Content uploaded by Richmond Aryeetey
Author content
All content in this area was uploaded by Richmond Aryeetey on Jan 23, 2023
Content may be subject to copyright.
https://doi.org/10.18697/ajfand.116.23075
22248
Afr. J. Food Agric. Nutr. Dev. 2023; 23(1)22248-22274 https://doi.org/10.18697/ajfand.116.23075
DRIVERS OF ANAEMIA REDUCTION AMONG WOMEN OF REPRODUCTIVE
AGE IN THE EASTERN AND UPPER WEST REGIONS OF GHANA: A
SECONDARY DATA ANALYSIS OF THE GHANA DEMOGRAPHIC AND
HEALTH SURVEYS
Tetteh A1*, Adanu RMK1,2, Folson G3,
Agyabeng K4, Dwomoh D4 and R Aryeetey1
Afua Tetteh
*Corresponding author email: afua.aay@gmail.com
1Department of Population, Family and Reproductive Health, School of Public
Health, University of Ghana, Legon, Ghana
2Ghana College of Physicians and Surgeons, Accra, Ghana
3Noguchi Memorial Institute for Medical Research, University of Ghana, Legon
4Department of Biostatistics, School of Public Health, University of Ghana, Legon
https://doi.org/10.18697/ajfand.116.23075
22249
ABSTRACT
Anaemia among women of reproductive age (WRA) increases the risk of
pregnancy-related morbidity, mortality, and poor pregnancy outcomes. Globally,
there is growing interest to reduce anaemia among WRA. In Ghana, anaemia
among WRA declined at the national level between 2008 (59%) and 2014 (42%).
There were also important declines at the sub-national level. The Eastern region
(in the south) and Upper West region (in the north) provide an interesting
opportunity to understand the decline. Identifying the drivers of anaemia reduction
among WRA in Ghana provides important implementation science evidence for
designing effective interventions. This current study examined the drivers of
reduced anaemia prevalence in women of reproductive age using data from the
Ghana Demographic and Health Surveys for 2008 and 2014. Anaemia was
diagnosed as haemoglobin<12.0g/dl. Data were summarized using proportions
and 95% confidence intervals. A weighted binary logistic-based multivariate
decomposition technique was used to identify the potential drivers of anaemia
across surveys for 2003, 2008 and 2014. Sensitivity analysis was carried out to
test the robustness of the results of the decomposition analysis using haemoglobin
concentration. The results of the decomposition analysis were presented as
endowment and coefficient effects. Statistical analysis was carried out using Stata
version 15. There was an improvement in access to water and sanitation, health
services, family planning, and health insurance across surveys. Drivers of anaemia
reduction over time at the national level included wealth index and maternal age,
education, use of hormonal contraception and body mass index (BMI). In the
Eastern region, the drivers of change were household wealth index, maternal age,
hormonal contraceptive use and BMI. The drivers of change in the Upper West
region, were household access to water, maternal education and BMI. The findings
suggest that multi-level interventions are needed across sectors to further reduce
anaemia among WRA.
Key words: Anaemia, Women, Nutrition, Reproductive-age, Ghana, Haemoglobin,
Regression, Decomposition analysis
https://doi.org/10.18697/ajfand.116.23075
22250
INTRODUCTION
Anaemia among women of reproductive age (WRA) is a key public health
challenge linked with adverse health, nutrition, social, and economic challenges
[1]. Anaemia increases the risk of pregnancy-related morbidity, mortality, and poor
pregnancy outcomes such as low birth weight and stunted growth in children [2,3].
There is growing global interest [4] to reduce anaemia among WRA by half
(Sustainable Development Goals indicator 2.2.3) [5].
A third (29.9%) of WRA are anaemic [6], representing more than half a billion
women worldwide. Anaemia prevalence is classified as severe and moderately
severe in many developing countries including Ghana. The anaemia prevalence is
higher in pregnant (36.5%) than in non-pregnant women (29.6%) [7]. Worldwide,
there was a slight decline in moderate (14% in 2000 to 13% in 2019) and severe
(2% in 2000 to 1% in 2019) anaemia prevalence among WRA in 2019 [8]. Key
drivers to this reduction included actions addressing both nutrition-specific (iron
supplementation and provision of micronutrient powders) and nutrition-sensitive
interventions (biofortification, distribution and use of insecticide-treated nets) [8–
10].
In Ghana, anaemia among WRA at the national level declined between 2008
(59%) and 2014 (42%) [11,12]. The decline was also observed at the sub-national
levels. The Eastern region (ER) in the southern part of Ghana had a decline from
58% in 2008 to 39% in 2014, and the Upper West region (UWR) in the north from
70% in 2008 to 36% in 2014. These two regions have been exposed to large
population-based nutrition-specific and sensitive interventions [13–16] whiles the
standard anaemia interventions are nationwide. This decline provides an
opportunity to understand the drivers of reduced anaemia prevalence among WRA
to inform programming at the sub-national level. Therefore, the current study
examined the drivers of reduced anaemia prevalence in WRA using data from the
Ghana Demographic and Health Surveys for 2003, 2008 and 2014.
MATERIALS AND METHODS
Dataset source
The current study used nationally-representative data from the Ghana
Demographic and Health Survey (GDHS) years 2003, 2008, and 2014. The GDHS
is a nationally representative survey of Ghana. The data comprised women of
reproductive age (15-49 years).
https://doi.org/10.18697/ajfand.116.23075
22251
Sampling design of the GDHS
A multi-stage stratified cluster sampling technique was used to select participants
for the surveys [9,10,17]. First, there was a selection of enumeration areas (EAs),
followed by stratification and proportionate sampling from each stratum using a
probability proportionate to size. Then, 20-30 households were systematically
sampled from each EA by listing the households in the selected EAs and
households were randomly selected from the list for the survey. Further details are
referenced here [9,10,17].
Eligible women were those between ages 15-49 years and were permanent
residents or visitors who stayed in the selected households the night before the
survey. Anaemia testing and anthropometric data (height and weight) were
collected from a subsample of the eligible women who consented to be interviewed
and tested [9,10,17].
The 2003 GDHS included haemoglobin concentration and anaemia data from 5278
WRA (92.7% response rate). The 2008 GDHS included haemoglobin concentration
and anaemia status of 4,758 WRA (response rate of 96.8%). The 2014 GDHS
included 4704 WRA’s anaemia and haemoglobin data (50.1% response rate). A
pooled dataset of n= 14,740 women aged 15-49 years was included in this current
analysis (Appendix 1).
Outcome measures
Anaemia status was used as the outcome measure for this study. Anaemia was
determined using haemoglobin (Hb) concentration measured using a battery-
operated portable HemoCue analyzer. Anaemia was classified using altitude-
adjusted haemoglobin (Hb) level for women not pregnant as Hb <12.0g/dl and for
pregnant women, anaemia was Hb <11.0g/dl. Anaemia status was further
classified as severe <7.0g/dl; moderate 7.0-9.9g/dl; or mild anaemia 10.0-11.9g/dl
for not pregnant and 10.0-10.9g/dl for pregnant women. Haemoglobin
concentration was used in sensitivity analysis to test the robustness of the results
of the decomposition analysis, using anaemia status.
