethnicity (Akan = 0)
Ga/Dangbe -0.020 -0.067 0.152* 0.077 0.078 0.028 0.058

(0.085) (0.121) (0.086) (0.135) (0.077) (0.120) (0.126)
Ewe -0.105 -0.205* 0.324*** 0.455*** 0.105 0.128 -0.106
(0.071) (0.107) (0.076) (0.120) (0.064) (0.104) (0.099)
** *
Guan 0.057 0.046 0.125 0.298 0.101 0.204 -0.041
(0.084) (0.108) (0.089) (0.126) (0.080) (0.112) (0.112)
Gruma 0.056 0.025 -0.012 0.070 0.028 0.069 -0.024
(Continued)
Article No~e00298
(1) (2) (3) (4) (5) (6) (7)
(0.077) (0.098) (0.080) (0.114) (0.071) (0.100) (0.098)

**
Mole-Dagbani -0.036 -0.016 0.106 0.224 0.032 0.134 -0.116
(0.063) (0.088) (0.067) (0.105) (0.058) (0.090) (0.084)
Grusi 0.097 0.154 -0.010 0.158 0.048 0.208* -0.152
(0.086) (0.109) (0.093) (0.132) (0.077) (0.109) (0.101)
* * *
Mande -0.063 0.054 0.210 0.332 0.086 0.274 -0.252**
(0.105) (0.131) (0.125) (0.172) (0.099) (0.141) (0.121)
Region (Western = 0)
Central -0.080 -0.234** 0.012 -0.192 -0.059 -0.291** 0.371***

(0.073) (0.119) (0.080) (0.164) (0.067) (0.132) (0.100)
Greater Accra -0.050 -0.218 0.017 -0.145 -0.024 -0.246 0.394**
(0.095) (0.179) (0.103) (0.230) (0.086) (0.192) (0.182)
** * * *
Volta -0.232 -0.317 0.199 -0.186 -0.050 -0.356 0.562***
(0.101) (0.168) (0.109) (0.224) (0.092) (0.183) (0.132)
** **
Eastern -0.037 -0.269 -0.030 -0.204 -0.051 -0.325 0.223*
(0.094) (0.131) (0.097) (0.164) (0.086) (0.138) (0.125)
Ashanti 0.021 -0.105 0.041 -0.237 0.044 -0.208 0.402***
(0.088) (0.145) (0.094) (0.186) (0.079) (0.152) (0.112)
Brong Ahafo 0.064 -0.112 0.143 -0.242 0.114 -0.262 0.597***
(0.086) (0.159) (0.095) (0.224) (0.079) (0.180) (0.113)
* ** *** *** *** ***
Northern -0.152 -0.278 -0.338 -0.398 -0.311 -0.440 -0.016
(0.084) (0.113) (0.091) (0.135) (0.078) (0.112) (0.110)
Upper East -0.266*** -0.529*** -0.184* -0.604** -0.292*** -0.750*** 0.617***
(0.091) (0.168) (0.101) (0.237) (0.084) (0.189) (0.117)
** *** ***
Upper West -0.197 -0.428 0.125 -0.313 -0.079 -0.526 0.615***
(0.090) (0.165) (0.099) (0.229) (0.083) (0.184) (0.115)
Instruments
Number of chicken 0.004***

(0.001)
Number of pigs 0.011***
(0.004)
Constant 0.342 0.335 -0.354 -0.590 -0.073 -0.259 0.118
(0.213) (0.294) (0.237) (0.388) (0.198) (0.303) (0.314)
(Continued)
Article No~e00298
(1) (2) (3) (4) (5) (6) (7)
N 6598 4727 6598 4727 6598 4727 4727
R2 0.055 0.045 0.178 -0.101 0.098 -0.200 0.500

F 9.581 8.250 34.969 16.654 16.851 7.816 326.701
Over ID (J-stat.) 0.625 1.059 1.899
[0.429] [0.303] [0.168]
Under ID (LM stat.) 17.293 17.293 17.293
[0.000] [0.000] [0.000]
Weak ID (F-stat.) 14.254 14.254 14.254
nutritional status by reducing dietary intake and intestinal absorption and

increasing catabolism and sequestration of nutrients which are essential for growth
(Brown, 2003). However, we fail to see this effect in the case of HAZ for children
who had a fever. We also observe a negative relationship between child’s age and
nutritional status.
As expected, household wealth positively correlates with height-for-age and