Covariates
The selection of the covariates was informed by the UNICEF conceptual
framework [18] of factors affecting undernutrition, and findings of published studies
reporting covariates of anaemia among WRA [19–25]. The data included
household- as well as individual-level variables. The household-level variables
included household size; sex of household head; socio-economic status;
urban/rural residence; region; access to improved water sources; household toilet
https://doi.org/10.18697/ajfand.116.23075
22252
facilities and the number of insecticides treated bed nets. The individual-level
variables were anaemia status, current age, religion, marital status, parity,
ethnicity, age, educational level, hormonal contraceptive use, and possession of a
valid National Health Insurance (NHI) card [9,10,17].
Statistical analysis
Data management and analyses were carried out using Stata version 15 [26]. All
statistical analyses accounted for the complex survey structure by adjusting for the
survey weighting for each survey year. For the pooled data analysis, the standard
survey weight was denormalised by dividing the survey's standard weight by the
survey sampling fraction that is the proportion of all women aged 15 to 49 years
who were interviewed during the survey year to all women at the time of the
survey). The GDHS datasets were used to determine the total number of women
aged 15 to 49 years who were interviewed during the survey year, and
OurWorldinData [27] was used to determine the total number of women aged 15 to
49 years who were living in the nation (Ghana) at the time of the survey. This is
represented by the formula:
women&sampling&weight&from&in&the&DHS ×total&females&aged&15 −49&in&the&country&during&the&survey&
number&of&females&aged&15 −49&interviewed&in&the&survey&
The analysis was carried out in two phases: the national level where data from all
women aged 15- 49 years who participated in the study across all the regions were
included. The second stage included participants from only the Eastern and Upper
West regions of Ghana.
Descriptive statistics were presented in terms of proportions and 95% confidence
intervals. Rao-Scott χ2 test statistics was used to test for associations between the
independent characteristics and anaemia status of the women. A weighted binary
logistic-based multivariate decomposition technique [28] was used to quantify the
contribution of selected variables to the change/reduction in anaemia prevalence
between 2008 and 2014 at the national level and for the Eastern and Upper West
regions.
Sensitivity analysis was carried out using haemoglobin concentration to test the
robustness of the results of the decomposition analysis. The results of the
decomposition analysis were presented as endowment and coefficient effects. The
endowment effect explains how much the changes in the level of the explanatory
variables between 2008 and 2014 contribute to the observed difference in anaemia
and mean Hb level between 2008 and 2014. The coefficient effect measures the
https://doi.org/10.18697/ajfand.116.23075
22253
contribution of differences in the coefficients (including differences in the intercept)
to the observed difference in the mean Hb level between 2008 and 2014.
The interaction term accounts for the fact that differences in endowments and
coefficients exist simultaneously between 2008 and 2014. Following binary
analysis, the variables -household wealth index, current age, maternal educational
level, hormonal contraceptive use and BMI were selected for decomposition
analysis for the national level and Eastern Region. For the Upper West Region,
variables selected were household access to improved water sources, maternal
education, and BMI. Contributions of each covariate were reported in terms of
coefficients and percentages.
RESULTS AND DISCUSSION
The findings provide background characteristics of the study population and the
changes in anaemia prevalence between 2003 to 2014 among selected
households and individual-levelled characteristics. The drivers (variables) of the
changes and decline in anaemia prevalence and haemoglobin concentration
between 2008 and 2014 for the national level, Eastern and Upper West regions of
Ghana are also presented.
Selected household and maternal/individual-levelled characteristics
The mean age of the household heads was 43.8 (± 13.7) years and a higher
proportion (61.6%) of households were headed by males. About four out of every
ten (44.4%) households had less than 5 members. More than two-thirds of the
households had access to improved drinking water (87.7%) Use of insecticide-
treated bednets, improved over the years from 2003 (19.3%) to 2014 (75.1%).
The mean age of the women was 29.2 (± 9.4) years. Close to half (47.6%) of the
women were currently married. More than half of the women had up to secondary
level education. Sixty percent of the women had normal body size while three in
every ten women were either overweight (BMI>25kg/m^2) or obese
(BMI>30kg/m^2). Most of the women had between one and four children each.
About 20% were currently breastfeeding. Appendix 2 provides further details of the
household and maternal/individual-levelled characteristics.
Levels, trends, and differences in maternal anaemia status between surveys
The prevalence of maternal anaemia at the national level has been fluctuating over
the past decades (Figure 1). In 2003, the prevalence was 44.6% [95%CI:42.8,46.4]
but increased to 58.7% [95%CI:56.7,60.7] in 2008. Thus, there was a 14.1% points
https://doi.org/10.18697/ajfand.116.23075
22254
increase in anaemia among women between 2003 to 2008. In 2014, the level of
maternal anaemia declined to 42.4% [95%CI: 40.5,44.2], thus, a 16.3% points
reduction from 2008. The reduction in maternal anaemia between 2003 and 2014
was 2.2% points (Appendices 3 and 4).
Overall, except for the sex of the household head, marital status, ethnicity, parity,
and the number of children alive, the prevalence of maternal anaemia was
significantly different across all levels of the variables studied (p<0.05).
Consistently, the prevalence of maternal anaemia was relatively lower among
women who use hormonal contraception. The prevalence of maternal anaemia
decreased among WRA with higher educational levels across survey years.
Appendix 3 provides details of the prevalence of anaemia among selected
households and individual-levelled characteristics, and Appendix 4, the changes in
reduction over the survey periods of study.
Figure 1: Changes in maternal anaemia prevalence between 2003 and 2014 at
the national level
Drivers of reduced anaemia (decomposition analysis results)
The multivariate decomposition analysis identified the drivers of the reduction in
maternal anaemia prevalence between the survey periods 2008 and 2014. At the
national level, changes in the characteristics (endowments) of the participants
were associated with a reduction in maternal anaemia by 0.01471 representing
44.6
58.7
42.4
0
10
20
30
40
50
60
70
GDHS 2003 GDHS 2008 GDHS 2014
Anaemia prevalence (%)
Survey years
https://doi.org/10.18697/ajfand.116.23075
22255
9.04% (p<0.001) whiles the size of the effect of all the characteristics (coefficients)
was associated with a decline of 0.14806 representing 90.96%. Changes in the
proportion of hormonal contraceptive use were associated with a 1.5% reduction in
maternal anaemia. Changes in the distribution of maternal BMI status were
associated with a 5.05% reduction in maternal anaemia. In the sensitivity analysis
using haemoglobin concentration as the outcome, 9.8 % of the changes in
characteristics (endowments) were associated with a reduction in maternal
anaemia by 0.047 whiles the size of the effect (coefficients) was associated with a
reduction of 0.592 representing 90.2%. Details of the decomposition analysis are
shown in Table 1 and Appendix 7.
In the Eastern Region, the variables studied were associated with changes in the
distribution of the characteristics (endowment) to the decline of anaemia by
0.01164 representing a 5.99% reduction while the size of the effect of all the
characteristics (coefficients) was associated with a decline by 0.18271
representing 94.01% of maternal anaemia between 2008 and 2014 survey periods.
Changes in the proportion of hormonal contraceptive use were associated with a
0.002 reduction (1.03%) of maternal anaemia. In the sensitivity analysis using
haemoglobin concentration as the outcome, 6.1 % of the changes in
characteristics (endowments) were associated with a reduction in maternal
anaemia by 0.015 whiles the size of the effect (coefficients) was associated with a
reduction of 0.601 representing 93.9% (Appendices 5 and 7).
In the decomposition analysis for the Upper West Region, the variables studied
were associated with a 0.0173 (5.48%; p<0.050) reduction to changes in the
distribution of the characteristics (endowment) of maternal anaemia between 2008
and 2014; whiles the size of the effect of all the characteristics (coefficients) was
associated with 0.29836 (94.52%) decline (Table 3). Changes in the proportion of
access to improved water sources were associated with a 1.09% reduction in
maternal anaemia. Maternal education was associated with 1.49% of the reduction
in maternal anaemia. For the sensitivity analysis using haemoglobin concentration
as the outcome, 6.7 % of the changes in characteristics (endowments) were
associated with a reduction of maternal anaemia by 0.019 whiles the size of the
effect (coefficients) was associated with a reduction of 1.149 representing 93.3%
(Appendices 6 and 7).
The current study examined and described the drivers of reduced anaemia
prevalence among WRA at the national and sub-national levels in the northern and
southern parts of Ghana using data from the Ghana Demographic and Health
Surveys for 2003, 2008 and 2014. Determining the drivers at the national and sub-
https://doi.org/10.18697/ajfand.116.23075
22256
national levels provides an opportunity to understand the variance at the sub-
national level. The findings will inform sub-national specific interventions to
promote the reduction of anaemia among WRA.
The drivers associated with a decline in anaemia between 2008 and 2014 at the
national and sub-national levels comprised both household and maternal
characteristics similar to other studies [21,24,25]. At the national level, the factors
included household wealth index and maternal level characteristics (maternal age
and education, BMI, and hormonal contraceptive use). At the sub-national level -
Eastern Region, the decline in anaemia among WRA was linked to household
wealth index, maternal age, maternal educational level, BMI, and hormonal
contraceptive use. In the Upper West Region, the key drivers explaining the
decline in anaemia prevalence included household access to improve water
sources, maternal education, and BMI.
Household wealth index is a measure of the socioeconomic status (SES) of the
household, the ability of the household to access basic needs and healthcare
services and provide for the needs of household members. It was identified as a
driver of the decline in anaemia at the national and the Eastern Region. Over the
period of the study, socioeconomic changes [29] were observed in the country.
These included an increase in access to information and communication
technology (ICT) [30], poverty reduction, improved trade policies, improved
healthcare infrastructure (expansion in Community-based Health and Planning
Services (CHPS)) [15,31], increasing access to healthcare services; increase in
water, sanitation and hygiene (WASH) services (drilling of boreholes and improved
toilet facilities) [32,33]; and provision of agricultural inputs and services [34].
Other services were an increase in social safety net programmes such as
Livelihood Empowerment Against Poverty (LEAP), the school-feeding programme,
health insurance, the distribution of iron-folate supplements to pregnant women
and adolescent girls, and provision of Sulfadoxine/pyrimethamine (SP) for
presumptive treatment of malaria during pregnancy [35]. These were enabled
through government, development partners and individual resources who worked
with community leaders to improve the livelihoods of households in the regions
[32,36]. All these changes were likely to have influenced household socioeconomic
status. According to the 2021 Population and Housing Census [29], the Eastern
Region with a wider population experienced more socioeconomic changes and
could therefore explain the changes observed as compared to the Upper West
Region. Household wealth index was noted to influence a decline in anaemia
among WRA in other developing settings [10,25].
https://doi.org/10.18697/ajfand.116.23075
22257
The changes in the SES [20,37] of the country preceding and during the period of
study could explain increased access to services including healthcare, and WASH
which could have led to better health outcomes such as decline in anaemia noted
in this study. For example, there were expansion in access to improved toilet
facilities and drilling of boreholes from the activities of non-governmental
organisations such as World Vision [38]. These interventions might likely have
improved household livelihoods and reduced the risk of infections and sickness
which could compromise the nutrition of women as observed by Kothari et al. [21].
The evidence of an association between selected WASH indicators and anaemia is
well established [10,21,25]. Kothari et al. [21] observed that in both WRA and
children, the presence of water in the household appeared to be protective against
anaemia (lower anaemia prevalence and odds). Nguyen et al., [25] also reported a
9% decline in anaemia prevalence among pregnant women due to improved
sanitation in India. Heckert et al., [24] noted a 12% decline in anaemia prevalence
among WRA in Tanzania due to a reduction in the percentage of households using
open defecation. These studies highlight the benefit of integrating WASH into
nutrition interventions specific to addressing anaemia, and therefore suggest an
increased access to these services to aid in the reduction of anaemia among
WRA.
Hormonal contraceptive use as a driver, reduces the risk of anaemia among WRA
possibly through a decrease in monthly menstrual cycle flow and birth spacing [10].
For this study, at the national level, the expansions in healthcare infrastructure [31]
and access to healthcare services such as family planning which could have
impacted access to hormonal contraceptives by WRA and therefore its uptake,
influencing birth spacing. The Eastern Region [39] was observed to have benefited
a lot from sexual reproductive health (SRH) services support such as enhanced
staff training and women’s education on SRH and promotion of free contraceptives
from donors and development partners which could explain the use of hormonal
contraceptives in the Eastern Region as compared to the Upper West. Hormonal
contraceptive use was also noted in other studies to be a driver of anaemia decline
among WRA [19,24,40]. Lakew et al. [40] from Ethiopia noted that contraceptives
use among lactating women was protective against anaemia (-7.7%). Heckert et al.
[24] study in Tanzania also reported a 30% change in anaemia prevalence a result
of contraceptive use.
A higher educational level among WRA as a driver of the decline in anaemia was
noted in other studies [24,25]. An increase in education was likely to have led to a
better understanding of anaemia and its associated factors among WRA. With an
https://doi.org/10.18697/ajfand.116.23075
22258
increase in the SES of the country and improved ICT, WRA were better positioned
to increase their knowledge about anaemia. Increased access to healthcare and
possible visits to the health facility and interaction with health staff were likely to
have informed dietary choices about the consumption of iron-rich foods, and
supplementation, therefore leading to the decline observed. Nguyen et al. [25]
noted a 24% decline in anaemia among WRA due to improved maternal education.
Heckert et al., [24] also noted a 36% change in anaemia among WRA as a result
of improved maternal education.
Increased maternal age was a driver of reduced anaemia among WRA. Exposure
to activities, projects and interventions as the woman ages were likely to have
enhanced the woman’s knowledge about anaemia and empowered her to make
informed decisions about her dietary habits, and other issues such as family
planning and birth spacing. This is consistent with literature [10].