weight-for-height. Children born to families within higher wealth quintiles tend to
have better nutritional status as compared to those born to households within the
poorest wealth quintile. For HAZ and WAZ, the effect of household wealth and
child nutritional status falls between those in the second and middle quintiles but
increases between the fourth and the richest quintiles. This suggests that the
relationship between wealth and child health may not be linear. Conversely,
weight-for-height is worse among children in the second and middle wealth
quintiles as compared to those in the poorest wealth quintile.
Regional fixed effects are significant in explaining child nutrition. With the
Western region as the base, we observe that except for Greater Accra, children in
the other regions have lower weight-for-age than those in the Western region.
Similarly, children in Northern, Upper East, and Upper West regions are worse off
in terms of weight-for-height than their counterparts in the Western region.
Interestingly the difference between the anthropometric score of children in the
Western region and the three northern regions tends to be higher. This could be
attributable to the high poverty and deprivation rates in this region.
Article No~e00298
3.5. Proportional selection assumption test

Following Oster (Oster, 2014) we test whether the omission of unobservable
factors may significantly bias the observed coefficients of dietary diversity in our
models. The test assumes that an inference can be made about the possible bias that
could be caused by the omission of unobservable factors by observing the
movement of the coefficient by successively including observable independent
variables. The idea of the test is that if the inclusion of additional explanatory
variables improves the R2 but leaves the coefficient relatively unchanged, then one
can be confident that the coefficient is relatively stable. Based on this, it can be
concluded that the coefficient will remain relatively unchanged if the unobservable
factors were added. Table 4 contains the results of the test. We select an R2 cutoff
of 30 percent, because studies in this are usually report R2 around 20 percent (see
Arimond and Ruel (2004) for a cross country study and the respective R2). We
found that, except for the WHZ model, the identified set always excludes zero.
This is means that even though these models do not include all the potential
explanatory variables, the effect of dietary diversity will be different from zero if
we were to observe these variables.
4. Discussion and conclusion

Because of the long-lasting effect of malnutrition on human development at later
stages in life, children’s nutrition has engaged the attention of policy makers and
researcher for several years. This has led to the prescription of various
interventions to deal with child malnutrition, of which dietary diversity is an
integral part. The association between dietary diversity and child health outcomes
has been explored by researchers in some developing countries. In view of the
evidence from these studies, the current study sought to investigate the causal
effect of dietary diversity on child health outcomes in Ghana.
Table 4. Proportional Selection Assumption.
Weight-for-Height Weight-for-Age Height-for-Age
IYCF uncontrolled 0.045 -0.029 -0.132

IYCF controlled 0.027 0.045 0.027
2
R uncontrolled 0.004 0.002 0.026
2
R controlled 0.054 0.093 0.172
Identified set [0.027, -5.406] [0.045, 0.934] [0.027, 0.349]
N 6655 6694 6643
δ ¼ 1; R2max ¼ 3.
Article No~e00298
Food diversity, the independent variable of interest, was captured as a continuous

variable from the count of the number of food groups from which a child had food
in the previous day before the survey. These food groups are based on the WHO’s
IYCF guidelines. The dependent variable, child health, was measured with three
anthropometric indicators: weight-for-age (wasting), weight-for-height (stunting)
and weight-for-height (underweight). We employed both bivariate and multivariate
analyses to evaluate the effect of dietary diversity on child health. We found that
there is a general trend towards higher dietary diversity as the child grows in age.
We consider this as a good trend for child nutrition, as parents and caregivers see
the need to meet the nutritional requirements of their wards in the growth process.
Dietary diversity is also significantly higher for children from mothers with higher
education. This was consistent with our apriori expectation since educated mothers
know the importance of a balanced diet for their children. This also puts education
as one of the important tools that policy makers can adopt to improve nutritional
adequacy among infants and young children. This can be achieved by making
nutrition awareness an integral part of the school curriculum, especially at the basic
level. Since more than half of the respondents in our sample had had no formal
education, we recommend that nutrition awareness should also be created outside
the classroom through avenues like the mass media and informal education. The
government of Ghana can take advantage of the existing information service
department as well the flourishing community information centers to provide
informal education on nutrition, especially in rural Ghana.
In general, our analysis shows that dietary diversity causes improvements in child
health among under five-year-olds in Ghana. We interpret this results with caution
because dietary diversity was measured over a one-day recall period which may
not be an accurate reflection the dietary pattern for a longer period. Despite this
limitation, our results show a consistent relationship across all the three
anthropometric indicators. We recommend that public health officials should
educate parents and caregivers on the importance of dietary diversity to their
children. The government could also take advantage of the structures of the
national school feeding program to diversify the diet of the school children. Our
analysis of the effects of each individual food groups on child health shows that
vitamin A rich food (pumpkin, yellow yam, green vegetables [kontomire] mangoes
and pawpaw), eggs and other vegetables have a positive association with the
weight-for-age score. Hence, given the right nutritional education, malnutrition
among children in rural areas could be minimized at a relatively lower cost because
these food items tend to be inexpensive in rural farming communities.
Article No~e00298
Declarations
Author contribution statement
Raymond Boadi Frempong, Samuel Kobina Annim: Conceived and designed the
experiments; Performed the experiments; Analyzed and interpreted the data;
Contributed reagents, materials, analysis tools or data; Wrote the paper.
Funding statement
This research did not receive any specific grant from funding agencies in the
public, commercial, or not-for-profit sectors.
Competing interest statement