Body mass index was a driver of reduced anaemia at the national and sub-national
levels. Women who were normal or overweight/ obese had reduced anaemia
compared to underweight women. With improvement in the SES of the country
(ICT, reduced poverty), these could have informed dietary habits and patterns and
was likely to have led to a decline in anaemia. The findings from this study are
similar to those from other studies [14,15,35].
The findings of this study demonstrate the roles of nutrition actions and
interventions (nutrition-specific and nutrition-sensitive) in addressing anaemia
among WRA. The evidence shows addressing the burden of anaemia among WRA
requires multiple actions across different sectors [10]. It also shows evidence of
actions and initiatives of the Government of Ghana and development partners that
precede the decade under investigation to help explain the observed decline in
anaemia among WRA. For example, the development and publication of the
National Plan of Action on Food and Nutrition 1995-2002 by the Government of
Ghana to address malnutrition served several ministries aside from health
(nutrition).
The current study had some limitations. The study used secondary data from
cross-sectional studies for analysis. Data for cross-sectional studies are carried out
at a single time point and are subject to respondent recall bias therefore it is
challenging to establish causality as the outcome and exposure variables are
measured at the same time. Secondly, the secondary data did not include
variables such as haemoglobin measurement of pregnant women at registration
and delivery, and malaria intermittent preventive treatment (IPT) doses which are
https://doi.org/10.18697/ajfand.116.23075
22259
important indicators about anaemia among WRA to inform decision-making and
planning of interventions. And, therefore, recommend the collection of these data
in future surveys. Thirdly, the current study used data from the last DHS -2014 in
the decomposition analysis to determine the drivers of the decline. It is likely, there
have been changes in the explanatory variables up to date which might help
explain the decline or otherwise. Despite these limitations, to the best of our
knowledge, this study is the first of its kind to examine, identify and describe the
drivers of anaemia reduction among WRA at the national and sub-national levels in
Ghana. The use of a large nationally representative dataset increased the study
sample size and power. The use of the multivariate decomposition technique, a
rigorous method enabled the determination of the effect of explanatory variables
contributing to the decline observed.
CONCLUSION
The current study provides evidence about the drivers of reduced anaemia
prevalence at the national and sub-national levels in Ghana. The decomposition
analysis identified changes at the national level in household wealth index,
maternal age and education, BMI, and hormonal contraceptive use as key drivers
of change. At the sub-national levels - Eastern Region, household wealth index,
maternal age and education, BMI, and the use of hormonal contraceptives were
the drivers. In the Upper West Region, household access to improved water
sources, maternal education, and BMI were the noted drivers. The observed
changes in decline in anaemia prevalence could be attributed to improvement in
the SES of Ghana and policy and programme interventions such as LEAP, school
feeding, and health insurance. Other interventions are WASH, iron-folate
supplementation for pregnant and adolescent girls and food fortification. Working
engagement across the different sectors from the national to the sub-national level
between government and development partners implementing these interventions
was also a contributory factor. Further declines in anaemia reduction require more
multisectoral coordination and targeted interventions which address the drivers of
anaemia among WRA reduction. Identifying the drivers of anaemia reduction
among WRA at the national and sub-national levels has highlighted the context-
specific drivers needed to tackling anaemia prevalence. It provides an opportunity
to understand the variance at the sub-national level which may inform sub-national
specific interventions to promote the reduction of anaemia among WRA.
ACKNOWLEDGEMENTS
This paper was made possible with the support of the Canadian Queen Elizabeth II
Diamond Jubilee Scholarships: Advanced Scholars (QES-AS) Program. "The
https://doi.org/10.18697/ajfand.116.23075
22260
Canadian Queen Elizabeth II Diamond Jubilee Scholarships (QES) is managed
through a unique partnership of Universities Canada, the Rideau Hall Foundation
(RHF), Community Foundations of Canada (CFC) and Canadian universities. The
QES-AS is made possible with financial support from IDRC and SSHRC". The
authors are grateful to the Demographic and Health Survey programme and the
Ghana Statistical Service for granting access to the 2003, 2008 and 2014 Ghana
Demographic and Health Survey data.
Ethical approval
The protocol for biomarker data (haemoglobin) testing was approved by the Ghana
Health Service Ethical Review Committee in Accra and the ORC Macro
Institutional Review Board in Calverton, Maryland, USA [9,10,17].
https://doi.org/10.18697/ajfand.116.23075
22261
Table 1: Multivariate decomposition showing contributions to the reduction
in maternal anaemia attributed to differences in endowments and
differences in coefficients at the national level, Ghana DHS 2008, and
2014
Endowment
Coefficient
Coefficent
Percent
Coefficent
Percent
Wealth index
Poorest
-0.00031
0.19
-0.0045
2.77
Poorer
-0.00056**
0.34
0.00613
-3.76
Middle
0.0002
-0.12
0.00657
-4.04
Richer
0.00031*
-0.19
-0.00893
5.49
Richest
-0.00001
0.01
0.00023
-0.14
Maternal current age (years)
Under 18
-0.00084**
0.52
0.00119
-0.73
18-34
0.00014
-0.09
0.00233
-1.43
35+
-0.00074*
0.45
-0.00432
2.65
Maternal education
No education
-0.00064
0.39
0.00289
-1.77
Primary
-0.00036
0.22
-0.00966
5.93
Secondary
0.00008
-0.05
-0.00345
2.12
Higher
-0.00125
0.77
0.00147
-0.9
Hormonal contraceptive use
Do not use hormonal contraceptive
-0.00121***
0.74
0.02732
-16.78
Use hormonal contraceptive
-0.00124***
0.76
-0.00257
1.58
Body mass index (kg/m2)
Underweight
-0.00141*
0.87
0.00175
-1.07
Normal
-0.00227*
1.39
-0.00095
0.58
Overweight
-0.00177*
1.09
-0.00328
2.02
Obese
-0.00277*
1.7
-0.00039
0.24
Overall
-0.01471***
9.04
-0.14806
90.96
P-value notation: ***p<0.001, **p<0.01, *p<0.05; kg/.m2= Kilogram/ metre-squared
https://doi.org/10.18697/ajfand.116.23075
22262
REFERENCES
1. Victora CG, Christian P, Vidaletti LP, Gatica-Dominguez G, Menon P
and RE Black Revisiting maternal and child undernutrition in low-income
and middle-income countries: variable progress towards an unfinished
agenda. Lancet (London, England) 2021;6736.
2. Young MF Maternal anaemia and risk of mortality: a call for action. Lancet
Glob Heal, 2018;6:e479–80.
3. Kumari S, Garg N, Kumar A, Guru PKI, Ansari S, Anwar S, Singh PK,
Kummari P, Mishra PK, Gipta BK, Nehar S, Sharma AK, Raziuddin M
and M Sohali Maternal and severe anaemia in delivering women is
associated with risk of preterm and low birth weight: A cross sectional study
from Jharkhand, India. One Heal, 2019;8:100098.
4. World Health Assembly. Sixty-ninth World Health Assembly Resolution
WHA69.8: United Nations Decade of Action on Nutrition (2016–2025). 2016;
7–9.
5. United Nations. SDG indicator metadata. 2021.
6. Development Initiatives. 2021 Global Nutrition Report: The State of Global
Nutrition. Bristol, UK: 2021.
7. World Health Organization. WHO Global Anaemia estimates, 2021 Edition.
Global anaemia estimates in women of reproductive age, by pregnancy
status, and in children aged 6-59 months. Geneva World Heal. Organ. 2021.
8. Stevens GA, Paciorek CJ, Flores-Urrutia MC, Borghi E, Namaste S,
Wirth JP, Suchdev PS, Ezzati M, Rohner F, Flaxman SR and LM Rogers
National, regional, and global estimates of anaemia by severity in women
and children for 2000–19: a pooled analysis of population-representative
data. Lancet Glob Heal, 2022;10:e627–39.
9. da Silva Lopes K, Takemoto Y, Garcia-Casal M and E Ota Nutrition-
specific interventions for preventing and controlling anaemia throughout the
life cycle : an overview of systematic reviews ( Protocol ). Published Online
First: 2018.
10. Owais A, Merritt C, Lee C and ZA Bhutta Anemia among Women of
Reproductive Age : An Overview of in Low- and Middle-Income Countries.
Nutrients, 2021;13:2745.
https://doi.org/10.18697/ajfand.116.23075
22263
11. Ghana Statistical Service (GSS), Ghana Health Service (GHS), ICF
Macro. Ghana Demographic and Health Survey 2008. Accra, Ghana: 2009.
12. Ghana Statistical Service (GSS), Ghana Health Service (GHS), and ICF
International. Ghana Demographic and Health Survey 2014. Rockville,
Maryland, USA: 2015.
13. Marquis GS, Colecraft EK, Kanlisi R, Aidam BA, Atuobi-Yeboah A, Pinto
C and R Aryeetey An agriculture–nutrition intervention improved children’s
diet and growth in a randomized trial in Ghana. Matern Child Nutr,
2018;14:1–10.
14. Atuobi-Yeboah A, Marquis GS, Colecraft E, Kanlisi R, Aryeetey R and M
Klevor An integrated nutrition-sensitive health and agriculture intervention to
increase egg consumption among infants and young children in Upper
Manya Krobo, Ghana. FASEB J, 2016;30.
15. Japan International Cooperation Agency (JICA). Health Sector: Project
for Improvement of Maternal and Neonatal Health Services Utilizing CHPS
System in the Upper West Region. 2011;4.
16. USAID, Ghana Health Service, Maternal and Child Health Integrated
Program and Child Survival Program. Successful Practices to Increase
Intermittent Preventive Treatment in Ghana. 2011.
17. Ghana Statistical Service (GSS), Noguchi Memorial Institute for Medical
Research (NMIMR), ORC Macro. Ghana Demographic and Health Survey
2003. Calverton, Maryland: 2004.
18. United Nations Children’s Fund. UNICEF’s approach to scaling up
nutrition for mothers and their children. Discussion paper. Programme
Division, UNICEF. New York: 2015.
19. Iruhiriye E, Olney DK, Ramani GV, Heckert JM, Niyongira E and EA
Frongillo Stories of Change- Rwanda, Final Report. 2019.
20. Aryeetey R, Atuobi-Yeboah A, Billings L, Nisbett N, van den Bold M and
M Toure Stories of Change in Nutrition in Ghana: a focus on stunting and
anemia among children under-five years (2009 – 2018). Food Secur
2022;14:355–79.
https://doi.org/10.18697/ajfand.116.23075
22264
21. Kothari MT, Coile A, Huestis A, Pullum T, Garett D and C Engmann
Exploring associations between water, sanitation, and anemia through 47
nationally representative demographic and health surveys. Ann. N. Y. Acad.
Sci. 2019;1450:249–67.
22. Birhanu A, Id T and GA Tesema Anemia and its associated factors among
women of reproductive age in eastern Africa : A multilevel mixed-effects
generalized linear model. 2020;85:1–16.
23. Yalew M, Adane B, Damtie Y, Kefale B, Arefaynie M and T Yasin Trend
and determinants of anemia change among pregnant and/or lactating
women in Ethiopia: A multivariate decomposition analysis. PLoS One
2020;15:1–14.
24. Heckert J, Headey D, Ndiaye B, Brero M and V Assey Analysis of the
drivers of change in women’s anemia in Tanzania 2005-2015. IFPRI -
Discuss Pap 2019.
25. Nguyen PH, Scott S, Avula R, Tran LM and P Menon Trends and drivers
of change in the prevalence of anaemia among 1 million women and
children in India, 2006 to 2016. BMJ Glob Heal, 2018;3:1–12.
26. StataCorp. Stata Statistical Software. Release 15. Coll. Station. TX
StataCorp LLC. 2017
27. Roser M, Ritchie H and E Ortiz-Ospina World Population Growth. Our
World Data. 2013.
28. Dwomoh D, Amuasi S, Agyabeng K, Incoom G, Alhassan Y and AE
Yawson Understanding the determinants of infant and under-five mortality
rates : a multivariate decomposition analysis of Demographic and Health
Surveys in. 2019;1–20.
29. Ghana Statistical Service. Ghana 2021 Population and Housing Census
General Report: Economic Activity. Accra: 2021.
30. Ghana Statistical Service. Ghana 2021 Population and Housing Census
General Report: Information and Communication Technology. Accra: 2022.
31. Ministry of Health. National Community-Based Health Planning and
Services (CHPS) Policy. Accra, Ghana: 2016.
32. World Vision Ghana. World Vision: 2017 Annual Report Ghana. Accra: 2017.
https://doi.org/10.18697/ajfand.116.23075
22265
33. Community Water and Sanitation Agency. Upper West Region Water
Safety Plan for Jirapa Water Community. 2020.
34. Ministry of Food and Agriculture Upper West Region. Ministry of Food
and Agriculture Upper West Region Regional Performance Draft Report
Annual - 2016. 2016.
35. Ministry of Gender Children and Social Protection. Draft National School
Feeding Policy. 2015.
36. Ghana Health Service. Family Health Division Annual Report. Accra: 2016.
37. Ghana Statistical Service. Ghana Living Standards Survey Round 7
(GLSS7): Poverty trends in Ghana; 2005-2017. Accra: 2018.
38. World Vision International. World Vision: Global Annual Report 2021.
2022.
39. Eastern Regional Directorate-Ghana Health Service. Reproductive and
Child Health 2018 Annual Report. 2018.