The authors declare no conflict of interest.
Additional information
No additional information is available for this paper.
References
Adedini, S.A., Odimegwu, C., Imasiku, E.N., Ononokpono, D.N., 2015. Ethnic
differentials in under-five mortality in Nigeria. Ethn. Health 20 (2), 145–162.
Annim, S.K., Imai, K.S., 2014. Nutritional Status of Children. Food Consumption
Diversity and Ethnicity in Lao PDR (No. DP2014-17).
Annim, S.K., Awusabo-Asare, K., Amo-Adjei, J., ICF International, August 2013.
Household Nucleation, Dependency and Child Health Outcomes in Ghana. DHS
Working Papers. http://www.measuredhs.com/publications/publication-wp98-
working-papers.cfm.
Arimond, M., Ruel, M.T., 2004. Dietary diversity is associated with child
nutritional status: evidence from 11 demographic and health surveys. J. Nutr. 134
(10), 2579–2585.
Bagriansky, J., 2010. The economics consequences of malnutrition in Albania.

UNICEF, Albania.
Bain, L.E., Awah, P.K., Geraldine, N., Kindong, N.P., Sigal, Y., Bernard, N.,
Tanjeko, A.T., 2013. Malnutrition in Sub − Saharan Africa: burden, causes, and
prospects. Pan Afr. Med. J. 15, 120.
Brown, K.H., 2003. Diarrhea and Malnutrition. J. Nutr. 133 (1), 328S–332S.
Article No~e00298
Cuesta, J., 2007. Child Malnutrition and the Provision of Water and Sanitation in
the Philippines. Journal of the Asia Pacific Economy 12 (2), 125–157.
Dewey, K.G., Adu-Afarwuah, S., 2008. Systematic review of the efficacy and
effectiveness of complementary feeding interventions in developing countries.
Matern. Child Nutr. 4 (Suppl. 1), 24–85.
Fuse, K., 2010. Variations in attitudinal gender preferences for children across 50
less-developed countries. Demographic Research 23 (36), 1031–1048.
Ghana Statistical Service, 2011. Ghana Multiple Indicator Cluster Survey with an
Enhanced Malaria Module and Biomarker, (Final Report). Ghana Statistical
Service, Accra, Ghana.
Hoddinott, J., Maluccio, J.A., Behrman, J.R., Flores, R., Martorell, R., 2008. Effect
of a nutrition intervention during early childhood on economic productivity in
Guatemalan adults. Lancet 371 (9610), 411–416.
ICF Macro, 2010. Trends in demographic, family planning, and health indicators in
Ghana, 1960-2008: Trend analysis of demographic and health surveys data. ICF
Macro.
Saha, K.K., Frongillo, E.A., Alam, D.S., Arifeen, S.E., Persson, L.Å., Rasmussen,
K.M., 2008. Appropriate infant feeding practices result in better growth of infants
and young children in rural Bangladesh. Am. J. Clin. Nutr. 87 (6), 1852–1859.
Strauss, J., Thomas, D., 1998. Health, nutrition, and economic development. J.
Econ. Lit 36 (2), 766–817.
Swindale, A., Bilinsky, P., 2006. Household dietary diversity score (HDDS) for
measurement of household food access: indicator guide. Food and Nutrition
Technical Assistance Project, Academy for Educational Development, Washing-
ton, DC.
UNICEF, 2013. Improving child nutrition: The achievable imperative for global
progress. UNICEF, New York.
UNICEF, WHO, World Bank, 2015. Levels and trends in child malnutrition: Key
findings of the 2015 edition. Global Database on Child Growth and Malnutrition.
UNICEF, WHO, World Bank, 2016. Levels and trends in child malnutrition: Key
findings of the 2016 edition. Global Database on Child Growth and Malnutrition.
Urke, H.B., Bull, T., Mittelmark, M.B., 2011. Socioeconomic status and chronic
child malnutrition: Wealth and maternal education matter more in the Peruvian
Andes than nationally. Nutr. Res. 31 (10), 741–747.
Article No~e00298
Vaahtera, M., Kulmala, T., Hietanen, A., Ndekha, M., Cullinan, T., Salin, M.L.,
Ashorn, P., 2001. Breastfeeding and complementary feeding practices in rural
Malawi. Acta Paediatr. 90 (3), 328–332.