40. Lakew Y, Biadgilign S and D Haile Anaemia prevalence and associated
factors among lactating mothers in Ethiopia: Evidence from the 2005 and
2011 demographic and health surveys. BMJ Open, 2015;5:1–7.
https://doi.org/10.18697/ajfand.116.23075
22266
Appendix 1: Summary of Ghana Demographic and Health Surveys included
in this study’s analysis
Year
Reference
period
Number of
households
interviewed
Number of
women aged 15-
49 years
interviewed
Eligible
women's
response
rate
Number of
respondents
with measured
anaemia status
2003
1999 - 2003
6251
5,691
95.7
5,278
2008
2004 - 2008
11778
4,916
96.5
4,758
2014
2010 - 2014
11835
9,396
97.3
4,704
Appendix 2: Prevalence of household and selected maternal/individual-
levelled characteristics of survey respondents across surveys
in 2003, 2008, and 2014
Characteristic
2003 GDHS
2008
GDHS
2014 GDHS
2003 and
2014
%
%
%
%
Sex of household head
Female
37.5
39.2
38.4
38.4
Wealth index
Poorest
17.0
16.0
17.0
16.6
Poorer
16.9
18.2
17.2
17.5
Middle
19.2
20.2
20.9
20.0
Richer
21.9
22.9
22.1
22.3
Richest
25.0
22.8
22.8
23.5
Type of residence
Rural
52.1
51.8
46.1
50.5
Region
Eastern
9.9
9.8
8.9
9.6
Upper West
2.6
2.5
2.4
2.5
Household size
<5 residents
39.3
46.1
48.5
44.4
5-7 residents
40.3
38.6
37.5
38.9
8+ residents
20.4
15.3
14.0
16.7
Household access to improved water
source
Improved
81.9
89.7
92.2
87.7
Household access to improved toilet
Improved not shared
14.2
12.5
14.4
13.5
Improved shared
62.3
57.0
58.5
59.2
Number of insecticide-treated bednets in household
No net at all
80.7
46.2
24.9
52.7
1 net
12.3
27.0
23.2
21.1
2 nets
4.1
16.7
26.3
14.8
3+ nets
2.8
10.2
25.6
11.5
Maternal current age (years) [Mean ± SD]
29.1 ± 9.7
28.9 ±8.3
29.9 ± 10.0
29.2 ± 9.4
https://doi.org/10.18697/ajfand.116.23075
22267
Under 18
12.4
12.4
10.6
11.9
18-34
56.3
56.3
55.6
56.1
35+
31.3
31.3
33.8
31.9
Religion
Christian
77.4
77.6
80.1
78.1
Islam
15.5
15.0
15.4
15.3
Traditional/spiritualist/other
2.6
4.3
2.2
3.2
No religion
4.5
3.2
2.2
3.4
Marital status
Never married
28.0
32.3
32.1
30.8
Currently married
54.4
45.3
42.2
47.6
Cohabiting
8.3
13.2
15.1
12.0
Previously married*
9.2
9.2
10.6
9.5
Educational level
None
28.2
21.3
19.5
23.2
Primary
20.1
20.3
18.2
19.7
Secondary
49.3
54.8
56.6
53.4
Tertiary
2.4
3.6
5.8
3.8
Body mass index (kg/m 2)
Underweight
8.8
8.1
5.7
7.8
Normal
63.2
61.4
53.8
60.1
Overweight
16.9
20.7
25.0
20.5
Obese
7.3
9.3
15.4
10.1
Missing
3.7
0.6
0.2
1.6
Parity
None
31.0
33.4
30.6
31.9
1-4
46.2
47.9
50.1
47.9
5-7
17.1
14.6
15.9
15.8
8+
5.7
4.1
3.3
4.5
Children alive
None
32.3
34.1
31.2
32.8
1-4
49.8
50.9
53.1
51.1
5-7
15.5
13.1
14.2
14.2
8+
2.3
1.8
1.5
1.9
Maternity status
Neither breastfeeding nor pregnant
68.5
71.5
70.2
70.1
Breastfeeding
23.9
21.2
22.4
22.4
Pregnant
7.6
7.4
7.3
7.4
Hormonal contraceptive use
Do not use hormonal contraceptive
91.2
91.4
89.2
90.8
Use hormonal contraceptive
8.8
8.6
10.8
9.2
Health insurance
No
-
60.0
37.7
51.6
Mean ± SD = standard deviation; GDHS= Ghana Demographic and Health Survey;
Previously married = divorced, widower, separated; kg/.m2 = Kilogram/ metre-squared
https://doi.org/10.18697/ajfand.116.23075
22268
Appendix 3: Maternal anaemia prevalence by selected household/individual-levelled characteristics, Ghana 2003,
2008, and 2014
2003
2008
2014
2003 and 2014
% [ 95% CI]
P-value
% [ 95% CI]
P-value
% [ 95% CI]
P-value
% [ 95% CI]
P-value
Overall
44.6 [42.8,46.4]
58.7 [56.7,60.7]
42.4 [40.5,44.2]
49.9 [48.7,51.1]
Sex of household head
0.608
0.066
0.127
0.135
Male
44.3 [42.2,46.4]
60 [57.5,62.5]
43.4 [41.2,45.7]
50.5 [49.1,52]
Female
45.1 [42.3,48]
56.8 [54,59.5]
40.6 [37.6,43.7]
49 [47.2,50.7]
Wealth index
<0.001
0.020
<0.001
<0.001
Poorest
53.2 [49.6,56.7]
61 [56.6,65.2]
43.6 [39.6,47.6]
53.9 [51.4,56.3]
Poorer
45.2 [41.2,49.3]
63.1 [59.3,66.7]
50.5 [46.9,54.1]
54.2 [51.8,56.7]
Middle
47.1 [43.4,50.9]
59.3 [55.4,63]
45.2 [41.2,49.4]
51.7 [49.4,54.1]
Richer
40.6 [37.3,43.9]
58 [54.2,61.7]
37.2 [33.2,41.4]
47.2 [44.9,49.5]
Richest
39.9 [36.6,43.3]
53.9 [49.4,58.4]
37.7 [33.3,42.3]
45 [42.5,47.5]
Type of residence
0.003
0.001
0.509
<0.001
Urban
41.7 [39.1,44.4]
55.3 [52.3,58.4]
41.8 [39.1,44.5]
47.3 [45.5,49]
Rural
47.2 [44.8,49.7]
61.9 [59.4,64.4]
43 [40.5,45.6]
52.5 [50.9,54.2]
Region
<0.001
<0.001
0.038
0.017
Western
38.4 [32.8,44.4]
71.3 [64.5,77.3]
42.6 [38.3,47.1]
51.9 [47.1,56.6]
Central
38.6 [33.4,44.1]
63.5 [59.3,67.6]
46.7 [40.1,53.5]
51.2 [47.3,55.1]
Greater Accra
47.1 [43.3,50.9]
50.8 [45.9,55.6]
42.4 [37.1,48]
47.3 [44.5,50.1]
Volta
48.8 [41.4,56.2]
58.1 [52.1,63.9]
48.7 [43.7,53.7]
52.8 [48.8,56.7]
Eastern
48.6 [42.6,54.6]
58.5 [51,65.7]
38.9 [33.8,44.3]
50.6 [46.5,54.7]
Ashanti
46.3 [42.6,49.9]
59.9 [55.8,63.9]
40.5 [36.3,44.8]
50.8 [48.2,53.4]
Brong Ahafo
33.3 [27,40.4]
57.8 [50,65.3]
36.4 [31.8,41.2]
43.8 [39.2,48.5]
Northern
48.2 [43.7,52.6]
59.5 [53.4,65.3]
47.5 [41.5,53.6]
53 [49.5,56.5]
Upper East
51.3 [44.9,57.7]
48.3 [41,55.6]
39.6 [33.8,45.7]
47.6 [43.5,51.8]
Upper West
48.3 [41.3,55.4]
67.1 [62.5,71.3]
35.6 [30.5,41]
53.2 [48.4,57.9]
Household size
0.479
<0.001
0.182
0.001
<5 residents
43.9 [41.3,46.5]
55.2 [52.5,57.8]
40.8 [38.1,43.6]
47.9 [46.3,49.5]
5-7 residents
44.4 [41.8,47]
61.1 [58.2,63.8]
43.4 [40.6,46.3]
51 [49.3,52.8]
8+ residents
46.3 [43.1,49.5]
63.6 [59.2,67.8]
45 [40.8,49.3]
52.6 [50.2,55.1]
https://doi.org/10.18697/ajfand.116.23075
22269
Household access to
improved water source
<0.001
0.080
0.472
0.004
Improved
43.3 [41.3,45.3]
58.2 [56.1,60.3]
42.2 [40.2,44.1]
49.4 [48,50.7]
Not improved
50.6 [47.1,54.2]
63 [57.9,67.9]
44.7 [38.3,51.2]
54.0 [51,57]
Household access to
improved toilet
0.011
0.552
<0.001
<0.001
Improved not shared
42.8 [37.9,47.8]
56.6 [51,62]
44.7 [39.9,49.7]
48.6 [45.5,51.7]
Improved shared
43.4 [41.2,45.7]
59.2 [56.6,61.7]
39.4 [37,41.8]
48.8 [47.2,50.3]
Not improved
49.7 [46.6,52.8]
59.9 [56.8,63]
47.6 [44.5,50.7]
53.9 [52,55.9]
Missing
41.9 [32.5,51.9]
36.1 [27.2,46]
46.1 [9.53,87.4]
39.7 [33,46.9]
Number of insecticide-
treated bednets in
household
0.237
0.101
0.180
<0.001
No net at all
44 [42.1,45.9]
57.3 [54.6,60]
38.8 [34.8,43]
48.2 [46.6,49.8]
1 net
47.5 [43.4,51.7]
60.3 [57.1,63.4]
43.3 [39.7,46.9]
53.2 [50.9,55.4]
2 nets
49.4 [41.7,57]
61.8 [57.9,65.5]
43.6 [40.1,47.2]
52.7 [50,55.3]
3+ nets
43.3 [34.8,52.3]
56 [50.9,60.9]
43.7 [40.4,47]
48.2 [45.5,50.9]
Maternal current age (years)
0.613
0.001
<0.001
<0.001
Under 18
45.2 [40.9,49.5]
66.4 [62.1,70.5]
53.5 [48.3,58.6]
56.1 [53.3,58.9]
18-34
44 [41.7,46.2]
57.6 [55.1,60]
41.5 [39.1,43.9]
49 [47.6,50.5]
35+
45.5 [42.8,48.2]
57.8 [54.8,60.7]
40.3 [37.4,43.3]
49.2 [47.4,50.9]
Religion
0.002
0.194
0.001
Christian
43.4 [41.3,45.5]
57.9 [55.6,60.2]
41.8 [39.7,44]
49 [47.6,50.4]
Islam
47.1 [43.8,50.4]
62.3 [57.6,66.9]
42.9 [39.1,46.9]
52.3 [49.7,54.9]
Traditional/ spiritualist/
other
55.5 [47.5,63.3]
57.5 [51.3,63.4]
51.7 [41.5,61.8]
56 [51.6,60.3]
No religion
50.6 [44.3,57]
62.9 [54.8,70.4]
47.8 [37.8,58]
55 [50.4,59.5]
Educational level
0.007
<0.001
0.007
<0.001
None
48.4 [45.9,50.9]
59.9 [56.2,63.4]
45.5 [42.2,49]
52.2 [50.3,54.1]
Primary
44.3 [40.9,47.7]
63.5 [59.9,67]
44.6 [40.5,48.8]
52.6 [50.3,54.9]
Secondary
42.9 [40.4,45.4]
57.6 [55,60.1]
41.6 [39.4,43.8]
48.8 [47.3,50.4]
Tertiary
37.8 [28.9,47.6]
42.9 [32.9,53.5]
32.1 [24.9,40.3]
37.7 [32.3,43.4]
BMI (kg/m 2)
<0.001
<0.001
<0.001
Underweight
47.6 [42.8,52.4]
63.5 [58.2,68.5]
51 [44.2,57.8]
55.1 [51.8,58.3]
https://doi.org/10.18697/ajfand.116.23075
22270
Normal
47.1 [45,49.2]
61 [58.8,63.2]
46.2 [43.9,48.6]
52.8 [51.4,54.2]
Overweight
38.7 [35,42.4]
53.3 [49.7,57]
36.5 [32.8,40.4]
44.2 [41.9,46.5]
Obese
34.9 [29.4,40.9]
51 [45.4,56.5]
35.2 [30.5,40.1]
41.1 [38,44.3]
Missing
41.2 [35.3,47.3]
70.4 [51.4,84.2]
36.9 [14.9,66.2]
45.6 [39.7,51.5]
Parity
0.660
0.620
0.169
0.390
None
43.8 [40.8,46.9]
58.7[55.8,61.5]
45 [41.7,48.4]
50.6 [48.7,52.4]
1-4
44.3 [42,46.6]
58 [55.3,60.6]
40.8 [38.3,43.4]
49.1 [47.5,50.7]
5-7
46.6 [42.8,50.5]
60.7[56.5,64.7]
41.4 [37.1,45.7]
50.7 [48.3,53.1]
8+
45.1 [39,51.3]
61.2[53.7,68.1]
46 [36.9,55.3]
51.4 [47.1,55.7]
Children alive
0.871
0.153
0.127
0.223
None
43.8 [40.9,46.8]
58.5[55.6,61.3]
44.9 [41.5,48.3]
50.4 [48.6,52.3]
1-4
44.7 [42.5,47]
58.3 [55.7,61]
40.4 [38,42.9]
49.3 [47.7,50.8]
5-7
45.4 [41.3,49.5]
59.2[54.9,63.4]
44 [39.7,48.3]
50.3 [47.8,52.9]
8+
47 [37,57.4]
71.4 [61,80]
42.4 [29.5,56.5]
55.9 [49.2,62.5]
Maternity status
<0.001
<0.001
0.166
<0.001
Neither breastfeeding
nor pregnant
41.6 [39.4,43.8]
56.6[54.4,58.9]
41.3 [39,43.6]
47.9 [46.5,49.3]
Breastfeeding
48 [45.1,51]
61.8[58.4,65.1]
45 [41.4,48.7]
52.7 [50.7,54.7]
Pregnant
61.1 [55.7,66.2]
70.2[64.7,75.2]
44.6 [38.4,51]
60.8 [57.4,64.2]
Hormonal contraceptive use
<0.001
0.022
<0.001
<0.001
Do not use hormonal
contraceptive
46.1 [44.2,47.9]
59.3[57.3,61.3]
43.6 [41.7,45.6]
51 [49.8,52.2]
Use hormonal contraceptive
29.4 [25,34.2]
52.8[47.1,58.4]
32.2 [27.9,37]
39.3 [36.3,42.4]
Health insurance
0.059
0.465
<0.001
No
-
60 [57.5,62.4]
43.2 [40.3,46.1]
55.5 [53.4,57.5]
Yes
-
56.8 [53.9,59.5]
41.9 [39.6,44.1]
49.6 [47.7,51.5]
MA = maternal anaemia; P-value notation: ***p<0.001, **p<0.01, *p<0.05., Other = other ethnic groups; Previously married = divorced, widower, separated; BMI = Body mass