Dietary Diversity and Child Malnutrition in Ghana: Raymond Boadi Frempong, Samuel Kobina Annim

Uploaded by

Copyright:

Available Formats

Dietary Diversity and Child Malnutrition in Ghana: Raymond Boadi Frempong, Samuel Kobina Annim

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Dietary Diversity and Child Malnutrition in Ghana: Raymond Boadi Frempong, Samuel Kobina Annim

Uploaded by

Copyright:

Available Formats

Received:

Raymond Boadi Frempong a, * , Samuel Kobina Annim b

Keywords: Medicine, Pediatrics, Public health

Following UNICEF’s (1990) conceptual framework on the causes of malnutrition,

Whilst UNICEF’s (1990) conceptual framework links food security to child

2.2. Measurement of dietary diversity

2.3. Definition of nutrition status (anthropometric indicators)

2.4. Model specification and estimation technique

yij ¼ β0 þ β1 sexi þ β2 agei þ β3 f everi þ β4 numchildren þ β5 urbani

f ood i ¼ α0 þ α1 sexi þ α2 agei þ α3 f everi þ α4 numchildreni þ α5 urbani

yij ¼ ϕ0 þ ϕ1 sexi þ ϕ2 agei þ ϕ3 f everi þ ϕ4 numchildreni þ ϕ5 urbani

3. Results and discussion

Table 1. Bivariate analysis of socioeconomic variables and anthropometric

Stunted Wasted Underweight

Child ill with fever in last 2 weeks

Food diversity Score

Main source of drinking water

Type of toilet facility

Wealth index quintile

Poorest 33.1 514 7.9 124 19.9 314

164.96 (Pr = 0.000) 16.91 (Pr = 0.002) 106.60 (Pr = 0.000)

Ethnicity of household head

Total 22.5 1,550 6.4 444 13.3 925

Source: Authors’ Computation from MICS4, 2011.

3.1. Feeding pattern in Ghana by child age groups

3.2. Maternal education and food diversity

Table 2. Dietary diversity across levels of maternal education.

None Primary Middle/JHS Secondary Mean

None 0.00 3.00

3.3. Regional distribution of food diversity among children

Fig. 3. Distribution of Food Diversity by Administrative Regions in Ghana. (Source: Authors’Compu-

3.4. Estimation results

We also observe that an increase in the number of children in the household

(1) (2) (3) (4) (5) (6) (7)

IYCF score 0.029** 0.179 0.028** 0.613** 0.036*** 0.518**

Child Age (0–5 = 0)

6–11 -0.465*** -0.777** -0.444*** -1.427*** -0.546*** -1.369*** 1.707***

Mother’s Educ. (None = 0)

Primary 0.016 -0.004 0.021 0.024 0.024 -0.000 0.113**

Middle/JSS -0.034 -0.009 0.029 -0.064 -0.013 -0.058 0.120*

Wealth quintile (Poorest = 0)

Second 0.023 -0.003 0.123** 0.075 0.086** 0.039 0.139***

Water source (Pipe = 0)

Protected -0.004 -0.026 -0.036 -0.052 -0.033 -0.055 -0.045

Toilet type (Flush = 0)

Pit latrine -0.034 0.082 -0.168** -0.157 -0.113* -0.022 -0.091

Ga/Dangbe -0.020 -0.067 0.152* 0.077 0.078 0.028 0.058

(0.077) (0.098) (0.080) (0.114) (0.071) (0.100) (0.098)

Central -0.080 -0.234** 0.012 -0.192 -0.059 -0.291** 0.371***

Number of chicken 0.004***

N 6598 4727 6598 4727 6598 4727 4727

R2 0.055 0.045 0.178 -0.101 0.098 -0.200 0.500

nutritional status by reducing dietary intake and intestinal absorption and

As expected, household wealth positively correlates with height-for-age and

3.5. Proportional selection assumption test

4. Discussion and conclusion

Table 4. Proportional Selection Assumption.

Weight-for-Height Weight-for-Age Height-for-Age

IYCF uncontrolled 0.045 -0.029 -0.132

IYCF score 0.029 0.179 0.028 0.613 0.036* 0.518**

6–11 -0.465* -0.777 -0.444* -1.427* -0.546* -1.369* 1.707***

Second 0.023 -0.003 0.123 0.075 0.086 0.039 0.139***

Central -0.080 -0.234 0.012 -0.192 -0.059 -0.291 0.371***