index; (kg/m 2) = kilogram/metre-squared
https://doi.org/10.18697/ajfand.116.23075
22271
Appendix 4: Estimate of changes in prevalence of selected variables
associated with maternal anaemia between surveys, Ghana
2003, 2008, and 2014
Percent of change/ differences between surveys
2003 and 2008
2008 and 2014
2003 and 2014
Wealth index
Poorest
7.8
-17.4
-9.6
Poorer
17.9
-12.6
5.3
Middle
12.2
-14.1
-1.9
Richer
17.4
-20.8
-3.4
Richest
14
-16.2
-2.2
Household access to improved water source
Improved
14.9
-16
-1.1
Not improved
12.4
-18.3
-5.9
Number of insecticide-treated bednets in household
No net at all
13.3
-18.5
-5.2
1 net
12.8
-17
-4.2
2 nets
12.4
-18.2
-5.8
3+ nets
12.7
-12.3
0.4
Maternal current age (years)
Under 18
21.2
-12.9
8.3
18-34
13.6
-16.1
-2.5
35+
12.3
-17.5
-5.2
Educational level
None
11.5
-14.4
-2.9
Primary
19.2
-18.9
0.3
Secondary
14.7
-16.0
-1.3
Tertiary
5.1
-10.8
-5.7
Body Mass Index (kg/m 2)
Underweight
15.9
-12.5
3.4
Normal
13.9
-14.8
-0.9
Overweight
14.6
-16.8
-2.2
Obese
16.1
-15.8
0.3
Missing
29.2
-33.5
-4.3
Parity
None
14.9
-13.7
1.2
1-4
13.7
-17.2
-3.5
5-7
14.1
-19.3
-5.2
8+
16.1
-15.2
0.9
Children alive
None
14.7
-13.6
1.1
1-4
13.6
-17.9
-4.3
5-7
13.8
-15.2
-1.4
8+
24.4
-29.0
-4.6
Hormonal Contraceptive use
https://doi.org/10.18697/ajfand.116.23075
22272
Do not use hormonal contraceptive
13.2
-15.7
-2.5
Use hormonal contraceptive
23.4
-20.6
2.8
Other = other ethnic groups; Previously married = divorced, widower, separated; (kg/m 2 = kilogram/metre-squared
Appendix 5: Multivariate decomposition showing contributions to the decline
in maternal anaemia attributed to differences in endowments
and differences in coefficients for the Eastern region GDHS
2008, and 2014
Endowment
Coefficient
Coefficient
Percent
Coefficient
Percent
Wealth index
Poorest
-0.00003
0.02
-0.01156
5.95
Poorer
-0.00029*
0.15
0.02827
-14.55
Middle
-0.0003
0.16
-0.00569
2.93
Richer
0.00052
-0.27
-0.01998
10.28
Richest
-0.00091
0.47
0.01223
-6.3
Maternal current age (years)
Under 18
-0.00013
0.07
-0.0185
9.52
18-34
-0.00004
0.02
0.01827
-9.4
35+
-0.00002
0.01
0.03197
-16.45
Maternal educational level
No education
-0.00025
0.13
0.00568
-2.92
Primary
-0.00086
0.44
0.02238
-11.52
Secondary
0.00004
-0.02
0.0711
-36.58
Higher
-0.00291
1.5
-0.00743
3.82
Hormonal contraceptive use
Do not use hormonal contraceptive
-0.00192*
0.99
0.0639
-32.88
Use hormonal contraceptive
-0.002*
1.03
-0.00542
2.79
Body mass index (kg/m2)
Underweight
-0.00041
0.21
-0.00468
2.41
Normal
-0.00043
0.22
0.02511
-12.92
Overweight
0.00052
-0.27
0.00765
-3.94
Obese
-0.00202
1.04
-0.0006
0.31
Overall
-0.01164
5.99
-0.18271***
94.01
P-value notation: ***p<0.001, **p<0.01, *p<0.05; kg/.m2= Kilogram/ metre-squared
https://doi.org/10.18697/ajfand.116.23075
22273
Appendix 6: Multivariate decomposition showing contributions to the decline
in maternal anaemia for the Upper West region, GDHS 2008, and
2014
Endowment
Coefficient
Coefficient
Percent
Coefficient
Percent
Household access to
improved water source
Improved
-0.00171
0.54
0.08379
-26.55
Unimproved
-0.00175
0.55
-0.0016
0.51
Maternal education
No education
-0.0016
0.51
0.02977
-9.43
Primary
-0.00008*
0.02
-0.00818
2.59
Secondary
-0.00048
0.15
0.02245
-7.11
Higher
-0.00256
0.81
-0.00147
0.47
Body mass index (kg/m2)
Underweight
-0.00068
0.21
0.00085
-0.27
Normal
-0.0048
1.52
0.06396
-20.26
Overweight
-0.00128
0.4
0.0094
-2.98
Obese
-0.00225
0.71
-0.005
1.58
Overall
-0.0173*
5.48
-0.29836***
94.52
P-value notation: ***p<0.001, **p<0.01, *p<0.05; kg/.m2= Kilogram/ metre-squared
https://doi.org/10.18697/ajfand.116.23075
22274
Appendix 7: Multivariate decomposition showing contributions to the decline in haemoglobin concentrations at the
national and sub-national level-Eastern and Upper West regions, GDHS 2008, and 2014
Components of decomposition
National
Eastern
Upper West
Mean Hb prediction for 2014
11.968
12.018
12.184
Mean Hb prediction for 2008
11.317
11.375
10.972
The difference in mean Hb prediction (2014-2008)
0.652***
0.643
1.212
Contribution to the difference in the mean Hb prediction due to the
endowment effect
0.047***
0.015
0.019
Contribution to the difference in the mean Hb prediction due to
coefficient effect
0.592***
0.601
1.149
Due to interaction
0.012
0.028
0.044
Endowment as a percent of total
9.8%
6.1%
6.7%
Discrimination as a percent of total
90.2 %
93.9%
93.3%
Source: Computed by the author from 2008 and 2014 Demographic and Health Survey conducted in Ghana; Note:
R = E + C + I
, ***p< 0.001, **p<0.01, *p<0.05
